Tool/software:
Hello
I would like to perform a compensation of the offset voltage caused by OPA2626.
Is there a typical value of the drift? As for offset voltage drift, it is +/-4uV/℃. Does that mean it has 2 directions.
For input offset current drift, it is positive?
Thanks!
I checked different datasheet. it seems sometimes the value is marked with +/-, sometimes not.
if this value could be positive or negative, it means a typical compensation is not possible.
Hello,
I apologize for delay here, not sure why your thread only showed up this morning.
You are correct it will always be +/- for offset voltage and drift which means it can be in either direction due to the inherent mismatch of the input transistors.
For example:
For any amplifier, offset voltage and drift will always be positive or negative. However, some datasheets will often omit the +/-, but it should always be assumed it is either direction.
We have an app note that covers how to compensate for high precision applications: https://www.ti.com/lit/an/slyt204/slyt204.pdf?.
For bias current and drift, it will depend on architecture if it is positive or negative. Usually in our datasheets, we will have a footnote where it specifies in/out which will refer to polarity. For example, in a bipolar amplifier, input bias current is the current flowing into the base of each transistor in the input pair. Or if an amplifier has an internal bias current cancellation, then bias current can have either polarity. To compensate, it is suggested to add bias current compensation via adding a resistor equal to the other path.
Specifically for the OPA2626, input bias current is positive.
Thank you,
Sima
Hello,
No problem! For offset current, there isn't a really precise way to compensate since this is the difference between the two bias currents, other than to try to match the impedances seen by each input of the amplifier. But, there will be an inherent difference between the two which can not be matched exactly.
"Ideally, the two input bias currents would be equal to each other and would cancel. In reality, though, they are not equal, and the difference of these currents is defined as input offset current. If the input offset current is low, it’s possible to match the impedances connected to each input and cancel the offset developed from the input bias currents."
By looking at the OPA2626, you could do a gaussian type calibration if you application can handle that type of +/-sigma deviation estimation:
And, offset current drift will be positive, since this value will increase with temperature
Thank you,
Sima
