a FEATURES Ultralow Bias Current: 60 fA max (AD549L) 250 fA max (AD549J) Input Bias Current Guaranteed Over Common-Mode Voltage Range Low Offset Voltage: 0.25 mV max (AD549K) 1.00 mV max (AD549J) Low Offset Drift: 5 mV/8C max (AD549K) 20 mV/8C max (AD549J) Low Power: 700 mA max Supply Current Low Input Voltage Noise: 4 mV p-p 0.1 Hz to 10 Hz MIL-STD-883B Parts Available APPLICATIONS Electrometer Amplifiers Photodiode Preamp pH Electrode Buffer Vacuum lon Gage Measurement PRODUCT DESCRIPTION The AD549 is a monolithic electrometer operational amplifier with very low input bias current. Input offset voltage and input offset voltage drift are laser trimmed for precision performance. The AD549’s ultralow input current is achieved with “Topgate” JFET technology, a process development exclusive to Analog Devices. This technology allows the fabrication of extremely low input current JFETs compatible with a standard junctionisolated bipolar process. The 1015 Ω common-mode impedance, a result of the bootstrapped input stage, insures that the input current is essentially independent of common-mode voltage. The AD549 is suited for applications requiring very low input current and low input offset voltage. It excels as a preamp for a wide variety of current output transducers such as photodiodes, photomultiplier tubes, or oxygen sensors. The AD549 can also be used as a precision integrator or low droop sample and hold. The AD549 is pin compatible with standard FET and electrometer op amps, allowing designers to upgrade the performance of present systems at little additional cost. The AD549 is available in a TO-99 hermetic package. The case is connected to Pin 8 so that the metal case can be independently connected to a point at the same potential as the input terminals, minimizing stray leakage to the case. *Protected by Patent No. 4,639,683. Ultralow Input Bias Current Operational Amplifier AD549* CONNECTION DIAGRAM GUARD PIN, CONNECTED TO CASE NC OFFSET NULL 8 1 AD549 V+ 7 INVERTING 2 INPUT 6 OUTPUT 3 NONINVERTING INPUT 5 4 OFFSET NULL V– 10kΩ 5 1 4 –15V VOS TRIM NC = NO CONNECTION The AD549 is available in four performance grades. The J, K, and L versions are rated over the commercial temperature range 0°C to +70°C. The S grade is specified over the military temperature range of –55°C to +125°C and is available processed to MIL-STD-883B, Rev C. Extended reliability PLUS screening is also available. Plus screening includes 168-hour burn-in, as well as other environmental and physical tests derived from MIL-STD-883B, Rev C. PRODUCT HIGHLIGHTS 1. The AD549’s input currents are specified, 100% tested and guaranteed after the device is warmed up. Input current is guaranteed over the entire common-mode input voltage range. 2. The AD549’s input offset voltage and drift are laser trimmed to 0.25 mV and 5 µV/°C (AD549K), 1 mV and 20 µV/°C (AD549J). 3. A maximum quiescent supply current of 700 µA minimizes heating effects on input current and offset voltage. 4. AC specifications include 1 MHz unity gain bandwidth and 3 V/µs slew rate. Settling time for a 10 V input step is 5 µs to 0.01%. 5. The AD549 is an improved replacement for the AD515, OPA104, and 3528. REV. A Information furnished by Analog Devices is believed to be accurate and reliable. However, no responsibility is assumed by Analog Devices for its use, nor for any infringements of patents or other rights of third parties which may result from its use. No license is granted by implication or otherwise under any patent or patent rights of Analog Devices. One Technology Way, P.O. Box 9106, Norwood, MA 02062-9106, U.S.A. Tel: 617/329-4700 Fax: 617/326-8703 AD549–SPECIFICATIONS (@ +258C and V = +15 V dc, unless otherwise noted) S Model Min INPUT BIAS CURRENT1 Either Input, VCM = 0 V Either Input, VCM = ± 10 V Either Input at TMAX, VCM = 0 V Offset Current Offset Current at TMAX AD549J Typ Max 150 150 250 250 Min 11 50 2.2 AD549K Typ Max 75 75 100 100 10 32 32 15 INPUT VOLTAGE NOISE f = 0.1 Hz to 10 Hz f = 10 Hz f = 100 Hz f = 1 kHz f = 10 kHz 4 90 60 35 35 4 90 60 35 35 INPUT CURRENT NOISE f = 0.1 Hz to 10 Hz f = 1 kHz 0.7 0.22 INPUT IMPEDANCE Differential VDIFF = ± 1 Common Mode VCM = ± 10 OPEN-LOOP GAIN VO @ ± 10 V, RL = 10 k VO @ ± 10 V, RL = 10 k, TMIN to TMAX VO = ± 10 V, RL = 2 k VO = ± 10 V, RL = 2 k, TMIN to TMAX INPUT VOLTAGE RANGE Differential3 Common-Mode Voltage Common-Mode Rejection Ratio V = +10 V, –10 V TMIN to TMAX OUTPUT CHARACTERISTICS Voltage @ RL = 10 k, TMIN to TMAX Voltage @ RL = 2 k, TMIN to TMAX Short Circuit Current TMIN to TMAX Load Capacitance Stability G = +1 FREQUENCY RESPONSE Unity Gain, Small Signal Full Power Response Slew Rate Settling Time, 0.1% 0.01% Overload Recovery, 50% Overdrive, G = –1 0.15 1.0 1.9 20 100 100 2 10 10 15 AD549L Typ 40 40 4.2 30 1.3 INPUT OFFSET VOLTAGE2 Initial Offset Offset at TMAX vs. Temperature vs. Supply vs. Supply, TMIN to TMAX Long-Term Offset Stability 0.5 Min Max Min 60 60 2.8 20 0.85 0.3 0.25 0.4 5 32 32 5 10 10 15 AD549S Typ Max Units 75 75 100 100 fA fA 420 30 125 0.3 0.5 0.9 10 32 32 10 10 32 15 pA fA pA 0.5 2.0 15 32 50 mV mV µV/°C µV/V µV/V µV/Month 4 90 60 35 35 4 90 60 35 35 µV p-p nV/√Hz nV/√Hz nV/√Hz nV/√Hz 0.5 0.16 0.36 0.11 0.5 0.16 fA rms fA/√Hz 1013i1 1013i1 1013i1 1013i1 ΩipF 1015i0.8 1015i0.8 1015i0.8 1015i0.8 ΩipF 6 300 1000 300 1000 300 1000 300 1000 V/mV 300 100 800 250 300 100 800 250 300 100 800 250 300 100 800 250 V/mV V/mV 80 200 80 200 80 200 25 150 V/mV ± 20 +10 –10 80 76 90 80 –12 –10 15 9 20 90 80 +12 –12 +10 35 –10 15 9 4000 0.7 2 1.0 50 3 4.5 5 2 ± 20 +10 –10 100 90 20 90 80 +12 –12 +10 35 –10 15 9 4000 0.7 2 1.0 50 3 4.5 5 2 ± 20 +10 –10 100 90 20 2 1.0 50 3 4.5 5 2 –2– 90 80 +12 –12 +10 35 –10 15 6 4000 0.7 ± 20 +10 –10 0.7 2 100 90 20 V V dB dB +12 V +10 35 V mA mA 4000 pF 1.0 50 3 4.5 5 MHz kHz V/µs µs µs 2 µs REV. A AD549 Model Min POWER SUPPLY Rated Performance Operating Quiescent Current TEMPERATURE RANGE Operating, Rated Performance Storage PACKAGE OPTION TO-99 (H-08A) Chips AD549J Typ Max Min 618 0.70 65 +70 +150 0 –65 ± 15 65 0.60 0 –65 AD549K Typ Max Min 618 0.70 65 +70 +150 0 –65 ± 15 AD549JH AD549JChips AD549L Typ Max Min 618 0.70 65 +70 +150 –55 –65 ± 15 0.60 AD549KH 0.60 AD549S Typ Max Units 618 0.70 V V mA +125 +150 °C °C ± 15 AD549LH 0.60 AD549SH, AD549SH/883B NOTES 1 Bias current specifications are guaranteed after 5 minutes of operation at T A = +25°C. Bias current increases by a factor of 2.3 for every 10°C rise in temperature. 2 Input offset voltage specifications are guaranteed after 5 minutes of operation at T A = +25°C. 3 Defined as max continuous voltage between the inputs such that neither input exceeds ± 10 V from ground. Specifications subject to change without notice. All min and max specifications are guaranteed. Specifications in boldface are tested on all production units at final electrical test. Results from those tests are used to calculate outgoing quality levels. ABSOLUTE MAXIMUM RATINGS 1 METALIZATION PHOTOGRAPH Supply Voltage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ± 18 V Internal Power Dissipation . . . . . . . . . . . . . . . . . . . . . . 500 mW Input Voltage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ± 18 V2 Output Short Circuit Duration . . . . . . . . . . . . . . . . . Indefinite Differential Input Voltage . . . . . . . . . . . . . . . . . . +VS and –VS Storage Temperature Range (H) . . . . . . . . . . –65°C to +125°C Operating Temperature Range AD549J (K, L) . . . . . . . . . . . . . . . . . . . . . . . . 0°C to +70°C AD549S . . . . . . . . . . . . . . . . . . . . . . . . . . –55°C to +125°C Lead Temperature Range (Soldering 60 sec) . . . . . . . . +300°C Dimensions shown in inches and (mm). Contact factory for latest dimensions. NOTES 1 Stresses above those listed under “Absolute Maximum Ratings” may cause permanent damage to the device. This is a stress rating only and functional operation of the device at these or any other conditions above those indicated in the operational section of this specification is not implied. Exposure to absolute maximum rating conditions for extended periods may affect device reliability. 2 For supply voltages less than ± 18 V, the absolute maximum input voltage is equal to the supply voltage. CAUTION ESD (electrostatic discharge) sensitive device. Electrostatic charges as high as 4000 V readily accumulate on the human body and test equipment and can discharge without detection. Although the AD549 features proprietary ESD protection circuitry, permanent damage may occur on devices subjected to high energy electrostatic discharges. Therefore, proper ESD precautions are recommended to avoid performance degradation or loss of functionality. REV. A –3– WARNING! ESD SENSITIVE DEVICE AD549–Typical Characteristics +VIN 10 –VIN 5 0 +25°C RL = 10k 5 10 15 SUPPLY VOLTAGE ± V –VOUT 10 5 5 10 15 SUPPLY VOLTAGE ± V 500 110 100 90 20 3000 1000 ∆ |VOS| – µV –25 5 35 65 TEMPERATURE – °C 95 Figure 7. Open-Loop Gain vs. Temperature 125 300 0 30 50 25 45 15 5 10 15 SUPPLY VOLTAGE ± V 20 40 35 30 25 5 –55 100k Figure 6. Open-Loop Gain vs. Supply Voltage 10 100 100 1k 10k LOAD RESISTANCE – Ω 1000 –10 0 +10 +15 INPUT COMMON-MODE VOLTAGE – V 20 300 10 100 –15 Figure 5. CMRR vs. Input Common-Mode Voltage Figure 4. Quiescent Current vs. Supply Voltage 5 80 INPUT CURRENT – fA 5 10 15 SUPPLY VOLTAGE ± V 10 3000 70 0 15 Figure 3. Output Voltage Swing vs. Load Resistance OPEN-LOOP GAIN – V/mV 600 VS = ±15 VOLTS 20 20 120 COMMON-MODE REJECTION RATIO – dB AMPLIFIER QUIESCENT CURRENT – µA 0 Figure 2. Output Voltage Swing vs. Supply Voltage 700 25 0 0 20 800 OPEN-LOOP GAIN – V/mV +VOUT 15 Figure 1. Input Voltage Range vs. Supply Voltage 400 OUTPUT VOLTAGE SWING – Volts p-p 15 0 30 20 OUTPUT VOLTAGE SWING ± V INPUT VOLTAGE ± V 20 20 0 0 1 2 3 4 5 WARM-UP TIME – Minutes Figure 8. Change in Offset Voltage vs. Warm-Up Time –4– 6 7 –10 –5 0 5 10 COMMON-MODE VOLTAGE ± V Figure 9. Input Bias Current vs. Common-Mode Voltage REV. A AD549 50 40 35 30 25 120 100 80 60 40 0 5 10 15 POWER SUPPLY VOLTAGE ± V 10 20 80 80 35 60 60 40 40 20 20 0 PHASE MARGIN – ° 40 OUTPUT VOLTAGE SWING – V 100 –20 –40 10 100 1k 10k 100k FREQUENCY – Hz 1M 100 1k 10 10Hz BANDWIDTH 1 AMPLIFIER GENERATED NOISE 100k 1M 10M 100M 1G 10G 100G SOURCE RESISTANCE – Ω Figure 12. Noise vs. Source Resistance –40 10M 100 80 30 60 25 20 15 40 20 10 0 5 –20 0 10 Figure 13. Open-Loop Frequency Response 100 1k 10k FREQUENCY – Hz 100k 1M Figure 14. Large Signal Frequency Response 10 100 1k 10k 100k FREQUENCY – Hz OUTPUT VOLTAGE SWING – V 100 80 + SUPPLY 60 40 – SUPPLY 20 10mV 5 5mV 1mV 0 10mV 5mV –5 1mV 0 –20 –10 10 100 1k 10k 100k FREQUENCY – Hz 1M 10M Figure 16. PSRR vs. Frequency REV. A 0 1 2 3 SETTLING TIME – µs 4 Figure 17. Output Voltage Swing and Error vs. Settling Time –5– 1M 10M Figure 15. CMRR vs. Frequency 10 120 PSRR – dB 100 10k Figure 11. Input Voltage Noise Spectral Density 100 –20 1kHz BANDWIDTH RESISTOR JOHNSON NOISE 1k FREQUENCY – Hz Figure 10. Input Bias Current vs. Supply Voltage 0 10k WHENEVER JOHNSON NOISE IS GREATER THAN AMPLIFIER NOISE, AMPLIFIER NOISE CAN BE CONSIDERED NEGLIGIBLE FOR THE APPLICATION 0.1 20 20 OPEN LOOP GAIN – dB 140 CMRR – dB INPUT CURRENT – fA 45 100k INPUT NOISE VOLTAGE – µV p-p NOISE SPECTRAL DENSITY – nV/√Hz 160 5 AD549 Figure 18. Unity Gain Follower Figure 19. Unity Gain Follower Large Signal Pulse Response Figure 20. Unity Gain Follower Small Signal Pulse Response Figure 21. Unity Gain Inverter Figure 22. Unity Gain Inverter Large Signal Pulse Response Figure 23. Unity Gain Inverter Small Signal Pulse Response The AD549 has been optimized for low input current and offset voltage. Careful attention to how the amplifier is used will reduce input currents in actual applications. The amplifier operating temperature should be kept as low as possible to minimize input current. Like other JFET input amplifiers, the AD549’s input current is sensitive to chip temperature, rising by a factor of 2.3 for every 10°C rise. This is illustrated in Figure 24, a plot of AD549 input current versus ambient temperature. 1nA 100pA 10pA 1pA However, heavy output loads can cause a significant increase in chip temperature and a corresponding increase in input current. Maintaining a minimum load resistance of 10 Ω is recommended. Input current versus additional power dissipation due to output drive current is plotted in Figure 25. 6.0 NORMALIZED INPUT BIAS CURRENT MINIMIZING INPUT CURRENT 5.0 BASED ON TYPICAL IB = 40fA 4.0 3.0 2.0 1.0 100fA 0 25 50 75 100 125 150 175 200 ADDITIONAL INTERNAL POWER DISSIPATION – mW 10fA Figure 25. AD549 Input Bias Current vs. Additional Power Dissipation 1fA –55 –25 5 35 65 TEMPERATURE – °C 95 125 CIRCUIT BOARD NOTES There are a number of physical phenomena that generate spurious currents that degrade the accuracy of low current measurements. Figure 26 is a schematic of an I-to-V converter with these parasitic currents modeled. Finite resistance from input lines to voltages on the board, modeled by resistor RP, results in parasitic leakage. Insulation resistance of over 1015 Ω must be maintained between the amplifier’s signal and supply lines in order to capitalize on the AD549’s low input currents. Standard PC board material Figure 24. AD549 Input Bias Current vs. Ambient Temperature On-chip power dissipation will raise chip operating temperature causing an increase in input bias current. Due to the AD549’s low quiescent supply current, chip temperature when the (unloaded) amplifier is operated with 15 V supplies, is less than 3°C higher than ambient. The difference in input current is negligible. –6– REV. A AD549 does not have high enough insulation resistance. Therefore, the AD549’s input leads should be connected to standoffs made of insulating material with adequate volume resistivity (e.g., Teflon*). The surface of the insulator’s surface must be kept clean in order to preserve surface resistivity. For Teflon, an effective cleaning procedure consists of swabbing the surface with high-grade isopropyl alcohol, rinsing with deionized water, and baking the board at 80°C for 10 minutes. mized. Input capacitance can substantially degrade signal bandwidth and the stability of the I-to-V converter. The case of the AD549 is connected to Pin 8 so that it can be bootstrapped near the input potential. This minimizes pin leakage and input common-mode capacitance due to the case. Guard schemes for inverting and noninverting amplifier topologies are illustrated in Figures 28 and 29. Figure 28. Inverting Amplifier with Guard Figure 26. Sources of Parasitic Leakage Currents In addition to high volume and surface resistivity, other properties are desirable in the insulating material chosen. Resistance to water absorption is important since surface water films drastically reduce surface resistivity. The insulator chosen should also exhibit minimal piezoelectric effects (charge emission due to mechanical stress) and triboelectric effects (charge generated by friction). Charge imbalances generated by these mechanisms can appear as parasitic leakage currents. These effects are modeled by variable capacitor CP in Figure 26. The table in Figure 27 lists various insulators and their properties.1 Material Volume Resistivity (V–CM) Teflon* Kel-F** Sapphire Polyethylene Polystyrene Ceramic Glass Epoxy PVC Phenolic 1017–1018 1017–1018 1016–1018 1014–1018 1012–1018 1012–1014 1010–1017 1010–1015 105–1012 Minimal Triboelectric Effects W W M M W W W G W Minimal Resistance Piezoelectric to Water Effects Absorption W M G G M M M M G G G G M M W W G W Figure 29. Noninverting Amplifier with Guard Other guidelines include keeping the circuit layout as compact as possible and input lines short. Keeping the assembly rigid and minimizing sources of vibration will reduce triboelectric and piezoelectric effects. All precision high impedance circuitry requires shielding against interference noise. Low noise coax or triax cables should be used for remote connections to the input signal lines. OFFSET NULLING The AD549’s input offset voltage can be nulled by using balance Pins 1 and 5, as shown in Figure 30. Nulling the input offset voltage in this fashion will introduce an added input offset voltage drift component of 2.4 µV/°C per millivolt of nulled offset (a maximum additional drift of 0.6 µV/°C for the AD549K, 1.2 µV/°C for the AD549L, 2.4 µV/°C for the AD549J). G–Good with Regard to Property M–Moderate with Regard to Property W–Weak with Regard to Property Figure 27. Insulating Materials and Characteristics Guarding the input lines by completely surrounding them with a metal conductor biased near the input lines’ potential has two major benefits. First, parasitic leakage from the signal line is reduced since the voltage between the input line and the guard is very low. Second, stray capacitance at the input node is miniFigure 30. Standard Offset Null Circuit 1 Electronic Measurements, pp. 15–17, Keithley Instruments, Inc., Cleveland, Ohio, 1977. *Teflon is a registered trademark of E.I. DuPont Co. **Kel-F is a registered trademark of 3-M Company. REV. A The approach in Figure 31 can be used when the amplifier is used as an inverter. This method introduces a small voltage referenced to the power supplies in series with the amplifier’s –7– AD549 positive input terminal. The amplifier’s input offset voltage drift with temperature is not affected. However, variation of the power supply voltages will cause offset shifts. In an inverting configuration, the differential input capacitance forms a pole in the circuit’s loop transmission. This can create peaking in the ac response and possible instability. A feedback capacitance can be used to stabilize the circuit. The inverter pulse response with RF and RS equal to 1 MΩ appears in Figure 34. Figure 35 shows the response of the same circuit with a I pF feedback capacitance. Typical differential input capacitance for the AD549 is 1 pF. COMMON-MODE INPUT VOLTAGE OVERLOAD Figure 31. Alternate Offset Null Circuit for Inverter AC RESPONSE WITH HIGH VALUE SOURCE AND FEEDBACK RESISTANCE Source and feedback resistances greater than 100 kΩ will magnify the effect of input capacitances (stray and inherent to the AD549) on the ac behavior of the circuit. The effects of common-mode and differential input capacitances should be taken into account since the circuit’s bandwidth and stability can be adversely affected. The rated common-mode input voltage range of the AD549 is from 3 V less than the positive supply voltage to 5 V greater than the negative supply voltage. Exceeding this range will degrade the amplifier’s CMRR. Driving the common-mode voltage above the positive supply will cause the amplifier’s output to saturate at the upper limit of output voltage. Recovery time is typically 2 µs after the input has been returned to within the normal operating range. Driving the input common-mode voltage within 1 V of the negative supply causes phase reversal of the output signal. In this case, normal operation is typically resumed within 0.5 µs of the input voltage returning within range. Figure 34. Inverter Pulse Response with 1 MΩ Source and Feedback Resistance Figure 32. Follower Pulse Response from 1 MΩ Source Resistance, Case Not Bootstrapped Figure 35. Inverter Pulse Response with 1 MΩ Source and Feedback Resistance, 1 pF Feedback Capacitance DIFFERENTIAL INPUT VOLTAGE OVERLOAD Figure 33. Follower Pulse Response from 1 MΩ Source Resistance, Case Bootstrapped In a follower, the source resistance and input common-mode capacitance form a pole that limits the bandwidth to 1/2 π RSCS. Bootstrapping the metal case by connecting Pin 8 to the output minimizes capacitance due to the package. Figures 32 and 33 show the follower pulse response from a 1 MΩ source resistance with and without the package connected to the output. Typical common-mode input capacitance for the AD549 is 0.8 pF. A plot of the AD549’s input currents versus differential input voltage (defined as VIN+ –VIN–) appears in Figure 36. The input current at either terminal stays below a few hundred femtoamps until one input terminal is forced higher than 1 V to 1.5 V above the other terminal. Under these conditions, the input current limits at 30 µA. –8– REV. A AD549 100µ IIN– 10µ IIN+ INPUT CURRENT – Amps 1µ 100n 10n 1n Figure 39. Input Voltage Clamp with Diodes 100p 10p SAMPLE AND DIFFERENCE CIRCUIT TO MEASURE ELECTROMETER LEAKAGE CURRENTS 1p 100f 10f –5 –4 –3 –2 –1 0 1 2 3 4 5 DIFFERENTIAL INPUT VOLTAGE – V (VIN– – VIN–) Figure 36. Input Current vs. Differential Input Voltage INPUT PROTECTION The AD549 safely handles any input voltage within the supply voltage range. Subjecting the input terminals to voltages beyond the power supply can destroy the device or cause shifts in input current or offset voltage if the amplifier is not protected. A protection scheme for the amplifier as an inverter is shown in Figure 37. RP is chosen to limit the current through the inverting input to 1 mA for expected transient (less than 1 second) overvoltage conditions, or to 100 µA for a continuous overload. Since RP is inside the feedback loop, and is much lower in value than the amplifier’s input resistance, it does not affect the inverter’s dc gain. However, the Johnson noise of the resistor will add root sum of squares to the amplifier’s input noise. There are a number of methods used to test electrometer leakage currents, including current integration and direct current to voltage conversion. Regardless of the method used, board and interconnect cleanliness, proper choice of insulating materials (such as Teflon or Kel-F), correct guarding and shielding techniques and care in physi-cal layout are essential to making accurate leakage measurements. Figure 40 is a schematic of the sample and difference circuit. It uses two AD549 electrometer amplifiers (A and B) as current-to voltage converters with high value (1010 Ω) sense resistors (RSa and RSb). R1 and R2 provide for an overall circuit sensitivity of 10 fA/mV (10 pA full scale). CC and CF provide noise suppression and loop compensation. CC should be a low leakage polystyrene capacitor. An ultralow leakage Kel-F test socket is used for contacting the device under test. Rigid Teflon coaxial cable is used to make connections to all high impedance nodes. The Figure 37. Inverter with Input Current Limit In the corresponding version of this scheme for a follower, shown in Figure 38, RP and the capacitance at the positive input terminal will produce a pole in the signal frequency response at a f = 1/2 π RC. Again, the Johnson noise RP will add to the amplifier’s input voltage noise. Figure 38. Follower with Input Current Limit Figure 39 is a schematic of the AD549 as an inverter with an input voltage clamp. Bootstrapping the clamp diodes at the inverting input minimizes the voltage across the clamps and keeps the leakage due to the diodes low. Low leakage diodes, such as the FD333’s should be used, and should be shielded from light to keep photocurrents from being generated. Even with these precautions, the diodes will measurably increase the input current and capacitance. REV. A Figure 40. Sample and Difference Circuit for Measuring Electrometer Leakage Currents –9– AD549 use of rigid coax affords immunity to error induced by mechanical vibration and provides an outer conductor for shielding. The entire circuit is enclosed in a grounded metal box. The test apparatus is calibrated without a device under test present. A five minute stabilization period after the power is turned on is required. First, VERR1 and VERR2 are measured. These voltages are the errors caused by offset voltages and leakage currents of the current to voltage converters. VERR1 = 10 (VOSA – IBA × RSa) VERR2 = 10 (VOSB – IBB × RSb) Figure 42. Photodiode Preamp DC Error Sources Once measured, these errors are subtracted from the readings taken with a device under test present. Amplifier B closes the feedback loop to the device under test, in addition to providing current to voltage conversion. The offset error of the device under test appears as a common-mode signal and does not affect the test measurement. As a result, only the leakage current of the device under test is measured. Input current, IB, will contribute an output voltage error, VE1, proportional to the feedback resistance: VE1 = IB × RF The op amp’s input voltage offset will cause an error current through the photodiode’s shunt resistance, RS: VA – VERR1 = 10[RSa × IB(+)] I = VOS/RS VX – VERR2 = 10[RSb × IB(–)] Although a series of devices can be tested after only one calibration measurement, calibration should be updated periodically to compensate for any thermal drift of the current to voltage converters or changes in the ambient environment. Laboratory results have shown that repeatable measurements within 10 fA can be realized when this apparatus is properly implemented. These results are achieved in part by the design of the circuit, which eliminates relays and other parasitic leakage paths in the high impedance signal lines, and in part by the inherent cancellation of errors through the calibration and measurement procedure. PHOTODIODE INTERFACE The AD549’s low input current and low input offset voltage make it an excellent choice for very sensitive photodiode preamps (Figure 41). The photodiode develops a signal current, IS equal to: IS = R × P where P is light power incident on the diode’s surface in Watts and R is the photodiode responsivity in Amps/Watt. RF converts the signal current to an output voltage: VOUT = RF × IS Figure 41. Photodiode Preamp DC error sources and an equivalent circuit for a small area (0.2 mm square) photodiode are indicated in Figure 42. The error current will result in an error voltage (VE2) at the amplifier’s output equal to: VE2 = ( I + RF/RS) VOS Given typical values of photodiode shunt resistance (on the order of 109 Ω), RF/RS can easily be greater than one, especially if a large feedback resistance is used. Also, RF/RS will increase with temperature, as photodiode shunt resistance typically drops by a factor of two for every 10°C rise in temperature. An op amp with low offset voltage and low drift must be used in order to maintain accuracy. The AD549K offers guaranteed maximum 0.25 mV offset voltage, and 5 mV/°C drift for very sensitive applications. Photodiode Preamp Noise Noise limits the signal resolution obtainable with the preamp. The output voltage noise divided by the feedback resistance is the minimum current signal that can be detected. This minimum detectable current divided by the responsivity of the photodiode represents the lowest light power that can be detected by the preamp. Noise sources associated with the photodiode, amplifier, and feedback resistance are shown in Figure 43; Figure 44 is the spectral density versus frequency plot of each of the noise source’s contribution to the output voltage noise (circuit parameters in Figure 42 are assumed). Each noise source’s rms contribution to the total output voltage noise is obtained by integrating the square of its spectral density function over frequency. The rms value of the output voltage noise is the square root of the sum of all contributions. Minimizing the total area under these curves will optimize the preamplifier’s resolution for a given bandwidth. The photodiode preamp in Figure 41 can detect a signal current of 26 fA rms at a bandwidth of 16 Hz, which assuming a photodiode responsivity of 0.5 A/W, translates to a 52 fW rms minimum detectable power. The photodiode used has a high source resistance and low junction capacitance. CF sets the signal bandwidth with RF and also limits the “peak” in the noise gain that multiplies the op amp’s input voltage noise contribution. A single pole filter at the amplifier’s output limits the op amp’s output voltage noise bandwidth to 26 Hz, a frequency comparable to the signal bandwidth. This greatly improves the preamplifier’s signal to noise ratio (in this case, by a factor of three). –10– REV. A AD549 tracter section’s gain for positive and negative inputs matched over temperature. Frequency compensation is provided by R11, R12, and C1 and C2. The bandwidth of the circuit is 300 kHz at input signals greater than 50 µA, and decreases smoothly with decreasing signal levels. To trim the circuit, set the input currents to 10 µA and trim A3’s offset using the amplifier’s trim potentiometer so the output equals 0. Then set I1 to 1 µA and adjust the output to equal 1 V by trimming R10. Additional offset trims on the amplifiers A1 and A2 can be used to increase the voltage input accuracy and dynamic range. Figure 43. Photodiode Preamp Noise Sources The very low input current of the AD549 makes this circuit useful over a very wide range of signal currents. The total input current (which determines the low level accuracy of the circuit) is the sum of the amplifier input current, the leakage across the compensating capacitor (negligible if polystyrene or Teflon capacitor is used), and the collector to collector, and collector to base leakages of one side of the dual log transistors. The magnitude of these last two leakages depend on the amplifier’s input offset voltage and are typically less than 10 fA with 1 mV offsets. The low level accuracy is limited primarily by the amplifier’s input current, only 60 fA maximum when the AD549L is used. Figure 44. Photodiode Preamp Noise Sources’ Spectral Density vs. Frequency Log Ratio Amplifier Logarithmic ratio circuits are useful for processing signals with wide dynamic range. The AD549L’s 60 fA maximum input current makes it possible to build a log ratio amplifier with 1% log conformance for input current ranging from 10 pA to 1 mA, a dynamic range of 160 dB. The log ratio amplifier in Figure 45 provides an output voltage proportional to the log base 10 of the ratio of the input currents I1 and I2. Resistors R1 and R2 are provided for voltage inputs. Since NPN devices are used in the feedback loop of the frontend amplifiers that provide the log transfer function, the output is valid only for positive input voltages and input currents. The input currents set the collector currents IC1 and IC2 of a matched pair of log transistors Q1 and Q2 to develop voltages VA and VB: VA, B = – (kT/q) ln IC/IES where IES is the transistors’ saturation current. The difference of VA and VB is taken by the subtractor section to obtain: Figure 45. Log Ratio Amplifier VC = (kT/q) ln (IC2/IC1) VC is scaled up by the ratio of (R9 + R10)/R8, which is equal to approximately 16 at room temperature, resulting in the output voltage: VOUT = 1 × log (IC2/IC1) V. R8 is a resistor with a positive 3500 ppm/°C temperature coefficient to provide the necessary temperature compensation. The parallel combination of R15 and R7 is provided to keep the sub REV. A The effects of the emitter resistance of Q1 and Q2 can degrade the circuit’s accuracy at input currents above 100 µA. The networks composed of R13, D1, R16, and R14, D2, R17 compensate for these errors, so that this circuit has less than 1% log conformance error at 1 mA input currents. The correct value for R13 and R14 depends on the type of log transistors used. 49.9 kΩ resistors were chosen for use with LM394 transistors. Smaller resistance values will be needed for smaller log transistors. –11– The pH probe output is ideally zero volts at a pH of 7 independent of temperature. The slope of the probe’s transfer function, though predictable, is temperature dependent (–54.2 mV/pH at 0 and –74.04 mV/pH at 100°C). By using an AD590 temperature sensor and an AD535 analog divider, an accurate temperature compensation network can be added to the basic pH probe amplifier. The table in Figure 47 shows voltages at various points and illustrates the compensation. The AD549 is set for a noninverting gain of 13.51. The output of the AD590 circuitry (point C) will be equal to 10 V at 100°C and decrease by 26.8 mV/°C. The output of the AD535 analog divider (point D) will be a temperature compensated output voltage centered at zero volts for a pH of 7, and having a transfer function of –1.00 V/pH unit. The output range spans from –7.00 V (pH = 14) to +7.00 V (pH = 0). TEMPERATURE COMPENSATED pH PROBE AMPLIFIER A pH probe can be modeled as a mV-level voltage source with a series source resistance dependent upon the electrode’s composition and configuration. The glass bulb resistance of a typical pH electrode pair falls between 106 and 109 Ω. It is therefore important to select an amplifier with low enough input currents such that the voltage drop produced by the amplifier’s input bias current and the electrode resistance does not become an appreciable percentage of a pH unit. The circuit in Figure 46 illustrates the use of the AD549 as a pH probe amplifier. As with other electrometer applications, the use of guarding, shielding, Teflon standoffs, etc., is a must in order to capitalize on the AD549’s low input current. If an AD549L (60 fA max input current) is used, the error contributed by input current will be held below 60 µV for pH electrode source impedances up to 109 Ω. Input offset voltage (which can be trimmed) will be below 0.5 mV. PROBE TEMP A (PROBE OUTPUT) B (A 3 13.51) C (590 OUTPUT) D (10 B/C) 0 258C 378C 608C 1008C 54.20 mV 59.16 mV 61.54 mV 66.10 mV 74.04 mV 0.732 V 0.799 V 0.831 V 0.893 V 1.000 V 7.32 V 7.99 V 8.31 V 8.93 V 10.00 V 1.00 V 1.00 V 1.00 V 1.00 V 1.00 V C1073a–10–10/87 AD549 Figure 47. Table Illustrating Temperature Compensation Figure 46. Temperature Compensated pH Probe Amplifier OUTLINE DIMENSIONS Dimensions shown in inches and (mm). 0.370 (9.40) 0.335 (8.50) 0.335 (8.50) 0.305 (7.75) 0.040 (1.0) MAX SEATING PLANE INSULATION 0.05 (1.27) MAX 8 LEADS 0.2 (5.1) TYP 3 0.185 (4.70) 0.165 (4.19) REFFERENCE PLANE 2 4 8 6 1 PRINTED IN U.S.A. TO-99 (H) Package 45° EQUALLY SPACED 5 7 0.500 (12.70) MIN 0.019 (0.48) DIA 0.016 (0.41) 0.034 (0.86) 0.028 (0.41) 0.045 (1.1) 0.020 (0.51) BOTTOM VIEW –12– REV. A