Post Go back to editing

How to calculate offset voltage of AD628


I'm using the difference amplifier AD628 (exact model AD628ARMZ) in our circuits and we are having issues which seem to be caused by the offset voltage. To identify if the parts used here are within their spec, I would like to calculate the offset voltage based on the information given in the datasheet.

Now it is not fully clear to me, how to calculate the offset voltage given by the datasheet correctly. The datasheet gives values for the output amplifier only, for the difference amplifier only and for the combination of both.

I want to calculate the offset voltage for an operating voltage of +-15V and a common mode voltage of 43V. Furthermore, the external resistors for setting the gain of the output amplifier are chosen with 2kR and 118kR which results in a maximum gain of 60.2 (resistor tolerances are taken into account here!).

According to the formula mentioned in note 2 on page 4 of the datasheet, I should have a total output offset voltage of

V_offset = (0.1 * 43V) / (10^3.75) * 60.2 = 46.03mV

Of course, temperature has also an impact, but this is neglected for the moment.

Now my question: Is the calculation above correct for the complete amplifier (i.e. the combination of the difference amplifier and the output amplifier)? Does the formula mentioned in note 2 refer to the combination of both amplifiers?

Or, is the calculation above wrong and the 46.03mV caused by the increased common mode voltage and calculated above add up to the already existing offset voltage at V_CM = 0V (in my case would that be +-1.5mV * 60.2 = 90.3mV offset at a common mode voltage of 0V)?

  • Hi RMosig,

    Thank you for considering AD628 on your application.
    You're already on the right track with regards to getting the total offset voltage for AD628.
    The additional error formula you see on the datasheet page 4 is just the error due to your common-mode voltage:

    There are other factors that you may need to consider such as offset voltage and the effect of Bias current on your system.
    For example, looking at the offset voltage of the output amplifier and the gain you set (60.2), you'll have an additional 9.03mV offset (0.15mV*60.2). 

    Another is for the effect of the bias current you have 3nA*2kohm*60.2 that yields around 0.36mV :

    These additional offset are on top of the one you calculated at around 46.03mV that's due to the common-mode voltage. so overall, you may have around 55.42mV total offset voltage. 

    Though on your application what's the offset voltage you're seeing and what's the target accuracy? Can we see the schematic diagram in case we're missing something. If you can give us more insight on your specific application we may be able to help you out further. 

    Thanks and Best Regards,

  • Hi Dann,
    Thanks for your answer.

    Do I read your answer correctly that my previously mentioned 90.3mV (caused by 0V common mode voltage) would not add up to the common mode error voltage of 46.03mV (at a common mode voltage of 43V)? With the other factors mentioned, you're right, they would need to be taken into account. But as they do not have a big impact, I would also neglect them here.

    For a better understanding, here how the part is used here. Due to confidentiality and the circuitry is spread over several pages, I condensed it. The amplifier is used in a current sensing application, i.e. the 0.1 Ohm resistor shown here is the shunt.

    My basic problem is that we had some devices showing an increased offset voltage under increased common mode voltage. We took affected devices from the board and measured it stand alone under several common mode voltages (input pins being short-circuited to have same voltage on both pins):

    Common mode voltage Output voltage
    10V 0.0167V
    20V 0.0364V
    30V 0.0569V
    40V 0.079V
    43V 0.0853V

    Now it's the question, if these devices are within their specifications or not. For this purpose I wanted to be sure on my calculations based on the datasheet.

    Best regards,



    I just additionally did some error calculations to get the maximum offset voltages:

    CM voltage CM error Output offset error Bias Error Temperatur Error Total Error
    10V 10.71mV 9.03mV 0.36mV 0.72mV 20.83mV
    20V 21.42mV 9.03mV 0.36mV 0.72mV 31.54mV
    30V 32.13mV 9.03mV 0.36mV 0.72mV 42.25mV
    40V 42.84mV 9.03mV 0.36mV 0.72mV 52.96mV
    43V 46.06mV 9.03mV 0.36mV 0.72mV 56.17mV

    This table should include all error sources: The error due to common mode (based on the formula on page 4), the error due to output offset (based on your reply), the error due to bias error (based on your reply) and the temperature error due to 15K temperature rise.

    If we compare now the calculated maximum values, they are lower than the meaured results. If we assume, I did all the calculations correctly and no other error sources are forgotten, our parts are not within in spec, right?

  • Hi Raphael,

    The offset voltage you mentioned which is +/-1.5mV will also add up to the overall system offset, but since it mentioned that this is the RTI from the inputs for both the Differential Amplifier and output amplifier, the overall gain will only be 6 (Differential Amplifier's Gain x Output Buffer Gain = 0.1 * 60 = 6, this is shown on table 5 page 6 of the datasheet) and should only add up around 9mV, sorry I missed this one on the previous response.)

    Though considering this additional offset it would seem that you're still getting higher error starting at 30V common mode. When you mentioned that only some device, does this mean that some of the AD628s you have are within spec? 

    I'll continue to look into this, but on the devices that have this unexpectedly high offset, can you also measure the voltage at pin 4? This is basically the output of the differential amplifier and we'll be able to see if the unexpected offset is coming from the differential amplifier or the output buffer.

    Thanks Raphael.

    Best Regards,

  • Hi Dann,

    ok, so finally the worst case error at a common mode voltage of 43V will be 56.17mV (from the table above) + 9mV (as mentioned in your last post) = 65.17mV.

    So far we had this issue with devices with datecode #1935 and #1936. After replacing the problematic component with newer components, the problem seems to be solved. Might there be a problem with the batch?

    We did some more measurements with a better multimeter which showed again that the offset voltage is higher than expected:

    V_CM V_Pin4 V_Pin5
    0V 0.027mV -0.0027mV
    10V 0.349mV 19.32mV
    20V 0.679mV 39.15mV
    30V 1.022mV 59.75mV
    40V 1.369mV 81.01mV
    43V 1.470mV 87.33mV

    Best regards,


  • Hi Dann,

    did you have the opportunity yet to have a further look into this topic?

    Best regards,