[crk]

Hi expertz
I've just bought an oscilloscope. While trying to track down why my CH47 was (and still is) oscillating rather than amplifying. I noticed the waveform when the amp was working looked really bad. It showed clipping and those straight triangular waveforms suggesting something was slew rate limited.
The output from all my other amps (which really do amplify all the time) looked just as bad, so I checked the waveform from the source and it looked just as bad! So I checked all my other sources and, surprise, they all looked similar. I happened to be using track 2 of Coldplay's Parachutes at the time. This track seems to particularly full of these artifacts.
The three sources were: MP3 from a Nomad Jukebox, CD on a Rat Shack CD-3904 and the CD on my Sony C333ES.

crk

 

[Joobu]

I hope you don't take any offense at these questions:

1. How much experience do you have with a 'scope?

1. Have you tested the 'scope with a simple function generator such as a sine/square wave generator?

 

[tangent]

Joobu's right: when trying to look for oscillation and other distortions on a scope, you need to know what the waveform is supposed to look like to begin with. That means using a sine wave or something else that's nice and clean. Ideally, you'd use a dual-channel scope with the signal generator's output going to one channel and the amp's output going to the other. Scale the generator's signal, and then overlay it on the amp's signal and see if they match.

 

[crk]

Joobu
1. Loads
2. The scope's fine (checked and calibrated)

The waveform from an audio disk is typically a jumble of things that look vaguely sinusoidal, i.e. curved. You don't often see straight lines unless something is clipping or slew rate limited. Indeed this is the case with almost all audio tracks. . Presumably, signals go through some sort of a limiter when being recorded to a medium or digitized so a bit of flattening on large amplitudes might be expected. But the track I mentioned seems full of these clipping/slew like artifacts.

As to my CH47, it is still oscillating! I'm using 2134's. I noticed that Erix published a schematic where he used one amp for both L&R gain stages and the other for the L&R buffers. I was doing the opposite, i.e. one 2134 for the left channel the other for the right.

Thanks again
crk

 

[Joobu]

There are MANY different schematics for amps. The original basic CMOY would use one 2134 for the entire amp. The current-doubled amp which has various names, usually CHA47, uses one 2134 per channel.

I've looked at several audio waveforms on a scope. They usually look alright. However, since music is aperiodic, I would be careful about your trigger settings. I'd probably use auto trigger set at 0V.

There are several ways of stabilizing the amp but first isolate the exact point of the problem. Since you do have a scope I would suggest a sine wave or square wave generator to debug your amp. That way you can trace through the signal path, input, amp, output.

 

[crk]

Joobu
Thanks for your help. This is where I've got to. The amp is currently working fine but it required a 100pf across the feedback resistor on the gain (x2 on my amp) stage to get this stability.

The best I could achieve without the caps would be the following. The amp would oscillate (about 2MHz) with the volume at minimum with headphones connected. (It was stable with nothing connected.). As the volume was turned up (roughly a third of the awy, the oscillation would stop and the amp would work (scope and listening) fine. This transition did vary with the headphone connected. My SR-80's were much worse than the Sporta-pros, for example.

My current theory is that the CHA47 design is very sensitive to the capacitive load. This was suggested by looking at the 2134 data sheet which shows large overshoots for small capacitive loads when the amp is configured with low (~1) non-inverting gain. Certainly adding a small capacitor across the output with no headphone would cause the amp to oscillate. Also adding a resistance (75 ohms) in series with the headphone improved stability.

But if I am correct, I would expect everybody to have this problem.....

crk

 

[Joobu]

Many people appear to have oscillation problems. Using low gain designs definitely does not help. The stock CMOY has moderate gain so it's usually not as big an issue.

Maybe a feedback capacitance or output resistor should be included in the schematics of the current doubled designs.