Using oscilloscopes to spot signal irregularities in analog systems

[Alan]

You've got your oscilloscope connected and you're looking at waveforms that should be straightforward, but something isn't right, there's noise, clipping, or some strange ripples. So, what settings and probe techniques can help you track down the source of the problem in older analog equipment without just swapping parts and crossing your fingers?

 

[barfle]

There's a whole lot of "that depends," of course. If it's clipping, follow the signal back until it's clean, then figure out what's causing a bad element. I'd use a scope setting that showed me a few cycles of the test signal, at about 75% of full screen.

Back in the early 1980s, we were using Motorola 6800 processors fairly extensively. When they were glitched just right (and I've forgotten how), they just counted through all the addresses, which made finding data and address line shorts pretty easy with a scope. Anything halfway was a dead giveaway.

I pretty much always use the x10 probe settings, as they tend to minimize the influence of the testing equipment on the tested equipment.

 

[Alan]

Great info, thank you! That 6800 trick is a fantastic bit of old-school wisdom, and I really appreciate you sharing that kind of experience with everyone.

 

[voltagepathmaker]

There's a whole lot of "that depends," of course. If it's clipping, follow the signal back until it's clean, then figure out what's causing a bad element. I'd use a scope setting that showed me a few cycles of the test signal, at about 75% of full screen.

Back in the early 1980s, we were using Motorola 6800 processors fairly extensively. When they were glitched just right (and I've forgotten how), they just counted through all the addresses, which made finding data and address line shorts pretty easy with a scope. Anything halfway was a dead giveaway.

I pretty much always use the x10 probe settings, as they tend to minimize the influence of the testing equipment on the tested equipment.

This advice is spot on. It's clear, practical, and filled with real-world experience. It's a treasure for anyone facing signal issues.

 

[Phil]

Good approach is using AC coupling to reveal small ripple and narrowing the time base around the anomaly, then compare before/after stages so you can pinpoint exactly where the waveform starts degrading.

 

[oldgeek]

Books have been written on this topic. But here are a few basics:
- Know what you are looking for. The problems one can encounter with a class A/B audio amp are very different than one will encounter with a 400MHz RF power amp and they require different tools and techniques. Let's focus on audio.

Key parameters for any audio amp, from the mic to the speaker, are frequency response, distortion (harmonic and intermodulation) and crossover distortion (applicable to class A/B power amps, opamps--anything with a push-pull output stage. But we also have other things to look out for: clipping, slewing, compression (more of an RF phenomenon), DC offset, noise immunity, RFI rejection, EMI rejection. The last few sound very RF-ish, and they are--but when you take your latest guitar pedal project on the road and it proves to be better at receiving the local AM broadcast station than what you intended, you will wish you had tested for it.

All I can say in a short post that people will read is that an o'scope is one must-have tool for any of these tests. Some will require a low distortion audio oscillator and/or an RF sweep generator with amplitude modulation capability as well, but first and foremost a good o'scope with dual timebase and delayed sweep are absolutely necessary in hunting down small errors in large signals.