swinging CCS PP output stage

[peterr]

I am working on a 807 PP amp.
It is a two stage affair going from a CCS’ed LTP ECC81 into pentode connected 807’s using partial feedback more or less like Gary Pimms 47PP.
To bias the 807’s I have a negative supply at around –25V to give 65mA per valve at 350V B+. Works fine.

Next experiment has been to CCS the output stage as well. To do this I built a DiyAudio CCS set to 130mV to give me the same 65mA per valve and this is where the weird stuff starts.

The CCS does its job and keeps the 130mV steady but the way these 130mV are divided over the two 807’s isn’t steady at all. It is not just that the valves draw an uneven current but it keeps changing constantly from let’s say 80mA for valve A and 50mA for valve B to 60mA for valve A and 70mA for valve B a minute later. Trying to level the two valves using some negative bias doesn’t affect this behaviour.
Before I put the CCS in the 807’s could be biased easily and remained stable without any problem.

Is anybody able to shed some light here? Particularly:
Is this normal?
What causes it?
Is it a problem anyway?

Thanks

[hemgjord]

"Is this normal?#

No

"What causes it?"

Good question

"Is it a problem anyway?"

Probably.

Try to post a schematic, it is always easier to discuss over something concrete.
The only thing I can think of is some HF oscillation somewhere.

/Olof

[Boris_The_Blade]

Maybe you have some LF instability. Do the currents fluctuate with no feedback?

[Svein_B]

peterr, where is the voltage source for your CCS?

If the negative side of the CCS is connected to 0V it may be a problem. If you do not have sufficient voltage "headroom" for the CCS to do its work, the negative bias for the 807 grids might be the reason for limiting the curent in the 807's.

You could try to reduce the negative grid bias and see if this helps to keep the balance more stable.

SveinB

[planet10]

Originally posted by peterr
The CCS does its job and keeps the 130mV steady but the way these 130mV are divided over the two 807’s isn’t steady at all. It is not just that the valves draw an uneven current but it keeps changing constantly from let’s say 80mA for valve A and 50mA for valve B to 60mA for valve A and 70mA for valve B a minute later.

AFAIK you can only get away with a single CCS if you are running Class A triode which gives you a fully differential Allen Wright-style amplifier

In Pentode or Class AB you need 2 CCS like in Gingertube's Baby Huey

dave

[peterr]

Thanks for your answers.
I'll try to do a schematic but that will take a while.

Originally posted by Boris_The_Blade
Maybe you have some LF instability. Do the currents fluctuate with no feedback?

I took the feedback of to test this. No difference.
I also decoupled the first stage to make sure it doesn't affect things. No difference.

Originally posted by Svein_B
peterr, where is the voltage source for your CCS?

If the negative side of the CCS is connected to 0V it may be a problem. If you do not have sufficient voltage "headroom" for the CCS to do its work, the negative bias for the 807 grids might be the reason for limiting the current in the 807's.

You could try to reduce the negative grid bias and see if this helps to keep the balance more stable.

SveinB

The CCS goes to -15V. I've tried both with negative bias and with the grids earthed. No difference. In the first configuration the CCS has about 20V to work and in the second around 35V. Actually in earthed grid configuration the swing seems to become even larger.

Originally posted by planet10


AFAIK you can only get away with a single CCS if you are running Class A triode which gives you a fully differential Allen Wright-style amplifier

In Pentode or Class AB you need 2 CCS like in Gingertube's Baby Huey

dave

I know you can't do class AB but why not pentode? Look at the 47PP from Gary Pimm for instance, he does the same thing.
Gary's 47pp schematic Go to Rev 9 to see the CCS.

[oshifis]

I also tried CCS in my 807 PP amplifier. It was a separate LM317T in each cathode connected in CCS mode with a resistor and bypassed with a capacitor. But I changed back to fixed bias for several reasons: cathode bias limits the drive voltage swing, the stabilizer blocks were too hot, and I wanted to avoid the bypass capacitor. Fixed bias is quite stable. The operation point in my amp is a bit different than yours: the tubes are connected in triode, B+ is 425V, Vg1 is about -36V to -40V, Ik = 50mA. The negative bias supply is -50V.

[peterr]

Originally posted by oshifis
I also tried CCS in my 807 PP amplifier. It was a separate LM317T in each cathode connected in CCS mode with a resistor and bypassed with a capacitor. But I changed back to fixed bias for several reasons: cathode bias limits the drive voltage swing, the stabilizer blocks were too hot, and I wanted to avoid the bypass capacitor. Fixed bias is quite stable. The operation point in my amp is a bit different than yours: the tubes are connected in triode, B+ is 425V, Vg1 is about -36V to -40V, Ik = 50mA. The negative bias supply is -50V.

I have had it in triode too but the pentode/partial feedback version really is a best of both worlds I think. By the way isn't 425V in triode a bit much for the screens?

[peterr]

Today I took the whole thing apart to try to find the problem. I'm afraid it's rather simple... I went back to non CCS fixed bias and it turned out the 807's ran away in that configuration too and after putting in a fresh set all was well. Then I reapplied the CCS and now it is nicely stable. Listening to it now and sounding really really good

The weird thing is that before I tried the CCS it was in fixed bias too and stable but after going back to the exact same schematic it isn't any more.
Does anybody know what causes a valve to run away?

[planet10]

Originally posted by peterr
I know you can't do class AB but why not pentode? Look at the 47PP from Gary Pimm for instance, he does the same thing.

Looks indeed like Gary is using it in Pentode... i was just going by what Allen Wright told me (hecame up and spent 3 days up here after the VSAC where he introduced the concept)

I am very happy to learn that pentode is fine.. i really like what this kind of feedback did for the SE EL84 (RH84) and have been saving parts to build up a PP RL84 version. Being able to get 10W instead of 4 will be a big boon.

dave