Forums » General Electronics Discussion

I've designed (I think!) a simple op-amp differentiator:

https://www.circuitlab.com/circuit/zjgvgk/lab-4-differentiator/

The output for the Sine and Square inputs makes sense to me, since for one it looks like Cosine and for the other it looks like an impulse function.

However, the triangle generator should (I believe) be producing a square output form. It isn't.

Any suggestions?

That's because you have not designed a pure differentiator.

You have designed a high pass filter. Note the same thing.

The 50Hz input is far too close to the low frequency rolloff of the filter of about 100Hz for it to see much in the way of differentiation.

Run a Frequency Domain sim with Vin as the input. You can leave the other Frequency Domain sim settings at their default values and then plot Output.

Run the sim for R1 = 2.2k and for R1 = 0 (you can do this automatically using the Sweep Parameter option).

Do the same thing with your Time Domain sim.

Note the difference in responses.

See also:

https://www.circuitlab.com/forums/audio-electronics/topic/ux89tq97/mide-side-microphone-matrix-with-op-amp/#comment_4204

:)

More:

http://www.ti.com/lit/an/sboa092a/sboa092a.pdf

Thank you for the guidance!

I'm a little confused, or a lot probably.

In my understanding, R1 is necessary in a practical application to avoid adverse behavior.

That's explained much better here than I could attempt:

http://www.ecircuitcenter.com/Circuits/opdfr/opdfr.htm

RL is there to simulate a load resistance. Rx is probably superfluous, but I don't believe would adversely change the behavior of the circuit.

Now, at a 50Hz frequency, we're looking at 0.02s for a full cycle. I though as long as RC << 0.01s, I should be safe that the capacitor would be able to fully charge/discharge over each half of the cycle. I have a feeling this is a big part of what I'm misunderstanding!

Thank you so much for taking the time to explain what you already have to me. It's much appreciated. Any further help would be great. I'm going to run the simulations you suggested as soon as I get home.

Very helpful! Thank you so much!

:)

Another way to think of a differentiator is that for it to work properly you need to be operating in a +20db/decade slope a very long way from any change of slope at the low or high frequency extremes.