Odd Square Wave Behavior in Lab Experiment

[Feldoh]

So the other day in a lab my group was working with a signal and some noise and the signal we got was a bit interesting...

Essentially we were using a white noise with 0.5 V_pp and a 1kHz square wave with the 10 times the amplitude (5V) running them through a 741 op-amp summing amplifier.

We were supposed to experiment with different ratios of approximate voltage and looking at FFTs of the data we collected. The first ratio was ~1/1 so we were dividing the square wave signal by a factor of 10.

However just running the signal (no noise) through the amplifier gave us an interesting display and I'm still not particularly sure why it is happening:

Here's are some simulations that show pretty much the same thing we observed in the lab
http://filer.case.edu/pal25/square%20wave.JPG -- Output on oscilloscope
http://filer.case.edu/pal25/circuit.JPG -- Schematic of circuit (essentially it's two function generators in a summing amplifier)

There are spikes on the square wave if we run it through the amplifier and I'm not particularly sure why we've tried using a better op-amp (OP42G) and we still get effectively the same display which leads me to believe it's something else.

 

[torquil]

I'm no expert, but it seems to me that this could more easily occur in your inverting summing amp than in a noninverting version?

I saying this because in your case, the input square wave is resistively coupled "directly" to the output of the amp, so any slowness in the amps feedback loop would result in an overshoot spike if the rise time of your signal generator is faster than the opamp.

Maybe you should try a noninverting configuration?:
http://www.allaboutcircuits.com/vol_3/chpt_8/8.html

Another possibility is to se if you can reduce the rise time of your square wave generator. Either on the apparatus, or by introducing a capacitor across the signal generator output? But please find out beforehand if this can cause damage to the generators output.

Did you notice a difference in the width of the overshoot spike for your two types of op-amp? I would expect it to be wider for the slower op-amp.

Torquil