Very odd behaviour of preamp tube in vintage Gibson amp ("dancing voltage")

[slidincharlie (Carlo P)]

That confirms my theory regarding the power supply.
With the added caps the effect of the volume pot should be much weaker if any.

So annoying... Brand new expensive caps not doing their job...
Do you suggest to leave the whole assemblage (two parallel caps, total = 38 uF instead of 10 uF as per the schematic) for each filter stage, which is a bulky gizmo? Or may I try a single new 22 uF cap for each stage? Or rather buy two new 30-40 uF caps (if this value ain't too much compared to the original 10 uF shown in the schematic)?

A bit of this noise might be normal with these amps.
Does it really bother you with a normal/sensible tremolo setting?

Actually no, considering the weak ticking obtained with the increased filtering. I never use this amp at full volume, not to mention full tremolo depth.

[Helmholtz]

So annoying... Brand new expensive caps not doing their job...
Do you suggest to leave the whole assemblage (two parallel caps, total = 38 uF instead of 10 uF as per the schematic) for each filter stage, which is a bulky gizmo? Or may I try a single new 22 uF cap for each stage? Or rather buy two new 30-40 uF caps (if this value ain't too much compared to the original 10 uF shown in the schematic)?

Your 16µ caps might be good.
But increasing capacitance improves filtering especially at low frequency.
If disconnecting the 16µ caps doesn't make things worse, the 22µ caps alone should be enough.

[The Dude]

I'd also check bias. If your bias is too much on the "hot" side, any existing noise will be more audible.

[Chuck H]

FWIW I've had a few "new" electrolytic caps purchaced for filters that were bad on arrival. Atom caps on two occasions. I've quit using them because of this. If bad, new filter caps is ultimately confirmed as your problem I surely sympathize. But I wouldn't be surprised.

[slidincharlie (Carlo P)]

I have replaced the supposedly bad 16 uF filter caps (actually a 16+16 multisection cap) with two 22 uF caps (much cheaper, hopefully working). The ticking noise is still there and rather strong. Stronger than during the test with higher filtering (16+22 uF in parallel in each filter stage). I am disappointed, I thought that replacing the caps would have done the trick of decreasing (if not eliminating) the put-put.
As regards possibly unmatched power tubes: I know how to read the idle current on both tubes combined through the cathode resistor voltage drop, but how do I measure it on each tube separately? May I use the output transformer shunt method in a cathode-biased amp? I know that this method is risky but I have used it many times in the past in fixed-bias amps and feel quite confident.

[Helmholtz]

No reason to be disappointed.
We're trying to improve the amp's performance over its original condition.
Remember this is a very basic amp.

Your last test shows that 22µ caps are not sufficient for your demands, so you need to increase the capacitance to 32µ or even 47µ. Maybe its enough to increase only one of the caps.

The problem is caused by 2 independent effects:
1) The tremolo oscillator modulating the preamp supply voltages. This is where increasing filter capacitance helps.
2) Some imbalance in the power stage preventing perfect rejection of the tremolo modulation frequency.

While the OT shunt method would work, I don't recommend it.
Use the OT resistance method instead.

[Chuck H]

There's no guarantee that idle current tests will reveal what an individual tube is doing while conducting. (<period) That said...

You can plug in each power tube, one at a time, and read idle current for that one tube. Don't play the amp this way. Just take a bias reading. This would not be the ideal bias because you only have one tube plugged in but it does give you a single tube measurement. Test each tube this way. If they're the same brand/type/ and hopefully batch they should conduct in a similar way. So if the individual idle bias readings are grossly different you can guess that a mismatch in the power tubes is a big part of the problem.

[g1]

You can plug in each power tube, one at a time, and read idle current for that one tube. Don't play the amp this way. Just take a bias reading.

I think with cathode biased this can be problematic. At least for approximating idle currents.

[Chuck H]

I think with cathode biased this can be problematic. At least for approximating idle currents.

Ah. Sure. Because it's self correcting to a degree. I'll stand by it though as a means of seeing the tubes are grossly off. My own amp is cathode biased at idle (I do have a clamping diode that kicks in at higher conduction levels) and I individually test and match tube "batches" for idle current with definable readings and good results so far.

EDIT: Thinking on the matter... It might help accuracy in the case of doing this with cathode bias to monitor plate voltage in each test as well. Then it's still apples to apples.

[Helmholtz]

When you pull one of the power tubes, the remaining one will increase its current (and overdissipate).
I don't understand what information could be gained from doing this.

[Chuck H]

When you pull one of the power tubes, the remaining one will increase its current (and overdissipate).
I don't understand what information could be gained from doing this.

Dissipate more perhaps. No problem with glowing screens or plates doing this so far. Though I've never tried it with an amp other than mine. It has high-sh Vp and cathode biased about 85%. So I can't say if tubes would overdissipate with an amp biased at 100% or more. Maybe.

FWIW there are several tests suggested here than involve pluggin in one power tube at a time and I don't remember any one other than me, right now taking heat for it.?.

The information would be whether or not each tube, when tested individually, is drawing the same current at idle. Which is not a perfect way to match tubes but may tell us if they are grossly different.

[Helmholtz]

Dissipate more perhaps. No problem with glowing screens or plates doing this so far. Though I've never tried it with an amp other than mine. It has high-sh Vp and cathode biased about 85%. So I can't say if tubes would overdissipate with an amp biased at 100% or more. Maybe.
.

As current must increase with only one tube , the remaining tube will overdissipate when the amp was biased around 100%.
The increased current will not give reliable info on normal idle current.
Two tubes matched or balanced at some current may drift apart at a different bias.

What's wrong with just measuring idle currents using the OT resistance method or even add 1R individual cathode resistors?

FWIW there are several tests suggested here than involve pluggin in one power tube at a time and I don't remember any one other than me, right now taking heat for it.?.

Sure, but that's with fixed bias which will not change with one tube only (because it's fixed ).

[Chuck H]

What's wrong with just measuring idle currents using the OT resistance method or even add 1R individual cathode resistors?

Nothing at all. Except that I've seen people seem to have trouble wrapping their head around the OT resistance method and installing the 1R resistors means you need to have 1%/1R resistors on hand. It's also more trouble than just plugging a tube in. Since measuring cathode voltage one tube at a time has worked out well for me I thought to suggest it. Matching my batched tubes at what might seem like an arbitrary current has never resulted bad matching so far.

Sure, but that's with fixed bias which will not change with one tube only (because it's fixed ).

Ah. Of course.

[slidincharlie (Carlo P)]

Here's my report on power tube bias.
The bias with both tubes on is on the cold-ish side, with 33 mA drawn from each tube (corresponding to abut 8.4 W or 70% of the maximum dissipation for an old Marconi 6V6 tube).
According to the "one-tube-at-a-time" method suggested by Chuck H, the tubes displayed very close readings - so, if I understand correctly Chuck's words, the seller (a trusted one) sold me a well matched pair.
Then I tried the OT resistance method as per Helmholtz's advice, but I did not succeed because I could get no ohm reading between the center tap (probe on the first filter cap or on the rectifier lug, alternatively) and either tube plate. My Fluke multimeter gave a flashing 0.L (=no reading). Why is that?

[Tom Phillips]

... My Fluke multimeter gave a flashing 0.L (=no reading). Why is that?

That's strange. Are you sure the meter was set to the DCV function?
Could it need a new battery?