AD8067 Current to Voltage Converter

[ocw]

I just completed construction and initial testing of a prototype current to voltage converter.  The first test results look good!

I decided to use an AD8067 after not being satisfied with the single polarity of a OPA380.  I found that the first AD8067ARTZ which I used exceeded its specification sheet open loop gain figure by enough to permit the accurate use of a 10M ohm feedback resistor.  That provided resolution of the readings down to 0.01 pA while using the AD8067 circuit feeding my 34401A meter.  Admittedly the "meter zero" using the 10M resistor was 13.02 pA (8.61 pA when using a 1M feedback resistor).  Correcting for this zero, the accuracy of the current readings using the AD8067 for readings down to 100 pA was under 1%.

It was tough for me to find an accurate current to use as a reference standard for these measurements.  The 10 nA resolution on my 34401A obviously was inadequate.  I didn't have a more appropriate meter available (that's the reason for this project).  So my calibration was made using an accurately measured voltage (via the 34401A) and 300M, 1G, 5G and 10G ohm 1% resistors.  I also have a 40G 5% resistor available.  However, I didn't consider its accuracy as being adequate.  So, 1% accuracy was the best accuracy which I could measure.  Because of that, the actual accuracy of my AD8067 circuit is probably better than what I measured.  For higher current levels I have a variety of 0.1, 0.01% and better resistors which were used.

I didn't try and measure the burden voltage or what should be called the burden resistance of my AD8067 circuit.  It obviously wasn't significant at least compared to the limited accuracy of my reference current.  Correcting for the meter zero was the only post reading change which was made.  When that correction was not made, the converter had better than 1% accuracy for 900+ pA readings.  I haven't decided if I am going to add a zeroing circuit to the converter or not.  It is probably simpler to simply let the 34401A zero the reading when there isn't any current through the input of the converter.

I have selectable 10M, 1M, 100k, 10k and 1k ohm feedback resistors in my prototype circuit.  That permits measurement of currents as low as those mentioned up to 10 mA.  For the 1M (1 uA = 1 V output) and 1k (1 mA = 1 V output) I have precision calibration added.  For example, I have a 976k resistor in series with a 50k trimmer instead of a 1M resistor.  That permits precise calibration of that range.  There's a similar combination in place of the 1k resistor.  So, hopefully those two ranges are more accurate than with the 0.01% resistors used on the 10k and 100k ranges.

[TiN]

Schematics, PCB photos, test rig setup photos, test data charts or any data you can post does miracles on thread posts like yours 

[ocw]

My AD8067 current to voltage converter is still a work in progress.  Its schematic is still just on a "Dave Pad" [TM].  It is a real world hand-drawn schematic, complete with additions, corrections, etc.  Its prototype board reflects some design changes.  It is electrically sound with close mounted bypass capacitors but doesn't have a finished product appearance.  I still need to add a power supply to it rather than relying on my bench supply.  I'm probably going to add a Traco TDN1-1233WI DC/DC converter to provide +/-12 VDC with isolation from either a single battery or AC adapter.

I have found that my circuit's zeroing issue was actually a noise problem.  When you have a circuit which provides 1 VDC output when fed a 100 nA current you have noise possibilities when measuring pA currents.  I've shortened the test leads, done some other things to reduce the noise and have rerun my circuit's accuracy tests.  The results are shown on the attached spreadsheet.  The AD8067 outputs shown didn't have any extra changes made to reflect zeroing.  Once the noise level was reduced no extra zeroing was required.

The more accurate readings shown with the 1k resistor used for about a 5 mA measurement, and with the 1M resistor and 5 uA reflect that the converter had been hand calibrated very near those points.  The 0.01% accuracy at about 50 nA using the 100M resistor reflects the precision resistors used and more so the luck in the extra precision in the resistor actually used.  No extra calibration was involved with that reading.  Most of the other accuracy figures reflect about I would expect given the accuracy of the resistors used and the circuit which was used.

I want to put some more effort into further noise reduction.  It was still too high for readings under 500 pA.  But, that noise may be coming from the power supply producing the current which was measured.  I need to repeat those tests using battery voltage.  I currently have a 1.1 pF capacitor in parallel with the feedback resistor.  I may look into increasing the value of that and perhaps adding a mild noise filter to the AD8067's output.

The schematic of the circuit is not anything special.  The important parts are simply the IC and selectable feedback resistors.  It mirrors that found in a lot of publications.  The circuit is simple and left me wondering why I didn't do this long ago.  Examples of the resistors used for calibration include the Caddock USF340-1.00M-0.01%-5ppm and the Ohmite SM103035007FE 5G ohms 1% resistor.  The Caddock resistor obviously has its important specs as part of its part number.

Anyway, while not perfect, I've found this to be a useful circuit which can be built at a low cost.

[Kleinstein]

The AD8067 is a rather unusual choice for low current measurement. A more natural choice would be something like am LMC6001 or TLC271. There is no real need for high speed - such a fast OP could be more sensitive to RF pickup and also reacts more sensitive to capacitive load at the output.

Generally one need a resistor and / or inductor at the input to avoid excessive capacity at the inverting node. Also at least a little capacitance in feedback is needed to compensate for the input's capacity and maybe do some filtering, so that 50 Hz pickup will not so easy saturate the amplifier.

With a low power OP, operation form 9 V batteries is realistic, at least for the lower current ranges up to about 1 mA.

[ocw]

The AD8067 is a rather unusual choice for low current measurement.

You aren't being imaginative enough!  I've just been mentioning DC calibration.  I've just measured 50 nA of 400 Hz AC with pretty good accuracy!  That's not either the upper frequency or minimum current possible.  I've just started the AC evaluation.  I was using a 1M ohm feedback resistor for that measurement.

I wanted a dual voltage supply IC so that it would make AC measurements easier and avoid the need to worry about DC polarity.  I preferred a +/-12 VDC supply capability so that +/-10 VDC outputs are possible for greater dynamic range possibilities.  Plus, it is an easy number to remember for a maximum reading.  Analog Devices says on the AD8067 spec sheet, "The combination of low noise, DC precision, and high bandwidth makes the AD8067 uniquely suited for wideband, very high input impedance, high gain buffer applications. It is also useful in wideband transimpedance applications, such as a photodiode interface, that require very low input currents and DC precision."   Plus, it seemed like my best choice of the 152 different model op amps in my current inventory (I've done a lot of op amp evaluations lately).

The noise on my circuit is not 60 Hz noise, but rather a high frequency noise which is causing sub-pA current jumps in my readings.  I may be trying for unobtainable perfection as far as eliminating the noise is concerned.

[ocw]

I measured the frequency response of my current to voltage converter.  For the different feedback resistors the response relative to 100 Hz was:

1k      -3 dB  >20 MHz
10k    -3 dB    3.7 MHz
100k   -3 dB   370 kHz
1M      -3 dB   37 kHz    -1% 5 kHz     -3% 9 kHz
10M    -3 dB    4.35 kHz

That's higher than I was expecting.  I didn't want to change from my function generator to my RF generator to measure the exact figure for the 1k resistor.  It is obviously very like to be 37 MHz.  While its accuracy won't be 1% there, it is nice to be able to measure uA RF currents at 30 MHz.  It should have respectable accuracy up to 10 MHz.

Not surprisingly, noise was more of an issue with low level AC current measurements.  While accurate 50 nA readings were possible up to about 1 kHz, the noise level was too much for accurate 5 nA and below AC readings.  I didn't spend much time trying to minimize that.  The board needs to be mounted in a metal case before I'm going to worry too much about it.

[ocw]

I've slowly been improving the design of my circuit using an AD8067.  I've reduced the noise level for AC current to voltage conversion and added the selection of a 1G ohm resistor for lower current DC conversion.  There isn't significant AC frequency response with the 1G resistor but the DC sensitivity is much better.  It's nice to have 0.1 fA resolution even if the noise level prevents that digit from being meaningful.  The attachment shows the stability of a 17.45 pA reading as determined by using a 100 G ohm 1% resistor connected to 1.745 volts.

I also compared the reverse current of two diodes connected to 15 VDC and measured it as 4.6617 pA for a BAV199 diode as compared to 4.3597 nA for a BAV99.

I still need to make some additional refinements and remove some of its prototype appearance.

[Rerouter]

Another person that would be very interested in even the davecad of something like this,