Post Go back to editing

Adjusting for input offset error using AD8230 Figure 35

I need to adjust for a wide offset voltage error coming from a strain gauge bridge.  The common mode output of the bridge is 2.5V.  The normal differential output range from the bridge is 0-15mV, but initial offset error from the bridge can be as high as +/- 20mV.  I am using the AD8230 because of its remarkable input offset voltage specification and what I hope to be a very small offset voltage drift as the part warms up.  I plan to correct for bridge offset error by doing a calibration step at the beginning of operation.  The amplifier is configured for a gain of 250, so my normal amplifier output voltage swing would be 0-3.75 volts.  So far so good.  However, when bridge offset error is considered, the output swing would be -5V to +8.75V.  My supply rails are only +/- 5V.  Here are my questions.  If I were to use the method of output offset adjustment shown on Figure 35 of the datasheet to inject an output offset voltage adjustment signal in place of the return connection, as indicated by VR', would I be able to adjust for this much offset without saturating the first stage of the amplifier?  Also, are there any phase reversal issues that I might need to contend with?  In other words, would the amplifier behave properly and just saturate high or low depending on the polarity of the offset until I have a chance to inject the correct offset signal at VR'?  This would be a real win for my design if I could solve my problem in this manner.  Thanks.

Parents
  • Hello Carl,

    If you inject current into the feedback node RG, you are doing so before the large gain is applied to the signal, so you shouldn't have any issues with what you are describing. You can deduct from the gain equation on page 11 that only twice the input voltage appears across RG (figure 32). Injecting or stealing current from this node will modify the current supplied by the output of the inamp. You can make a similar analysis with an opamp in non-inverting configuration, only that the feedback node will be at the input voltage instead of twice that much (and the input signal is referenced to the "ground" where the feedback is connected instead to a differential input).

    Now, keep in mind that if you expect a bipolar offset, adjusting the output to zero will require that you either inject or subtract current from this node. So, unless I'm missing something, if you plan to use, say a single-supply 5V DAC for your adjustment, you will need to use an amplifier to make it a bipolar voltage source. Then I would size the resistor Ro to use as much resolution from the DAC, while keeping some room for component variatios. Some calculations on the back of the envelope:

    Rg = 100Ohm

    RF = 12.4kOhm

    With a +/-20mV offset you'll get +/-40mV at the feedback node which would translate into +/-400uA that you need to inject plus about +/-1% to compensate the error voltage. So, if you have +/-5V (after generating your bipolar voltage source), Ro=10kOhm will give you about 500uA for adjustment. You can use that or you can make Ro slightly smaller to get better margin (although you should already have about 20% margin for error).

    By the way, there are no phase reversal issues that you need to worry about. Even if you apply the wrong polarity to correct the offset, the amplifier will just remain saturated (with the correct polarity).

    Good luck and let me know if you have any more questions.

    Regards,

    Gustavo

Reply
  • Hello Carl,

    If you inject current into the feedback node RG, you are doing so before the large gain is applied to the signal, so you shouldn't have any issues with what you are describing. You can deduct from the gain equation on page 11 that only twice the input voltage appears across RG (figure 32). Injecting or stealing current from this node will modify the current supplied by the output of the inamp. You can make a similar analysis with an opamp in non-inverting configuration, only that the feedback node will be at the input voltage instead of twice that much (and the input signal is referenced to the "ground" where the feedback is connected instead to a differential input).

    Now, keep in mind that if you expect a bipolar offset, adjusting the output to zero will require that you either inject or subtract current from this node. So, unless I'm missing something, if you plan to use, say a single-supply 5V DAC for your adjustment, you will need to use an amplifier to make it a bipolar voltage source. Then I would size the resistor Ro to use as much resolution from the DAC, while keeping some room for component variatios. Some calculations on the back of the envelope:

    Rg = 100Ohm

    RF = 12.4kOhm

    With a +/-20mV offset you'll get +/-40mV at the feedback node which would translate into +/-400uA that you need to inject plus about +/-1% to compensate the error voltage. So, if you have +/-5V (after generating your bipolar voltage source), Ro=10kOhm will give you about 500uA for adjustment. You can use that or you can make Ro slightly smaller to get better margin (although you should already have about 20% margin for error).

    By the way, there are no phase reversal issues that you need to worry about. Even if you apply the wrong polarity to correct the offset, the amplifier will just remain saturated (with the correct polarity).

    Good luck and let me know if you have any more questions.

    Regards,

    Gustavo

Children
No Data