电子工程师手册（英文版五）：ch089

Sherr, S., Durbeck, R.C., Suryn, W., Veillette, M. “Input and Output” The Electrical Engineering Handbook Ed. Richard C. Dorf Boca Raton: CRC Press LLC, 2000 89 Input and Output 89.1 Input Devices Keyboards ? Light Pen ? Data Tablet (Graphics, Digitizer) ? Mouse ? Trackball ? Joystick ? Touch Input ? Scanners ? Voice ? Summary ? Advantages and Disadvantages 89.2 Computer Output Printer Technologies Classification of Printer Technologies ? Page Printer Technologies ? Serial Nonimpact Printer Technologies ? Impact Printer Technologies 89.3 Smart Cards Hardware Architecture ? Contact ICC, Contactless ICC ? Operating Systems ? Standards ? Applications ? Readers ? Card-to-System Solutions?Trends 89.1 Input Devices 1 Solomon Sherr Input devices are those portions of computer, data processing, and information systems that perform the essential function of providing some means for entering commands and data into the system. Therefore, input devices are found in all such systems, but are treated here as a separate equipment group, independent of the total system configuration. However, the place of input devices in a representative computer system may be clarified by reference to Fig. 89.1(a), which shows the interface of the main input device categories in relation to the portions of the generalized system that accept the inputs. The categories and the devices listed in Table 89.1 are the subject of this section. Keyboards Keyboards are essentially electromechanical devices, and are still ubiquitous, in spite of the inroads of other input devices. The primary type of keyboard in use as an input device is the alphanumeric (A/N) form, well known in its typewriter application, but with various additions and expansions consisting of numeric and special function keys. This type of keyboard is shown in Fig. 89.1(b) with a standard QWERTY format, so named because of the layout of the top left alpha keys, for the A/N portion, a separate numeric set to the right, and a group of function keys at the top. Other layouts for the A/N portion have been proposed and at least one (Dvorak) accepted by the American National Standards Institute (ANSI), but it has not received much use in spite of its advantages in increased efficiency. At present, the overwhelming majority of system keyboards still use the QWERTY layout, and it is the only one considered here. As illustrated in Fig. 89.1, a keyboard consists of a number of keyswitches whose exact structure is of prime importance in keyboard design. The relevant characteristics of keyswitch operation are life, actuation force, travel distance, and feedback. Accepted values are shown in Table 89.2 for different keyswitch designs. The elastomer type is preferred to a limited extent over the other two when the electronic audio feedback is included. This indicates that some type of audio feedback is desirable. One form of keyswitch design using an elastomer 1 The material contained in this section is a shortened version of that which appears in Electronic Displays, 2nd ed., by Sol Sherr, Chapter 6, Section 6.1, 1993, published by John Wiley & Sons, Inc., and is reprinted here by permission. Solomon Sherr Westland Electronics Robert C. Durbeck IBM Corporation Witold Suryn Gemplus Michel Veillette Gemplus ? 2000 by CRC Press LLC ? 2000 by CRC Press LLC A RT OF C OMPILING S TATISTICS Herman Hollerith Patented January 8, 1889 #395,781 n excerpt from Herman Hollerith’s patent application: Having thus described my invention, what I claim as new is (1) The improvement in the art of compiling statistics, which consists in first preparing a series of separate record-cards, each card representing an indi- vidual or subject; second, applying to each card at predetermined intervals circuit-controlling index points arranged according to a fixed plan of distribution, to represent each item or characteristic of the individual or subject, and third, applying said separate record-cards successively to circuit-controlling devices acted upon by the index-points to designate each statistical item represented by one or more of said index-points, substantially as described. This patent, along with two others, describes a system for tabulating statistical items represented by holes punched in cards. The 1890 U.S. census was completed $5 million under budget and two years ahead of schedule because of Hollerith’s system. The punch card system with encoded holes (the code for representing alphanumeric characters with holes was named after Hollerith) was widely used for sorting, counting, and tabulating even into the 1980s. Hollerith’s original Tabulating Machine Company was the forerunner to the computer giant, IBM. (Copyright ? 1995, DewRay Products, Inc. Used with permission.) A TABLE 89.1 List of Input Devices Category Designation Operation Mode Keyboard Alphanumeric Electromechanical Keyboard Function Electromechanical Pointing Light pen Screen pointing Pointing Touchscreen Screen pointing Pointing Pen tablet Tablet pointing Coordinates Digitizer X-Y conversion Coordinates Data tablet X-Y location Cursor Mouse Movement Cursor Trackball Movement Cursor Joystick Movement Image Scanner Conversion Verbal Voice Conversion FIGURE 89.1 (a) Generalized display-system block diagram. (Source: After S. Sherr, Electronic Displays, New York: John Wiley & Sons, 1979.With permission.) (b) Alphanumeric keyboard. (Courtesy of Key tronic.) TABLE 89.2 Keyboard Parameter Values Parameter Snap Switch Elastomer Foam Pad Key travel 3.8 mm 3.2 mm 3.8 mm Force >60 gm >50 gm >30 gm Life 10 million cycles 10 million cycles 10 million cycles Feedback Audio mechanical Audio electric Tactile ? 2000 by CRC Press LLC or “molded boot” is shown in Fig. 89.2(a), in which the boot consists of two collapsible domes. In this design, the internal movement of the keyswitch is completely silent so that some source of sound must be added to achieve the desired audible feedback. The snap switch design shown in Fig. 89.2(b) has built-in sound and achieves a small reduction in insertion errors over the elastomer design with audio feedback. The life requirement is estimated on the basis of workstation users operating at approximately half the accepted rate of 20 million actuations per key used for electronic typewriters. The actual layout and content of the keyboard may vary greatly, ranging from the standard typewriter arrangement, through different com- binations of alphanumerics and symbols, to the special-function keyboards that contain legends and symbols specific to the particular application. However, the outputs of each type are the same in that they must contain coded signals that relate the action to be performed by the information system to that defined by the key being operated, in terms of the input code of the system. Thus, many of the keyboards output the ASCII code, and the system is usually designed so that it can accept this type of standard code. Incidentally, ASCII, the acronym for American Standard Code for Information Interchange, is the standard means for encoding alphanumerics and a group of selected symbols for transmission to a display system, among others. It is the standard code used in the United States and most other English-speaking countries and corresponds to the ISO seven-bit code. The seven-bit ASCII is usually used, and it should be noted that for serial data transmission an eighth bit is added for parity. Various keyboard arrangements are possible, and many variants are found in particular applications. The means for coding the key operation may be through magnetic reed relays, solid-state circuits, or more exotic devices such as Hall effect sensors. These device characteristics are only incidental to the operation and beyond the scope of this chapter. Similarly, we do not discuss the human-factors aspects of keyboard design, not because they are not important, but because, apart from the visual considerations, the other factors have to do with tactile and physical features best left to others. Light Pen The light pen initially was a very popular means for accomplishing manual input to the random deflection information display systems, but fell out of favor when raster systems became more popular due to its being FIGURE 89.2 (a) Elastomer-type keyswitch. (b) Snap switch. (Source: After H. Brunner et al., “Effects of key action design on keyboard preference and throughput performance,” Micro Switch. With permission.) ? 2000 by CRC Press LLC somewhat difficult to use with raster systems. This device goes by a misleading name, as it does not emit light and is not a pen other than being somewhat similar to one in its physical appearance, as shown in Fig. 89.3(a). However, when we consider its functional characteristics, the validity of the term becomes apparent, as it is used to cause the electron beam to “write” patterns on the cathode ray tube (CRT) that are defined by the motion of the light pen on the CRT faceplate. The light pen operates by sensing the existence or nonexistence of a pulse of light at the point on the screen of the CRT or surface of any other light-emitting device where the point of the pen is placed. This is accomplished by means of the circuit shown in Fig. 89.3(b), where the light pulse is collected and transmitted through the fiber optics to a light-sensitive device that converts the light pulse into an electrical pulse which is shaped by some form of electronics (of which a Schmitt trigger is one example). We need not concern ourselves with the exact form of the electronics except to note that this pulse is then sent to the computer, as shown in Fig. 89.4, and provides a complete, closed-loop system. As the electronic pulse occurs at the time when the light pulse passes under the light pen, the computer is informed of the location at which the designated operation is to be performed and may proceed accordingly. Thus, the light pen is a pointing device that designates a point on the display screen and can be used as an input device. Various light pen programs have been written to expand the capabilities of the original one, and it should be noted that the light pen is coming back into favor as improvements in accuracy, ease of operation, and reliability occur. There are two characteristics of light pen operation that affect the capabilities of this input device. The first is the sensitivity, given by FIGURE 89.3 (a) Light pen. (Courtesy of FTG Data Systems.) (b) Light pen schematic. (Source: After S. Sherr, Electronic Displays, New York: John Wiley & Sons, 1979, p. 388. With permission.) ? 2000 by CRC Press LLC S = E L m p A p A m m s m f t L (89.1) where E L = illuminance at photodetector, m p = photodetector sensitivity, A p = preamplifier gain, A m = main amplifier gain, m s = Schmitt trigger sensitivity, m f = flip-flop sensitivity, and t L = optical loss. Equation (89.1) may be used to calculate the light output required from the display surface, which may be a CRT or other light-emitting device, but with the limitation that most of the flat panel units are matrix driven and must track the drive sequence in order to know the location of the light pen from the drive pulse timing. When phosphors are involved as for the CRT, vacuum fluorescent displays (VFDs), thin-film electroluminescent (TFEL) units, and color liquid crystal displays (LCDs), the phosphor delays must be entered into the timing, and the total delay is given by E o = E i (1 – e – t/t) (89.2) where E o = voltage at triggering element, E i = voltage equivalent of phosphor light output, t = time, and t = sum of all delays. These delays set limits to the positional accuracy, as the computer tracking the signal will be in error by this amount. Other inaccuracies are due to the dimensions of the optical pickup surface, all of which somewhat negate the simplicity of operation. The result is the parameter values shown in Table 89.3. Data Tablet (Graphics, Digitizer) A very convenient means for data entry, retaining some of the ease of operation of the light pen but with much better accuracy, are the various forms of data tablets available. These tablets differ from the light pen in another significant way in that they do not require a moving spot of light to detect the location of the beam or direct it to a new location. This need for a moving light spot made the light pen difficult to use with the data tablets initially designed to overcome this limitation while still using a device with a pen-like input. The first successful example was the Rand tablet, a digital device that used an X–Y assembly from which a wand placed above some point on the X–Y wire matrix could pick up pulse generator output that fed X and Y electrical pulses into the matrix. By determining the number of pulses in a time period, the location of the wand is established. Another similar device used magnetostrictive rather than electrical signals to accomplish the same result, and this location is converted into display coordinates used to position a cursor on the CRT screen. The cursor may then be FIGURE 89.4 Block diagram of light pen computer system. (Source: S. Sherr, Electronic Displays, New York: John Wiley & Sons, 1979, p. 389. With permission.) TABLE 89.3Light Pen Data Field of View Response Time Sensitivity 0.02–0.08 in. 120–150 ns 0.02–0.04 ft.L ? 2000 by CRC Press LLC used as a visual feedback element so that the operator can correct the position of the wand until the cursor is properly placed. At this time the information from the tablet may also be transferred to either the host computer or the resident desktop or portable computer, as desired. Since the cursor is not used to signal its position to a pickup device, as is the case with the light pen, it may be used with any type of display system, including the non-light-emitting flat panel displays. Another advantage of the tablet is that it may be used to position cursors in the blank areas of the display, where no light pulses are available unless they are specially generated by the light pen. There have been numerous improvements and new developments using a variety of technologies that include magnetostrictive, electromagnetic, electrostatic or capacitive, scanned X–Y grid, resistive, and sonic. Of these, electromagnetic tablets dominate the digitizer market, and sonic is of interest because it does not require a tablet, but most of the other technologies are essentially restricted to touch input devices covered later. As noted previously, electromagnetic is the most popular technology for high-performance digitizer tablets. Operation is based on transformer principles, whereby a conductor carrying ac creates a magnetic field around it that induces a current in a second conductor. The digitizer tablet uses the amplitude and phase of the induced current to determine digitizing data. The tablet contains an X–Y pattern of conductors beneath its surface, in a manner similar to the Rand Tablet, but instead of counting pulses in a time period a circular conductor is used as the pick-up element for the induced current. This coil is placed on the tablet surface, and its position is determined by measuring the phase and amplitude of the current in the coil. Its center is interpolated by sweeping through the X–Y grid lines and demodulating the signal in the coil to determine the phase reversal point, or by calculating this point using digitized data fed into a microprocessor. The X–Y coordinates may be resolved to better than 0.025 mm using either of these two techniques. Figure 89.5(a) is a photograph of a representative digitizer tablet. Another digitizer technology is the one that uses the measurement of the time required for sound waves to travel from a source to movable microphone pickups.This sonic technology has the advantage that no special digitizing board is required, and either a stylus or a cursor can be used as the digitizer. Two sonic sources are contained in an L frame so that both X and Y coordinates can be determined by calculating the time it takes for the sound wave to reach the microphones contained in the pickup device. This calculation is made on the basis of sound traveling at 345 m/s at 20°C, and the accuracy is dependent on stable ambient conditions. This tends to limit the resolution to about 300 lpi, and the accuracy to ±0.1%. The device may also be implemented with a single sonic source as the digitizing means and a pair of microphones located outside the digitizing area. In this case the location of the transducer is calculated by triangulation and converted into Cartesian coordinates. Digitizers are used primarily for inputting accurate coordinate data from maps and engineering drawings. Their high accuracy requirements have led to relatively high prices. Alternative means for inputting data are the data and graphics tablets that meet most input requirements at a lower cost and accuracy. The main technology is still electromagnetic, and the units are essentially the same as the digitizers, but with lower accuracies. However, several of the other technologies have also been used to achieve lower costs. Most successful among them are the capacitive and resistive versions, which may also be used as digitizers. The capacitive units, also termed electrostatic, use capacitive coupling where the coupling between the tablet and the cursor or stylus is determined by the capacitance made up of the tablet surface as one plate and the pickup element as the other. In this case, the capacitance is given by C = f (eA/d) (89.3) where C = capacitance, e = permittivity of dielectric, A = relative area of two plates, d = distance between plates, and f = proportionality factor. A scanned grid approach is used to determine the location of the cursor. As in the electromagnetic tablet, an X–Y grid of conductors is embedded in the tablet, with semiconductor switches on each line providing contact on a scanned basis. The charge flowing from each capacitance is summed through a summing amplifier as shown in Fig. 89.5(b). The resultant voltage peaks twice, once for the X and once for the Y lines, as they are scanned. The peak positions are digitized by means of a counter that starts at the beginning of the scan, and runs at some multiple of the scan rate. The digital values represent the coordinates of the cursor location. ? 2000 by CRC Press LLC Mouse The mouse has gone a long way from its original invention by Engelbart in 1965, through its redesign at Xerox and introduction by Apple as a main input device, and its general acceptance by computer users as an important addition to the group of input devices. It should be noted, in passing, that the mouse is essentially an upside- down trackball, although the latter is now being referred to as an upside-down mouse. However, the trackball came first and is described further in the next section. FIGURE 89.5 (a) Digitizer tablet. (Courtesy of Numonics.) (b) Capacitive technology. (Source: After T. E. Davies et al., “Digitizers and input tablets,” in Input Devices, S. Sherr, Ed., New York: Academic Press, 1988, p. 186. With permission.) ? 2000 by CRC Press LLC Mice contain motion-sensing elements and are operated by moving mechanical or optical elements. One form uses wheels and shafts to drive the sensing elements, as shown schematically in Fig. 89.6. The angular velocity (w) of the wheel and shaft is given by w = V r /R rad/s (89.4) where V r = velocity of wheel and R = wheel radius. The rotation angle (q) is given by (q) = X/R rad (89.5) where X = distance moved. This type of mouse has two sets of wheels and shafts, one for horizontal and the other for vertical motion. A more popular type of mechanical mouse is the one that uses a ball for the motion sensing device, as shown in Fig. 89.7. Again, the velocity of the ball circumference equals the velocity of the mouse, and the angular velocity is given by w = V/R 1 rad/s (89.6) where R 1 = shaft radius. The smaller the shaft the more rapid its rotation for a given mouse velocity. Another form of the ball-and-shaft mouse is the one that uses an optical interrupter, as shown in Fig. 89.8. In this form, the light from the light-emitting diodes (LEDs) is interrupted by the coded disks that are rotated by the shafts, and is then picked up by the phototransistors and converted into the digital signal that represents the disk rotation. An optical interrupter is also used for the optomechanical mouse, and here the interrupter contains a set of slots; as the inter- rupter rotates quadrature signals are created that correspond to the shaft rotation. In addition to the shaft and optomechanical mice, an early form of mouse used multiturn potentiometers connected to the wheels, and the output voltage that represented the motion varied in direct proportion to the mouse motion. The voltage was then converted by means of an analog-to-digital converter into digital form for input to the computer. Finally, there are the true optical mice that use a special surface that is printed with a set of geometric shapes, usually a grid of lines or dots, that are illuminated and focused on a light detector. The most common form uses a grid made up of orthogonal lines, with the vertical and horizontal lines printed in different colors. These colors absorb light at different frequencies so that the optical detectors can differentiate between horizontal FIGURE 89.6Wheel showing velocities and slip angle. (Source: After C. Goy, “Mice,” in Input Devices, S. Sherr, Ed., New York: Academic Press, 1988, p. 225. With permission.) FIGURE 89.7Ball and shaft. (Source: C. Goy, “Mice,” in Input Devices, S. Sherr, Ed., New York: Academic Press, 1988. With permission.) ? 2000 by CRC Press LLC and vertical movement of the mouse. If such a structure is used as the mouse, then the photodetector will pick up a series of light-dark impulses consisting of the reflections from the mirror surface and the grid lines and convert them into square waves. A second LED and photodetector that is mounted orthogonally to the first is used to detect motion in the orthogonal direction, and the combination of the two inks avoids confusion between the two directions of motion. The system then counts the number of impulses created by the mouse motion and converts the result into motion information for the cursor. This type of mouse has the advantage that no mechanical elements are required. Trackball As noted previously, the trackball uses technology similar to the mouse, but preceded it as an input device. Thus, the comment that it is an upside-down mouse should be reversed. The movable element is housed in an assembly as is shown in Fig. 89.9, and the assembly remains stationary so that much less desk space is required than for the mouse. In addition, the trackball may be mounted on a keyboard so that very little additional desk space is needed. The movable element can be the same as used in the mouse, and the output can be a set of bits corresponding to the coordinates to which the cursor should be driven, or where the command should be carried out. The output format is essentially equivalent to that used for the mouse, and the same protocols are used. The typical trackball has an X and Y optical encoder that generates a pulse for each 0.76 mm of incremental motion of the ball. This means that the pulse train may range from 10 to 2500 pulses per second (pps), depending on how fast the ball is rotated. This is much more rapid than required for satisfactory updates, which need not be greater than about 100 times per second. This can easily be accomodated by the RS-232 protocol using an eight-bit word. Thus, the trackball is an excellent alternative for the mouse, and is rapidly returning to a preferred position as an input device. Joystick The joystick has not achieved much acceptance as an input device for electronic display systems, except for video games, although it has been the preferred control for many types of aircraft. However, it can be used to some extent in display systems other than those used in video games, and therefore warrants inclusion in this section. There are two basic types of joysticks, termed “displacement” and “force-operated”. A typical displace- ment joystick is shown in Fig. 89.10, and may have two or three degrees of freedom. The activating means may vary from as few as four switches mounted 90 degrees apart, to full potentiometers for analog output, and optical encoders for digital output. A third axis may be added by allowing the handle to rotate and drive a third potentiometer. Spring forces of 5 to 10 lbs. are usual for the other two axes, and displacements go from 6 to 30 degrees. FIGURE 89.8Optical interrupter. (Source: C. Goy, “Mice,” in Input Devices, S. Sherr, Ed., New York: Academic Press, 1988, p. 229. With permission.) ? 2000 by CRC Press LLC The force joystick operates by responding to pressure on the handle to generate the X–Y coordinates. It may be either a two- or three-dimensional version, with the same types of handles as for the displacement joysticks. However, it is difficult to use a rotating handle for the third dimension because some force is usually transmitted to the other dimensions causing crosstalk. Therefore, a separate lever is preferred. The force is detected by means of piezoelectric sensors that are bonded to the handle rod, and a voltage source is applied across the network, as shown in Fig. 89.11. The output is taken from the strain gauge and the analog voltage will be proportional to the amount of force. The same type of protocol and output circuitry may be used as for the displacement unit, and both can generate either position or rate data. An exponential curve with a dead zone threshhold is preferred for pulse rates in order to avoid starting pulse rate uncertainties, with the first pulse starting as soon as the threshhold is exceeded. Touch Input Touch input devices come in two basic forms, either placed directly on the display surface, or as a separate panel attached to the computer system. In its second form it is essentially a data tablet differing mainly in that it acts as another display unit with some form of a touch-sensitive surface. In this implementation it is the FIGURE 89.9Trackball. (Courtesy of CH Products,Vista, Calif.) FIGURE 89.10Three-axis displacement joystick. (Courtesy of CH Products, Vista, Calif.) FIGURE 89.11Schematic connections in a force joystick. (Source: After D. Doran, “Trackballs and joysticks,” in Input Devices, S. Sherr, Ed., New York: Academic Press, 1988, p. 260. With permission.) ? 2000 by CRC Press LLC same as the Touchscreen input device, and this discussion concentrates on the technologies used for Touch- screens. There are five different technologies used for touch input devices, which are capacitive or resistive overlays, piezoelectric, light beam interruption, and surface acoustic wave. The system may be divided into the sensor unit, which senses the location of the pointing element, and the controller that interfaces with the sensor and communicates the location information to the system computer. Since the controller is an electronic device that does not use technology different from the computer it is not covered here. The main differences among the different touch input devices are due to the choice of sensor technology, and the discussion concentrates on these technologies. Capacitive. Capacitive overlay technology is illustrated in Fig. 89.12 where a transparent metallic coating is placed over the display screen and the finger or stylus capacitance is sensed to determine the touch location. The overlay may consist of a group of separate sections etched into the surface with each separate section connected to the controller, or a continuous surface con- nected at the four corners. The first form is termed discrete capac- itive, and touch location is determined by having each section sequentially connected to an oscillator circuit where the frequency of oscillations is affected by the pointing device. The oscillation frequency is measured and compared to a stored reference fre- quency. If the frequency difference is large enough then it is rec- ognized as a touch at that location. It is a simple system, but suffers from low resolution and slow response so that it is only practical for menu selection. The analog capacitive system uses the same metallic overlay, but the metallic surface is continuous rather than etched. The connec- tions at the four ends are each connected to a separate oscillator, and the frequency of each is measured and stored. Then when the overlay is touched the change in capacitance will have a different effect on the frequency of each oscillator. These are measured and the differences are used to determine the coordinates of the touch by means of an algorithm. This technique is capable of much higher resolution (250 ′ 250) than the digital approach and is preferred for graphics or other high-density displays. Resistive.Resistive overlay technology requires a more complex assembly consisting of two layers, as illustrated in Fig. 89.13. The layers both contain transparent metallic surfaces and are separated by spacers so that an air gap exists between the layers in the absence of any pressure on the touch panel. The metallic layers face each other and when the outer panel is pressed the metallic layers make contact and form a conductive path at the point of contact. When a voltage is applied between the top of the outer layer and the bottom of the inner layer, the two layers act as a voltage divider, and the voltage at the point of contact may be measured in the X and Y directions by applying the voltage in first one and then the other direction. The measured voltages are then transmitted to the controller where they are converted into coordinates which are then sent to the computer. The panel may be discrete, in which the conductive coating on the top layer is etched in one direction and that on the bottom layer in the other direction, or analog, where the conductive coatings in both layers are continuous. In the discrete case, the panel then acts as an X–Y matrix, and the resolution is determined by the number of etched lines. The analog configuration requires the addition of linearization networks on each edge of the panel so that a large-area resistor is created with a voltage drop in one direction. Other linearization techniques are also possible, but only the four-element system is described here as shown in Fig. 89.14. In this arrangement, one of the layers acts as the large-area resistor and the other as a voltage probe where either can function in either role. For the Y coordinate value the top layer is the voltage probe, and the voltage is applied by the controller to the bottom layer. Similarly, the X coordinate is found by connecting the voltage to the top layer and making the bottom layer into the voltage probe. In either type of system, the resolution can be very high, but the transmissivity is reduced to under 80% due to the multiple layers. FIGURE 89.12Capacitive overlay technology. (Source: After A. B. Carrell and J. Carstedt, “Touch input technology,” SID Sem. Lecture Notes, p. 15.30, 1987. With permission. Courtesy Society for Information Display.) ? 2000 by CRC Press LLC Piezoelectric. The piezoelectric technology uses pressure-sensitive transducers as the means for determining the location of the touch, as shown in Fig. 89.15. The sensor is a glass plate with transducers connected to the four corners. Pressure on the plate causes readings to occur at each of the transducers, which depend on the location of the pressure. Thus, the controller can measure the readings and obtain the coordinates by means of a proper algorithm. This technique allows a high-transmissivity plate to be used that can be curved to follow the CRT face plate curvature, but it allows only a limited number of touch points to be used. Light Beam Interruption.This is a fairly straightforward technology that requires a matrix of light sources and detectors facing each other in the X and Y directions. When the beams from the X and Y light sources are interrupted, this is sensed by the facing light detectors and the signals are sent to the controller. The light beams FIGURE 89.13Resistive overlay technology. (Source: After A. B. Carrell and J. Carstedt, “Touch input technology,” SID Sem. Lecture Notes, p. 15.31, 1987. With permission. Courtesy Society for Information Display.) FIGURE 89.14Four-wire analog resistive. (Source: A. B. Carrell and J. Carstedt, “Touch input technology,” SID Sem. Lecture Notes, p. 15.32, 1987. With permission. Courtesy Society for Information Display.) FIGURE 89.15Piezoelectric technology. (Source: A. B. Carrell and J. Carstedt, “Touch input technology,” SID Sem. Lecture Notes, p. 15.34, 1987. With permission. Courtesy Society for Information Display.) ? 2000 by CRC Press LLC are turned on sequentially by pulsing the LEDs and thus create a full matrix of light beams without requiring each of them to be on continuously. This system does not reduce the screen transmissivity as there is no obstruction of the screen output, but it is limited in resolution to the number of LED detector pairs that can be placed on the periphery of the screen. Another approach to light interruption is to use a rotating beam of light, which has the advantage that only one light source and detector pair is required. This technology is depicted in Fig. 89.16 and consists of a LED and a light detector placed inside a rotating drum which has a slit that allows light to be transmitted outside the drum. The light is swept across the surface and strikes the retroreflectors that sends it back directly to the detector. The beam scan is sampled 256 times on each scan, and Fig. 89.16 shows how two angles of interruption are created, angle B by direct interruption, and angle C by mirror reflection interruption. The result is that the location of the interruption can be calculated by comparing the two angles. Again, there is no obstruction of the screen but a moving element must be added, and parallax errors may occur. Pen-Based Computing.This is an application for touch input devices that is growing at a rapid rate. The input device comes in several forms, each of which can recognize hand printing with the special operating system and software recognizing this type of input. The pen-based input device comes in several forms, of which the one termed TouchPen? can function both as a digitizer with a touch tablet, and as the touch input device with a touch input pen-based computer system. A second one is that developed by Wacom, Inc., primarily for the GO Systems computer, but used by other pen-based systems as well. Finally, a third unit is that made by Scriptel Corp. and used by Wang Laboratories in its system. TouchPen? was developed by Microtouch Systems, Inc., initially for use in GridPad made by the Grid Systems Corp. It is essentially a high-resolution digitizer consisting of an all-glass tablet that can be used with a number of stylus input operating systems to digitize handwriting. It is basically a touch input device using resistive techniques to digitize the handwriting appearing on the display surface of pen-based computer systems. The glass tablet is placed on the display surface and the system pen is used to transmit the digitized data to the computer. As noted previously, the tablet may also be used as a standard touch input device. The second form of pen-based input device is one that uses electromagnetic technology and consists of a grid of wires that transmit radio waves that are picked up by a tuned circuit in the stylus. This circuit resonates at its own frequency and transmits that signal back to the wires at the grid location it is touching. The pen also transmits its signal to the computer, which turns off the grid transmission, and locates the position of the pen by determining which of the grid wires pick up the pen signal. The pen does not need to actually touch the display surface and does not require any power, which is an advantage somewhat counteracted by the higher cost. Finally, the Scriptel unit is similar to that made by Microtouch, but differs in that it uses electrostatic technology and is also similar to the capacitive touch panel. FIGURE 89.16 Rotating infrared beam technology. (Source: A. B. Carrell and J. Carstedt, “Touch input technology,” SID Sem. Lecture Notes, p. 15.34, 1987. With permission. Courtesy Society for Information Display.) ? 2000 by CRC Press LLC Surface Acoustic Wave (SAW). This technology is more recent than the others and has not received wide acceptance as yet. It is based on the transmission through the glass of SAWs generated by transducers mounted on the glass overlay. These waves are detected by receivers also mounted on the glass, and the time of arrival of the waves at the receivers is known because the wave velocity is known. The placing of a finger on the glass weakens the signal and the location of the finger can be deter- mined by the difference in its effect on the SAW. There are two types of SAW systems in use, namely those using reflective techniques and those using atten- uation as the source of position information. The reflec- tive systems are similar to sonar where the time from the source to the pointing finger and then from the finger to the receiver is measured to arrive at finger location. The attenuation technology is illustrated in Fig. 89.17 and consists of two transducers, two receivers, and four reflector strips, all mounted on a glass substrate. One transducer-receiver pair is used for X and the other for Y location. Figure 89.17 shows the X axis pair, and the transducer transmits a burst of acoustic energy in a hor- izontal wave. The wave is partially reflected by the top reflector strips and travels down to the bottom strip where the reflectors are at an angle such that it is reflected to the lower left corner receiver. The wave now has a long rectangular shape, and each point in time corresponds to a specific vertical path across the substrate. The Y axis is scanned in the same fashion after the X wave dies out. Then, when the finger touches the substrate, its water content absorbs some of the energy in the wave, and the wave is attenuated. The dip in the wave amplitude corresponds to the amount of absorbed energy, and the time of the lowest point can be determined, allowing the location of the finger to be calculated. Finally, in addition to the X and Y coordinates, a Z coordinate can be determined, depending on how hard the user presses. This depends on surface contact, which affects the amount of attenuation. The advantages of this system are high resolution, speed of transmission, and the availability of a Z axis component. Its main disadvantages are the variation in moisture content in fingers and sensitivity to local moisture on the substrate. However, it is being used in developmental units and should be considered as another input device technology. Scanners Scanners are a means for inputting text and/or images directly into the computer system, thus avoiding the need for retyping and redrawing information contained in other sources. It is a relatively convenient way to avoid repetition if the data to be entered already exist in readable form. This is done by special image-recognition software that accompanies the scanning hardware, and can transfer an entire image containing both text and illustrations, but without the capability to modify the image. However, the addition of optical character recognition (OCR) software allows the entered text to be modified as if it were entered by typewriter. This can greatly simplify entering and editing text from some preexistent source and has resulted in a proliferation of devices that can perform this function. These devices come in two main forms, hand-held and page scanners, with or without OCR software in addition to the standard image-recognition software. A typical hand-held scanner is shown in Fig. 89.18 and it consists of a light source, a light-sensitive device such as a charge-coupled device (CCD) array, and the electronics to actuate the elements of the array sequentially under software control. The scanner window is placed over the page, and is moved down or across the page so that the window covers as much of the page as falls within the capability of the software. The light source is reflected from the page to the CCD and the charge in the CCD is modified by the reflectivity of the printed material. FIGURE 89.17Attenuation SAW technology. (Source: A. B. Carrell and J. Carstedt, “Touch input technology,” SID Sem. Lecture Notes, p. 15.35, 1987. With permission. Courtesy Society for Information Display.) ? 2000 by CRC Press LLC The window area ranges from 4 to 5 in. in width by 0.5 in. in height and may be moved through 14 to 20 in., so that a fairly large area may be covered in a single manual scan. Images wider than the maximum window may be scanned in two passes, and the OCR software can stitch the two scans together into a single image, although this procedure requires considerable care in scanning so that the scans line up properly. Therefore, when images wider than the window of the hand-held scan- ner are to be scanned, it is advisable to use a flatbed scanner of the type shown in Fig. 89.19 which can handle a full 8.5 in. by 11 in. page, or some of the larger scanners than can accept large drawings and input them into the computer system. Resolutions of 400 dpi and higher, with up to 250 levels of gray and 24 bits of color resolution are available. Thus, scanners offer a wide variety of choice and performance capabil- ities, and are powerful input devices when prepared data in visual form is to be entered into the computer system. Voice Voice input is an intriguing approach to data input, with particular attractiveness to managers who want a simple and direct means for inputting data and commands. For many years, this technology tended to promise more than it could achieve, but recent developments have brought it to the point where it can be considered as a viable input means. This has been due to new developments in software that make it possible to minimize the amount of training required and increase the success rate to close to 100%. One basic approach to speech recognition is represented by the block diagram shown in Fig. 89.20. This is a system that is built around a special chip developed by Texas Instruments. This system uses templates and special algorithms for recognizing the input speech patterns. The system is speaker dependent, with the capability of storing up to 32 word templates and user-defined phrases. The output portion may be superfluous when the system is used only for inputting data and commands, but can be a useful adjunct to the visual response. Other techniques such as speaker-independent and phoneme-recognition systems are also available. Vocabularies range from 50 to 5000 active words, and both isolated and connected words can be recognized, although the larger numbers tend to be associated with isolated word systems. In general, it seems feasible that a combination of speech input and pen-based computing may find a viable market. FIGURE 89.18Hand-held scanner. (Courtesy of Logitech, Freemont, Calif.) FIGURE 89.19Page scanner. (Cour- tesy of Chinon, Torrance, Calif.) ? 2000 by CRC Press LLC Summary The multiplicity of input devices that are available makes it difficult to determine which is most suitable for any specific set of requirements. However, the limited functional comparison of the input devices covered in this section shown in Table 89.4 may be of some use, and in any event is a starting point in this evaluation. It should be noted that what appears best at one time may become unpopular or obsolete at a later time, as occurred for light pens and trackballs, both of which have come back into favor. In addition to the generalized evaluation shown in Table 89.4, it is also of interest to examine representative performance parameters. These are shown in Table 89.5 and while representative do not necessarily cover the range of performance parameters offered. More data may be obtained from the vendors of these devices. Advantages and Disadvantages Input devices make up one of the functional groups of the display systems, and their technical characteristics are covered in some detail at the beginning of this chapter, with performance information provided in Table 89.5 containing characteristic parameter values for each type, as available. The following material expands somewhat on that information by placing these devices in the context of a full graphics display system and evaluating the functions that the various types of input devices perform in that type of system in terms of their advantages and disadvantages. It is of some interest to compare the advantages and disadvantages of each type at this point, as listed in Table 89.6. This is an imposing list and may be used to aid in choosing the best input devices for specific applications. It also concludes this section on input devices. FIGURE 89.20Block diagram of speech recognition and synthesis chip. (Source: After M. Leonard, “Speech poised to join man-machine interface,” Electronic Design, pp. 43–48, September 26, 1991. With permission.) TABLE 89.4Input Device Functional Evaluation Function Input Device Control Data/Text Data/Graphics Total Keyboard E E P 9 Light Pen G G E 10 Tablet E G E 11 Mouse E F E 11 Trackball E G E 11 Joystick F F G 5 Touchscreen G F G 8 Scanner F E G 9 Voice G F P 6 Total 29 23 28 80 E = Excellent = 5; G = Good = 4; F = Fair = 3; P = Poor = 2 ? 2000 by CRC Press LLC Defining Terms Data tablet/digitizer: A device consisting of a surface, usually flat, and incorporating means for selecting a specific location on the surface of the device and transmitting the coordinates of this location to a computer or other data processing unit that can use this information for moving a cursor on the screen of the display unit. Joystick: An input device somewhat in the form of the navigation control device found in early aircraft and operating in a somewhat similar manner by generating series of pulses whose frequency or number depend on how far, with what force, and in what direction the control stick is moved from the central position. Keyboards: Electromechanical devices consisting of sets of keys labeled with alphanumeric, numeric, and functional designations that enable the user to describe and define the operation to be performed. Light pen: Neither a pen or a light source but rather an input device in the shape of a pen that operates by sensing the existence or nonexistence of light pulses at specific locations on the surface of a display device and uses this information to signal the computer as to the location of the pen. Mouse: An input device based on a much older type known as a trackball and fancifully named because it bears only a casual resemblance to a mouse. It consists of a roller ball that is moved on a flat surface and causes orthogonal potentiometers or other types of X–Y-position signal generators to move and produce electrical signals defining the desired coordinates of the cursor on the screen so that the cursor can be moved to that position. TABLE 89.5 Representative Performance Parameters Input Device Parameter Value Light pen Response time 150–500 ns Spectral response 4200–9500 A Luminous sensitivity 0.03–0.7 nts Field of view 0.02–0.1 in. Ambient rejection 350 nts Data tablet (digitizers) Resolution (l/in.) 100–2000 Accuracy (in.) 0.0005–0.02 Active area (in.) 12 ′ 12–60 ′ 120 Active height (in.) 0.02–2.5 Digitizing rate (pps) 100–350 Transducers Stylus, puck, cursor Mouse Resolution 10–1000 dpi Speed 1–20 in./s Accuracy 25–1000 dpi Trackball Resolution 100–1000 cpi Speed 1200–9600 BPS Accuracy 100–1000 dpi Ball diameter 1.5–2.5 in. Joystick Travel 25–30° Accuracy 5–10% Repeatability 1% Touchscreen Resolution 256 ′ 256–4096 ′ 4096 Transmissivity 60–100% Viewing area (in.) 3 ′ 4.5–15 ′ 20 Speed 80–200 touch pts./s Scanner Resolution (dpi) 75–1600 Scan rate (in./s) 0.5–2.0 Scanning width (in.) 4.1–36 gray shades 32–256 Scan time (s/page) 1–30 Voice Active vocabulary 13–5000 words ? 2000 by CRC Press LLC Scanners: Means for converting hard copy into electrical signals that can be entered into a computer or data processing system. The usual means for accomplishing such conversion is to move a light beam over the surface containing the data either by hand or automatically and using arrays of light-sensitive devices to convert the reflected light into electrical pulses. Touch input: A means for selecting a location on the surface of the display unit using a variety of technologies that can respond to the placing of a finger or other pointing device on the surface. These are essentially data panels placed either on the display surface or between the user and the display surface. Trackball: The earliest version of an input device using a roller ball, differing from the mouse in that the ball is contained in a unit that can remain in a fixed position while the ball is rotated. It is sometimes referred to as an upside-down mouse, but the reverse is more appropriate as the trackball came first. Voice: Means for enabling a computer or data processing system to recognize spoken commands and input data and convert them into electrical signals that can be used to cause the system to carry out these commands or accept the data. Various types of algorithms and stored templates are used to achieve this recognition. Related Topic 89.2 Computer Output Printer Technologies TABLE 89.6 Input Devices—Advantages and Disadvantages Device Advantages Disadvantages Keyboard Simple operation Requires many keys Well known Requires training Standard interface No graphics Light pen Eye-hand coordination Arm fatigue Low cost models Limited resolution No desk space required May block display Graphic tablet Natural hand movements Eye-hand conflict Screen not blocked Requires desk space No parallax Breakable stylus Good for graphics Poor for A/N entry Mouse Small space needed Some space needed Low cost Slow transmission Screen viewing Low resolution Any surface may be used Grid for optical (Optical) noiseless Mechanical noise Trackball High resolution Poor for A/N input Fixed desk space Slow transmission Screen viewing Mechanical noise Tactile feedback 3-D difficult Joystick Fixed desk space Low accuracy Low fatigue Low resolution Low cost No A/N input Touchscreen Eye-hand coordination Arm fatigue Minimal training May block display Minimum input errors Varied resolution User acceptance Parallax No special commands Slow data entry Scanner Full A/N page input Hand scanner width Color scan input High cost for color High resolution Slow input OCR software Compatibility Voice Ease of use Limited words Minimal training Machine training No special devices Graphics difficult ? 2000 by CRC Press LLC References H. Brunner et al., “Effects of key action design on keyboard preference and throughput performance,” Micro Switch. A.B. Carrell and J. Carstedt, “Touch input technology,” SID Sem. Lec. Notes, pp. 15.30–15.35, 1987. T.E. Davies et al., “Digitizers and input tablets,” in Input Devices, S. Sherr, Ed., New York: Academic Press, 1988, p. 186. D. Doran, “Trackballs and joysticks,” in Input Devices, S. Sherr, Ed., New York: Academic Press, 1988, pp. 251–262. C. Goy, “Mice,” in Input Devices, S. Sherr, Ed., New York: Academic Press, 1988, pp. 225–232. M. Leonard, “Speech poised to join man-machine interface,” Elec. Des., pp. 43–48, Sept. 26, 1991. S. Sherr, Electronic Displays, New York: Wiley, 1979, pp. 323, 388–389. Further Information Electronic Displays, 2nd ed., by Sol Sherr and published by John Wiley & Sons, Inc., contains an extensive and detailed discussion of other aspects of display systems and technology, as well as a somewhat expanded version of this section. In addition, Input Devices, edited by Sol Sherr, and Output Hardcopy Devices, edited by Robert C. Durbeck and Sol Sherr, both published by Academic Press, include extensive discussions of a wide variety of devices. The Society for Information Display (SID) sponsors a yearly symposium at which a large amount of information on new developments in information display as well as tutorials and seminars on basic information display topics are presented and made available in published form. In addition, it publishes two journals, namely, Proceedings of the Society for Information Display and Information Display. Other relevant meetings and publications are those sponsored by the Computer Society and Electron Devices groups of the IEEE, the SIGGRAPH group of the Association for Computing Machines (ACM), and the National Computer Graphics Association (NCGA). 89.2 Computer Output Printer Technologies Robert C. Durbeck Electronic printers for computer output represent a very important part of the computer industry. They range from small, inexpensive printers for personal computers and workstations to very large and fast page printers used for bulk printing output for large-scale computer systems. The technologies employed for this wide scope of printing requirements are diverse: some based on “impact” methods to transfer ink from a sheet or ribbon to paper, others based on more sophisticated “nonimpact” methods. Today, there is no single technology which completely dominates. A wide range of user needs has led to the present proliferation of printer technologies. The most prevalent are discussed in the following. Classification of Printer Technologies Table 89.7 illustrates the two main classifications of printer technologies and also the wide range of extant technologies. Those technologies listed under line impact and page printing are used for large computer system printing; by far the most popular are fully formed character and electrophotographic (so-called laser printers). The most common printing technologies used for personal/workstation computer systems are electrophoto- graphic, ink jet, and serial wire matrix. Emphasis in the following is directed to these favored technologies, with brief descriptions of the others included. Page Printer Technologies By far the most important page printer technology is electrophotography (EP). EP, as well as the much less employed ionographic and magnetographic technologies, uses “powder toning” development of an intermediate ? 2000 by CRC Press LLC “image” created in the process. Liquid toners are also used to develop electrostatic images. Thermal page printers employ a full-page-width linear array of thin-film heater elements to melt and transfer ink from a ribbon to the paper or to mark special thermal-sensitive paper. Electrophotographic Printing EP printers use essentially the same technology found in most “plain paper” copiers, the major exception being the printhead. Instead of using page- or line-imaging optics as in a copier, the printhead utilizes a solid-state laser (usually GaAlAs) or gas laser (typically HeNe) to scan across and expose a photoconductor drum or belt to create a “latent image” (see below). A few EP printers use stitched arrays of light-emitting GaAs 1–x P x diodes (LEDs) with Selfoc? glass fiber optics or an array of liquid crystal shutters, the latter to modulate light from a bright line light source. Other possibilities are electroluminescent, magnetooptic or electrooptic arrays, but these have not been commercialized to any extent. There are basically six major steps employed in the EP printing process (see Fig. 89.21): (1) uniform charging of the photoconductor (PC) electrostatically; (2) exposing the PC to the image light pattern, which results in selective discharge of the area charge created in Step 1, creating an electrostatic image; (3) developing the PC by bringing electrostatically charged toner particles (black or colored) to the surface of the PC where they selectively adhere to appropriately charged regions; (4) electrostatically transferring the toned image from the PC to the final medium (usually paper); (5) thermal fusing of the toner to the paper; and (6) cleaning residual toner from the surface of the PC to allow reinitiation of the six step cycle. TABLE 89.7Types of Printer Technologies Impact Nonimpact Line impact Page printing Fully formed character Electrophotographic Dot band matrix Ionographic Shuttle hammer matrix Magnetographic Electrostatic Thermal Serial impact Serial nonimpact Fully formed character Ink-jet Serial wire matrix Continuous Piezoelectric/impulse Thermal/bubble-jet Thermal Direct (thermal paper) Thermal transfer Resistive ribbon Figure 89.21The six basic electrophotographic printer process steps: charge, image, develop, transfer, fuse, and clean. ? 2000 by CRC Press LLC Step 1—Charging the PC.The most common approach used is corona charging. One or more thin corona wires (typically tungsten) are supported directly above the PC and are energized to 5–8 kV. The resultant high electric field surrounding the wire causes electrons in the immediate region to be accelerated to energies sufficient to ionize local air molecules. Either positive or negative ions are then attracted to the outer surface of the PC (which, when unexposed to light, acts for a short period of time as an insulator) depending on the sign of the potential difference. At the same time, a counter-image charge is formed on the inner side of the PC. The two corona structures most commonly used are the corotron and the scorotron (see Fig. 89.22). The grid on the scorotron is used to more precisely control the resultant voltage charge level on the PC (approximates the grid voltage). Both dc and ac designs are used; the latter usually include a glass sleeve around the corona wire to reduce localized high-emission spots on the wire due to contaminants. To save on cost for very low- cost EP printers and to reduce corona by-products (e.g., ozone), a lower-voltage conductive elastomer charge roll in direct contact with the PC has also been used in place of the corona wire. Step 2—Exposing the PC.The wavelength of the exposing light source must match the spectral sensitivity of the PC. If the PC is discharged in areas that will be printed white, the overall process is termed charge area development (CAD); if the discharged area will be printed black (or color), the process is called discharge area development (DAD). Both CAD and DAD processes are used in EP printers but CAD is the only common process used in copiers. Selective discharge of the PC involves two steps: (1) photogeneration of electron-hole pairs and (2) transport of the electrons and holes in opposite directions under the influence of a high-dc bias field, locally dissipating the surface charges created in Step 1 (see Fig. 89.23). Both organic and inorganic PC materials are used (see Fig. 89.24). A variety of charge-generation and charge- transport material systems have been developed for organic PCs; most use separate layers for charge generation and charge transport. Examples of efficient organic charge-generation materials sensitive at both GaAlAs (7800 ?) and HeNe (6328 ?) wavelengths include squarylium and thiopyridium dyes in an appropriate binder layer (?0.5 mm thick). The charge-transport layer (CTL) consists of a thicker layer (20–30 mm) of a charge-transport Figure 89.22Two types of coronas for charging the photoconductor: the scorotron and the corotron. Figure 89.23Light is used to selectively discharge the photoconductor. Electron-hole pairs are photogenerated in the charge-generation layer, followed by charge transport under a dc bias field and then selective neutralization of the surface charges, thereby creating an electrostatic image. ? 2000 by CRC Press LLC molecule dispersed in an inert binder, or simply a good charge-transporting polymer (e.g., polyvinylcarbazole). Since most polymer transport materials are essentially hole-transport materials, this requires that the PC surface charge produced in Step 1 be negative. The CTL must be transparent at the imaging wavelength. Examples of inorganic PC materials include amorphous chalcogenide alloys such as a-Se, a-As 2 Se 3 and Te- doped Se. Most do not have a high sensitivity at GaAlAs wavelength, but a-Si does. a-Si also offers other desirable properties (e.g., durability and lack of fatigue), but is quite expensive to produce. The laser beam is scanned linearly over the PC in a direction orthogonal to the PC motion; the combined motions covering a “page”. The most common gas laser scanning technique employs a high-speed rotating polygon mirror along with beam-expanding optics, an acoustic-optic modulator (not required for solid-state lasers) and an f-q imaging lens. To produce a quality image, the multifaceted scanning mirror system must be essentially free of facet defects, up-and-down wobble, variations in polygon rotational velocity and lack of synchronization with the pixel clock. Some nonplanarity of the facet surfaces can be corrected with anamorphic optics. LED arrays and liquid crystal shutter systems do not have these same technical challenges but, so far, they are more expensive to produce. Step 3—Developing the PC. There are basically three development techniques: dual component, monocom- ponent and liquid development. The first two use powder toner. Until the advent of EP printers for personal and workstation computers, the most common method of development was dual component, where polymer- coated magnetic carrier beads are mixed with the toner particles and development is done with a “magnetic brush.” This technique is still prevalent in high-end printers. With this approach, the 5–20 mm toner particles (consisting mainly of resin plus carbon particles or colorant) are triboelectrically charged by repeated contacts during mixing with the much larger (60–250 mm) magnetic carrier beads. The toner particles then electrostat- ically adhere to the opposite-sign charged carrier beads. Charge control agents (e.g., complex organometallic salts) are often included in the toner composition to control the charge level, rate of charging, and consistency of charge. The mix is mechanically directed to a nonmagnetic rotating shell which has fixed magnets located within its core, adjacent to the gap between the PC and the shell (see Fig. 89.25). The gap is typically 0.5 to 6 mm, depending on the specific system. As the shell rotates, the mixture is carried to the gap, and chains of carrier beads (coated with toner—the “magnetic brush”!) form along the local magnetic field lines. These field lines are approximately perpendicular to the shell at the smallest gap. In addition, a development voltage (200–500 V) is applied between the PC and shell. This provides a high field in the gap whose local value is determined by the applied voltage, gap dimension and the electrical properties of the material mix in the gap. Figure 89.24Spectral absorption characteristics of several photoconductor materials: (a) squarylium dye, (b) SeTe, (c) As 2 Se 3 , and (d) a-Si. ? 2000 by CRC Press LLC The electric field at the end of the last carrier bead (next to the PC) in a chain may be up to 50 times the nominal unfilled gap field, the value being greatest with uncoated carrier beads. It can be shown that for coated carrier beads, the mass of toner per unit area developed on the PC is approximately (89.7) where v r /v p is the surface speed ratio between the developer roll (shell) and the PC surface, V the voltage across the gap L g , C m the toner charge-to-mass ratio, e 0 the permittivity of free space, r c the carrier bead radius, r t the toner radius, k the dielectric constant of the bulk toner, and d c the thickness of the carrier polymer coating. Monocomponent development is used almost exclusively for low-end printers because this process does not require carrier beads, toner concentration sensors or toner replenishment hardware, resulting in much lower manufacturing costs. This approach has also allowed the use of replaceable toner/developer cartridges which, although more costly on a supplies cost-per-page basis, adds greatly to the user-perceived reliability of the system (if it fails—just replace the cartridge!). A rotating donor/development roll with appropriate charging properties is employed to charge the toner by touch-and-rubbing contacts (see Fig. 89.26). The toner electro- statically adheres to the donor roll and is transported to contact the PC at the nip. Here, in the presence of a development bias field, the toner is selectively transferred to those areas on the PC with opposite sign charge. Liquid development employs a high-resistivity hydrocarbon dispersion of very fine toner particles (<1 mm) that are charged naturally in the solvate. Mechanical means are used to bring the liquid into contact with the PC, and the toner is then electrophoretically transferred to the latent image areas on the PC. Color can be accomplished by using multiple development stations, one each for the subtractive colors (cyan, yellow and magenta) plus black. Toners are colored by either dyes or pigments. The four-colored images may Figure 89.25Dual-component magnetic brush developer. Toner particles, adhering electrostatically to much larger mag- netic carrier beads, are transported into the photoconductor-developer gap to tone the electrostatic image. (Source: R. C. Durbeck and S. Sherr, Eds., Output Hardcopy Devices, San Diego, Calif.: Academic Press, 1988, p. 242. With permission.) m A V LC rr vv gm ct c rp =+× p kd e 0 2 {}/[(/) ] /** ? 2000 by CRC Press LLC be accumulated on the PC (see Fig. 89.27) or alternatively on an intermediate belt or drum, or even on the paper itself. A wide range of colors and the visual illusion of gray scale can be achieved by the use of laser pulse- width modulation and the use of “super-pixels” consisting of N 3 M arrays of binary pixels to provide digital half-toning. Step 4—Transfer of Toner to Paper.Here a corona is used to charge the back side of the paper and the toner is transferred from the PC to the paper. Typically, some small fraction of the toner usually has a charge of the wrong sign. It is at this step where most of this wrong-sign toner is removed, minimizing “background” on the page, since primarily only toner with the correct sign gets transferred. Step 5—Fusing Toner on the Paper. After transfer the toner is only loosely held (electrostatically) onto the paper; it must be fixed by fusing. There are several fusing techniques: (1) hot-roll using hot fuser rolls under pressure, (2) cold-roll, (3) solvent vapor, (4) flash lamp, and (5) radiant heating. Hot-roll fusing is predomi- nantly used but all five techniques have been commercialized. The most efficient approach is hot-roll but all Figure 89.26Monocomponent developer system. Figure 89.27Four developers are used to produce color plus black. Four photoconductor drum rotations are needed to accumulate the four-color toner images. ? 2000 by CRC Press LLC of the thermal approaches require 0.5–2.0 J/cm 2 energy at the paper surface. The energy required is determined by the contact duration (related to the nip compliance and roll speed), pressure, paper water content, and the melt-flow rheological characteristics of the toner. Flash fusing has only been used in mid-range to high-end printers because expensive power supplies and capacitor banks are needed. A xenon flash lamp is typically pulsed for a few milliseconds to provide the required energy. Thermal efficiency improves with very short pulses, but undesired toner degradation volatiles increase with the higher temperatures produced. Care must be taken with radiant fusing to insure that excessive paper overheating does not occur if the paper is stopped under the incandescent lamp. Step 6—Cleaning the PC. To restart the overall EP process, the PC must be cleaned of residual toner and contaminants. Fiber brushes, scraper blades, and “magnetic-brush” cleaners are used. For low-end printers, scraper blades suffice and are replaced (in some systems) with each new cartridge. Rotating soft-fiber brushes with air flow collection are used commonly in mid-range to high-end printers. Magnetic-brush cleaning is very similar to magnetic-brush development (see above). Ionographic Printing Ionographic printer devices also use powder toner technology and the overall technology is very similar to EP. One major difference, however, is the creation of the electrostatic image for subsequent toner development. The ionographic process uses an ion source (high-voltage drive electrode) and a page-wide array of control and screen electrodes to gate ions directed toward a drum coated with a thin-film dielectric material (e.g., Al 2 O 3 ), thus creating a charged image on the drum. In present commercial ionographic printers, the transfer and fuse steps of EP are also replaced with a “transfix” process where the paper sheet is squeezed between the toner-developed dielectric drum and a compliant cold pressure roll. The ionographic technology has two fewer process steps (vis-à-vis EP) but the ion printhead represents significant technical challenges (cost and lifetime). Magnetographic Printing Magnetographic printing is also similar to EP but employs magnetic powder toner and a magnetic printhead. The magnetic head is an array of individually addressable magnetic write gaps, each representing a pixel location on a line across the page. The head writes a magnetic “image” on a belt or drum coated with a magnetic material such as g-Fe 2 O 3 , Co:P or Co:Cr. Toner is attracted to those areas on the drum where flux reversals (and, hence, magnetic fields external to the the magnetic media) are present. The other process steps are essentially the same as used with EP but toner charge levels must be kept under control to ensure that electrostatic forces do not dominate over magnetic forces. Electrostatic Printing There are two basic approaches that have been developed: (1) one using a special dielectric paper, and (2) the other using plain paper. In the first approach, a page-wide linear array of electrode discharge pins are inde- pendently pulsed to charge the surface of a moving, conductive-base, dielectric-coated paper. Both powder and liquid toners may be used to develop the image; thermal fusing of the image is required for powder toner. With the plain paper version, either a precharged or uncharged dielectric drum or belt can be used depending on whether the array of electrode pins are used to charge or discharge the media surface. The other process steps are essentially the same as with the nominal EP printing process. Cost savings can be realized by multiplexing the electrode driver lines because there is a process threshold of hundreds of volts. This requires, however, that segmented counter electrodes be positioned behind the receiving surface (drum, belt, or paper). Thermal Page Printing Thermal transfer page printing is accomplished by using a stationary page-wide linear array of thermal heating elements coupled with a page-wide transfer ribbon roll (see Serial Thermal Printing below for discussion of thermal printhead technologies). The ribbon is positioned between the printhead and the receiving paper. The ribbon typically consists of a polycarbonate or polyester film substrate (10–20 mm) coated with a waxy ink layer. Heat from the thermal printhead elements must penetrate through the substrate to heat (melt) the dye/wax coating. The melted ink layer is pressed into the paper surface by printhead pressure, and after the paper and ribbon move together away from the printhead, the partially cooled ink layer adheres better to the paper than ? 2000 by CRC Press LLC to the ribbon substrate. The ribbon is then peeled away from the paper, leaving the desired image on the paper. Considerable energy must be applied to melt the ink layer (e.g., 2–4 J/cm 2 ) which does restrict speed. Also, comparably smooth paper (￡50–100 ml/min Sheffield roughness) is required for quality printing with minimal gaps or voids. Print resolution of better than 8 pixels/mm can be achieved with smooth paper. Sublimable transfer dyes have been used for high-quality colored images since the amount of dye transferred can be controlled somewhat by the energy supplied to the head heater elements. Also, since the sublimable dyes tend to penetrate into the paper sizing and fibers, a less smooth paper can be used than with waxy materials. The energy required for dye transfer is similar to that required for waxy ribbons. Serial Nonimpact Printer Technologies Ink-jet and thermal printers represent the two major classes of serial nonimpact technologies used for personal and workstation computer printer output. As shown in Table 89.7, the ink-jet technologies can be subclassified as continuous, piezoelectric, and thermal/bubble-jet. The thermal technologies can be subdivided into direct thermal, thermal transfer, and resistive ribbon. Serial Ink-Jet Printing Ink-jet technologies have evolved over the last two decades with continuous ink-jet being the first to be developed, followed by piezoelectric and then thermal or bubble-jet technologies. All three have been commer- cialized, although bubble-jet is by far the most popular today and continuous ink-jet is used primarily for speciality page printer or high-quality color graphics and image applications. Both piezoelectric and bubble- jet are “drop-on-demand” or “impulse” technologies, i.e., a drop is ejected from the printhead only when desired. With continuous ink-jet, a continuous stream of droplets is generated by the printhead and undesired drops are deflected electrostatically away from the paper (or vice versa). Ink requirements for these technologies are very demanding and the development of appropriate inks is as important (and difficult) as the development of the printer and printhead hardware. Demands on the ink are both extensive and conflicting, and include the requirements of nonclogging in the nozzle but fast dry time on the paper, water-based but water-resistant after drying on the paper, and quality printing on a wide range of papers (requires minimal feathering and controlled spot size). Nonaqueous solid inks have also been used; the ink is solid at room temperature but is liquid at an elevated head temperature. Continuous Ink-Jet Printing.With this technology, ink is continuously jetted from a small-diameter nozzle under pressure (see Fig. 89.28). Although the resultant jet stream will naturally break into small drops, this phenomenon is assisted and stabilized by the inclusion of a piezoelectric perturbation transducer, driven at the desired drop rate. Lord Rayleigh was the first to determine that the dimensionless instability factor (89.8) (where g r is the growth rate of jet instability, s the surface tension, r the fluid density, and d the jet diameter) is maximized for l/d = 4.51, where l is the perturbation wavelength. Operation at this design point produces Figure 89.28(a) Continuous drop ink-jet, and (b) continuous spray ink-jet technologies. g d r (/) / sr 312 ? 2000 by CRC Press LLC very consistent drop breakoff and diameter. Operation at frequencies in excess of 100 kHz is possible with drop velocities of typically 25 m/s with a 50-mm nozzle diameter. Because the drops must be electrostatically charged at the breakoff point, conductive ink must be used. The ink source is typically grounded and a controllable voltage electrode is placed at the breakoff point (usually surrounding the jet stream). The charge level on each drop is then proportional to the applied voltage. Each charged drop can then be deflected by parallel downstream deflection plates with a field of typically 10 kV/cm. This deflection (assuming parallel plates and a uniform electric field between the plates) can be approximated by: (89.9) where d l is the deflection length, q the drop charge, m the drop mass, E the electric field, v the drop velocity, l p the deflection plate length, and l z the distance from the upstream end of the deflection plates to the paper. Drop deflection can be either binary or analog. With the former, the drop either reaches the paper or is directed into a collection gutter. With analog deflection, the drop may be deflected linear (e.g., any position over the height of a printed character). Smaller satellite drops may be also produced between the primary drops. To eliminate their effect, the excitation system is designed to produce forward-merging satellites as these charge simultaneous with the preceding drop; thus the forward-merging (through “drafting”) and subsequent drop coalescence will not alter the charge-to-mass ratio of the augmented drop. Compensation must be provided for both aerodynamic and electrostatic interactions. Two primary examples are that (1) the first drop in a sequence of drops encounters much greater aerodynamic drag than subsequent drops, and (2) the charge on a drop is influenced by the charge on the previous few drops. One approach to greatly reduce electrostatic drop interaction and to stabilize merging effects is to include noncharged drops between charged drops. This obviously reduces the effective drop rate by a factor of two and requires a design configuration where only charged drops reach the paper. Charging electrode voltage adjustment algorithms based on voltages applied to prior drops are also used to reduce the electrostatic interaction. An alternate approach is the continuous spray design where a smaller nozzle (10–20 mm) is used. Much smaller drops are produced at higher velocities (?40 m/s). It is often called the Hertz method [see Fig. 89.28(b)]. As with the continuous drop approach, a controlled voltage electrode is positioned around the breakoff point of the jet stream, a conductive ink is used, and the smaller droplets are charged proportionally to the applied voltage. The stream of droplets is directed to the paper when no voltage is applied. When a voltage is applied, the resulting electrostatic charge on the droplets produces strong mutual repulsion forces and the stream transforms into a spray, the cone angle of which is determined by the applied voltage. The spray is intercepted by a collecting surface surrounding the collection orifice which allows only uncharged or low-charged droplets through to the paper. With analog voltage control, the amount of spray that passes through the orifice can be varied, thus providing gray-scale capability. This approach (with multiple orifices) has been commercialized for very high-quality color image and graphics applications. Piezoelectric Ink-Jet Printing. Piezoelectric ceramic transducers are employed with this technology. These materials (e.g., lead zirconate titanate and barium titanate), when polarized, change their physical dimensions when subject to an electric field—usually applied through surface electrodes. Deflections of several angstroms per volt are typical. When the transducer is pulsed with a voltage, the deflection generates a pressure wave in an adjacent ink chamber, resulting in the ejection of a single drop—hence the descriptors, “impulse” or “drop- on demand”. Four implementations are shown in Fig. 89.29. Often arrays of these devices are integrated into a serial printhead which allows the printing of a one-character-high-per-head pass across the paper. Color printing can be accomplished by assigning one or more nozzles per color. Only a very small (100–1000 ?) deflection of the piezoelectric transducer is needed to create the ink chamber pressure wave if the displacement is very rapid (10–100 ms). Hence, these devices are very efficient; only a few microjoules per drop are required for robust operation. Drop ejection rates of over 20 kHz have been demonstrated in the laboratory but commercial devices are typically designed for the 5–10 kHz range. Drops that produce a d l qE mv ll l p zp = ? è ? ? ? ÷ × 2 2 2 (–) ? 2000 by CRC Press LLC spot size of 150–200 mm on paper can be achieved with an orifice diameter of about 50 mm. Even greater device efficiency can be obtained by synchronizing the arrival of a direct wave and a reflected wave at the nozzle (e.g., a negatively reflected wave caused by an initial expansion pulse plus the direct wave from a following compres- sion pulse). These two waves can be made to reinforce one another at the time and place of drop ejection. Figure 89.29(a) shows a piezoelectric/membrane laminate design with a disk-shaped piezoelectric transducer which, upon being pulsed, deforms (assumes a finite radius of curvature) and thus creates a pressure wave in the ink chamber. Figure 89.29(b) illustrates an “oil-can” version of the laminate approach. In Figure 89.29(c), a tubular transducer is used to squeeze the ink chamber. Figure 89.29(d) represents a push-rod or piston design where the piezoelectric material is used in the extensional mode. Attempts have also been made to use modern semiconductor planar/etching processing techniques to create low-cost arrays of devices. Bubble-Jet Printing.This technology has also been called thermal-jet. With this approach, very small thermal resistors on the ink chamber wall are electrically pulsed. Joule heating of the resistor causes the temperature of the ink adjacent to the heater to rise to 350–400°C (see Fig. 89.30). Because the ink becomes locally superheated, nucleation of tiny bubbles takes place on the surface over the heater. These bubbles coalesce and very rapidly form a single expanding bubble which, by displacement (like a piston), propels a single drop of ink out at the orifice. The electrical pulse must be short (typically 3–6 ms) to insure low conductive heat losses; however, the power density is extremely high (~500 MW/m 2 ). The energy applied per drop is 30–50 mJ but only a small fraction (a few percent) represents the kinetic energy of the drop. The remaining energy is thermally dissipated in the ink and device structure. When the thermal energy in the superheated layer is depleted, the bubble begins to collapse. The total cycle (nucleation plus bubble growth and collapse) is normally complete in about 20 ms. Drop rate, however, is typically limited to less than 10 kHz, mainly because of the limits of thermal dissipation. Cavitation damage can occur to the heater structure if the bubble collapse is too violent. Proper design of the ink chamber geometry can provide the necessary damping and minimize this problem. Heater element materials used for this tech- nology include HfB 2 , ZrB 2 , Ta 2 Al and TaN. Passivation over-layers (e.g., SiC, SiO 2 , Si 3 N 4 , plus certain metals such as tantalum) are deposited on top of the thin-film heaters to provide protection from chemical and mechanically enhanced corrosion. Materials, structures and thin-film deposition processes must be carefully Figure 89.29Four approaches to piezoelectric drop-on-demand ink-jet technology: (a) piezoelectric/membrane laminate design, (b) “oil-can” version of (a), (c) piezoelectric squeeze-tube approach, and (d) piezoelectric push-rod method. ? 2000 by CRC Press LLC designed to allow billions of heater pulse cycles without failure at a local power density of 500 MW/m 2 . Also, very special ink must be used so that very minimal chemistry occurs at the hot interface layer, e.g., so that the chemicals in the ink do not break down and form a thick carbonaceous film over the heater. These carbonaceous films, when more than a few hundred angstroms thick, destroy the device velocity and thermal efficiency characteristics. The ink must also not etch the overcoats. Very low-cost, compact, low-power printers have been developed based on this technology because energy requirements are very low and the printhead can be made inexpensively using semiconductor-like planar processing techniques. A proliferation of these printer products (many with color) have appeared in the market over the past five years. Serial Thermal Printing Thermal technologies have been used in many low-cost serial printer applications. There are three basic approaches: (1) use of a heat-sensitive special paper, (2) thermal transfer of ink from a ribbon to paper, and (3) a variant on (2) with a special ribbon structure and printhead that improve thermal efficiency and allow printing on a wide range of standard papers (including the relatively rough office bond papers). Direct Thermal Printing. The key aspects to this technology are the thin-film resistive serial printhead and the special paper required. A typical array structure (see Fig. 89.31) consists of photolithographically defined and deposited heater material (e.g., Ta 2 N or TaAl) on a contoured insulating substrate. One or more protective layers (e.g., SiO 2 , Ta 2 O 5 or SiC) are deposited on top of the heater layer for abrasion resistance and electrical insu- lation. Contours (raised areas or bumps) on the surface under the heater areas are constructed by a raised glass “glaze” of about 40 mm and provide both improved contact with the paper and short time constant thermal insulation to the substrate. This structure is often called the thin-film head structure. Alternates include (1) the silicon mesa technology, where a two-dimensional array of silicon mesas is fabricated from monolithic silicon, and (2) the thick-film technology. With the former, each mesa contains its own resistor-transistor/diode on the base of the silicon chip. Joule heating occurs when voltage is applied to the resistor via the transistor/diode. With the thick-film approach, a resistor Figure 89.30Thermal-jet device showing one heater-nozzle package. Joule heating of the thin-film heater causes nucleate boiling in the ink adjacent to the heater. A bubble forms and pushes a drop of ink out of the nozzle. Figure 89.31Thin-film thermal head struc- ture. (Source: R.C. Durbeck and S. Sherr, Eds., Output Hardcopy Devices, San Diego, Calif.: Academic Press, 1988, p. 282. With permission.) ? 2000 by CRC Press LLC paste is typically screened onto a ceramic substrate. Materials for the resistor paste include borosilicate glass and lanthanum glass-ruthenium oxide. The elements are typically 25 mm thick and slowly wear down by contact with the paper. Since, with each approach, the resistive elements must be allowed to cool before the head is ready to “write” the next pixel position on the paper, the heating and cooling time constants of the head primarily determine the attainable print speed. Special thermal paper typically has a coating with a leuco dye plus a phenol developer in a polymer binder which react at the elevated temperature provided by the printhead to form a colored species. The resulting optical density is a function of the temperature reached so that some gray-scale capability is possible. Energy density requirements are in the range of 2–6 J/cm 2 . Thermal Transfer Printing.The desire to eliminate costly special thermal papers even for low-cost serial printers led to the development of the thermal transfer printing technology where a separate transfer ribbon is interposed between the thermal printhead and the paper (see Thermal Page Printing above for discussion of ribbon technology). Resistive Ribbon Printing. This technology grew from the need to provide quality thermal printing at higher speeds and on a wide range of papers, including standard office bonds. With this technology, the heating function is repositioned from within the printhead into the ribbon itself. The ribbon (see Fig. 89.32) includes an additional aluminum conductive and heating thin-film layer, sandwiched between the “standard” thermo- plastic ink layer and the ribbon substrate. The substrate is made conductive (350–900 V/M) by the incorpo- ration of sufficient (20–30%) carbon black in the polycarbonate film. Joule heating occurs as current flows from the addressable head pixel-size electrodes through the aluminum layer to the large single-ground return electrode. Current density is maximum directly beneath the addressable electrodes so that temperatures there (but not elsewhere) are sufficient to cause ink layer melt and transfer. In addition, a thin release layer can be added between the ink layer and the aluminum heating layer to provide enhanced and smooth-edged pixels; Figure 89.32Resistive ribbon head and ribbon system. ? 2000 by CRC Press LLC this provides sharp character edges and allows more detail in images. This technology allows (1) 2–3 speed increase (vis-à-vis conventional thermal transfer printing) because printhead thermal time constants are much less of a factor, (2) very high-resolution printing (up to 40 pixels/mm), and (3) printing on office-quality bond papers. Also, already deposited ink can be selectively lifted off the paper by reducing the printhead power and using a fresh ribbon area. Offsetting these advantages is the fact that the ribbon is more complex and costly than conventional transfer ribbons. Impact Printer Technologies The earliest electronic printing technologies were impact devices employed on the early plug-board program- mable accounting machines that used punched-card input. Before that there were typewriters and teleprinters. Line impact printers are used on mid-range to high-end computer systems; their continuing appeal is based first on the ability to print multipart forms, and secondly on lower cost of printing and greater reliability (as compared with electrophotographic printers). Serial impact printers historically represent the greatest sales volume for all computer printer technologies by a wide margin. By far the largest volume has been for personal computer output applications. Their appeal stems from products offering very low cost (device and supplies), high reliability and, again, multipart forms. More recently, however, thermal ink-jet and low-end EP printers have largely supplanted these impact products. Line Impact Printer Technologies As shown in Table 89.7, there are three distinct technology approaches for line impact printers: (1) fully formed character, (2) dot band matrix, and (3) shuttle hammer matrix. The first method dominates in high-end computer applications. The other two approaches are primarily used with mid-range computer systems. Fully Formed Character Printing.Various line printer technologies have evolved over the past but the most popular extant technology makes use of a band of etched engraved characters (plus timing and position marks) in a continuously moving loop configuration (see Fig. 89.33). This band is positioned between a page-wide bank (typically 132) of hammers-actuators and the paper/ribbon set. With timing and position marks, the location of the characters (replicated several times on the band loop) can be easily tracked and the appropriate hammers can be asynchronously fired to provide a full line of print in tens of milliseconds. A limit to the throughput P r (lines per minute [lpm]) of such a printer may be approximated by: (89.10) where V b is the band velocity (in./min), T i the time to increment the paper to the next print line position (min), P c the character pitch (in.), and N c the number of characters in the character string. This relation assumes that the hammer settle-out time is less than the paper advance time, the latter normally not the limiting speed factor. The usual limiting speed factor is related to print quality, i.e., well-registered, sharp-edged, high-optical- density characters. In addition, because not all lines on a page are usually printed, with line skipping the actual throughput may be significantly faster than indicated above. Several factors are important to achieve high-quality print characters. First, the hammer-actuator must provide enough impact force and energy to produce optically dense print. Second, it must impact the ribbon/paper set at Figure 89.33Line impact printer mechanism. (Source: R.C. Durbeck and S. Sherr, Eds., Output Hardcopy Devices, San Diego, Calif.: Academic Press, 1988, p. 130. With permission.) P V VTPN r b bi cc = + ? 2000 by CRC Press LLC just the right instant to provide good character registration. Third, it must have a short impact time. The impact force and energy are determined by the kinetic energy transferred by the hammer; this is the main factor in determining the amount of ink transfer to the paper. The hammer energy is typically 4–8 mJ. The total time from electronic print impulse to impact with the ribbon/paper set must be closely controlled to achieve good character registration, increasing in importance with band loop speed. The flight time variance must not be more than 1.7 ms to yield a print registration error of no greater than 0.05 mm with a loop speed of 30 m/s. This can best be done by automatically and periodically measuring the time to impact (using a piezoelectric impact bar) and then making microcode/electronic impulse delay adjustments. To minimize “slur” (i.e., the blurring of a character caused by lateral relative motion between the engraved character and the ribbon/paper set during impact), short impact time is important. It may be shown that this impact time is inversely proportional to the square root of the hammer mass (for given hammer kinetic energy), and inversely propor- tional to the hammer velocity (for given ribbon/paper set thickness and compliance). For a given amount of acceptable slur, the hammer velocity must increase proportional to the band loop velocity, and the hammer mass (for fixed kinetic energy) must decrease—inversely proportional to the square of the band loop velocity. Hammer-actuator systems have been designed with one, two and even three moving piece components. With the one-piece design, the mass impacting the paper is largest, limiting this design to slower printers (￡650 lpm). To have less mass in the hammer impacting the ribbon/paper set, separate parts are used for the armature and the impacting hammer (two-piece design). In this case, most of the kinetic energy from the pivoting armature is transferred to the hammer without the large mass handicap of the ferromagnetic armature. Of course, the residual energy in the arma- ture must be absorbed and dissipated within the print- head. Further design improvements are possible by interposing a push rod between the the armature and the hammer pieces (see Fig. 89.34). Since the lengths of adjacent push rods can be made alternately short and long, close packing of these actuator assemblies having a single pivot axis is possible. With the most efficient designs, printing speeds of over 5000 lpm are possible. Both moving-coil and stored-energy actuators are used, the latter using a “bucking” coil and a stored-energy flexible spring. The bucking coil, when energized, cancels the magnetic flux from a permanent magnet, holding the actuator in the “cocked” position, and thus converts the stored spring energy to kinetic energy. Dot Band Matrix Printing.This technology has been employed on some printers used for mid-range com- puter systems, and, in a sense, is a hybrid technology combining attributes of high-end line band printers with features found in serial wire matrix printers (see below). Present designs also use a moving metal band but incorporate pixel-size raised bumps instead of etched characters. These bumps (typically 120 on a band) are positioned at the apex of chevron-shaped springs etched into the band (see Fig. 89.35). Also, timing/position slots are etched into the band above the spring slots. Instead of one full line of characters being printed in one cycle, one full line of dots is printed with this technology. Also, each hammer covers typically three character positions across the page; thus, 45 hammers can cover a page-width of 135 character positions. Hammer cycle time is typically 1.2 ms and the band speed can be 0.28 m/s or higher. After a line of dots is printed, the paper is advanced N times until a full character height is achieved. This technology is much less expensive than typical engraved band technology, and allows the use of a faster draft mode using fewer dots per character; however, normal throughput (lpm) is generally much less. Shuttle Hammer Matrix Printing. Transverse shuttling a reduced number of horizontally spaced hammers to cover a full print line is another way to implement matrix line printing. Here, a pixel-size raised bump is Figure 89.34 Three-piece hammer-actuator design. (Source: R.C. Durbeck and S. Sherr, Eds., Output Hardcop Devices, San Diego, Calif.: Academic Press, 1988, p. 143. With permission.) ? 2000 by CRC Press LLC incorporated into the strike surface of each hammer. One design includes 33 hammers (each spaced 1 cm from its neighbor) covering 132 character positions. With an oscillatory shuttle motion of 1 cm, all line dot positions are covered, and the shuttle return time can be used for paper advance. Moderate printing speeds (600–900 lpm) can be accomplished with 1 / 4 -cm hammer spacing; 300 lpm is possible with 1-cm spacing. Opposite moving counterweights have been employed to reduce shuttle vibration. Serial Impact Printing There are two classes of technologies presently employed for serial impact printing: fully formed character and wire matrix. The former category includes both daisywheel and typeball technologies, the latter now in little use for computer output except for a few dedicated word processing applications. Daisywheel technology is used in many dedicated word processing systems and for some personal computer systems. With these two technologies, font change can be implemented by simply changing a wheel or ball, but all-points-addressable printing for image and graphics is basically not possible. Serial wire matrix technology has for years been the most popular output device for personal computers, but has now lost most of its market share to low-end laser printers and thermal ink-jet. Daisywheel Printing. This technology makes use of a “petal-like” rotating print wheel with typically 96 spring- like fingers or “petals” radiating outward from a central core. An engraved raised character is present at the end of each petal. The wheel is rotated by a dc servomotor to the desired character. A single hammer-actuator then impacts the back side of the petal, forcing the raised character into the ribbon/paper set. The simplest designs have a constant carriage speed and a fixed time delay to allow for up to 180° rotation of the wheel between strokes. These systems are typically limited to about 30 characters per second (cps). More modern designs have incorporated microcode logic which slow down the carriage motor when the next character position is more than, for example, 30° away. This can increase the net average speed to as high as 60 cps. Variants on this design use thimble- and cup-like rotary devices with petals bent parallel to the axis of rotation. This lowers the rotational inertia and allows for more petals (and hence more characters—up to 128). Typeball Printing. Typeball technology was first developed for typewriters in the 1950s, but has also been used for low-speed, correspondence-quality printers. The golf ball-size sphere, rotated and tilted to reach the Figure 89.35 Dot band printer technology. (Source: R.C. Durbeck and S. Sherr, Eds., Output Hardcopy Devices, San Diego, Calif.: Academic Press, 1988, p. 198. With permission.) ? 2000 by CRC Press LLC desired character position, has four rows of 11 characters, repeated on both hemispheres, making a total of 88 character positions. The usual operating speed is 10 cps but up to 15 cps has been achieved. Serial Wire Matrix Printing.This technology employs an array of guided wires (often tungsten) that are individually driven into the ribbon/paper set (see Fig. 89.36). The array may form a single-row configuration at the plane of ribbon impact, may have two in-line rows, or may have two staggered rows. Nine wire designs became very popular in the 1980s for low-cost personal computer systems, but the demand for higher print quality has moved the “standard” design point to 24 wires. Early technology limited wire cycle repetition rates to about 500 Hz. More advanced designs can perform at 2500 Hz, but most available products operate in the 1000–1500 Hz range. Actuators to drive these wires are quite robust and can print through many layers (up to 6) of a multipart form. The kinetic energy needed to print through 4–6 layers is normally about 0.5 mJ for 200-mm diameter wires, 1.0 mJ for 300-mm wires. For 2-layer forms, these energies drop to about 0.3 mJ and 0.45 mJ, respectively. There are two common approaches to driving the guided wires. One is a pivot-type actuator and the other is a direct solenoid design (see Fig. 89.37). The magnetic circuit in all cases must be designed to maximize either (1) the moving wire kinetic energy in the case of a free flight wire configuration or (2) the kinetic energy of the combined armature and moving wire in the case where these two components are either permanently connected or remain in contact during the drive cycle. Other actuator designs use a stored energy approach where a preloaded leaf spring is held by a permanent magnet until released by a “bucking coil” to counteract the static flux. Also, experimental stacked piezoelectric transducers have been investigated using high lever ratios (e.g., 30:1) which have operated at 3 kHz and above. Synthetic ruby or ceramic wire guide holes are sometimes used to combat wear. Several hundred million cycles lifetime operation for each wire is typically required for today’s devices. Figure 89.36Wire matrix serial printhead. (Source: R.C. Durbeck and S. Sherr, Eds., Output Hardcopy Devices, San Diego, Calif.: Academic Press, 1988, p. 187. With permission.) Figure 89.37Solenoid and pivot-type actuators for serial wire matrix printing. ? 2000 by CRC Press LLC Higher draft-quality speed (e.g., 200 cps) can be achieved without decreasing wire cycle time by simply increasing the carriage speed. Higher than normal print quality can be accomplished by both slowing down the carriage and interlacing horizontal rows of dots. This near letter-quality printing is usually at greatly reduced speed (e.g., 48 cps). Most printers also print during the carriage return. Color can be produced by shifting four-color ribbons. Defining Terms Line printing: A printer prints one full line width of characters or dots at a time. The paper is then moved into the next print line position, ready for the next line of characters or dots. The printer may pause or stop between lines. Printing speed is often given in units of lines per minute (lpm). Page printing: The information to be printed on a page is electronically composed and stored before shipping to the printer. The printer then prints the full page nonstop. Printing speed is usually given in units of pages per minute (ppm). Pixel: The nominal printed spot area or “picture element” addressed by a particular printing device. It is sometimes called “pel.” Serial printing: Printing is done one character at a time. The print head must move across the entire page to print a line of characters. The printer may pause or stop between characters. Printing speed is usually given in units of characters per second (cps). Related Topic 89.1 Input Devices References R.C. Durbeck and S. Sherr, Eds., Output Hard Copy Devices, San Diego, Calif.: Academic Press, 1988. J. Heinzel and C.H. Hertz, Advances in Electronics and Electron Physics, vol. 65, P.W. Hawkes, Ed., San Diego, Calif.: Academic Press, 1985, pp. 91–285. L.B. Schein, Electrophotography and Development Physics, Berlin: Springer-Verlag, 1988. J.M. Sturge et al., Eds., Imaging Processes and Materials, New York: Van Nostrand Reinhold, 1989. D. Winkelmann et al., Ullmann’s Encyclopedia of Industrial Chemistry, vol. A13, Weinheim, Germany: VCH Publishers, 1989, pp. 571–660. Further Information Proceedings of the SPSE and IS&T International Congresses on Nonimpact Printing, published by the Society for Imaging Sciences & Technology, 7003 Kilworth Lane, Springfield, VA 22151. Proceedings of the Society for Information Display, 1980–1996, published by The Society for Information Display, 1526 Brookhollow Drive, Suite 82, Santa Ana, CA 92705–5421. 89.3 Smart Cards Witold Suryn and Michel Veillette The smallest laptop in the world. That is the frequently used expression describing intelligent chip cards. The word “intelligent” plays an important role, as there are also other, very popular chip cards that can hardly be named that way — memory cards containing defined amounts of available memory and some access control logic. One can find them today as the most consumed commodity articles: prepaid services cards. A good example would be a telephone card, preloaded with some electronic “money” when manufactured, debited each time one makes a phone call and, finally, thrown out when emptied. Even if interesting as technology solution this type of card won’t be a subject of next chapters as this article is rather dedicated to intelligent members of chip card’s family — Smart Cards. ? 2000 by CRC Press LLC The history of this device is long with plenty of defeats. Invented in 1974 by Roland Moreno in France (there were also earlier announcements: in 1968 in Germany and 1970 in Japan), it was refused any right of recognition as a “usable invention.” These years the world was dominated successfully by a still-new fashion — plastic cards with a magnetic stripe. The very important questions are: did such a status quo change? Where? And why? Depending on the region of the world, smart cards replaced magnetic-stripe cards more or less successfully. In Europe, you could hardly find a single magnetic stripe card in use, while in America, intelligent chip cards (ICC) still experience their childhood. The reason for such a worldly footprint of ICC technology is due to the market’s demand for security measures as well as for diverse functionality. Last, but not least, the costs of on- line telecommunication services should be mentioned. Magnetic stripe offers about 140 bytes of available memory, almost nonexistent security, and in most cases, requires costly on-line service while ICC may be used in both on- and off-line environments, grants a very high, sophisticated level of secure operation and, finally, allows multifunctionality. Could it be explained more vividly? Yes, if you make comparison between a piece of magnetic tape and a computer. Because a smart card is a computer. In following chapters, we will attempt to present to the reader a complete overview of the world of smart cards. Beginning with internal hardware architecture through card software applications and ending by complete system solutions, we hope to prove the exceptionality of ICC technology both from technical and applicability points of view. Hardware Architecture Year 1998 state of the art recognizes two main hardware architectures of the ICC: 8-bit and 32-bit microcon- trollers. Why not 16-bit? Because silicon chip technology goes faster than “bits-and-bytes” maturity fights of smart card software developers. 16-bit did not catch its historical moment, however, from time to time there are successful attempts to re-use this technology. This statement may sound strange for nowadays’ software programmers used to easy “consumption” of megabytes of memory but for those who began their developer careers in early 1980s, the mastership of putting every required functionality in magic “64k” number of bytes must sound familiar. 8-bit Microcontroller Based ICC The architecture presented in Fig. 89.39 is one of the most obvious schemas known in computer science. Things become less obvious when put in ICC context. What are the roles of blocks forming ICC when compared to “full-blooded” computer? ? Microprocessor is a typical CISC (Complete Instruction Set Computer) 8-bit unit driven by an external clock of 3.57 MHz frequency you might find in early standalone systems. ? I/O — typical input/output serial communication device. ? RAM — the same memory you find today in any type of a machine. FIGURE 89.38 Pocket computer — Smart Card (courtesy of Gemplus Corporation). ? 2000 by CRC Press LLC ? ROM — system kernel in ICC, BIOS in PC (observe the difference). ? EEPROM — ICC’s hard disk. It may happen that some readers now feel a bit surprised, but the surprise will become bigger when we look at “numbers.” By default, 8-bit processor maximum memory space is 64 kilobytes (2 16 ), but in most manufactured ICCs that space is not fully used. Why so, if even 64k seems to be today less than nothing? The answer is simple: price. The device being sold yearly in hundreds of millions must be cheap, but memory forms a considerable part of the overall cost of the chip. So, as much software as possible is needed in as little memory as a developer can possibly accept. Let’s then come back to numbers: ROM — average 8k, RAM — 256 to 1024 bytes (no, we did not forget “kilo”), EEPROM — 4 to 32k (we can imagine the ironical smile of the reader — 32 kb of hard disk). And that’s the whole wealth that the developer has for his needs. Now, dear reader you can understand why developers fight for every bit. Sometimes even for a half of it. And why the ICC software world is so specific, closed and exacting. The last important feature of chips being used is the voltage of power supply (Vcc). The most commonly used voltages are 5V and 3V, however, lower (1V, 0.5V) voltages are being strongly considered. Let’s have a quick look at the industry. Most of manufactured ICCs are based on one of two industrial patterns: Intel’s I8051 microcontroller and Motorola’s 6805 one. These patterns were broadly accommodated by the world’s biggest hardware manufacturers like Siemens, Hitachi, SGS Thomson, Texas Instruments and of course — Motorola. These five names are the leaders of the smart card industry. 32-bit Microcontroller Based ICC Thirty-two bit microcontrollers are the discovery of the past year. Still new and fresh, they attract card manufacturers’ attention by its potential processing power and much larger memory space. FIGURE 89.39 The general hardware architecture of 8-bit ICC, where: Vcc — +5V power supply line Gnd — 0V (ground) line I/O — bi-directional communication line Clk — external clock line RST — external reset signal line RAM — random access memory ROM — read only memory EEPROM — electrically erasable programmable ROM ? 2000 by CRC Press LLC The internal structure of the chip is very similar to this shown in the Fig. 89.39, however, some differences may be observed: ? Microprocessor is 32-bit RISC (Reduced Instruction Set Computer) type. ? EEPROM may be replaced by faster and easier to re-program Flash ROM. ? Two I/O lines. ? In newest releases, ordinary RAM is being replaced by FeRAM (20 times quicker if used in 8-bit chip, several hundred times in 32-bit chip). ? Clock rate is 33MHz (for Vcc = 5V). ? Memory is split into Banks (pages). Now again numbers: ROM — average 56kb, RAM — 2kb, EEPROM — up to 64kb. The first impression we might have is that such a machine should give the output several tens of times better than its 8-bit sister. Surprise! The recent benchmarking based on Java Card applications (Java Card will be discussed later) have shown that the performances of both cards may be comparable, but not the price, as the 32-bit card is several times more expensive. So why go in this direction? There are many reasons among which some are crucial. Although the 8-bit solution has almost reached its top performance, that does not mean that it’ll vanish overnight. As in the world of cars, there are fans of Rolls Royce, Cadillac, and Geo Metro which exactly reflect the future of 8-bit cards. Many existing and incoming applications will still be satisfied by an 8-bit solution, offering to card issuers and clients something fulfilling their needs and at the same time remaining affordable. The second crucial reason is 32-bit technology maturity being on its learning curve. In certain, software developers as well as chip manufacturers will reach the level of expertise high enough to push prices from the “Cadillac” level to the “Geo Metro” level. The 8-bit card shall survive even if in the “bicycle” position. The market continuously creates a third important factor: multi- application requirement. Today most of us have his/her wallet filled up with dozens of credit and/or debit cards, ID cards, healthcare cards, loyalty program cards, etc. Card issuers try to attract their customers by a vision of merged functionalities installed on one card, requiring a need for more sophisticated technology. The perception today of such a solution directs issuers’ as well as card manufacturers’ interest toward 32-bit technology. There is still a future for now since there are few 32-bit chip manufacturers known: NEC, Philips, Siemens, and Hitachi, however, the rest of world’s biggest players are close to launching their modern products. Contact ICC, Contactless ICC The distinction of contact and contactless cards brings us on the next level of smart card technology. “Contact” or “contactless” define the method in which the card communicates with the external world. There are cards combining both methods sometimes called “combi” cards. Contact ICC (Fig. 89.38) is equipped with a small field of ohmic contacts to which all necessary signals are being applied. To operate such a card, a reader would have to possess the ability of physical touch. Contacless ICC uses “untouchable” media such as microwave transmission, optical transmission, capacitive coupling and inductive coupling, but — due to physical constraints — the latter one is most commonly accommodated. Before we proceed further, two facts should be strongly expressed: there is no power supply on board the card and the transmission is unidirectional (terminal to card). In such a case, four problems arise: ? How to power up the card. ? How to transmit clock. ? How to transmit data to the card. ? How to transmit data from the card. From a purely electrical point of view, the solution seems to be simple: both card and terminal are equipped with coupling loops (electrical coils) allowing transmission of energy similar to a coreless transformer (Fig. 89.40). The card’s loop is embedded in a plastic body to avoid getting destroyed or touched. The main carrier is a sinusoidal magnetic field (f = 4.9152MHz) that transmits the power supply to the card. To transmit the data/clock to the card, the terminal uses frequency modulation of the main carrier, being easily ? 2000 by CRC Press LLC filtered and decoded by the card’s I/O device. Transmission of the data from the card to the terminal is based on the passive amplitude modulation of the main carrier, which is executed by changing the card’s I/O circuit impedance differently for “0” and “1.” To read the information, a reader must be able to detect a potential drop in the impedance in its internal AC generator. Inside the contactless type of cards there are two sub-groups recognized: close coupling cards and remote coupling cards. The distinction between the two is based on the distance terminal-card and the power that may be transmitted to the card. Regardless of which communication carrier is used, the transmission speed between the card and the terminal is, in most cases, one of the following: 9600 bit/s, 19,200 bit/s, or 115 kbit/s. Operating Systems Operating system (OS) is the next layer of technology that makes the microprocessor card a smart card. The term operating system applied to a smart card is in some sense misleading as in many cases even highly experienced programmer/analysts would not be able to find a clear borderline between “system” and “applica- tive” layers. Due to continuous memory constraints smart card software designers chose the formula of “nested functionalities” that finally resulted in placing real functionality inside OS services (or the opposite, depending on your point of view). Each card manufacturer has its own family of operating systems offering different dedicated functionalities to the user that are characteristic of the specific manufacturer. That type of OS forms a vast group of software products called proprietary OS. In the mid-1990s, another approach to smart card OS emerged: open OS. By its definition the open OS is quite similar to familiar DOS with the exception of being placed in the card’s ROM instead of the computer’s hard disk. For either of them some principles apply: ? The stable part of software application is placed in ROM during the masking (chip manufacturing) process. For proprietary OS, it makes all the functions regardless of whether “system” or “applicative” form the basic dedicated platform. For open OS, in most cases ROM keeps, at a minimum, basic OS. ? The changeable part of software application is always loaded to EEPROM. In both cases, EEPROM keeps “add-ins” stored in ROM. For proprietary OS, this is the “filter” consisting of either additional func- tionalities or corrective patches. For open OS, real functional software is stored. EEPROM contents may be loaded to the card during personalization process (when a real, customer-dedicated card is manu- factured) or in any moment of a card’s life cycle. ? RAM is used to store only temporary data during transaction time. No application code is allowed to be loaded to RAM for a card may be withdrawn from the terminal at any moment causing a permanent loss of important code. ? Communication with the external world is processed by means of specific interpreted code structures called APDU (Application Protocol Data Unit) and data exchange (for further details see [1–3]). ? The way in which a card performs required functions is based on strictly-defined rules concerning memory file structure, data structure and format, file management, data management, access rights (read/write/exe- cute control), execution control, and atomic (backup) methods (for further details see [1–3]). FIGURE 89.40 Card-terminal communication principle. ? 2000 by CRC Press LLC ? Of the pair, the terminal-card card is always a “slave.” Such a situation results from the simple fact that it’s the terminal that “wakes up” a card by applying power and reset signals to the card. In response, the card sends back ATR (Answer To Reset) APDU which identifies itself to a terminal. From this moment on the real session starts. The session may be terminated either by a card or by a terminal. In the following two sections, we will try to present both types of OS and discuss their pros and cons. Proprietary Operating Systems Proprietary OS is always targeted to fulfill a defined scope of ded- icated functionalities. The general, however specific, structure of the proprietary OS is shown on the Fig. 89.41. Such a structure may give the impression that proprietary OS is created in a “fuzzy” manner. This is incorrect. All the memory is perfectly structured, with areas storing data, application code, and basic OS functions as well as native functions (procedures perform- ing most basic functions as I/O protocol, DES algorithm, basic math, atomic functions, etc.). What are not that well-structured are functionalities performed by the card. Let’s take the example of an electronic purse. From the user’s point of view, the card should perform at minimum the following functions: ? User authenticate (user proves to the card that he/she is the owner of the card). ? Credit the card within pre-defined limit. ? Debit the card within pre-defined limits. ? Disable any attempt to break into the contents of the card. In “nested” OS, none of the above functions are standalone. All of them are, in great part or even totally, performed by OS. In most cases, the reader may find in an application area “script(s)” per- forming needed function(s) by calling necessary subroutines within OS, but rather rarely a standalone block of software per- forming an independently new function. What are the pros and cons of such a structure? Let’s investigate: PROS ? High “density” of the software or, in other words, good functionality/memory consumption ratio, ? Good functionality/price ratio, ? Good functionality/performance (speed) ratio, ? Good way to build a stable relationship between the card issuer and card manufacturer (watch the interpretation of this point in CONS). CONS ? Applications are not portable, sometimes even within the same family of operating systems. ? Debugging process requires tremendous efforts as, in the worst case, the entire card software has to be verified. ? There are only two ways to correct found bugs: add the “filter” in EEPROM (not always possible, but always temporary) or remask the chip. Specifically, the second possibility in some cases may be called the disaster as when the bug is found in the card that has been deployed into the market in 100 million volume. ? Good relationship created by a chain Card Manufacturer → Functionality → Card Issuer is more and more often perceived as a master-slave relationship. Card issuers prefer to be supplier independent. The conclusion drawn is somewhat obvious: proprietary OS is a handicapped solution, but there are hundreds of millions of such cards issued each year that imply the message: more flexible technologies are still to come. FIGURE 89.41 Proprietary OS structure. ? 2000 by CRC Press LLC Open Operating Systems Smart Card open operating systems (OOS) are being developed continuously since the early 1990s reaching the position today of emerging technology. The basic assumptions for OOS development are: ? Portability of applications. ? Application load/delete function. ? Multi-application ability. ? Upward/downward compatibility of OOS within the family. ? Development environment similar to or being a subset of “full-blooded” computer software development environment. ? High-level programming language. ? High level of security, both inside the application and outside (firewalls). ? Shorter development time (time to market). The general structure of OOS smart card is presented on Fig. 89.42. How much clearer is such a structure when compared to proprietary OS? The OOS forms a well-defined group of layers over which the real applications reside. Bottom-up direction shows higher and higher levels of abstraction of consecutive layers: ? Native functions and system resources — all the basic functions that are time critical (I/O subroutines, basic cryptographic functions, atomicity, basic file management) developed in assembly language. ? Virtual Machine — on-card interpreter of bytecodes. ? API — application programming interface — set of generic libraries. ? Specific API — nonmandatory API dedicated to specific card functionalities. ? Card Executive — file and application manager (loading/removing applications). ? Applications — developed in high-level languages, stored on the card in bytecode format. Applications developed for such a type of the card are commonly called applets or, better, cardlets. Usually, the cardlet is developed on the PC using one of several higher-level languages in a not necessarily dedicated development environment. Dedicated environment comes on the stage in the moment of the first compilation FIGURE 89.42 General structure of OOP Smart Card, where: Appl1,2,..n — card applications, API — application programming interface ? 2000 by CRC Press LLC and debugging. What is important to observe as the difference between PC-dedicated compilation and card- dedicated compilation is the result: for PC you may obtain the executable file (*.exe), for card it’s always bytecode (*.cap) as the card interprets application code by code. Debugging of the cardlet may be done in a simulated, PC-based environment as well as in a natural card environment. Ideally, during card life cycle, cardlets may be downloaded, added, or removed changing the profile of the card. All three possibilities generate immediate problems of a different nature: ? Co-existence of two or more cardlets requires high security measures to firewall cardlets against cross- corruption in any form as well as raises the needs of re-usability of common objects. ? Any type of manipulation of the cardlets creates responsibility/legal issues to be solved in the near future such as Who’s got the right to manipulate cardlets? Card issuer (bank?), card manufacturer, or card user? How to maintain cardlets and their life cycle, etc. There are more issues of both a technical and legal nature, to be answered, so those interested in smart card technology may get closer to it by observing the works of international standard bodies (e.g., ISO), industry associations (e.g., SCIA) or international forums (best known is Smart Card Forum). PROS ? Any developer knowing higher-level programming languages may develop an application. No “bits-and- bytes” fighter knowledge necessary. ? Development environment is PC-like or purely PC. ? Debugging may be off-card as well as on-card. ? Development time is incomparably shorter than this for proprietary OS. ? As applications are loaded by default to EEPROM, any in-the-field bug may be fixed by simply reloading the cardlet. ? Time to market dramatically shorter. ? Portability of applications. ? The customer is supplier independent. CONS ? Technology still not mastered, both in hardware and software domains. ? Development/debugging tools in a childhood phase. ? Performance still low (due to interpretation mode of operation and slow hardware). ? Due to evolution phase of standards, portability of cardlets merely applicable. ? High cost of cards. The conclusion might be: first steps are always the heaviest, but to explode, OOS technology needs more maturity rather than technical improvements. Next five years are forecasted as the real startup for OOS smart cards. Standards Smart card is one of the best standard-guarded pieces of equipment. The set of ISO 7816 regulations defines all the aspects of smart card technology and life cycle. Most of them are stable, however, due to continuous software and hardware technological improvements all of them are being amended. The complete list of applicable standards is presented below: ? ISO 7816-1 — Identification cards — Integrated circuit(s) cards with contacts — Part 1: Physical characteristics. Defines physical characteristics of cards as well as tests to be applied. ? ISO 7816-2 — Identification cards — Integrated circuit(s) cards with contacts — Part 2: Dimensions and location of the contacts. Defines dimensions and locations of contacts and magnetic stripe as well as tests to be applied. ? ISO 7816-3 — Identification cards — Integrated circuit(s) cards with contacts — Part 3: Electronic signals and transmission protocols. Defines basic electronic characteristics of ICCs, power supply, data elements in ATR and transmission protocol T = 0. ? 2000 by CRC Press LLC ? ISO 7816-4 — Identification cards — Integrated circuit(s) cards with contacts — Part 4: Interindustry command for interchange. Defines internal software structure of ICCs, security requirements, formats of APDUs, communication codes. ? ISO 7816-5 — Identification cards — Integrated circuit(s) cards with contacts — Part 5: Numbering system and registration procedure for application identifiers. Defines system for clear identification of national and international applications in ICC. ? ISO 7816-6 — Identification cards — Integrated circuit(s) cards with contacts — Part 6: Interindustry data elements. Defines data elements and structures used in retrieval of data from ICC. ? ISO 7816-7 — Identification cards — Integrated circuit(s) cards with contacts — Part 7: Enhanced interindustry commands. Defines additional commands added to Part 4 for SQL access, encryption, secure messaging. ? ISO 7816-8 — Identification cards — Integrated circuit(s) cards with contacts — Part 8: Interindustry security architecture. Defines security architecture of ICCs. ? ISO 7816-9 — Identification cards — Integrated circuit(s) cards with contacts — Part 9: Additional interindustry commands and security attributes. Describes coding of the life cycle of the cards and related objects, coding of security attributes of card related objects; functions and syntax of interindustry commands not defined in other parts of ISO/IEC 7816; data elements associated with these commands; and a mechanism for initiating card-initiated messages. ? ISO 7816-10 — Identification cards — Integrated circuit(s) cards with contacts — Part 10: Electronic signals and Answer To Reset for synchronous cards. Defines basic electronic characteristics, power supply, data elements in ATR for synchronous cards. ? ISO 7816-11 — Identification cards — Integrated circuit(s) cards with contacts — Part 11: Card structure and enhanced functions for multi-application use. Defines an enhanced logical structure for multi-appli- cation cards to support security with applications loaded after issuance, prerequisites for loading of an application after issuance, a mechanism to support an application life cycle independent of the card life cycle, interapplication security issues, including storage partitioning. For more details you may refer either directly to International Standards Organization or to ISO Web site: www.iso.org. Applications The purpose of this chapter is mostly descriptive, aiming to show the wealth and diversity of existing and upcoming applications deployed all over the world. The smart card industry is formed by a few big manufacturers covering practically all of the world’s needs (Gemplus, Schlumberger, Motorola, Orga, De La Rue, Gisecke&Devrient, and Oberthur). Each of them offers either applications, platforms, or both. The short overview of possible applications will be more persuasive if done by market sectors: ? Banking: credit/debit card, electronic purse, loyalty programs, individual financial programs. ? Retail: loyalty programs, retailers’ credit cards. ? Healthcare: personal medical treatment databases (medical files). ? Wireless communication: SIM (Subscriber Identity Module) card, OTA (Over The Air) services, e-mail services. ? IT (Information Technology): security access to IT systems, encrypting/decrypting files, securing e-mail, Internet transactions (e-wallet, e-commerce). ? Universities: multi-functional ID cards (credit, electronic purse). ? Government and agencies: ID cards, internal electronic purse, time control. ? Physical access: ID cards, zone access control. ? Manufacturing: electronic tags. ? Transportation: ticketing, air-miles programs, driving licenses. Most of the above-mentioned applications are based on a proprietary OS, however, OOS early birds are also observed. The first steps on the way to commercial OOS have been recently done by some of the industry ? 2000 by CRC Press LLC leaders. The joint efforts of Sun Microsystems, Gemplus, and VISA-created Java-based Visa Open Platform (VOP), Schlumberger has created its Java-based Open Platform — Cyberflex, Gemplus has its Java-based GemXpresso and its own OOS — Nucleos, and last but not least, MULTOS — OOS created by consortium MAOSCO/Mondex. Though still not matured, all of these OOSs have already provoked the chain reaction: creating cardlets’ development industry. Readers A smart card is of no use if you cannot communicate with it and the reader is the element that will give life to this hand-held device. The market offers several kinds of readers. Readers are categorized in three classes: standalone, peripheral, and contactless. Standalone Readers Standalone readers are modules that offer a complete interface to operate and communicate with the smart card. That kind of reader can be a part of a complex machine such as an automatic teller machine or presented as a hand-held device. For an application such as an automated bank teller, the smart card reader is then a programmable device integrated into a larger system with several peripherals such as display, keyboard, real- time clock, and printers. In many cases, the complete system also includes several security mechanisms from protected housings to sensors for detecting attacks and security modules. A security module is a smart card that holds the terminal secret keys, which are used to create card-specific keys. A standalone reader usually has a communication link to operate as a peripheral of a larger transaction server that will verify the user identi- fication and take charge of the information transmission, reconciliation of transactions, etc. Other applications require a reader that will participate in a secure information exchange transaction between participants. Thus, desktop systems are smaller and have their use in business application to provide secured services. Along with security modules, some readers will accept several cards to authenticate, for instance, the owner and the receiver of a confidential information. In healthcare application, the patient will authorize access to his files while the doctor will provide the proof that he has the right to access this kind of information. Several security modules will ensure security and protection for different applications executed through the same reader, treated separately one from another. Other suitable applications for those readers are payment transactions, electronic purses, and loyalty programs. The integration of the keyboard and the display to the security of the process ensures that the user’s PIN and exchanged information are not tapped by another software. Hand-held systems offer the flexibility of mobile systems. Integrated with telecommunication devices and hand-held computer technology, these systems can present touchscreen interfaces and provide facilities such as gaming, fax, e-mail, telephone, e-purse, loyalty, healthcare or driver’s license programs, security, authenti- cation, identification, electronic payment, secure access to networks, secure data transfer over those networks, etc. GSM is gaining more popularity by using the cell phone as a mobile reader to personalize your phone and provide the user with facilities such as e-mail and prepaid services. Peripheral Readers A peripheral reader provides a communication interface between a larger system and a smart card. Using standard interfaces, the reader is usually coupled to a control system that will send commands to the reader. A peripheral reader can be integrated into a larger peripheral, such as payphones, a vending machine, or a card personalization device. Peripheral readers are often attached to a computer to add functionality for card-aware applications. The market offers several types of connections for peripheral devices. Serial connections are suitable for most desktop computers and are standard and affordable. PC Card connections (pcmcia) are mainly for laptop computers. Although slow, keyboard readers, using the PS/2 connection, provide better security, mainly for the PIN management. When entering a pin, the reader intercepts the characters from the keyboard and replaces them with a neutral character that is sent to the system. This way no resident software can intercept the PIN. A keyboard reader can be integrated to the keyboard itself or added to the PS2 connector of the keyboard. USB readers are now coming into the market and will compensate for any deficiency of the serial reader. Some ? 2000 by CRC Press LLC attempts are also done to integrate a reader into a mouse or floppy disk. This last solution offers the user the possibility to use a smart card without adding new hardware to his machine. Finally, one can find specialized peripheral readers combining the functionality of other device types. For instance, a smart card reader and modem were combined to secure network access in Internet communication, electronic commerce, and home banking. Also, a smart card reader and a television signal processing module were combined for conditional access, electronic program guides, interactive applications, home shopping, and payment systems. The use of peripheral readers is an emerging market with the recent deployment of application for the electronic commerce, home banking, online gaming, health care, network security on the Internet, and tele- communications. Readers with biometrics interfaces have begun their introduction on the markets. Fingerprints are used instead of a PIN to authenticate the user. Contactless Reader A contactless reader (also called proximity reader) needs no physical contact between the card and the reader to exchange data. The card goes immediately into action when it is within a few inches of the reader. Trans- mission details were described earlier. Using a hard-wired algorithm and random number generation, it is possible to provide high-level security. The reader and the card use a mutual authentication and a stream ciphering. Such system must also provide anticollision management to operate with multiple cards in the communication field. Contactless smart card readers are perfectly suited for physical access control and smart tracking applications. The use of Smart Labels (or tags) injected or embedded into a product allows the identification and tracking of product for either authentication or automated manipulation. Since data can also be written into a contactless card, payment functions can be implemented for fare collection in public transportation, for instance. When the distance between communications is a concern, systems using intelligent transponders enable the payment of tolls from equipped vehicles travelling on motorways at normal speed. Card-to-System Solutions Naturally, communication with a smart card would be impossible without several software layers that transform the request of the user into a sequence of bits transmitted to the device. At the lowest level, the drivers provide the basic functionality by transferring a sequence of bytes to the reader. Then layers of software are built over the driver to form application program interfaces (APIs). API offers to the application developer functions that combine basic operations to hide specificity of a particular device and to ease the development. Drivers To communicate with a reader, the basic transmission is handled by three protocol layers: the physical layer, the transport layer, and the command layer. The physical layer is related to the data transmission itself through the physical port (serial, pcmcia, usb, etc.). The transport layer specifies the transmission type, the message addressing, and handles the message validation. It defines how each byte transmitted by the physical layer is interpreted and ensures that it forms a consistent and valid message. Finally, the command layer specifies commands that are transmitted to the reader to execute a set of functions. These functions will allow the communication with a smart card and the management of events such as insertion and removal of the card. In the communication process, the driver is a piece of software that hides the treatment related to the physical and the transport layers. With documented interfaces, the user can use the command layer to send commands to the reader and to communicate with the card. Usually, the operating system integrates drivers that take charge of the physical layer. The device vendor will then write a filter driver that is placed over the other one to wrap up the transport layer, transmitting commands into well-defined packages. The driver will allow basic commands to the reader: setting parameters of the readers (communication speed, voltage levels, card type used), setting communication parameters between reader and card (protocol, speed), power on, power off, transmitting bytes to the card. Moreover, the driver must manage the events coming from the system and control the states of the device. For instance, the insertion or the removal of the reader involves some processes of initialization of the communication and registration of the connection. The driver must also allow the sharing ? 2000 by CRC Press LLC of the resource by multiple applications or threads. Security issues regarding those accesses must be considered. One must preserve access to the information on the card from the application that was not allowed. With only the driver, the development of an application requires the knowledge of the command format to send to the card and of the mapping of the smart card. To ease and speed up his development, a programmer needs tools that hide specificity of cards and readers. API and middleware are coming to his rescue. APIs and Middleware Application program interfaces (API, also called middleware) usually take the form of libraries that provide the user with high-level functions to ease the development and to hide disparities between different technologies. Basic APIs try to hide the command layer to the user and regroup under common interfaces the communication with different readers. At a higher lever, an API hides the smart card operating system particularities, providing the user with card services and with abstract concepts that ease the use of the information on the card. Thus, card services will allow the user to select a file, to read, write, and erase objects on the card, and to authenticate the user. API can also regroup specialized functions such as cryptographic functions, purse functions, etc. Usually, APIs are proposed by smart card vendors to offer an easy-to-use environment for the development of applications on their products. It offers a wide variety of commands and responses. However, it makes the developer dependent of a product and of the support of these API. Efforts were made to standardize some API. PKCS#11[5] is an example of cryptographic libraries. This standard API is defined by RSA (RSA Data Security, Inc.) and permits the use of data objects in the form of tag-length-value. Standardizing Approach With the lack of interoperability between operating systems, API, readers, and smart cards, some efforts were made to offer more compatibility and harmonization. PC/SC and the OpenCard Framework are the main stream. PC/SC PC/SC[4] is an effort to provide interoperability of the smart card technology in the personal computer environment. The aims of PC/SC is to simplify the development of application by defining high-level program- ming interfaces that reduce the dependency of application on proprietary implementation. To achieve its goals, PC/SC distinguishes three layers between the hardware device and the application level: The IFDHandler Layer, the resource manager layer, and the service provider layer. The IFDHandler allows access to the smart card reader. It is the driver level and the hardware manufacturer that will normally provide it. PC/SC defines the interfaces that the IFDHandler must expose to take place in the structure. The IFDHandler is the implementation of the functionality required by the exposed functions to transmit commands to the reader and/or the card. The IFDHandler maps the native capabilities of the smart card reader to the IFDHandler interface. All distinctions between smart cards based on ISO protocol handling, whether synchronous or asynchronous, are hidden. Once installed, the IFDhandler registers itself to the Resource Manager. The Resource Manager is a key component of the PC/SC architecture. It is responsible for managing the other resources (such as service providers) within the system and for controlling access to the smart card readers and to the card itself. The Resource Manager is provided by the operating system vendor. Thus, the Resource Manager is responsible for identification and tracking of resources. This includes the tracking of the readers, of the smart cards, of the service providers, of the supported interfaces and of the smart card insertion and removal events. Thus, the Resource Manager can list the available resources to each application. The Resource Manager is also responsible for controlling the allocation of readers and across multiple applications with shared or exclusive modes of operations. Finally, it supports transaction primitives to allow multiple commands to be executed without interruption, ensuring that intermediate state information is not corrupted. The Service Provider regroups a set of functions exposed by a specific smart card and makes it accessible through high-level programming interfaces. The common functionalities are related to file access, authentica- tion, and cryptographic services. However, these interfaces may be extended to meet the needs of specific domains. ? 2000 by CRC Press LLC PC/SC divides the Service Provider into two independent components: the Smart Card Service Provider (SCSP) and the Cryptographic Service Provider (CSP). The CSP regroups cryptographic functionality through high-level programming interfaces. This distinction was brought to deal with security issues and with interna- tional laws regarding cryptographic devices. The Answer To Reset (ATR) string is used to perform the binding between a card and an interface. Thus, when a service provider is installed on a system, it registers itself to the Resource Manager with the ATR of cards it can serve. An application can define at run-time which Service Provider will be used since the Resource Manager enumerates the list of the available interfaces when the card is inserted. Although PC/SC is platform neutral, to date, Microsoft did the only implementation of this standard on Windows platform (even now the implementation still lacks of some features: “plug and play” and power management). Card services are exposed to an application through Component Objet Model (COM) interface. PS/SC v.1.0 supports only T = 0 and T = 1 cards and optionally T = 14 (protocols of communication between a card and a reader. For details see [1–3]). Next release of this standard will consider memory card and multiple interfaces readers such as keyboard, display, multiple cards, and SAM modules. OpenCard Framework OpenCard Framework (OCF) [6] objective is to offer to the programmer a transparency with regard to smart card operating systems, card terminals, and card issuers. To achieve this, OCF has defined a provision of functions to support installation, removal, enumeration, selection of applications on the card and of functions to perform name resolution for data files on the card. Several mechanisms allow a user to develop an application without knowing where the card issuer placed his applications. OpenCard Framework based its architecture on two main parts: the CardTerminal layer and the CardService layer. The CardTerminal layer plays the role of the driver and the reader manufacturer should provide it. It offers an interface that enables the seamless integration of the reader in the OCF environment. With OCF-compliant interfaces, a specific card terminal gives access to the reader and the smart card. A CardTerminal can also act as a bridge to provide a transparent access to a reader through the PCSC interface to take advantage of the existing components. The CardService layer is defined as a high-level application programming interface that hides the character- istics of a particular provider’s components (specific to a type of smart card and/or smart card reader) from the application and service developers. Card services are standard APIs regrouping functions to access resources of a card operating system. Thus, a card issuer will provide the users with a card services allowing access to the files (FileAccessCardService), to the signature functions (SignatureCardService), to PKCS#11 functions, to the ISO7816 file system, and to purse functions. Between those two layers, a specific card service, the application management card service, manages the card resident applications. The application management component lists the applications that a card can support, locates them or even installs or removes them from the card. It is a special card service, also provided by the card issuer, that can manage multiple applications on the same smart card. OCF essentially was created for the Java programming environment and the world of network computing. Because of this, it claims all the advantages that the Java language provides. Basically, any platform that is capable of running JAVA can exploit OCF right away. Moreover, OCF offers mechanisms to allow one to download from the network missing components for a particular card at hand and plug them into the framework. As it is for Java, OCF targets embedded systems such as automatic teller machines, point-of-sales terminals, and hand-held devices (phone, electronic agenda). Applications Developing an application is made easier and more flexible when based over standard API and middleware products. The challenge remains, however, to respond to the new needs of the industry. Applications must integrate security schemes to avoid holes and prevent piracy of a system. To secure their system, applications developers must be aware of the complete structure of the operating system they are using. For instance, the market proposes new software to protect access to a computer and to encrypt files on computers. Such a system requires low level control of the resources of a computer to block unexpected entry. Also, sensitive information must not be exposed. Encryption keys are never kept clearly in memory to prevent software attack. ? 2000 by CRC Press LLC Applications are also confronted with the Internet challenge. Security systems are facing new offensives and new security schemes that must be put in place. The Internet not only broadcasts the information at the speed of light; it represents to hackers a tremendous power for parallel computing. Competitions are organized on the net to break encryption keys: each participant downloads the required software and adds its contribution to the computing effort. Key size is lengthened, consequently. Secure transactions over the net also bring new problems. Applications are distributed, client-server schemes become more important, the number of trans- actions are growing and, consequentially, we do not know who is on the other side. For an electronic wallet application, this is crucial because nobody wants to give his credit card number without some credentials. Third party certificates are there to authenticate your invisible partner and mechanisms are set to establish a confidence path between all involved parties of a transaction. Trends Trends in smart card industry may, but rather should not be, seen only from an ICC perspective. The reason is amazingly simple: the most powerful, super fast and multi-functional card is nothing more than a piece of plastic and silicon if there is no IT system able to use it. That’s why the trends will be discussed in two groups: card technology and complete solutions. The general challenges driving card technology trends may be defined by four simple words: faster, cheaper, more, easier. Technical solutions to come should allow multi-application (more), flexible functional profile i.e., easy changeable dedicated applications (easier), shorter, nonspecific technology-tied development cycle (faster), shorter time to market (more, cheaper), higher security level (easier). All of the above points to a generally named card open OS solution. How serious this trend is depends upon industry reactions. Hardware manufacturers design more powerful chips dedicated to operate with higher clock speeds, offering larger memory space, in some cases injecting specific commands into the processor’s command list (Siemens’s picoJava concept encapsulating specific Java- Card commands in processor design), and adding specialized coprocessors (cryptographic functions). Card manufacturers improve their OOSs to meet multi-application requirements by adding new features up to recently imminent to pure IT (multi-thread, COM, firewalls, add/load/delete), system integrators adapt their solutions to absorb specific card management systems (CMS) allowing control over the live cycle of applications (and cards), card issuers design multi-applicative programs for their customers, and finally, software develop- ment companies begin to create competitive, cutting edge competence in smart card OOS-dedicated applica- tions development. The recent entry into the industry includes the software giant — Microsoft with its Smart Card for Windows (SCW). This entry establishes the seriousness of the smart card trend. To penetrate new markets, the smart card industry must now develop value-added solutions by enhancing the functionality of the conventional card. To be attractive to customers, the same card should be used for access control, securing transactions on the net, and for payment. This trend implies the coexistence of multiple applications on the same card. It also means the customization of the user card. The Java Card and, more recently, the Smart Card for Windows are supporting this stream where developers can download their own application into the card. Defining Terms APDU: Application Protocol Data Unit is a set of bytes forming a unitary “container” used to send to or receive data from a card. There are two types of APDUs: command APDU (sent to a card) and response APDU (sent by a card). Formats of both APDUs are different, however, no mismatch is possible as the communication between a card and the reader is ruled by a sequence: command (reader)-response (card). API: Application Programming Interface. In general terms, API is a set of functionalities (library) loadable or preloaded available for a programmer developing the application. In smart card technology, APIs are found on board the card (e.g., Java Card) as well as on the system side (reader APIs). The goal of creating API is to free the developer from card-related specific knowledge allowing him/her to develop on the higher level of abstraction. ? 2000 by CRC Press LLC ATR: Answer To Reset is a set of maximum 33 bytes returned by a card to a reader after RESET signal was applied. ATR consists of information necessary to authenticate a card to the system and the reader as well as to negotiate all communication parameters as protocol, transmission speed, etc. DES algorithm: Data Encryption Standard algorithm is the most popular encryption/decryption mathemat- ical engine. DES is being broadly applied in the security domain for encryption/decryption purposes, secure messaging, and authentication. Best known is 56-bit encryption/decryption key version. For more sophisticated security applications, 3DES (triple DES) is being applied. 3DES is a method using DES three consecutive times with a different encryption key used each time. Filter: Filter, in smart card technology, defines an additional block of code loaded to the card’s EEPROM whose function is to either enhance the existing card’s functionality or patch found bugs. In the first case, you can see it as the development “pilot” that allows the manufacturer to check new releases before its industrial launch. In the second case, a “filter” is the easiest way to debug and fix existing application or OS. When checked, verified, and approved, the “filter” is moved to ROM and remasked. Masking process: The masking process is the essential process of manufacturing the chip to be later embedded onto a card. The masked chip contains all the hardware components as well as all of the software stored in ROM. The remaining EEPROM and RAM are empty and not structured. Personalizing process: The personalizing process is performed by a card manufacturer. The goal is to make a masked chip card that the customer can use. The process covers loading onto the card all specific customer info (name, primary security keys, serial number) as well as building the internal memory structure (directories, file structures, file types). PC/SC: Personal Computer/Smart Card (PC/SC) is the “de facto” or industry standard introduced by Microsoft Corporation. The standard defines structure, functionality, and communication entries for smart card readers’ drivers being installed on all existing Windows platforms (95, 98, NT4, 2000). USB: Universal Serial Bus is the newest PC technology applied to serial communication solutions. The technology allows connecting to a PC several serial devices in a chain at the same time. The communi- cation protocol between PC and external serial devices is similar to this used in networking. USB becomes more and more in demand in PC-Smart Card solutions. References 1. Allen C., Barr W.J. Smart Card seizing strategic opportunity. Irvin, 1998. 2. Jurgensen T. Smart Card developer’s kit. Macmillan, 1998. 3. Rankl W., Effing W. Smart Card Handbook. John Wiley & Sons Ltd., 1995. 4. PC/SC Workgroup. Interoperability Specifications for ICCs and Personal Computer Systems. Revision 1.0, 1997, available electronically at: http://www.smartcardsys.com. 5. RSA Laboratories. PKCS#11: Cryptographic Token Interface Standard. RSA Laboratories technical notes, version 2.01, 1997, available electronically at: http://www.rsa.com. 6. OpenCard Consortium. OpenCard Framework 1.1 Programmer’s guide. 1998, available electronically at: http://www.opencard.org. Further Information Gemplus Web site (www.gemplus.com). ISO web site (www.iso.ch). ISO-IEC JTC1-SC17 Web site (www.funkster.com/ossian). Proceedings of CardTech/SecurTech 1998, CTST 1998. Smart Card Forum Web site (www.smartcrd.com). Smart Card Industry Association Web site (www.scia.org). ? 2000 by CRC Press LLC