Dancing scope traces

[Debug]

Hello.
I've been restoring/refurbishing my old Ampeg G20 which has long been one of my favorite amps. It fell into serious disrepair about 10 years ago & I finally got around to dealing with it. I'm so close to done that I can taste it but I have a behavior in the power section that I really can't explain. I don't _think_ it's normal but I wanted to get a sanity check & some advice on what to look at if it's not normal.

Here's a schematic for the amp. Not the best quality but it's what I've been working off of. http://www.turretboards.com/layouts_.../ampeg_g20.gif

It' a fixed bias amp & has no adjustment pot or current sense resistors at the moment. The bias voltage is about -60 which seems to scale with the other internal voltages that are also higher than listed on the schematic due to my ~125-128 VAC line voltage.
Signal coming out of the PI look clean and stable. All the coupling caps are brand new. This doesn't seem to be affected by how much signal I pump into it.


I shot some video & one still picture of the scope traces with one channel hooked to the plate of each 6L6. Notice how the trace "dances". I've endlessly tweaked the holdoff on the scope to get it this good but I can never get a stable trace. I have a still on my ipad that I'll post separately that shows what I get with the holdoff on auto. It's not pretty.

Video #1. Channel 1 of the scope attached to the plate of one output valve. Amp issue scope channel 1 - YouTube

Video #2. Channel 2. Amp issue scope channel 2 - YouTube

Video #3. Both channels in differential mode. Both channels in differential mode - YouTube


Anyway, I would deeply appreciate any insight into what I'm seeing. There is a little bit of harshness on the sound and I do not know if these crazy scope traces and the harshness are related and therefore I am not sure what direction to go in solving it. Being a PCB amp it's going to be kind of a pain to add bias adjustment hardware (I already converted it off of the 7199 PI & did not enjoy cutting traces) so if this behavior is something other than a biasing issue (or if it's absolutely normal which I would love further explanation of...) I would like to weed that out now. If bias adjustment is the ticket then I'll gladly go there.


Finally, I really enjoy reading this forum & hope to be able to contribute something useful at some point.

thank you so much

[Debug]

Here is the still. It's channel two with the hold off clicked. It's a bit hard to see but the trace jumps to about 20 different paths.

[olddawg]

So... You are using a 1khz tone. Line auto triggering the scope? If you back down the time base will it show that the 1khz is riding on a little 60hz ripple? Make sure your source generator does not have ripple as well. Ch1 the source, Ch2 the output. A ground loop between the generator and the amp can cause this too with nothing wrong.

[Enzo]

Yes^^^

What do your ears tell you about hum, anything?


With 1kHz showing only 2 or 3 cycles on the screen of the scope, a 60Hz image would be something like 16 times wider, so 16 screen-widths. That 60Hz hum signal won't sync up to the 1kHz, nor to the screen under those conditions. Your screen is synced to a repetitive 1kHz signal, the peak may appear in the same place, but you are actually seeing a series of peaks every 1/1000 of a second. So sometimes the 1kHz peak occurs when the 60Hz is on the low side, and sometimes on the high side, and sometimes in between. So the image is moving as the phase relationship changes between the two signals. Thinking about it, this is exactly the same thing as the wagon wheel spokes in an old western movie, when they appear to be moving slowly backwards as they roll forwards due to the strobe effect of the camera shutter. or perhaps a strobe light on a table fan makes the blade appear to turn slowly when it is really spinning fast.


I tend to doubt this is making the sound harsh directly, though the sound problems might be related to a soft power supply.

[Debug]

Wow. That was an eye opener. Here's the pic (channel 1 out channel 2 source).



I don't see any ripple on the input but clearly the output has some. I've been using this scope and signal generator for a couple of years now and have never felt like there was a ground loop or other problem. So I measured the ripple.

Scope shows more or less 30v pp ripple set for 5v/div with a 10x probe. Vtvm shows about 9vac. I think my stuff needs calibrating.

That seems like a lot of ripple to me but I can't find anything wrong with the components in the power supply which are all new. With that A rail hanging off the input cap I guess I shouldn't be surprised. Should I just leave well enough alone and assume that it's working to spec as built? The other rails have almost undetectable ripple.

Now that I understand what my scope was trying to tell me I feel like I can move on in search of the harshness.

When I'm done I think I will write up the adventures of servicing this amp as the G20 and it's relatives are somewhat obscure and I ran into some interesting stuff.

Thanks!!

[g1]

Also should mention that scoping power tube plates can be quite a safety hazard unless you have high voltage probes.

[Debug]

Truth! I have 600v probes, so they say anyway. Got about 100v of safety margin on this one when loaded with at least a few tubes. With no tubes the PT puts out real close to 600v so I won't go near it with the scope in that configuration.

[g1]

The issue I was warning of was when there is signal through the power tubes. Then you are probably quite a bit higher than the 600V the probes are rated for.