EL34 pp "fixed" bias is wandering

[OzMikeH]

First, apologies if this is a repeated thread. I have searched but all of the words I'm looking for are very common.

The amp is a push-pull EL34 with bias set by adjusting a pot near each tube.
I warm the amp up for 15 minutes, then set the bias for 40mA, if I check again 10 minutes later it can be anywhere between 36 and 50mA. so I re-bias it, listen for a while and check it again, and it's usually changed on one particular tube position. All positions vary, but one is worse than most. It's so bad I'm actually quite surprised if it hasn't changed.

I've replaced the bias pots, checked all connections, done limited tube swapping (changing positions, not new tubes).

I'm using Valve Art EL34s at the moment, should I try a new set of tubes?

[Boris_The_Blade]

If they are indeed new tubes, they make take a while to settle down. Give it perhaps another few dozen hours too see if they stabilize...

[Poindexter]

What value are your grid circuit ('bias') resistors, brah, and what kind of EL34s are you using?  Some EL34s can draw substantial grid current thermally, and the olde tyme 270K or greater value can give some bias drift.  My best audio buddy loves GE fatboy 6CA7s in triode, and those buggaz can really pull the bias around.

It would help also to know what the driver tubes are, and what the plate resistors are on them, in order to balance out the circuit paramaters.

Aloha,

Poinz

Edit; punctuation.  Sheesh.

[ray_moth]

I agree with Poinz that too high a value of "grid leak" resistor with fixed bias can cause bias drift. 220k or lower is usually pretty safe with EL34s, which are pentodes after all (with beam tetrodes it would normally need to be a lot lower than that). Don't forget to factor in the grid stoppers, because they form part of the "grid leak".

If you drop the values of the "grid leaks" you may need to increase the values of the coupling capacitors to compensate, to get the same roll-off frequency/time constant.

Of course, if you use direct-coupled cathode follower or source follower drivers, your "grid leak" resistors will already be low value and there will be no problem in that department.

[Miles Prower]

If you have access to a tube tester, you might want to test those EL34s. VTs that won't settle down under fixed bias could be gassy.

[OzMikeH]

In no particular order:
No chance of getting anywhere NEAR a tube tester.

Drivers are 6N8P via 0.1uF then 1k into pin 5. The plate resistors on the drivers are 40k to 50k.

Bias supply is 100k off the wiper of the 20k bias pot. Bias supply impedance is less than 121k worst case. the 100k connects between the 0.1uF and the 1k mentioned earlier.

The output tubes are new production Valve Art EL34B (chinese, slightly more plate disspation than other EL34) with about 30 hours on them. They aren't particularly closely matched. When I swap the push with the pull the bias goes to 32mA and 56mA

All of the tubes are brand new chinese tubes.

This amp can use KT88 or 6550 as well, power supplies have plenty of current.

Perhaps I should leave it on for a few days and get back to you.
How many hours until they settle? 100?

[Lowrider]

The same happened to me using JJ KT77, I changed to Shuguang KT66, and no more, now bias stays as I set, always...

[jon_010101]

I had that problem recently with a pair of 6AV5s. I put 'em back in their boxes and installed a new pair. Rock-sold-stable now

[ray_moth]

And the same happened to me with ersatz Svetlana EL34s (i.e. not =C=). The Gms were all over the place and eventually one tube packed in - behaved as if the screen was disconnected. Quality control was sadly lacking.

[OzMikeH]

I've had the amp on for about 6 hours so far this session.
I went out for 3 hours, before leaving I set all to 43mA.
One tube was 52mA, the rest were 45.
I've adjusted them all to 44.5, It's sounding bloody good. Giving it a bit of a workout with Imago by The Butterfly Effect with 2dB of boost at 32Hz.

Will leave the bias alone and report back tomorrow after a 15 minute warm up.