4
\$\begingroup\$

I am using an instrumentation amplifier (AD8220) to interface a pH probe. The pH probe produces a +/-414mV signal with large output impedance (>100MOhm). AD8220 has very low input bias current (~10pA) and low offset voltage (~500uV).

When I simulate with no pH probe impedance and a 1V common mode voltage, the output is as expected, with common mode totally removed.

enter image description here

However, when I add 100MOhm in series with my probe there is a considerable offset (~700mV), greater than the 10pA bias current can produce. I expect input impedance of the in-amp is very large (>1TOhm) and fairly balanced. Why is this offset produced?

enter image description here

\$\endgroup\$
3
  • \$\begingroup\$ First thing crossing my mind (your circumstances) was to use a JFET. I had no idea about the AD8220. But it's a JFET input opamp. So, I think you are playing in the right ballpark, anyway. I'll +1 the question, too. (I don't have an answer right now.) \$\endgroup\$
    – jonk
    Jun 16, 2022 at 23:27
  • \$\begingroup\$ Could you please share a link to the documentation of the pH probe? I'm not sure this should be a new question or an edit to this one but maybe there is a different/better way to model it. \$\endgroup\$
    – devnull
    Jun 16, 2022 at 23:58
  • 1
    \$\begingroup\$ The pH probe is Chinese made, and documentation is sparse: aliexpress.com/item/2255799999624524.html - \$\endgroup\$
    – Rherma
    Jun 17, 2022 at 0:00

2 Answers 2

3
\$\begingroup\$

Why is this offset produced?

Because this LTSpice model is taking a much larger input current than the datasheet indicates:

enter image description here

LTSpice usually has good models for AD and LT devices. Strange. Edit: new model, still large currents

According to the datasheet these would be the currents around 80C or so.

\$\endgroup\$
4
  • 1
    \$\begingroup\$ Yes - this is the answer... I did not consider the model is incorrect. Will create a ticket w/ ADI. \$\endgroup\$
    – Rherma
    Jun 16, 2022 at 23:32
  • \$\begingroup\$ @Rherma I don't think so. Just looking at the datasheet right now. I think your problem is not seeing "source imbalance" in the datasheet or following that up with an LTspice simulation that is set up correctly. \$\endgroup\$
    – jonk
    Jun 16, 2022 at 23:40
  • 1
    \$\begingroup\$ @jonk I did see that in the datasheet, and when I simulate with 100Meg on both sides of the input the common mode voltage disappears. However, surely the source impedance is in series with the device input impedance, which is several orders of magnitude larger. I am missing something, could not find any other discussion of "source imbalance" in other technical documents. \$\endgroup\$
    – Rherma
    Jun 16, 2022 at 23:43
  • \$\begingroup\$ Just informing that the above plot was obtained without the resistor (and the offset current is quite small, and almost identical to spec). \$\endgroup\$
    – devnull
    Jun 16, 2022 at 23:43
4
\$\begingroup\$

Try:

enter image description here

The data looks like this:

enter image description here

I suspect that if you start a ticket, they may suggest something similar to try out. They really are pretty good at vetting their models.

\$\endgroup\$
6
  • 1
    \$\begingroup\$ Yes, this configuration produces the expected output with acceptable common mode rejection. However, the input current to the in-amp is still unexpectedly high (almost 1nA). This is producing a large voltage drop over the 50Meg device impedance (but balanced by the other 50Meg on the inverting input). I must be missing something. \$\endgroup\$
    – Rherma
    Jun 16, 2022 at 23:50
  • 1
    \$\begingroup\$ Six of one, half dozen of the other :) I am not so sure it is a different question, since in the case you provided, the device impedance is "balanced". In reality, I am not so sure how to model the pH probe impedance. Physical construction leads me to think most of the impedance is produced by the various membranes and liquid junctions in the probe which surely is just in series with the voltage output. Whether this is an LTSpice question, or more fundamental about the operation of in-amps is to be answered. Hopefully ADI apps engineers will shine some much needed light on my question. \$\endgroup\$
    – Rherma
    Jun 16, 2022 at 23:54
  • 1
    \$\begingroup\$ @Rherma The output is certainly no longer the one you show in the question. So the presentation of your facts has changed. But sure. Agreed. Definitely talk with them! And/or just buy one and set it up and see. That's what I often did before calling in an FAE. (And yeah... just looking... an nA does seem excessive to me.) \$\endgroup\$
    – jonk
    Jun 16, 2022 at 23:56
  • 1
    \$\begingroup\$ Yes, this does reduce the common mode voltage (but by a factor of the difference in gain - 5x). Maybe this helps iron out questions regarding "source imbalance" and gain having an effect on the common mode rejection? Unsure - but I do agree, testing on the real chip would help. I do have some samples order now \$\endgroup\$
    – Rherma
    Jun 17, 2022 at 0:09
  • 1
    \$\begingroup\$ @Rherma Best wishes on the results. I'm hoping it meets what you perceive from the datasheet and that LTspice needs its model fixed. \$\endgroup\$
    – jonk
    Jun 17, 2022 at 0:13

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.