This circuit is used to:

  1. Read an encoder output and level shift it to 3V3 from 24V;
  2. Convert a 0-10V input signal to 0-3V3;

Both points 1 and 2 should be usable at the same time. Point 2 works correctly while point 1 has a strange output with a "notch".

Here's the schematic:

enter image description here

In both tests where GND is connected with GNDS and where not, there's no noticeable difference.

What oscilloscope is reading:

enter image description here

  • A capacitor on R6 doesn't make any difference.
  • A capacitor of between pin 2 and pin 3, create a sinusoid when signal should be low, even 1pF.
  • I can remove this "notch" by adding a capacitor between R58 endings of about 22pF and this solves my encoder problem, but then the 0-10V conversion doesn't work anymore because a fixed input voltage is converted to a sinusoidal output.

I've double checked the power supply and ground of my board. I have a decoupling capacitor near the amplifier, a solid ground plane under it, GNDs via near each ground point on schematics. Also a clear +24V square wave, generated by a lab equipment, on amplifier input pin.

Another test I've done is to change MCP634 with a similar one, but nothing changed (just slew rate.)

  • 2
    \$\begingroup\$ What does that specific opamp do with input voltages exceeding its supply rails? According to its datasheet. \$\endgroup\$
    – user16324
    Dec 7, 2020 at 13:27
  • \$\begingroup\$ Singed, if you are done here please accept my answer or raise a comment should you need further clarification. \$\endgroup\$
    – Andy aka
    Aug 1, 2022 at 12:52

1 Answer 1


You have probably got a significant common mode voltage problem i.e. the common mode voltage of the signal source is probably several volts different to the ground connection on your measurement circuit.

You get rid of this CM voltage by shorting both grounds and the circuit behaves.

Any differential amplifier has to deal with both the signal voltage extremes AND the common mode voltage together. This is usually achieved by making R58 and R4 much bigger in value. I can see you are on the way with this but, I suspect you might not have evaluated the true common mode voltage of the source relative to measurement ground.

I think the problem is amplified by you choosing the MCP634 - the highest voltage allowed on the input pins is Vdd - 1.3 volts and with you operating at 3.3 volts on your supply, the largest voltage allowed (before you get silly things happening) is 2 volts. So, if you say there is 2 volts across R5 then things go wrong when ENC1A is greater than 8 volts. Can this happen? Does it happen?


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.