Garrod, S.A.R. “D/A and A/D Converters” The Electrical Engineering Handbook Ed. Richard C. Dorf Boca Raton: CRC Press LLC, 2000 32 D/A and A/D Converters 32.1 D/A and A/D Circuits D/A and A/D Converter Performance Criteria?D/A Conversion Processes?D/A Converter ICs?A/D Conversion Processes?A/D Converter ICs?Grounding and Bypassing on D/A and A/D ICs?Selection Criteria for D/A and A/D Converter ICs Digital-to-analog (D/A) conversion is the process of converting digital codes into a continuous range of analog signals. Analog-to-digital (A/D) conversion is the complementary process of converting a continuous range of analog signals into digital codes. Such conversion processes are necessary to interface real-world systems, which typically monitor continuously varying analog signals, with digital systems that process, store, interpret, and manipulate the analog values. D/A and A/D applications have evolved from predominately military-driven applications to consumer- oriented applications. Up to the mid-1980s, the military applications determined the design of many D/A and A/D devices. The military applications required very high performance coupled with hermetic packaging, radiation hardening, shock and vibration testing, and military specification and record keeping. Cost was of little concern, and “low power” applications required approximately 2.8 W. The major applications up the mid- 1980s included military radar warning and guidance systems, digital oscilloscopes, medical imaging, infrared systems, and professional video. The applications requiring D/A and A/D circuits in the 1990s have different performance criteria from those of earlier years. In particular, low power and high speed applications are driving the development of D/A and A/D circuits, as the devices are used extensively in battery-operated consumer products. The predominant applications include cellular telephones, hand-held camcorders, portable computers, and set-top cable TV boxes. These applications generally have low power and long battery life requirements, and they may have high speed and high resolution requirements, as is the case with the set-top cable TV boxes. 32.1 D/A and A/D Circuits D/A and A/D conversion circuits are available as integrated circuits (ICs) from many manufacturers. A huge array of ICs exists, consisting of not only the D/A or A/D conversion circuits, but also closely related circuits such as sample-and-hold amplifiers, analog multiplexers, voltage-to-frequency and frequency-to-voltage con- verters, voltage references, calibrators, operation amplifiers, isolation amplifiers, instrumentation amplifiers, active filters, dc-to-dc converters, analog interfaces to digital signal processing systems, and data acquisition subsystems. Data books from the IC manufacturers contain an enormous amount of information about these devices and their applications to assist the design engineer. The ICs discussed in this chapter will be strictly the D/A and A/D conversion circuits. Table 32.1 lists a small sample of the variety of the D/A and A/D converters currently available. The ICs usually perform either D/A or A/D conversion. There are serial interface ICs, however, typically for high-performance audio and digital signal processing applications, that perform both A/D and D/A processes. Susan A.R. Garrod Purdue University ? 2000 by CRC Press LLC D/A and A/D Converter Performance Criteria The major factors that determine the quality of performance of D/A and A/D converters are resolution, sampling rate, speed, and linearity. The resolution of a D/A circuit is the smallest change in the output analog signal. In an A/D system, the resolution is the smallest change in voltage that can be detected by the system and that can produce a change in the digital code. The resolution determines the total number of digital codes, or quantization levels, that will be recognized or produced by the circuit. The resolution of a D/A or A/D IC is usually specified in terms of the bits in the digital code or in terms of the least significant bit (LSB) of the system. An n-bit code allows for 2 n quantization levels, or 2 n – 1 steps between quantization levels. As the number of bits increases, the step size between quantization levels decreases, therefore increasing the accuracy of the system when a conversion is made between an analog and digital signal. The system resolution can be specified also as the voltage step size between quantization levels. For A/D circuits, the resolution is the smallest input voltage that is detected by the system. The speed of a D/A or A/D converter is determined by the time it takes to perform the conversion process. For D/A converters, the speed is specified as the settling time. For A/D converters, the speed is specified as the conversion time. The settling time for D/A converters will vary with supply voltage and transition in the digital code; thus, it is specified in the data sheet with the appropriate conditions stated. A/D converters have a maximum sampling rate that limits the speed at which they can perform continuous conversions. The sampling rate is the number of times per second that the analog signal can be sampled and converted into a digital code. For proper A/D conversion, the minimum sampling rate must be at least two times the highest frequency of the analog signal being sampled to satisfy the Nyquist sampling criterion. The conversion speed and other timing factors must be taken into consideration to determine the maximum sampling rate of an A/D converter. Nyquist A/D converters use a sampling rate that is slightly more than twice TABLE 32.1 D/A and A/D Integrated Circuits Multiplying D/A Converter ICs Resolution, bvs. Fixed ReferenceSettling Time, ms Input Data Format Analog Devices AD558 8 Fixed reference 3 Parallel Analog Devices AD7524 8 Multiplying 0.400 Parallel Analog Devices AD390 Quad, 12 Fixed reference 8 Parallel Analog Devices AD1856 16 Fixed reference 1.5 Serial Burr-Brown DAC729 18 Fixed reference 8 Parallel DATEL DAC-HF8 8 Multiplying 0.025 Parallel National DAC0800 8 Multiplying 0.1 Parallel A/D Converter ICs Resolution, bSignal Inputs Conversion Speed, ms Output Data Format Analog Devices AD572 12 1 25 Serial & parallel Burr-Brown ADC803 12 1 1.5 Parallel Burr-Brown ADC701 16 1 1.5 Parallel National ADC1005B 10 1 50 Parallel TI, National ADC0808 8 8 100 Parallel TI, National ADC0834 8 4 32 Serial TI TLC0820 8 1 1 Parallel TI TLC1540 10 11 21 Serial A/D and D/A Interface ICs Resolution, bOn-Board Filters Sampling Rate, kHz Data Format TI TLC32040 14 Yes 19.2 (programmable) Serial TI 2914 PCM codec & filter 8 Yes 8 Serial ? 2000 by CRC Press LLC the highest frequency in the analog signal. Oversampling A/D converters use sampling rates of N times this rate, where N typically ranges from 2 to 64. Both D/A and A/D converters require a voltage reference in order to achieve absolute conversion accuracy. Some conversion ICs have internal voltage references, while others accept external voltage references. For high- performance systems, an external precision reference is needed to ensure long-term stability, load regulation, and control over temperature fluctuations. External precision voltage reference ICs can be found in manufac- turers’ data books. Measurement accuracy is specified by the converter’s linearity. Integral linearity is a measure of linearity over the entire conversion range. It is often defined as the deviation from a straight line drawn between the endpoints and through zero (or the offset value) of the conversion range. Integral linearity is also referred to as relative accuracy. The offset value is the reference level required to establish the zero or midpoint of the conversion range. Differential linearity is the linearity between code transitions. Differential linearity is a measure of the monotonicity of the converter. A converter is said to be monotonic if increasing input values result in increasing output values. The accuracy and linearity values of a converter are specified in the data sheet in units of the LSB of the code. The linearity can vary with temperature, so the values are often specified at +25°C as well as over the entire temperature range of the device. D/A Conversion Processes Digital codes are typically converted to analog voltages by assigning a voltage weight to each bit in the digital code and then summing the voltage weights of the entire code. A general D/A converter consists of a network of precision resistors, input switches, and level shifters to activate the switches to convert a digital code to an analog current or voltage. D/A ICs that produce an analog current output usually have a faster settling time and better linearity than those that produce a voltage output. When the output current is available, the designer can convert this to a voltage through the selection of an appropriate output amplifier to achieve the necessary response speed for the given application. D/A converters commonly have a fixed or variable reference level. The reference level determines the switching threshold of the precision switches that form a controlled impedance network, which in turn controls the value of the output signal. Fixed reference D/A converters produce an output signal that is proportional to the digital input. Multiplying D/A converters produce an output signal that is proportional to the product of a varying reference level times a digital code. D/A converters can produce bipolar, positive, or negative polarity signals. A four-quadrant multiplying D/A converter allows both the reference signal and the value of the binary code to have a positive or negative polarity. The four-quadrant multiplying D/A converter produces bipolar output signals. D/A Converter ICs Most D/A converters are designed for general-purpose control applications. Some D/A converters, however, are designed for special applications, such as video or graphic outputs, high-definition video displays, ultra high-speed signal processing, digital video tape recording, digital attenuators, or high-speed function generators. D/A converter ICs often include special features that enable them to be interfaced easily to microprocessors or other systems. Microprocessor control inputs, input latches, buffers, input registers, and compatibility to standard logic families are features that are readily available in D/A ICs. In addition, the ICs usually have laser- trimmed precision resistors to eliminate the need for user trimming to achieve full-scale performance. A/D Conversion Processes Analog signals can be converted to digital codes by many methods, including integration, succesive approxi- mation, parallel (flash) conversion, delta modulation, pulse code modulation, and sigma-delta conversion. Two of the most common A/D conversion processes are successive approximation A/D conversion and parallel or flash A/D conversion. Very high-resolution digital audio or video systems require specialized A/D techniques that often incorporate one of these general techniques as well as specialized A/D conversion processes. Examples ? 2000 by CRC Press LLC of specialized A/D conversion techniques are pulse code modulation (PCM), and sigma-delta conversion. PCM is a common voice encoding scheme used not only by the audio industry in digital audio recordings but also by the telecommunications industry for voice encoding and multiplexing. Sigma-delta conversion is an over- sampling A/D conversion where signals are sampled at very high frequencies. It has very high resolution and low distortion and is being used in the digital audio recording industry. Successive approximation A/D conversion is a technique that is commonly used in medium- to high-speed data acquisition applications. It is one of the fastest A/D conversion techniques that requires a minimum amount of circuitry. The conversion times for successive approximation A/D conversion typically range from 10 to 300 ms for 8-bit systems. The successive approximation A/D converter can approximate the analog signal to form an n-bit digital code in n steps. The successive approximation register (SAR) individually compares an analog input voltage to the midpoint of one of n ranges to determine the value of one bit. This process is repeated a total of n times, using n ranges, to determine the n bits in the code. The comparison is accomplished as follows: The SAR determines if the analog input is above or below the midpoint and sets the bit of the digital code accordingly. The SAR assigns the bits beginning with the most significant bit. The bit is set to a 1 if the analog input is greater than the midpoint voltage, or it is set to a 0 if it is less than the midpoint voltage. The SAR then moves to the next bit and sets it to a 1 or a 0 based on the results of comparing the analog input with the midpoint of the next allowed range. Because the SAR must perform one approximation for each bit in the digital code, an n-bit code requires n approximations. A successive approximation A/D converter consists of four functional blocks, as shown in Fig. 32.1: the SAR, the analog comparator, a D/A converter, and a clock. FIGURE 32.1 Successive approximation A/D converter block diagram. Control Clock Control Shift Register Successive Approximation Register (SAR) D/A Resistor Ladder Network Analog Comparator Analog Input Voltage Output Latch Digital Output Code Control Clocking Signals + – ? 2000 by CRC Press LLC Parallel or flash A/D conversion is used in high-speed applications such as video signal processing, medical imaging, and radar detection systems. A flash A/D converter simultaneously compares the input analog voltage to 2 n – 1 threshold voltages to produce an n-bit digital code representing the analog voltage. Typical flash A/D converters with 8-bit resolution operate at 20 to 100 MHz. The functional blocks of a flash A/D converter are shown in Fig. 32.2. The circuitry consists of a precision resistor ladder network, 2 n – 1 analog comparators, and a digital priority encoder. The resistor network estab- lishes threshold voltages for each allowed quantization level. The analog comparators indicate whether or not the input analog voltage is above or below the threshold at each level. The output of the analog comparators is input to the digital priority encoder. The priority encoder produces the final digital output code that is stored in an output latch. An 8-bit flash A/D converter requires 255 comparators. The cost of high-resolution A/D comparators escalates as the circuit complexity increases and as the number of analog converters rises by 2 n – 1. As a low-cost alternative, some manufacturers produce modified flash A/D converters that perform the A/D conversion in two steps to reduce the amount of circuitry required. These modified flash A/D converters are also referred to as half-flash A/D converters, since they perform only half of the conversion simultaneously. A/D Converter ICs A/D converter ICs can be classified as general-purpose, high-speed, flash, and sampling A/D converters. The general-purpose A/D converters are typically low speed and low cost, with conversion times ranging from 2 ms to 33 ms. A/D conversion techniques used by these devices typically include successive approximation, tracking, and integrating. The general-purpose A/D converters often have control signals for simplified microprocessor interfacing. These ICs are appropriate for many process control, industrial, and instrumentation applications, as well as for environmental monitoring such as seismology, oceanography, meteorology, and pollution monitoring. FIGURE 32.2 Flash A/D converter block diagram. + – + – + – + – + – + – + – +5 Volts Analog Input Voltage Analog Comparators Digital Output Code Analog Ground Resistor Ladder Network 4.375 Volts 1K ? 1K ? 1K ? 1K ? 1K ? 1K ? 1K ? 1K ? 3.75 Volts 3.125 Volts 2.5 Volts 1.875 Volts 1.25 Volts .625 Volts C MSB Octal Priority Encoder Output Latch LSB 1 2 3 4 5 6 7 B A ? 2000 by CRC Press LLC High-speed A/D converters have conversion times typically ranging from 400 ns to 3 ms. The higher speed performance of these devices is achieved by using the successive approximation technique, modified flash techniques, and statistically derived A/D conversion techniques. Applications appropriate for these A/D ICs include fast Fourier transform (FFT) analysis, radar digitization, medical instrumentation, and multiplexed data acquisition. Some ICs have been manufactured with an extremely high degree of linearity, to be appropriate for specialized applications in digital spectrum analysis, vibration analysis, geological research, sonar digitizing, and medical imaging. Flash A/D converters have conversion times ranging typically from 10 to 50 ns. Flash A/D conversion techniques enable these ICs to be used in many specialized high-speed data acquisition applications such as TV video digitizing (encoding), radar analysis, transient analysis, high-speed digital oscilloscopes, medical ultrasound imaging, high-energy physics, and robotic vision applications. Sampling A/D converters have a sample-and-hold amplifier circuit built into the IC. This eliminates the need for an external sample-and-hold circuit. The throughput of these A/D converter ICs ranges typically from 35 kHz to 100 MHz. The speed of the system is dependent on the A/D technique used by the sampling A/D converter. A/D converter ICs produce digital codes in a serial or parallel format, and some ICs offer the designer both formats. The digital outputs are compatible with standard logic families to facilitate interfacing to other digital systems. In addition, some A/D converter ICs have a built-in analog multiplexer and therefore can accept more than one analog input signal. Pulse code modulation (PCM) ICs are high-precision A/D converters. The PCM IC is often refered to as a PCM codec with both encoder and decoder functions. The encoder portion of the codec performs the A/D conversion, and the decoder portion of the codec performs the D/A conversion. The digital code is usually formatted as a serial data stream for ease of interfacing to digital transmission and multiplexing systems. PCM is a technique where an analog signal is sampled, quantized, and then encoded as a digital word. The PCM IC can include successive approximation techniques or other techniques to accomplish the PCM encoding. In addition, the PCM codec may employ nonlinear data compression techniques, such as companding, if it is necessary to minimize the number of bits in the output digital code. Companding is a logarithmic technique used to compress a code to fewer bits before transmission. The inverse logarithmic function is then used to expand the code to its original number of bits before converting it to the analog signal. Companding is typically used in telecommunications transmission systems to minimize data transmission rates without degrading the resolution of low-amplitude signals. Two standardized companding techniques are used extensively: A-law and m-law. The A-law companding is used in Europe, whereas the m-law is used predominantly in the U.S. and Japan. Linear PCM conversion is used in high-fidelity audio systems to preserve the integrity of the audio signal throughout the entire analog range. Digital signal processing (DSP) techniques provide another type of A/D conversion ICs. Specialized A/D conversion such as adaptive differential pulse code modulation (ADPCM), sigma-delta modulation, speech sub- band encoding, adaptive predictive speech encoding, and speech recognition can be accomplished through the use of DSP systems. Some DSP systems require analog front ends that employ traditional PCM codec ICs or DSP interface ICs. These ICs can interface to a digital signal processor for advanced A/D applications. Some manufacturers have incorporated DSP techniques on board the single-chip A/D IC, as in the case of the DSP56ACD16 sigma-delta modulation IC by Motorola. Integrating A/D converters are used for conversions that must take place over a long period of time, such as digital voltmeter applications or sensor applications such as thermocouples. The integrating A/D converter produces a digital code that represents the average of the signal over time. Noise is reduced by means of the signal averaging, or integration. Dual-slope integration is accomplished by a counter that advances while an input voltage charges a capacitor in a specified time interval, T. This is compared to another count sequence that advances while a reference voltage discharges across the same capacitor in a time interval, d T. The ratio of the charging count value to the discharging count value is proportional to the ration of the input voltage to the reference voltage. Hence, the integrating converter provides a digital code that is a measure of the input voltage averaged over time. The conversion accuracy is independent of the capacitor and the clock frequency since they affect both the charging and discharging operations. The charging period, T, is selected to be the ? 2000 by CRC Press LLC period of the fundamental frequency to be rejected. The maximum conversion rate is slightly less than 1/(2T) conversions per second. While this limits the conversion rate to be too slow for high-speed data acquisition applications, it is appropriate for long-duration applications of slowly varying input signals. Grounding and Bypassing on D/A and A/D ICs D/A and A/D converter ICs require correct grounding and capacitive bypassing in order to operate according to performance specifications. The digital signals can severely impair analog signals. To combat the electro- magnetic interference induced by the digital signals, the analog and digital grounds should be kept separate and should have only one common point on the circuit board. If possible, this common point should be the connection to the power supply. Bypass capacitors are required at the power connections to the IC, the reference signal inputs, and the analog inputs to minimize noise that is induced by the digital signals. Each manufacturer specifies the recommended bypass capacitor locations and values in the data sheet. The 1-mF tantalum capacitors are commonly recom- mended, with additional high-frequency power supply decoupling sometimes being recommended through the use of ceramic disc shunt capacitors. The manufacturers’ recommendations should be followed to ensure proper performance. Selection Criteria for D/A and A/D Converter ICs Hundreds of D/A and A/D converter ICs are available, with prices ranging from a few dollars to several hundred dollars each. The selection of the appropriate type of converter is based on the application requirements of the system, the performance requirements, and cost. The following issues should be considered in order to select the appropriate converter. 1. What are the input and output requirements of the system? Specify all signal current and voltage ranges, logic levels, input and output impedances, digital codes, data rates, and data formats. 2. What level of accuracy is required? Determine the resolution needed throughout the analog voltage range, the dynamic response, the degree of linearity, and the number of bits encoding. 3. What speed is required? Determine the maximum analog input frequency for sampling in an A/D system, the number of bits for encoding each analog signal, and the rate of change of input digital codes in a D/A system. 4. What is the operating environment of the system? Obtain information on the temperature range and power supply to select a converter that is accurate over the operating range. Final selection of D/A and A/D converter ICs should be made by consulting manufacturers to obtain their technical specifications of the devices. Major manufacturers of D/A and A/D converters include Analog Devices, Burr-Brown, DATEL, Maxim, National, Phillips Components, Precision Monolithics, Signetics, Sony, Texas Instruments, Ultra Analog, and Yamaha. Information on contacting these manufacturers and others can be found in an IC Master Catalog. Defining Terms Companding: A process designed to minimize the transmission bit rate of a signal by compressing it prior to transmission and expanding it upon reception. It is a rudimentary “data compression” technique that requires minimal processing. Delta modulation: An A/D conversion process where the digital output code represents the change, or slope, of the analog input signal, rather than the absolute value of the analog input signal. A 1 indicates a rising slope of the input signal. A 0 indicates a falling slope of the input signal. The sampling rate is dependent on the derivative of the signal, since a rapidly changing signal would require a rapid sampling rate for acceptable performance. Fixed reference D/A converter: The analog output is proportional to a fixed (nonvarying) reference signal. ? 2000 by CRC Press LLC Flash A/D: The fastest A/D conversion process available to date, also referred to as parallel A/D conversion. The analog signal is simultaneously evaluated by 2 n – 1 comparators to produce an n-bit digital code in one step. Because of the large number of comparators required, the circuitry for flash A/D converters can be very expensive. This technique is commonly used in digital video systems. Integrating A/D: The analog input signal is integrated over time to produce a digital signal that represents the area under the curve, or the integral. Multiplying D/A: A D/A conversion process where the output signal is the product of a digital code multiplied times an analog input reference signal. This allows the analog reference signal to be scaled by a digital code. Nyquist A/D converters: A/D converters that sample analog signals that have a maximum frequency that is less than the Nyquist frequency. The Nyquist frequency is defined as one-half of the sampling frequency. If a signal has frequencies above the Nyquist frequency, a distortion called aliasing occurs. To prevent aliasing, an antialiasing filter with a flat passband and very sharp roll-off is required. Oversampling converters: A/D converters that sample frequencies at a rate much higher than the Nyquist frequency. Typical oversampling rates are 32 and 64 times the sampling rate that would be required with the Nyquist converters. Pulse code modulation (PCM): An A/D conversion process requiring three steps: the analog signal is sampled, quantized, and encoded into a fixed length digital code. This technique is used in many digital voice and audio systems. The reverse process reconstructs an analog signal from the PCM code. The operation is very similar to other A/D techniques, but specific PCM circuits are optimized for the particular voice or audio application. Sigma-delta A/D conversion:An oversampling A/D conversion process where the analog signal is sampled at rates much higher (typically 64 times) than the sampling rates that would be required with a Nyquist converter. Sigma-delta modulators integrate the analog signal before performing the delta modulation. The integral of the analog signal is encoded rather than the change in the analog signal, as is the case for traditional delta modulation. A digital sample rate reduction filter (also called a digital decimation filter) is used to provide an output sampling rate at twice the Nyquist frequency of the signal. The overall result of oversampling and digital sample rate reduction is greater resolution and less distortion compared to a Nyquist converter process. Successive approximation: An A/D conversion process that systematically evaluates the analog signal in n steps to produce an n-bit digital code. The analog signal is successively compared to determine the digital code, beginning with the determination of the most significant bit of the code. Related Topic 15.1 Coding, Transmission, and Storage References Analog Devices, Analog Devices Data Conversion Products Data Book, Norwood, Mass.: Analog Devices, Inc., 1989. Burr–Brown, Burr-Brown Integrated Circuits Data Book, Tucson, Ariz.: Burr-Brown, 1989. DATEL, DATEL Data Conversion Catalog, Mansfield, Mass.: DATEL, Inc., 1988. W. Drachler, and M. Bill, “New high-speed, low-power data-acquistion ICs,” Analog Dialogue, vol. 29, no. 2, pp. 3–6, Norwood, Mass.: Analog Devices, Inc., 1995. S. Garrod and R. Borns, Digital Logic: Analysis, Application and Design, Philadelphia, Pa.: Saunders College Publishing, 1991, Chap. 16. J.M. Jacob, Industrial Control Electronics, Englewood Cliffs, N.J.: Prentice–Hall, 1989, Chap. 6. B. Keiser and E. Strange, Digital Telaphony and Network Integration, 2nd ed., New York: Van Nostrand Reinhold, 1995. Motorola, Motorola Telecommunications Data Book, Phoenix, Ariz.: Motorola, Inc., 1989. National Semiconductor, National Semiconductor Data Acquisition Linear Devices Data Book, Santa Clara, Calif.: National Semiconductor Corp., 1989. ? 2000 by CRC Press LLC S. Park, Principles of Sigma-Delta Modulation for Analog-to-Digital Converters, Phoenix, Ariz.: Motorola, Inc., 1990. Texas Instruments, Texas Instruments Digital Signal Processing Applications with the TMS320 Family, Dallas, Tex.: Texas Instruments, 1986. Texas Instruments, 1989. Texas Instruments Linear Circuits Data Acquisition and Conversion Data Book, Dallas, Tex.: Texas Instruments, 1989. Further Information Analog Devices, Inc. has edited or published several technical handbooks to assist design engineers with their data acquisition system requirements. These references should be consulted for extensive technical information and depth. The publications include Analog-Digital Conversion Handbook, by the engineering staff of Analog Devices, published by Prentice–Hall, Englewood Cliffs, N.J., 1986; Nonlinear Circuits Handbook, Transducer Interfacing Handbook, and Synchro and Resolver Conversion, all published by Analog Devices Inc., Norwood, Mass. Engineering trade journals and design publications often have articles describing recent A/D and D/A circuits and their applications. These publications include EDN Magazine, EE Times, and IEEE Spectrum. Research- related topics are covered in IEEE Transactions on Circuits and Systems, and also the IEEE Transactions on Instrumentation and Measurement. ? 2000 by CRC Press LLC