Post Go back to editing

AD8675/AD8676: Exact meaning of Maximal Differential Input Voltage in Datasheet

Hi everyone,

My Situation:

If build a circuit with some filter capacitors (10uF) an the input's of the OpAmp AD8676 and have the problem that after some time (days of power up/down) suddenly the OpAmp start to leak at it's positive input an sink about 4uA.

I'm searching now the problem and got aware of the maximum differential input voltage and the input current limit.

I think the problem comes with the power down sequence and the discharging of the capacitors.

While I measure the signals of both inputs during power down I see up to 5V voltage difference between the inputs.

And here starts my question:

In the datasheet of the AD8676 there is a statement about the Differential Input Voltage of +-0.7V in the "Absolute Maximum Ratings" section.

Does that mean, that I have to make sure that this 0.7V never be exceeded because some protection diodes at the input stage would start to conduct and the 10mA could flow in case of a an input circuit that can source that amount of current? (e.g. my capacitors?)

But how can it be that I see 5V between the inputs in this case? With diodes this should never be possible. 

So what exactly does this +-0.7V differential input voltage really mean?

Thanks for any advice

best regards


  • Dominik,

    Yes, the abs max of 0.7V and 10 mA means don't exceed either.  For the low noise BJT front ends, you will usually find

    anti-parallel diodes between the inputs to keep the EB junctions from zenering which would degrade the noise performance.

    The fact that you measure 5V means you blew out the tiny diode.

    I saw a circuit years ago with an integrator and a 2.2 uF feedback cap.  When it got charged up to 10 V, if you turned

    off the power at that exact instant, and the supplies ramped down very fast, the cap discharged back into the output.

    The current measured over 150 mA.     i = C dV/dt.

    Scope your V+ and V-  during turn on and turn off.

    I get nervous with any can over 1 uF on an input or an output pin.

    -- Change your filter components to reduce the C and increase R.

    -- Put 220-1k in series with the cap

    -- Add two more poles which allows the corners to be higher

    --  Ramp your power supplies up and down slowly.

    --  Ramp the input to filter to zero before turning the power off.


  • Dear Harry,

    Thanks a lot for your answer. Unfortunately you confirmed my apprehension of exact this situation.

    I will go for the slowing down of the power because I do not want do have higher resistors in my circuit for noise performance reason.