Differentiator Circuit R Values

[eev_carl]

Hi,

I built the attached differentiator circuit, but I expected sharper edges.  I verified that my R and C components match the listed schematic.  I also tried the LM6171 op am and actually got more noise than the  741 (in addition to the charging curves).

I get something that looks a lot more like a square wave if I pull out the 470 resistor but am not sure why.  I was wondering about the purpose of this resistor.

All my values match the schematic.  I was using +-9V for VCC and also tried +-10V.

Thanks,
Carl

[rstofer]

Your time constant is RC or 47 us.  It takes about 5 or 6 time constants (Tau) to level off and that seems pretty consistent with your trace.  It measures about 250 us, I believe.

You don't really have a differentiator, you have a low pass filter.  For a true differentiator, the input goes to the capacitor and for the op amp incantation, there is no input resistor.

http://www.electronics-tutorials.ws/rc/rc-differentiator.html

http://www.electronics-tutorials.ws/opamp/opamp_7.html

http://www.radio-electronics.com/info/circuits/opamp-circuits/operational-amplifier-differentiator.php

[rstofer]

Integrators are effectively a form of low pass filter - they reduce noise in the circuit.
Differentiators are a form of high pass filter - they amplify noise in the circuit.

In analog computing, we rearrange the equations to get away from differentiators and use only integrators and gain blocks.

I'm not sure whether, or how much, noise your circuit will have.

[rstofer]

I thought I would model this with LTSpice and I left the 47k resistor and this creates a lot of gain.  I made sure my signal output impedance was 50 Ohms because, when it wasn't, the noise was insane.

In any event, we can see the switching transients when the triangle wave changes direction.

Just for fun...

[eev_carl]

Thanks! I've pulled out the op amp and am experimenting with the different series RC circuits from your first link.

Are you able to export your netlist from the simulation?

[David Hess]

The input resistance is needed for stability reasons and to limit high frequency gain which would otherwise produce a very noisy output.  Draw a bode plot without the input resistor to see why.

[Zero999]

Yes, R1 is required for stability reasons. You can try reducing it, to get a better squarewave. I've just has a quick play with LTSpice and found that faster op-amps can tolerate having a lower R1.

Attached is the LTSpice file. If you've not got LTSpice, it's a free download from the Linear Technology website, which can be found using a search engine.

[rstofer]

It seems to me that LTSpice is an absolute requirement when working with circuits.  It is much easier to play "what if?" on a schematic than it is with wire and resistors.

I tend to use the UniversalOp device if I'm just messing around.  It comes with LTspice and the 741 model does not (AFAICT).  It would always be better to use a model for the exact device but sometimes it just isn't available.

One example of this is where there is no overshoot with the 741 model and there is with the UniversalOp model.

I dropped the series resistor down to 50 Ohms and the square wave is a lot more square.  It still isn't perfect and I suspect the difference is due to the more limited slew rate of the 741 versus the UniversalOp.

[David Hess]

It seems to me that LTSpice is an absolute requirement when working with circuits.  It is much easier to play "what if?" on a schematic than it is with wire and resistors.

I find SPICE useful for tuning component values in a well understood circuit but not so good for understanding an unknown circuit.  In this particular case for instance, I would not expect SPICE to model the stability or performance of the 741 well as the input resistance is changed while drawing a bode plot would reveal everything.

[rstofer]

It seems to me that LTSpice is an absolute requirement when working with circuits.  It is much easier to play "what if?" on a schematic than it is with wire and resistors.

I find SPICE useful for tuning component values in a well understood circuit but not so good for understanding an unknown circuit.  In this particular case for instance, I would not expect SPICE to model the stability or performance of the 741 well as the input resistance is changed while drawing a bode plot would reveal everything.

For the Bode' plot are you just going to substitute the triangle wave for a sine wave of varying frequency?  You had mentioned this before but I wasn't clear on how you intended to set up the input.

[David Hess]

For the Bode' plot are you just going to substitute the triangle wave for a sine wave of varying frequency?  You had mentioned this before but I wasn't clear on how you intended to set up the input.

I'm just going to doodle it out by hand on a piece of paper while calculating the various pole and zero frequencies.  SPICE can produce this also but I find it faster to write it out by hand.

[Audioguru]

A differentiator is a highpass filter. I do not know why people said it is an integrator or lowpass filter.