Colpitts conundrum

[cellularmitosis]

Hey all,

I've been fooling around with a very simple Colpitts oscillator, and ran into a surprise when I tried to throw an op amp in front of it to buffer its output.

First shot is of the unbuffered Colpitts oscillator circuit.

Second shot is of the pin I've labelled "output", as well as an FFT.  I found this pin to be the smoothest sine wave.

Shot three is the buffered circuit.  I've used an LM324 mod file.

Shot four is a plot of both the input and the output of the op amp.  As expected, they match almost exactly.  This is with a 3V supply powering the Colpitts, and 15V +/- rails for the op amp.

Shot five is where the problems begin.  The supply to the Colpitts has been turned up to 5V, and we are starting to see some distortion on the output of the op amp.

Shot five is with the Colpitts supply turned up to 9V.  What the heck is going on here?  At some point I'd expect to see clipping as I get closer to the op amps supply rails, but this output waveform has turned into a triangle!  Its as if it somehow no longer has the slew rate to keep up with a 33kHz input?!?

Can you pros hit this newb with a clue bat please?  What am I missing here?

(I've also tried this circuit with an LT1001 with similar results, so I don't think I have a bad LM324 model)

[Tac Eht Xilef]

Looking at the graph, a half-cycle is ~16us. The slew rate of an LM324 is low, ~0.5v/us. Look like you're running into that...

edit: specifically, at 0.5v/us the LM324 can manage about an 8v swing in 16us.

• At 3v to the oscillator, you're only asking it to do ~4v p-p in 16us. Not a problem...
• At 5v to the oscillator, you're asking it to do ~6.4v p-p in 16us. You're starting to see problems because the instantaneous rate of change at some points of the sine is faster than 0.5v/us.
• At 9v to the oscillator, you're asking it to do ~11.5v p-p in 16us, and it can't keep up. It's only managing to swing from ~2v to ~10v, or about 8v in total - which is what you'd expect from the specs...

edit2: I wasn't familiar with the LT1001, but looking at its specs it's got a slew rate of 0.25v/us - so it's even worse than a LM324.

Dig out the old trig and calculus, work out the highest instantaneous rate of change of your sine wave, and try an op-amp with a higher slew rate than that. Or rough it out, add a bit for fudge, then choose an op-amp with a better slew rate than that

[cellularmitosis]

Tac, thanks so much for laying that out for me!  I had no idea I was already that close to the lm324 slew rate limit

My op amp knowledge started out as simplifications (useful lies), which get chipped away at each time I run into a project which challenges them (which is the best way to make such a lesson stick!).

So glad to have encountered another such project!

Hmm, maybe that would make an interesting series of follow up videos to Daves op amp video: a collection of circuits which seem trivial at first glance, but fail because they violate one of the op amp simplifications.

[David Hess]

In the past if a discrete buffer was not used, a "video" operational amplifiers like the LM318 could be.

There are other ways to use operational amplifiers which will considerably increase their bandwidth and slew rate but today fast operational amplifiers are ubiquitous and inexpensive so such complications are not usually required.