Ghost Note Causes

[drewl]

Well I was running about 600hz to 1k, grids look clean, varying tone controls can alter the waveform

I can get around 20vac into 4 ohms clean, above that waveform peaks start getting ripples on them
Clean:



RIppley peaks:

[Helmholtz]

varying tone controls can alter the waveform

Not if you directly feed into the PI and your sig generator is low impedance. Otherwise always adjust for best sine when measuring output power.

I can get around 20vac into 4 ohms clean, above that waveform peaks start getting ripples on them

Is that RMS or p-p?

[drewl]

RMS This thing is a beast putting out like 150 watts peak.

I ran a preamp with guitar straight into the PI, still ghosted, tried driving each output tube grid with an amp but couldn't drive it enough to ghost.

I have to ask the owner if he left the output transformer connected in any way when he tried another OT.

The plate leads look like they were unsoldered, but the center tap HV doesn't, don't know if that would make a difference.

[Helmholtz]

As the amp can deliver at least 100W of (almost) clean power, I'd say the OT is good. At least has none of the usual issues.

Also seems clear now that the ghosting happens in the power amp.

[drewl]

Well I can run the preamp into an external power amp to make sure the problem is definitely in the power amp.

[nosaj]

RMS This thing is a beast putting out like 150 watts peak.

I ran a preamp with guitar straight into the PI, still ghosted, tried driving each output tube grid with an amp but couldn't drive it enough to ghost.

I have to ask the owner if he left the output transformer connected in any way when he tried another OT.

The plate leads look like they were unsoldered, but the center tap HV doesn't, don't know if that would make a difference.

What is your lead dress from the PI on to the outputs? Could one side of the PI be coupling the signal in to the otherside producing the Ghost in the Machine?
nosaj

[drewl]

Standard Marshall wiring from the PI through the power amp.

It does have some "mystery mod" that takes the master volume signal, runs it through a "cupcake" of goop and sends a boosted signal back to the bass pot.
The "cupcake" of goop has high voltage DC going into it.
I disconnected this mod but ghost notes are still present.

I tried another OT, no dice.


Thanks all.

[Helmholtz]

As long as the power amp (including PI) is well balanced, it has good ripple rejection regarding B+, screen node and bias ripple.
This is likely to change when it's overdriven and grid signals get asymmetrical (duty cycle shift and blocking distortion).
So make sure the screen supply has low ripple and the bias is clean.
Could even be a shorted choke.

[drewl]

Filter caps were all changed, I attached external cap to each node to double check, bias is clean Choke measure 105 ohms, unless it's breaking down under load.

Screen supply has these weird "peaks" of signal (?) When amp is cranked



They're about 1v p-p

[nosaj]

The extra gain stage is inserted before the cathode follower stage.
switching it off, removing the tube makes no difference.

The tubes are balanced within a couple mv, and I tried different ones.

Increasing the presence increases the ghosting/oscillation.

Signal hitting output tube grids is nice and clean, after OT it looks like this:

You know even with the tube out the wires can still act as antenna and pickup and inject signal. maybe scope some of those wires with tube out while ghost noting?
nosaj

[Helmholtz]

Choke measure 105 ohms,

A few (of hundreds) shorted turns would hardly show in DCR, but will significantly lower impedance.