 |
|
| If you can't view the Datasheet, Please click here to try to view without PDF Reader . |
|
|
|
| If you can't view the Datasheet, Please click here to try to view without PDF Reader . |
|
|
| Datasheet File OCR Text: |
| AN04 Application Note Voltage References for the CS5012A / CS5014 / CS5016 / CS5101A/ CS5102A / CS5126 Series of A/D Converters by Bruce Del Signore & Steven Harris +5V Analog Supply VA+ 10 VD+ BW BP/UP Mode Select Analog Signal Source Signal Conditioning CLKIN AIN Clock Source Serial Data Interface SDATA SCLK A/D CONVERTER D0-D15 V+ Data Processor EOC Voltage Reference + 10 uF 0.01 uF VREF EOT HOLD CAL INTRLV CS RD Control R AGND REFBUF VA-5V Analog Supply 10 A0 RESET TST DGND VD- Crystal Semiconductor Corporation P.O. Box 17847, Austin, TX 78760 (512) 445-7222 FAX: (512) 462-2723 http://www.crystal.com Copyright (c) Crystal Semiconductor Corporation 1996 (All Rights Reserved) JULY '94 AN4REV5 1 Voltage References for SAR-type A/D Converters INTRODUCTION This application note discusses voltage references for use with Crystal Semiconductor's successive approximation series of A/D converters. Reference design considerations, a design example and suggested reference circuits are explained in detail. Voltage references provide accurate voltages for use in data acquisition systems in order to establish a basis for conversion. In a data acquisition system, the value of the reference sets the gain of the A/D stage since the digital output corresponds to the ratio of the analog input signal to the reference voltage. In static applications, information is contained in the signal amplitude, therefore the absolute value of the reference voltage is important. In many signal processing applications, information is contained in the frequency and phase of the signal. Here, absolute value is not as important as the stability of the reference voltage during conversion. Zener-diode Reference There are two major varieties of voltage references. The first is the zener-diode based reference which uses a reverse-biased zener diode operated in its breakdown region. Most reference zeners breakdown at voltages between 6.0 and 7.0V, which limits the minimum supply voltage necessary for operation. When the diode is supplied with a constant current, it has a constant voltage drop. Zener references use a zener diode and an integrated feedback amplifier which provides constant current, gain, and buffering for the zener diode. Zener diodes exhibit two types of breakdown. The first is zener breakdown which has a negative temperature coefficient and is dominant at low current levels. The second, avalanche breakdown, occurs at higher current levels and has a 2 positive temperature coefficient. At some specific current level, these two effects cancel each other and the temperature coefficient of the zener breakdown voltage is zero. As the ambient temperature changes, one of the breakdown mechanisms becomes dominant and the the reverse-biased diode voltage will exhibit a temperature coefficient. Bandgap Reference The second major type of reference is the bandgap reference. This reference uses the baseemitter voltage (Vbe) of a bipolar transistor as a basis for operation. The Vbe has a negative temperature coefficient (-2mV/C). This negative temperature coefficient is balanced by a voltage with a positive temperature coefficient of the same magnitude. This voltage is usually obtained by using the difference of two Vbe's of transistors operating at different current densities. When both voltages are scaled and summed together, the result is a voltage which is less sensitive to temperature. The headroom required for bias and support circuitry is only a few volts over the output voltage. Reference Specifications Voltage references have six important specifications. These are absolute accuracy, temperature coefficient, long-term reference drift, power supply sensitivity, output impedance, and output noise. Absolute or untrimmed accuracy is the difference between the actual output voltage and the ideal output voltage. It is specified in millivolts. Temperature coefficient describes the drift in the output voltage with temperature. Since this drift is nonlinear, curve fitting is often used for all temperatures between those actually tested. Voltage references are available with temperature coefficients as low as 1 ppm/C. Inexpensive references are available with 10 to 50 ppm/C drift AN4REV5 Voltage References for SAR-type A/D Converters 0 -10 -20 Reference Voltage-30 Drift (ppm) -40 -50 250 500 750 1000 1250 1500 1750 2000 Time (hours) 2250 2500 Figure 1. - Long Term Stability of a Typical Zener Reference which is comparable to on-chip references of bipolar A/D converters. Temperature coefficient is specified in ppm/C. Long term stability is the drift in the reference voltage over time. Most references show minor deviations in voltage due to 1/f noise in circuit components. These deviations are usually small and are superimposed on a larger drift characteristic which is due to device aging. An example of this is seen in Figure 1. Long term drift is specified in ppm/1000 hrs. Power supply sensitivity (line regulation) is the change in output voltage due to a change in power supply. Most references have good power supply rejection at dc, but ac power supply rejection is also important when power supplies are subject to high frequency coupling or noise spikes. PSRR (Power Supply Rejection Ratio) is the ratio of the change in power supply to the change in output voltage. It is specified in dB. Output impedance is important because of the dynamic loads generated by successive-approxiAN4REV5 mation A/D converters. When the reference is sourcing or sinking current, its output voltage will change due to non-zero output impedance. This impedance must be low enough at all frequencies of interest so the deviation in reference voltage when sourcing current is negligible. Output impedance is specified in ohms. Output noise can lead to comparison errors in the A/D converter, and subsequently conversion errors. Reference noise is more evident with full scale inputs. It is specified in V peak-to-peak. Design Considerations When interfacing a voltage reference to an A/D converter, the specifications should be robust enough so that the reference does not become a source of conversion error. During conversion, each capacitor of the calibrated capacitor array in the ADC is switched between VREF and AGND in a manner determined by the successive approximation algorithm. The charging and discharging of the array results in a current load at 3 Voltage References for SAR-type A/D Converters the reference. The ADC's include an internal buffer amplifier to minimize the external reference circuit's drive requirement and preserve the reference's integrity. Whenever the array is switched during conversion, the buffer is used to pre-charge the array thereby providing the bulk of the necessary charge. This buffer enlists the aid of an external 0.1F ceramic capacitor which must be tied between its output, REFBUF, and the negative analog supply, VA-. The appropriate array capacitors are then switched to the unbuffered VREF pin to avoid any errors due to offsets and/or noise in the buffer. The external reference circuitry need only provide the residual charge required to fully charge the array after pre-charging from the internal buffer. This creates an ac current load as the ADC sequences through conversions. The reference circuitry must have a low enough output impedance to provide the requisite current without changing its output voltage significantly. As the analog input signal varies, the switching sequence of the internal capacitor array changes. The current load on the external reference circuitry thus varies in response with the analog input. Also with CS5012,4,6 converters, bits are converted at a 1MHz rate with a full speed (4MHz) clock. The reference must settle within one microsecond so that it will be accurate before the next bit is converted. Signal amplitude dependent loading and conversion settling time require the output impedance of the reference to remain low from dc to at least 1MHz in order to ensure good converter performance. The CS5012,4,6 series of converters can operate with a wide range of reference voltages, but signal-to-noise performance is maximized by using as wide a signal range as possible. All CS5012,4,6 converters can actually accept reference voltages up to the positive analog supply. However, the internal buffer's offset may increase as the reference voltage approaches VA+. This increases external drive requirements at VREF. Allowing 250mV headroom for the internal reference buffer is recommended. If the supplies are regulated specifically for the converter, 5.0 volt references may be used if the supply voltages for the ADC are kept between 5.25 and 5.5 volts. The magnitude of the current load presented to the external reference circuitry by the ADC's will vary with the master clock frequency. At full speed (4MHz clock), the ADC's require maximum load currents of 10A peak-to-peak (1A peak-to-peak typical). The voltage reference must supply this current and maintain adequate voltage regulation. The load currents scale proportionately with the master clock frequency. Slower clocks can be used to relax maximum output impedance specification of the reference. When driving multiple A/D converters from the same reference circuit, load currents will scale proportionally to the number of converters. Distribute the required decoupling components such that each ADC is locally decoupled. A reference with a maximum output impedance of 2 will yield a maximum error of 20V. This reference could drive a CS5016 (LSB=69V with Part # fclk (Vref=4.5V) (Vref=2.5V) (Vref=4.5V) 4MHz 27 15 7 2MHz 54 30 14 1MHz 108 60 28 500kHz 216 120 56 All units in ohms CS5012 CS5012 CS5014 CS5016 4 (Vref=4.5V) 2 4 8 16 AN4REV5 Table 1. - Maximum Output Impedance for 1/4 LSB Reference Deviation Voltage References for SAR-type A/D Converters a 4.5V reference) and maintain approximately 1/4 LSB deviation during conversion. Similarly for the CS5014 (LSB=276V with a 4.5V reference), and CS5012 (LSB=613V with a 2.5V reference), maximum impedances of 7 and 15 respectively will maintain adequate regulation. Table 1 defines maximum reference impedances allowed for each of the Crystal A/D's operating at different master clock frequencies in order to keep reference deviation approximately equal to 1/4 LSB. All precision references exhibit extremely low output impedance at dc. However, as frequency increases the impedance also increases. A large capacitor connected between VREF and AGND can provide sufficiently low output impedance at the high end of the frequency spectrum where the reference impedance is too high. For example, the impedance of an ideal 10F capacitor drops below 1 at frequencies greater than 16kHz. However, actual capacitors behave differently due to their physical structure. Tantalum-foil electrolytic capacitors begin to appear inductive at frequencies around 100kHz and as a result their impedance begins to rise at frequencies above this. Aluminum electrolytic capacitors appear inductive at frequencies around 10kHz. Ceramic-disk capacitors behave much closer to ideal and begin to appear inductive at frequencies around 5MHz, but 10F ceramic-disk capacitors are quite rare. Therefore, a high-quality tantalum V+ capacitor (10F) in parallel with a smaller (0.1F) ceramic capacitor is recommended. This combination yields low impedance up to frequencies around 50MHz. Peaking The presence of large capacitors on the output of some voltage references may cause peaking in the output impedance at intermediate frequencies. Care should be exercised to ensure that significant peaking does not exist or that some form of compensation is provided to reduce it. Most commercially available references use an integrated op-amp to buffer the actual reference generator. External capacitive loading will degrade performance of this op-amp. This degradation can be analyzed using classical analysis techniques. The open loop gain of an ideal op-amp is primarily determined by the internal compensation capacitor which generates a left-half-planepole (LHPP) at a very low frequency. The effect of this pole is to reduce the open loop gain by 20dB per decade and to add a -90 degree phase shift to the open loop transfer characteristic. Adding a capacitive load to the output of the op-amp generates another LHPP at a frequency inversely proportional to the capacitor's value. An additional 20dB per decade reduction in gain and -90 degree phase shift result from the second LHPP. Vref C1 10 F C2 0.01 F 28 VREF 29 REFBUF 0.1 F 30 VAADC R= 1 (C 1+ C 2 ) fpeak 2 R V- Figure 2. - Reference Connections AN4REV5 5 Voltage References for SAR-type A/D Converters The unity gain bandwidth of an op-amp (f0), is the frequency at which the open loop gain goes to unity. If the total phase shift reaches -180 degrees before f0 is reached the op-amp will become unstable. The closed loop frequency response peaks at f0. As the total open loop phase shift at f0 approaches -180 degrees, the closed loop peak at f0 approaches infinity. The point of critical damping is the point where the peaking is precisely zero. Any phase shift less than this results in no peaking, and phase shift greater than this results in increased peaking. Any peaking that might occur can be reduced by placing a small resistor in series with the capacitors (Figure 2). This resistor adds a left-halfplane-zero (LHPZ) to the open loop characteristic of the op-amp. This zero increases the gain by 20dB per decade, and adds a +90 degree phase shift. The resulting reduction in total phase shift at f0 reduces peaking in the closed loop characteristic. The equation in Figure 2 can be used to help calculate the optimum value of R for a particular reference. The term "fpeak" is the frequency of the peak in the output impedance of the reference before the resistor is added. 1000 Design Example Figure 3 shows the output impedance characteristic of an LT1019-5 reference trimmed to 4.5V. The three curves represent impedances of the stand-alone reference, the reference with a 10F tantalum and a 0.1F ceramic capacitor added in parallel to the output, and the reference with the capacitors and a 2.2 resistor in series with them (See Figure 2). Without loading, the reference impedance rises above 100 at 50kHz. Adding the capacitors, peaking can be seen, but the maximum impedance is about 13 at 4kHz. As shown in Table 1, 13 is sufficient for use with the 12-bit converters and for the 14 and 16bit converters with slow master clocks. With the addition of the 2.2 resistor, the peak is reduced to 6 and the impedance approaches 2.2 at high frequencies. Suggested Voltage Reference Circuits Nine reference circuits were characterized for use with the CS5012, CS5014, CS5016, CS5101, CS5102, CS5126 family of successive-approximation A/D converters. Important reference specifications such as output impedance and drift were measured for all references using standard test techniques. In addition, a Fast-Fourier Transform (FFT) test was performed to characterize the total dynamic performance of each reference circuit while driving a CS5016 converter. The same CS5016 was used for all tests yielding results which allow the comparison between different references. A summary of performance can be seen in the table at the end of this application note. During the FFT test, a pure sine wave is applied to the CS5016 and a "time record" of 1024 samples is captured and processed. The FFT algorithm analyzes the spectral content of the waveform and distributes its energy among 512 "frequency bins". Distribution of energy in bins outside of the fundamental and dc can be attributed to errors in the A/D converter's performance, the reference, or the input sine wave. AN4REV5 reference only 100 10 1 reference with capacitors and resistors 0.1 reference with capacitors 0.01 1 10 100 1K 10K 100K FREQUENCY (Hz) Figure 3. - Output Impedance Curves for LT1019-5 6 Voltage References for SAR-type A/D Converters Reference Type Bandgap Untrimmed Accuracy 2.5mV Max Impedance 6.5 @ 3.2KHz Total Output Drift 5ppm / C PSRR (50Hz to 500Hz) 90dB Long Term Stability Output Noise (dc to 1MHz) 250V p-p S / (N+D) (100Hz) 89dB S / (N+D) (1kHz) 89dB +10 +15V 2 IN LT1019-5 OUT +4.5V 6 + C1 TRIM GND 4 C2 5 R1 R2 R1: 25k Low Drift Pot C1: 10F tantalum C2: 0.1F ceramic R2: 3.9 1% Figure 4. LT1019-5 Reference Trimmable to 4.5V LT1019-4.5 +4.5V 6 C1 + +10 +15V IN 2 GND OUT C2 R2 4 C2: 0.1F ceramic R2: 2.6 1% C1: 10F tantalum Reference Type Bandgap Untrimmed Accuracy 3.0mV Max Impedance 3.1 @ 6.1KHz Total Output Drift 5ppm / C PSRR (50Hz to 500Hz) 90dB Long Term Stability Output Noise (dc to 1MHz) 150V p-p S / (N+D) (100Hz) 91dB S / (N+D) (1kHz) 90dB Figure 5. LT1019-4.5 Reference The result of the FFT test is the ratio of input signal amplitude to the combination of harmonic distortion and total integrated noise. It is referred to as S/(N+D) in all of the performance charts in Figures 4 to 10. This ratio is expressed in dB. If input sine wave distortion and the actual A/D converter's distortion and noise are assumed to be negligible, the S/(N+D) is due to the reference only. In reality, this assumption can not be made. In the case of the Great Reference (See Figure 11), performance matches or exceeds the capability of the test setup. S/(N+D) ratios of 72 and 82 dB are sufficient for the 12-bit and 14-bit converters. For the 16-bit converters, 88 to 94 dB is necessary. FFT tests were performed at 100Hz and 1kHz. The 100Hz test checks the output impedance of the reference chip itself which dominates at low frequencies. The 1kHz FFT test checks the output impedance at intermediate frequencies in the kHz range. The highest output impedance was seen in AN4REV5 all references at these intermediate frequencies. Since the reference capacitors dominate the impedance at high frequencies, high frequency FFT tests were not necessary. Although not tested, the best reference is likely to yield the best DNL performance. The least complicated reference circuit is the stand-alone reference chip with a passive compensation network. Its temperature drift and noise performance is equal to the reference chip itself since the compensation network does not change the dc output voltage. Keeping the output impedance low from dc to 1MHz is not trivial however, since there is no additional active circuitry added to perform this task. Five references were tested in the stand-alone configuration. Figures 4, 5, 6, 7, and 8 illustrate schematics and measured specifications for these references. All references are monolithic with the exception of the VRE104 reference which is a hybrid (available from Thaler (602) 742-5572). Notice that the VRE104 7 Voltage References for SAR-type A/D Converters Reference Type Bandgap Untrimmed Accuracy 1.25mV Max Impedance 4.0 @ 5.8KHz Total Output Drift 5ppm / C PSRR (50Hz to 500Hz) 90dB Long Term Stability Output Noise (dc to 1MHz) 100V p-p S / (N+D) (100Hz) 87dB S / (N+D) (1kHz) 89dB +7.5 +15V 2 LT1019-2.5 OUT +2.5V 6 C1 + IN C2 GND 4 R2 R2: 2.4 1% C1: 10F tantalum C2: 0.1F ceramic Figure 6. LT1019-2.5 Reference LT1021-5 OUT IN 2 +10 +15V Reference Type Untrimmed Accuracy 6 + C1 C2 Max Impedance R1 Total Output Drift TRIM PSRR (50Hz to 500Hz) GND 5 R3 R2 Long Term Stability 4 Output Noise C1: 10F tantalum R1: 27k 1% S / (N+D) (100Hz) R2: 2.2 1% C2: 0.1F ceramic R3: 50k Low Drift Pot S / (N+D) (1kHz) Figure 7. LT1021 Reference +5.0V +4.5V Zener 2.5mV 3.8 @ 5.0KHz 3ppm / C 86dB 15ppm / 1000hr 60V p-p 90dB 90dB Zener 500V 2.5 @ 20KHz 0.5ppm / C 100dB 6ppm / 1000hr 80V p-p 90dB 90dB +15V 11 VRE104 OUT 13 +PS + C1 C2 GND REFGND 6 7 R2 To AGND C1: 10F tantalum C2: 0.1F ceramic R2: 1.96 1% Reference Type Untrimmed Accuracy Max Impedance Total Output Drift PSRR (50Hz to 500Hz) Long Term Stability Total Output Noise S / (N+D) (100Hz) S / (N+D) (1kHz) Figure 8. VRE104 Reference +4.7 2 VIN VOUT R3 1k 6 3 +IN V+ 7 OUT 4 6 R4 3.57k C1 10F +15V CC 3.3F 2N2907A REF43 GND 4 2 OP90 -IN V- +4.5V + C2 0.1F R5 3.9 CF 0.1F R1 2k 0.1% R2 2.5k 0.1% Reference Type Untrimmed Accuracy Max Impedance Total Output Drift PSRR (50Hz to 500Hz) Long Term Stability Output Noise (dc to 1MHz) S / (N+D) (100Hz) S / (N+D) (1kHz) Bandgap 1.5mV 4.4 @ 1KHz 8.0ppm / C 60dB 400V p-p 88dB 88dB Figure 9. Low Power Supply Reference 8 AN4REV5 Voltage References for SAR-type A/D Converters +5.0V 1 VIN VOUT R1 1k 2 7 V+ OUT -5.0V 0.068F R6 4.7 10F 0.1F V4 CF 6 R5 560 2N2907A R4 3.3k C1 +4.5V + C2 -IN MC1403 2 GND 3 CN 10F 3 OP07 +IN R3 R2 2.49k 0.1% 1.96k 0.1% Reference Type Untrimmed Accuracy Max Impedance Total Output Drift PSRR (50Hz to 500Hz) Long Term Stability Output Noise (dc to 1MHz) S / (N+D) (100Hz) S / (N+D) (1kHz) Bandgap 50mV 6.9 @ 2KHz 25ppm / C 80dB 30V p-p 90dB 90dB Figure 10. Low Headroom Reference and the LT1019-4.5 require no trimming for 4.5V operation. The calculated value of R2 in each of the references above will change slightly between units. Since the actual variation is small, picking the closest 1% tolerance resistor to the calculated value should give similar performance for all references of a particular manufacturer's model. Other stand-alone voltage references with similar specifications include the AD584, REF02, REF03, REF10, and REF43. When designing with these references, the equation shown in Figure 2 should be used to calculate the appropriate value of R2 for each type of reference. For applications which use 5.0 volt supplies, the reference in Figure 9 can be used. This reference circuit, designed by engineers at the former Precision Monolithics, takes advantage of a low power op-amp in a novel feedback configuration to achieve a 4.5 volt reference which operates from 4.7 to 15 volt supplies. 2.14 volts. This output voltage is well within the maximum specification of the OP-90 op-amp. Other references can be substituted for the REF43 if different drift or noise specifications are required. The reference shown in Figure 10 is a low noise reference with less than 30V peak-to-peak of noise from dc to 1MHz. It uses a discrete output stage allowing Vref to come within 300mV of the positive supply. The filtering network R1,CN reduces the bandwidth of the reference and therefore reduces the total output noise. The OP-07 is a low noise op-amp which buffers the filtered reference. This op-amp contributes very little noise to the entire reference circuitry. The temperature coefficient of this reference is primarily due to to the matching of the gain resistors R2 and R3, so low temperature drift resistors should be used. Long term drift is dominated by the MC1403's drift. Other 2.5 volt references can be used to improve this specification. The output voltage can be changed by adjusting R2 and R3 according to the following equation: Vref = Vout((R2+R3)/R2). Resistors with 0.1% tolerance for R2 and R3 limit the reference's untrimmed accuracy only. Resistors with 1% or 5% tolerance can be used if untrimmed accuracy less than 50mV is not necessary. The supplies of the Since only a few microamps of quiescent current flows in the op-amp, it can be assumed that the only current flowing in R3 is the same as that flowing in R4. It can be shown that V6 = 3.57(Vin-V7). For an output of 4.5 volts, and a supply of 4.7 volts, the op-amp has a supply of approximately 4.0 volts and an output voltage of AN4REV5 9 Voltage References for SAR-type A/D Converters R2 2.21k 0.1% 6 R3 1.82k 0.1% 3 CN 10F +15.0V 0.1F 2 VIN VOUT = +IN V+ V4 -15.0V CF 0.01F R4 47k R1 R5 22 C1 16F + C2 0.1F +4.5V 7 OUT 6 ceramic + 10F tantalum REF10 GND 4 OP27 2 -IN -15 V 1.0k Reference Type Untrimmed Accuracy Max Impedance Total Output Drift PSRR (50Hz to 100Hz) Bandgap 25mV 0.5 @ 3kHz 9ppm / C 95dB Long Term Stability Output Noise (dc to 1MHz) S / (N+D) (100Hz) S / (N+D) (1kHz) 50ppm / 1000hr 5V p-p 92dB 92dB Figure 11. - Great Reference OP-07 should be bypassed with 0.1F capacitors to ground. The reference in Figure 11 exhibits very good noise, output impedance, and long term drift performance. It can be used in applications which have 15 volt supplies available. The reference has noise less than 10V peak-to-peak from dc to 1MHz. The filtering network R2, R3, and CN filters noise components greater than 10Hz from the output of the REF10 reference. The OP-27 is a very low noise op-amp with excellent input offset drift over time and temperature. The temperature coefficient of this reference is primarily due to the matching of the voltage divider R2 and R3, assuming that an appropriate low leakage capacitor is chosen for CN. Matched, low temperature drift resistors should be used when absolute accuracy is required. Temperature drift of the reference chip plus input offset drift +5V R1* 100 50 C1 TL431C R3 2k R2 0.1F R4 1.6k + C3 0.1F C2 10F +4.5V Reference Type Untrimmed Accuracy Max Impedance Total Output Drift PSRR (50Hz to 500Hz) Long Term Stability Output Noise (dc to 1MHz) S / ( N + D ) (100Hz) S / ( N + D ) (1kHz) Bandgap 30mV 50 @600Hz 30ppm/ C 85dB 100uV p-p 92dB 91dB * Can be operated from +12 or +15 volts if R1 is changed to 2K. Figure 12. -TL431 Shunt Reference 10 AN4REV5 Voltage References for SAR-type A/D Converters of the op-amp is about 9ppm/C. Other 10 volt references can be used in place of the REF10. The reference voltage can be changed by adjusting R2 and R3 according to the following equation. Vref = Vout(R3/(R2+R3)). This circuit has no protection against accidentally applying 15V to the VREF pin. This could occur if the OP27 fails. For applications where good dynamic performance is required, but only moderate dc accuracy, the TL431 shunt reference is an inexpensive solution. Figure 12 shows an example circuit, along with the excellent dynamic performance numbers. Miscellaneous Applications Information Noise from the voltage reference element may reduce system performance. Bandgap references tend to generate much more noise than zener diodes. To obtain the best noise performance from the reference element, it should be band-limited. Note the broadband noise for the LT1019-5 circuit (Figure 4, 250V) versus the noise of a similar bandgap reference with additional circuitry to band-limit the noise as in the Great Reference (Figure 11, 10V). Thermal temperature gradients due to power dissipation on the voltage reference die can create output voltage shifts. Keeping the entire chip on an isothermal plane is helpful. Reference load conditions should be kept very close to those specified, or degraded temperature performance will result. Some references specify a thermal regulation in ppm/mW. This can be used to calculate voltage drift for a specific power dissipation due to loading. Overall die temperature change can cause thermally induced output voltage variations which can exceed electrical effects. Shifts in power dissipation on the board level are the major conAN4REV5 tributor to this error. In critical applications, using a heat-sink is recommended to keep the reference temperature deviations small. Thermocouple effects between package leads can also cause excessive output voltage drift and noise. Differences between materials in IC leads and PC-board traces can cause thermoelectric effects. Ambient air turbulence around the leads causes mismatches in the temperature between the package leads. The resulting thermoelectric voltage contributes to noise. Using dual in-line packages (DIPs) is recommended over using TO5 type packages. The copper or Alloy 42 lead frames on DIPs are much less sensitive to thermocouple effects than the Kovar leads of the TO5 packages. Using an enclosure such as a polysulfone shield which blocks the air flow over the reference package will also reduce the problem by reducing air movement around the package leads. In reference circuits which have external gain setting resistors, tracking of the temperature coefficients of these resistors is vital. Wirewound resistors made of Evenohm or Mangamin have the lowest temperature coefficients. Ceramic film resistors such as Vishay are also good. Matching in resistor temperature coefficients as good as 0.4 ppm/C can be achieved. Arranging these resistors in close proximity to one another also helps matching. SIP or DIP resistors by Beckman and Vishay exhibit the best matching since all resistors are processed on the same substrate. 11 Voltage References for SAR-type A/D Converters Reference LT1019-5 LT1019-4.5 LT1019-2.5 LT1021-5 VRE104 Low Supply Low Headroom Great TL431 Shunt Type Bandgap Bandgap Bandgap Zener Zener Bandgap Bandgap Bandgap Bandgap Untrimmed Accuracy 2.5mV 3.0mV 1.25mV 2.5mV 500V 1.5mV 50mV 25mV 30mV Maximum Impedance 6.5 @ 3.2kHz 3.1 @ 6.1kHz 4.0 @ 5.8kHz 3.8 @ 5.0kHz 2.5 @ 20kHz 4.4 @ 1kHz 6.9 @ 2kHz 0.5 @ 3kHz 50 @ 600Hz Output Drift 5ppm / C 5ppm / C 5ppm / C 3ppm / C 0.5ppm / C 8ppm / C 25ppm / C 9ppm / C 30ppm / C PSRR (50Hz to 100Hz) 90dB 90dB 90dB 86dB 100dB 60dB 80dB 95dB 85dB Reference LT1019-5 LT1019-4.5 LT1019-2.5 LT1021-5 VRE104 Low Supply Low Headroom Great TL431 Shunt Long Term Stability* 15ppm / 1000hr 6ppm / 1000hr 50ppm / 1000hr - Output Noise (dc to 1MHz) 250V p-p 150V p-p 100V p-p 60V p-p 80V p-p 400V p-p 30V p-p 10V p-p 100V p-p S/(N+D) (100Hz) 89dB 91dB 87dB 90dB 90dB 88dB 90dB 92dB 92dB S/(N+D) (1kHz) 89dB 90dB 86dB 90dB 90dB 88dB 90dB 92dB 91dB *Taken from reference data sheets. All other parameters were measured. Performance Comparison Table 12 AN4REV5 |
Price & Availability of AN04
 |
|
|
|
|
All Rights Reserved © IC-ON-LINE 2003 - 2022 |
| [Add Bookmark] [Contact Us] [Link exchange] [Privacy policy] |
|
Mirror Sites : [www.datasheet.hk]
[www.maxim4u.com] [www.ic-on-line.cn]
[www.ic-on-line.com] [www.ic-on-line.net]
[www.alldatasheet.com.cn]
[www.gdcy.com]
[www.gdcy.net] |