offset problem of AD8629 OPAMP

Dear Sir/Madam,

Our customer have met one trouble as following.

(This discuss came from another sled cause my miss understood)

On my enclosed schematic Fig-1.pdf,

step-1 power off and keep waiting around 30 sec

step-2 power on

step-3 around 90msec offset voltage appear at out-1 pin

step-4 power off and on but the 90mV offset is not cleared

Also refer to my enclosed time chart "tme chart when power off and on.pdf".


*1 ; 30sec is not certainly.

*3 ; in some cases, it return to expected value

After once this trouble happen, out-2 is up and dowm on offset voltage 900mV(x10 gained value)


When we work with above action, does your AD8629 do such a motion?or not.

If he replace the thermopile(TP) to fixed resister 175kohm, then the result was same.

Please advice us before your holiday!!

Thank you in advance Kaos
  • Hi bcard,

    No, I could not fix this trouble.

    So I had recommended the another circuit for fix this trouble to our customer.

    It was put the chopping circuit. It means REDESIGN for the current circuit.. 

    In my case, thermopile have around 200kphm resistance. For AD8629, this value didnt bring a good result.

    To make matters worse, SOMETIMES happen this trouble when we power on -off frequently not always.

    So difficult to catch the reason why this trouble has comin....

    Thanks Kaos

  • Hi Bruce,

    I recommend to test by a lot of qty samples AD8552s.

    In fact I replaced it to another AD8629 so the trouble doesn't happen.

    I also tried to replace to ADA4522-x.

    But it brought the similar result.

    Note zero drift AMP works tricky motion. Maybe you need enough test.

    Thanks Kaos

  • Did you resolve the AD8629 offset issue?  We have found a similar problem when using the AD8629 in a thermopile amplifier: the offset changes after cycling the power to the circuit.  We use 2 identical amplifier circuits in a thermopile based gas detection.  The gas concentration is calculated from the ratio of the 2 signals.  The problem is worst when one of the offset increases while the other decreases.  The decrease does not occur very often, but when it does, the calculated gas concentration changes by as much as 150 ppm.    

  • Thanks for the quick response Harry.

    Sorry about the meager details.  I thought I was addressing Kaos.  I wanted to know if his customer resolved their problem before I started a new thread.


    Gas concentration measurement using a 2 ch IR sensor

    Circuit used:

    Each of the 2 channels is amplified by an identical amp circuit per above, but with a reduced gain: 6700 instead of 10,000.

    In our measurement range, the concentration is considered proportional to the Ln(Vch1 / Vch2)

    where Vch1 and Vch2 are the voltage signals from each amp circuit.

    The sensitivity of such an arrangement is low, and a slight change in the Vch1/Vch2 ratio yields a large change in the calculated gas concentration.


    After a power cycle, the output of the amp circuit [gain of 6,700] may jump either ~10 mV higher or ~10 mV lower than before.

    It does not happen every time.  Once in about 10 times, the 2 output circuits jump in opposite directions causing a large change in the gas concentration calculation.


    • The offset does not merely change by up to 10 mV, but jumps by 10 mV up or down from the baseline.
    • After most power cycles, the offset is at baseline.
    • An AD8552 in the same circuit does not exhibit the offset jumps, but the AD8629 is cheaper...


    The problem is not the 10 mV offset which is well within the published offset [after a gain of 6700].

    The problem is the uncertainty of a jump occurring and its direction after a power cycle.

    Any thoughts would be welcome.

  • Bruce,

      It's hard to trouble shoot a problem when there is no data provide other than "...the offset changes after power

    cyciling...."  How much?  If you want to start a separate thread with a schematic and some numbers, perhaps we

    can help.  You could also mmeasure only the op amp as a follower and see if it changes less than 10 uV over 

    temperature, then it meets spec.