Volume drop at 10 (DIY clone)

[BlueX]

There's a sudden volume drop right before I hit the max stop end on the High Treble volume pot on my newly built JTM45/100 clone. It's been consistent the few time I tested. Volume increases as expected when I turn the pot, until this drop happens.

It's also independent of guitar input volume, plugging in High or Low, or settings of Presence, Bass, Middle, and Treble. There's switchable tone stack (Lead/Bass) and NFB resistors (27k, 49k, and 100k, and 4, 8, and 16 Ohm) but different settings doesn't change the volume drop.

Could this be a faulty pot (new A1M CTS)? I need to order replacement if so.
Or could this be oscillation or similar, so I need to connect the oscilloscope?

Normal channel volume pot acts in the expected way, all the way to max. All voltages are normal, and except for the problem above the amp sounds very good. Also very low noise without guitar signal.

 

[Matthews Guitars]

First, check to see if the volume pot is OK. If it is, then I'd be suspecting that the amp is going into ultrasonic oscillation. Do you have a scope and can you use it to check for that?

If it's oscillating, wire dress changes may be all you need to fix it.

 

[BlueX]

First, check to see if the volume pot is OK. If it is, then I'd be suspecting that the amp is going into ultrasonic oscillation. Do you have a scope and can you use it to check for that?

If it's oscillating, wire dress changes may be all you need to fix it.

Thanks!
De-solder the pot and measure with DMM is probably the easiest first step. I might swap positions on Normal and HT volume pots, come to think of it.

I do have oscilloscope (and signal generator and isolator transformer), so I will probably spend some time at the bench if the pot seems OK.

 

[Marcomel79]

Also, after the fix, would be nice to hear a sample. Hope its an easy fix

 

[neikeel]

Sorry but I have lost track of the mods on your build.
It may well be oscillation. Does it squeak just before you lose volume?
If so I would check your lead dress in the preamp and whether you went for 4k7 or 5k6 grid stoppers on each pin 5 of your octals.
The stock 1k5 stoppers between pairs is adequate for standard 45/100s.

 

[BlueX]

Sorry but I have lost track of the mods on your build.
It may well be oscillation. Does it squeak just before you lose volume?
If so I would check your lead dress in the preamp and whether you went for 4k7 or 5k6 grid stoppers on each pin 5 of your octals.
The stock 1k5 stoppers between pairs is adequate for standard 45/100s.

I need written records to keep track myself. Here's the latest specification, but it's still missing how Heater CT elevation, switchable tone stack and NFB, and bias sampling have been arranged.

No noise/sound before volume drop, just gets muted. 470R to p4 and 1k5 to p5.

I will check volume pot first, and maybe switch place on Normal and High Treble, and report back.

Probably need some advise where to start scoping, if I have to go there, since I'm rather new to audio circuits.

 

[Pete Farrington]

My guess is ultrasonic oscillation.
If the issue goes away with the pin 2 end of C6 lifted, that would help to confirm it.

switchable tone stack and NFB

The implementation of these will be increasing the degree of unforeseen / unintended coupling between different sections of the circuit, so could be contributing to the issue.

Scoping the amp's output into a reactive load (with the amp controls set for the issue to manifest) should allow you to see it. But if the power amp is inside the positive feedback loop, the oscillation might put a lot of stress on the KT66 and screen grid supply. So this issue isn't ideal for getting used to the scope.

 

[BlueX]

I de-soldered both volume pots and they measure correct values, with nice slope/taper when I turn them both ways. Switched places and soldered them back. If same problem remains on HT channel, and Normal channel still works OK, I can rule them out.

Looks like I'm in for a deeper dive.

 

[BlueX]

My guess is ultrasonic oscillation.
If the issue goes away with the pin 2 end of C6 lifted, that would help to confirm it.

The implementation of these will be increasing the degree of unforeseen / unintended coupling between different sections of the circuit, so could be contributing to the issue.

Scoping the amp's output into a reactive load (with the amp controls set for the issue to manifest) should allow you to see it. But if the power amp is inside the positive feedback loop, the oscillation might put a lot of stress on the KT66 and screen grid supply. So this issue isn't ideal for getting used to the scope.

I will check C6 when I start testing again.

Regarding power amp oscillation: Would that consume considerable power?

My variac/LBL unit has a panel DMM that also shows power (W). Unexpected power consumption should show up there. Maybe a first test could be to check this, to try and avoid over-stress of tubes and other components?

 

[BlueX]

Did a short power-up.
Idle power with amp off: 3,5 W
Power on, but Standby open: 57 W
Power on, Standby closed, and both volumes on 0: 114 W
Turning Normal channel volume up to 10 does not change power consumption (still 114 W).

Turning HT channel volume up does not change power consumption until the volume drop happens. Power consumption then suddenly increases to 160 W.

Something happens inside the amp at full HT channel volume. Would the extra 46 W be enough to put too much stress on tubes or other components, or would this be something that is OK to probe with an oscilloscope?

 

[Pete Farrington]

Something happens inside the amp at full HT channel volume. Would the extra 46 W be enough to put too much stress on tubes or other components, or would this be something that is OK to probe with an oscilloscope?

The issue will probably be the screen grids / resistors overheating.
To help to mitigate this, is it feasible to keep the amp on the LBL whilst tinkering?

 

[playloud]

Turning HT channel volume up does not change power consumtion until the volume drop happens. Power consumption then suddenly increases to 160 W.

Something happens inside the amp at full HT channel volume. Would the extra 46 W be enough to put too much stress on tubes or other components, or would this be something that is OK to probe with an oscilloscope?

160W sounds fine to me. As Pete says, LBL would be a good safety measure (oscillation should still occur with sufficiently-large bulb), but it doesn't sound like a catastrophic amount of power. I sometimes play my 45/100 with a power meter downstream, and it often gets into the upper 200s when playing loud.

All signs point to ultrasonic oscillation here. Get your scope and see where the problem begins in the circuit. Removing the NFB loop is likely to be an initial troubleshooting step.

Edit: just saw your comment on needing advice on where to start. As Pete says, set the controls so the issue manifests, then start scoping the grids of tubes, following the signal from the input to the 'high treble' channel (pin 7 on V1). Worst case, you'll find it on the grids of the power tubes (pin 5).

 

[BlueX]

The issue will probably be the screen grids / resistors overheating.
To help to mitigate this, is it feasible to keep the amp on the LBL whilst tinkering?

160W sounds fine to me. As Pete says, LBL would be a good safety measure (oscillation should still occur with sufficiently-large bulb), but it doesn't sound like a catastrophic amount of power. I sometimes play my 45/100 with a power meter downstream, and it often gets into the upper 200s when playing loud.

All signs point to ultrasonic oscillation here. Get your scope and see where the problem begins in the circuit. Removing the NFB loop is likely to be an initial troubleshooting step.

Edit: just saw your comment on needing advice on where to start. As Pete says, set the controls so the issue manifests, then start scoping the grids of tubes, following the signal from the input to the 'high treble' channel (pin 7 on V1). Worst case, you'll find it on the grids of the power tubes (pin 5).

I've got a stash of light bulbs of different wattage, between 25 and 200W. Maybe start with a 100W bulb (to properly feed heaters). If I don't see this problem I can increase to first 150W, and maybe 200W bulb. Will report back.

 

[playloud]

I've got a stash of light bulbs of different wattage, between 25 and 200W. Maybe start with a 100W bulb (to properly feed heaters). If I don't see this problem I can increase to first 150W, and maybe 200W bulb. Will report back.

Honestly, any of those should be fine (I'd probably just go with 200W, since you know the problem will be reproduced). The purpose of the LBL is to stop any excess current surges if the instability changes/gets worse while you're poking around in the amp. The fuses should prevent any major damage, but you could still lose a tube or screen resistor. No need to make yourself 'jump' if you don't have to!

 

[BlueX]

Did some measurements with the oscilloscope. Nothing plugged into any of the inputs, and a resistive dummy load in speaker out.

Nothing happening after C2 or C3, but about 20 V (p-t-p) at 8 kHz after both C8 and C9 (on the tone stack side) when HT channel volume at max.

Readings on the screen was about 4 squares (p-t-p) with 5V/square and 2,5 squares (one full wavelength) with 50us/square. Wavelength should be 125 us, and frequency 8 000 Hz.

Maybe I should disconnect the switching functions for NFB as next step, and wire 27k between 16 Ohm tap and R19/C11 at PI. I've documented an oscillation, and I'm not sure I need to measure more right now. Better to try and get rid of it.

Any comments?

 

[Pete Farrington]

Maybe I should disconnect the switching functions for NFB as next step

Photos of the build would be most useful to advise on improvements.
How about "If the issue goes away with the pin 2 end of C6 lifted, that would help to confirm it."?

 

[playloud]

Did some measurements with the oscilloscope. Nothing plugged into any of the inputs, and a resistive dummy load in speaker out.

Nothing happening after C2 or C3, but about 20 V (p-t-p) at 8 kHz after both C8 and C9 (on the tone stack side) when HT channel volume at max.

Readings on the screen was about 4 squares (p-t-p) with 5V/square and 2,5 squares (one full wavelength) with 50us/square. Wavelength should be 125 us, and frequency 8 000 Hz.

Maybe I should disconnect the switching functions for NFB as next step, and wire 27k between 16 Ohm tap and R19/C11 at PI. I've documented an oscillation, and I'm not sure I need to measure more right now. Better to try and get rid of it.

Any comments?

Nice work. It sounds like the oscillation is occurring before PI (around the tone stack), so NFB wouldn't be the first priority.

Are the pics still as in your main build thread? I wouldn't be surprised if the extra path(s) for the tone stack switching is the culprit. The tone stack is a sensitive area in a cranked Marshall circuit. I've had some issues with a "Neil special" build in the past.

Can we get some better pics of that area/details on how exactly you've implemented it? A good debugging approach might be to get the amp stable with a single configuration first.

 

[Pete Farrington]

both C8 and C9

From the photos in the build thread, it seems that the layout puts the tonestack circuit in extremely close proximity to the input sockets.
There's 360° phase shift and a gain of maybe 35dB between those nodes.
Hence the slightest degree of coupling creates a positive feedback loop with no margin of stability.
My guess is that the C6 test will stop the instability.
And a screening plate around the input sockets will greatly help the margin of stability, perhaps so as to resolve the issue (ie with C6 back in circuit).

 

[BlueX]

Thanks for the input!

How about "If the issue goes away with the pin 2 end of C6 lifted, that would help to confirm it."?

Forgot about that yesterday evening. Will check in next step.

Are the pics still as in your main build thread? I wouldn't be surprised if the extra path(s) for the tone stack switching is the culprit. The tone stack is a sensitive area in a cranked Marshall circuit. I've had some issues with a "Neil special" build in the past.

Can we get some better pics of that area/details on how exactly you've implemented it? A good debugging approach might be to get the amp stable with a single configuration first.

Pic's in post #75 (p4) are still relevant, except for the added Heater CT elevation in post #78 (but nothing changed on the main board). I will get some close-ups.

From the photos in the build thread, it seems that the layout puts the tonestack circuit in extremely close proximity to the input sockets.
There's 360° phase shift and a gain of maybe 35dB between those nodes.
Hence the slightest degree of coupling creates a positive feedback with no margin of stability.
My guess is that the C6 test will stop the instability.
And a screening plate around the input sockets will greatly help the margin of stability, perhaps so as to resolve the issue (ie with C6 back in circuit).

Wouldn't coupling between input and tone stack create oscillation already after C3? Or maybe I didn't notice due to small amplitude?

 

[Pete Farrington]

Wouldn't coupling between input and tone stack create oscillation already after C3? Or maybe I didn't notice due to small amplitude?

Good point.
The signal would be almost 60 times smaller, so yes, you might not have noticed it.
Plus just probing there might have stopped the oscillation, due to the probe / lead / scope input loading acting to disturb conditions, rolling off the high frequency gain a fraction.