100W "Blackflag" presence/bias issue

[MHProd]

Hi,

So I have a nice 67 100w blackflag with dual rectifier here. Had a bad case of ghosting... Changed all the caps while raising the filerting a bit (from 32uf to 47uf) except for the preamp. Ghosting is pretty much gone, can still hear it a bit when volume is on 8 but nothing unmanageable or too distracting.

I changed the tubes too and biased them properly. Amps sounds good (seriously, this has to be the best clean sound I ever heard from a plexi... probably because of the Drake transformer). I let the tubes drift a bit to rebias. All good.

While playing the amp, I'm noticing the bias went up significantly after a little while. I re-adjust it no problem.

Next day I'm playing it again, check the bias, it's now too low... Anyway, I did this a few times and the bias always changed beyond what normal tube drift would do.

Then I noticed something I never encountered before: The presence pot impacts the bias big time. It seems to happen with the pot between 0 and 2, resulting in about 10ma different on the tubes... With the pot is on 0 Bias is hot, when on 2 (and all the way up to 10) bias gets cold.

Any ideas on what this could be and how to fix it?

[Roe]

check the .1u phase inverter caps and the bias circuit

[Paul G.]

Coupling caps between PI and Power tubes leaking DC?

[MHProd]

Thanks guys! Ill check and report back.

Right now I'm troubleshooting a new build. Ugh...

[MHProd]

All right back to the black flag! I forgot about it...

So I changed the Pi caps (all 3 .1uf), no change. I changed the bias caps, no change. Changed the .022uf in the Pi too and the presence cap. As soon as I turn the presence pot, the bias lowers significantly ( around 10ma). I also found out the same thing happens with the treble pot, however, it only does it with the pot between 5 and 7. Otherwise it goes back to normal.

With V2 removed, the bias is stable... with V3 removed, bias is stable it drops -10ma. tried to change the caps in V2, problem is still there.
What could it be?

[wpaulvogel]

The board may be conductive an leaking voltage between turrets.

[MHProd]

I’ve checked all the grounds, replaced the nfb wires so they are far from everything, tried changing all the caps, measured all resistors. No change.

I changed the presence pot, same...

I grounded the preamp and the issue is still there, so it comes from the power stage. If I lift the nfb resistor, the issue goes away, the bias is stable but lowered by a few ma which I think is normal. I mean that the issue actually increases the bias current... what could cause this?

If the issue was conductive board, wouldn’t lifting the nfb resistor have no effect?

[Roe]

I’ve checked all the grounds, replaced the nfb wires so they are far from everything, tried changing all the caps, measured all resistors. No change.

I changed the presence pot, same...

I grounded the preamp and the issue is still there, so it comes from the power stage. If I lift the nfb resistor, the issue goes away, the bias is stable but lowered by a few ma which I think is normal. I mean that the issue actually increases the bias current... what could cause this?

If the issue was conductive board, wouldn’t lifting the nfb resistor have no effect?

try cleaning the area between the 82k/100k/220k resistors with alcohol

[MHProd]

Cleaned the board with alcohol, underneath as well. No change...

If it’s an oscillation, where could it be coming from?

[Roe]

Some tubes drift during warmup and break in. Are you using new gold lion tubes?

[MHProd]

Nope, Mullard xf2 but put some ruby tubes after I discovered the issue. It still does it after a significant warm up period.

I’m at a loss...

[pdf64]

From the symptoms described, I’m pretty certain it’s oscillating at low presence settings.
Due almost certainly to the (ultrasonic) characteristics of that OT and a high degree of nfb; is it 27k fed from the 16ohm output?
I had that same issue with a JTM45 build that had a real RS DeLuxe.
To get it stable at all presence settings, I had to fit 100pF caps across the 82k LTP anode load and the 27k series feedback resistors.
The 47pF cap across the LTP anodes was useless, actually seemed to make the issue worse.

However, that the treble control setting also affects it is a difference, and indicates there may be another positive feedback loop in play. Perhaps leave the valve in V2 out until the power amp issue is resolved. My guess is that the loop that includes the treble control might then be stable, or at least be fixable by tweaking lead dress around the controls and OT (they’re awfully close together on Marshalls!).

My amp was fine with the stock circuit and other OTs, it was just that the RS had a bit too much phase shift and s resonant peak between 40 and 50kHz; other OTs didn’t have either of these.
Also consider that the amp would probably be find with the later type of 22k presence control, as there’s some degree of presence boost even at minimum.
I only really chased it down as a technical exercise, as it sounded great and the presence never goes below ‘9’

[Roe]

Try moving around the purple nfb wire

[MHProd]

I had tried moving the nfb wire, to the pot and to the selector. Chopsticked it in different spots, etc... I didn’t observe any change with the bias.

Actually, with low presence settings, the amp goes back to normal. With presence on 0, the bias current goes up. If I disconnect the nfb, the bias is normal and doesn’t change when I turn the treble pot.

I did solder a 47pf cap accross the nfb resistor and the bias is now perfectly stable.
It doesn’t seem like it changed the sound too much. Maybe I’ll go lower if I find a cap (1pf?).

The amp did ghost before but it’s also much better now (?). Waiting for caps to up the screens to 50uf from 25uf hoping it’ll take out the remaining ghosting.

These amps are unique sounding, best Marshall clean sound imo, with volume on 7, the amp breaks up a bit, but not much, great to play rhythm, very Jimmy Page style.

[pdf64]

...
The amp did ghost before but it’s also much better now (?)...

With AB amps that are idling at the margin of stability, with signal the anode current, gm and hence gain increases, which can lead the parasitic oscillation, ie an oscillation that only manifests when a 'real' signal is present.
That oscillation could cause more HT current to be drawn than the wanted signal might cause, leading to a greater degree of HT ripple and hence its ghosting intermodulation effect.
As you seem to have improved the margin of stability, you might also have damped down a parasitic oscillation, hence there being less ghosting apparent.
And also reduced the wear on the valves, so a triple win