Author Topic: DC op-amp drift  (Read 2015 times)

0 Members and 1 Guest are viewing this topic.

Offline ricko_ukTopic starter

  • Super Contributor
  • ***
  • Posts: 1015
  • Country: gb
DC op-amp drift
« on: November 28, 2019, 02:47:45 pm »
Hi,
We have two blocks shown in the schematic below (apologies for not posting the full schematic but we breadboarded it and don't have access to the CAD at the moment).
The idea is to have a two stage amplify and shift.
The output of the first stage feeds into the input of the second. BTW, we do have a cap between GND and the centre-tap of the offset pot.
The input signal into the first op-amp can vary from DC to max 20Hz (coming from a in-amp AD623 referenced to GND) with an amplitude of up to 80mV. Amplification of every stage is only 10 to 20X.
The op-amps used are MCP6002.

When we tested the system today we noticed that just by letting the system sit so that the input is DC the output shifts quite drammatically (up to 200mV) after just a couple of minutes. Very bizarre. Any idea what it could be?

Temperature on the PCB is stable (room temperature). No high currents anywhere in the system. All components are new.

Any suggestions would be greatly appreciated.

Thanks
Rik
 

Offline razvan784

  • Regular Contributor
  • *
  • Posts: 90
  • Country: ro
Re: DC op-amp drift
« Reply #1 on: November 28, 2019, 03:45:29 pm »
Bypass caps? Parasitic oscillation?
Amplified INA or sensor offset?
 

Offline ricko_ukTopic starter

  • Super Contributor
  • ***
  • Posts: 1015
  • Country: gb
Re: DC op-amp drift
« Reply #2 on: November 28, 2019, 04:27:15 pm »
Thanks Razvan,
plenty of bypass caps :)
Parassitic oscillation I doubt because it is a low noise design especially the front end.
The sensor itself is just 2 identical precision coils driven by a single source. So if there is any offset it would be in both. They are inside eachother and potted so no mechanical shifts are possible.
As I said it is a massive drift 200mV and it takes at least a couple of minutes to notice it. If you look at the signal across 5-10 seconds you cannot really see any drift. It is also not "predictable" or repetitive in its pattern/behaviour. Sometimes it drifts up some other times down. And the timing also change, sometimes a couple of minutes other times longer to see it.

Any other ideas?

Thanks :)
 

Offline razvan784

  • Regular Contributor
  • *
  • Posts: 90
  • Country: ro
Re: DC op-amp drift
« Reply #3 on: November 28, 2019, 05:19:32 pm »
Maybe the 5V supply you're using for the offset pot is drifting. That offset gets amplified a lot as I understand. The pot itself may be drifting... Is your signal really DC, can't you use AC coupling between stages?
 
The following users thanked this post: ricko_uk

Offline thinkfat

  • Supporter
  • ****
  • Posts: 2150
  • Country: de
  • This is just a hobby I spend too much time on.
    • Matthias' Hackerstübchen
Re: DC op-amp drift
« Reply #4 on: November 28, 2019, 06:14:52 pm »
What is the temperature coefficient of these pots? You're continually heating the 1k offset pot with 5mW.
Everybody likes gadgets. Until they try to make them.
 
The following users thanked this post: ricko_uk

Offline ricko_ukTopic starter

  • Super Contributor
  • ***
  • Posts: 1015
  • Country: gb
Re: DC op-amp drift
« Reply #5 on: November 28, 2019, 06:52:40 pm »
Thank you both :)

razvan784, I checked and the 5V is steady. What and how would AC coupling help? Perhaps it might be a solution but I don't understand what you suggest. Could you please explain. Thank you :)

thinkfat, I changed them to 50K and it is still there. :)
 

Offline razvan784

  • Regular Contributor
  • *
  • Posts: 90
  • Country: ro
Re: DC op-amp drift
« Reply #6 on: November 28, 2019, 07:33:02 pm »
If your signal does not have a DC component (you mentioned some coils earlier, those usually don't generate DC) you can use AC coupling between your amplifier stages so that you don't amplify their DC offsets. That way you shouldn't need an offset adjustment pot. Otherwise the offset voltage generated by the first stage gets amplified by all the following stages and can get quite large. In that case I suggest using parts having lower offset voltages.
 
The following users thanked this post: ricko_uk

Offline Gyro

  • Super Contributor
  • ***
  • Posts: 9410
  • Country: gb
Re: DC op-amp drift
« Reply #7 on: November 28, 2019, 09:30:10 pm »
I agree, it could well be the pots, especially f the direction of drift varies.

The simple way to eliminate them from suspicion would be to measure them and then replace with approximately the same values and ratios of metal film resistors and see if you still get the drift.
Best Regards, Chris
 
The following users thanked this post: ricko_uk

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 14072
  • Country: de
Re: DC op-amp drift
« Reply #8 on: November 28, 2019, 09:31:36 pm »
Depending on the overall gain, it could be just low frequency noise of the AD623. The AD623 is not especially low noise and maybe more suitable for higher impedance source. There is quite some 1/f noise, that can look like more or less random walks.  Also external magnetic fields can have an effect, if the 2 coils don't compensate.

I would do a test with just the MCP6002 stages to see the error source.
A gain of 100-400 on a 80 mV signal would be outside the 5 V supply to the MCP6002.
 
The following users thanked this post: ricko_uk

Offline ricko_ukTopic starter

  • Super Contributor
  • ***
  • Posts: 1015
  • Country: gb
Re: DC op-amp drift
« Reply #9 on: November 29, 2019, 09:05:30 am »
Thank you Kleinstein :)

What is a better low noise part to be used instead of the AD623 in these 2 scenarios:
1) directly connected across the coils
2) as a general in-amp to be used AFTER two low noise op-amps which are used as buffer amplifiers between the coils and said intrumentation amplifier?

In both applications the frequency used to drive the coils is 200 KHz (the DC signal mentioned previously is because we have been playing with low pass filtering - also to address razvan784 comment/suggestion).

The reason for asking about option 2 is that I am thinking of using two ultra low noise buffer amplifier in front of the instrumentation amplifier (perhaps with a tiny gain?):
1) would you recommend doing so? Pros and cons (i.e. maybe more noise introduced?)
2) if yes what op-amp part would you recommend?

Thanks :)
 

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 14072
  • Country: de
Re: DC op-amp drift
« Reply #10 on: November 29, 2019, 11:45:38 am »
The classical low noise In-amp is the AD620  (about 1/5 the noise of the AD623). However one has to check the supplies and common mode range.
An important factor is the source impedance and the frequency range of interest.  The noise figure depends on the source impedance and thus different amplifier are good at low and high impedance.

A second amplification stage is far less critical. So the AD623 would be well good enough. If going with OPs the MCP6002 could be just good enough. However there are other options, at low frequency even an LM358 may be better, though not rail to rail.

Rail to rail is tricky anyway as the input stage cross over can cause quite some error. So the really useful input range is often limited even with RR OPs :-BROKE. One could consider something like TLE2021 or LT1013 if price does not matter.

200 kHz would be a totally different ball park and would need much faster amplifiers.
 
The following users thanked this post: ricko_uk

Offline ricko_ukTopic starter

  • Super Contributor
  • ***
  • Posts: 1015
  • Country: gb
Re: DC op-amp drift
« Reply #11 on: November 29, 2019, 01:14:45 pm »
Thank you Kleinstein! :)

Following all the above we decided to redesign from scratch the front end and just started a new post here: https://www.eevblog.com/forum/metrology/eddy-current-sensor-for-metal-crackdefect-sensing/

If you have any suggestions/pointers (apart from all the good ones you so kindly already offered) I would be very much appreciate! :)

Thank you again for all your help so far!! :)
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf