3
\$\begingroup\$

I am trying to get this biased potentiostat to work. I have attached the circuit below. I am using the Alphasense Nitric Oxide NO-A4 sensor.

For the purposes of this example, you can ignore the Aux electrode, I am only taking into account the change on the working electrode.

When placed on zero gas, the reading settles, and appears good. After zeroing, the sensor is placed on span gas (5ppm), and the reading initially quickly climbs, as you would expect. However the reading does not settle, instead it drifts upwards slowly.

After moving back to zero gas, the reading is much higher, almost that of when initially placed on span gas. I cannot figure out where this drift is coming from, and could do with a second look at my circuit.

The 3v3Bat rail is continuously maintained, and therefore the sensor bias is maintained too. All voltages look good, and the general noise floor is very low.

Main Potentiostat:

enter image description here

Bias / virtual ground generation here:

enter image description here

Thanks!

\$\endgroup\$
4
  • \$\begingroup\$ What kind of capacitors are all the 100nF ones? \$\endgroup\$
    – EinarA
    Apr 26, 2019 at 0:03
  • \$\begingroup\$ Can you supply data on D1. I tried lt4001 and got listings for a phone. \$\endgroup\$
    – EinarA
    Apr 26, 2019 at 2:08
  • \$\begingroup\$ @EinarA Caps are just bog standard X7R Ceramics. Re D1, my bad. It should read LT1004. \$\endgroup\$
    – gecko242
    Apr 26, 2019 at 7:37
  • \$\begingroup\$ This just sounds like a sensor or test setup issue, I happened to work with sensors in ppm range and keeping things clean was an awful nightmare. Everything is somewhat porous, even steel! Anyhow ruling out electronics seems within reach, wire a couple of multiturn pots in place of the sensor so to drive your amplifiers and check. \$\endgroup\$
    – carloc
    Apr 28, 2019 at 7:14

3 Answers 3

1
\$\begingroup\$

Thankyou for the useful advice. In the end, it turned out to be something really simple! (doh).

The flow over the sensor is very critical to its performance. I have found (after a conversation with the sensor manufacturer) that anything over about 0.4 L/Min causes massive issues. I was running at more like 5L/Min; Still a tiny flow but not quite tiny enough! This flow had been fine for all other types of electrochemical sensors that we use, but not this one.

Thanks again for all of your responses!

\$\endgroup\$
0
\$\begingroup\$

Now, granted I haven't thoroughly analyzed your circuit but I'm not seeing any extreme low passes or anything that would "store" a value for several seconds or minutes.

I have to assume there is some kind of problem with the sensor itself, or the way you're using it. Perhaps it takes minutes or hours to stabilize after a change in conditions?

\$\endgroup\$
1
  • \$\begingroup\$ After the initial settling period, the response time should be under 20s. There may well be an issue with my circuit, since these can be quite finnicky devices to drive correctly. \$\endgroup\$
    – gecko242
    Apr 26, 2019 at 7:39
0
\$\begingroup\$

I went to the AD/LT website and looked at all the reference diodes. There is no LT4001 but there is a LT1004 and it requires a minimum of 10 uA. R5 supplies 12.6 and R6 and R7 draw off 9, so there isn't enough for D1. Decrease R5 to 18 to 20 K.

Using ceramic caps for C5,10-13 could be a problem. Dielectric absorbtion can cause long settling times and X7R has a lot. When I started in instrumentation in the 1980s all the literature said to use film caps and that is what I have always used so I don't actually know what symptoms other caps will produce. I would recommend polypropylene in this application but don't have any experience that will say it will be a cure. NP0 are also said to have low DA but I have not used them.

Other possiblities: D1 is oscillating. It may need a cap across it, or It may refuse to work with it's anode in the air. The opamp creating the virtual ground could also be oscillating. There might be some performance quirk with the opamps you have chosen that make them misbehave. I'm not really willing to dig through the data sheets to find out if any of these are realistic.

If there is a problem with the circuit, U2 is the most likely culprit. A common shortcoming of CMOS Opamps is a limited ability to handle capacitive loads. It seems like the Pstat cell could have enough capacitance to destabilize the amp. This can be fixed by adding a low value resistor (100 ohms) between the amp output and the cell. Keep C5 where it is. I think R3,4 are unnecessary and could be replaced with a direct connection. Same with R11,12.

\$\endgroup\$
4
  • 1
    \$\begingroup\$ Hi @EinarA, thanks for doing some detective work for me ;) Whilst that is clearly a potential problem, I don't believe that it is MY problem. I have just measured the Bias and VGnd rails, and they are rock solid throughout the entire sensor range. I will change the resistors to suit, but they do not appear to be the current problem. \$\endgroup\$
    – gecko242
    Apr 26, 2019 at 8:30
  • \$\begingroup\$ Knowing nothing about potentiostats I was unsure if an error in Vbias would affect it's behavior. But without sufficient current D1 is superfilous. I ( will) have added comments about the caps to the answer. \$\endgroup\$
    – EinarA
    Apr 27, 2019 at 1:26
  • \$\begingroup\$ Thanks for your comments, interesting notes about the caps, thats some very useful info. I will do a bit more research into that and see if it could be causing my problem. That being said, I had a similar (albeit non-biased) potentiostat working just fine with X7R's. re the oscilation issue - I did originally have an oscilation problem with the Vgnd op amp due to too much capacitance on the output, but have since fixed this. The diode appears to be happy, but in the next respin I will wack a cap across it. \$\endgroup\$
    – gecko242
    Apr 27, 2019 at 16:29
  • \$\begingroup\$ The effect of the X7R caps could be very subtle but I have never done a comparison test. Although I have focused on other things, U2 is the most likely problem and I have added some comments about it to my answer. \$\endgroup\$
    – EinarA
    Apr 28, 2019 at 4:29

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.