Valve amplifier service

[simon]

Well, there you go. (Weren't the words I actually used...)

I cleared my desk again and set up the audio stuff and took the bottom off the amp. The mains AC cables to the front on/off switch are fast-on type connectors - removing them with pliers and care, a wire came straight out of one of the spade connector thingies. Soldering it back on I made the circuit with clipleads and moved the harness away from the front to the side towards the rear and with some trepidation switched on. The louder hum on the left channel has reduced to similar to the left channel loud. Possibly a smidge louder using a BS calibrated ear. But a significant reduction nonetheless.

Fecking feck arse.

Dave might even find it listenable now, but the remaining hum would annoy me I think.

So, perhaps it's always hummed on the left channel? I can't really believe that this would be the case widely with this amp though - the WAD customers wouldn't have been happy.

Solutions.

The obvious one would be to put a new switch at the back out of the way, but that's not easy and not especially convenient.

Some sort of always-on low DC supply - might struggle for space, but might be able to use the same switch at the front?

"AC proof" V1 and V2? V1 (the gain stage) is closest and presumably most likely to be the offender?

[Mike H]

Well, there you go. (Weren't the words I actually used...)

[simon]

Well, I don't know.

I had a trip to Nick's yesterday for a scope lesson and the good news is it's operator error not scope error - the scope looks to be okay and the low level mush is most likely my inexperience.

So I had the bottom off the amp again today with the last lap in sight, but before I did anything I thought I should just check it out again, just so I knew what I was looking at. Except this time the hum didn't really change when I moved the AC wiring to the front switch. I know it did last time as I double checked it, though only with my ears for instrumentation.

This is what the scope picked up, probes on the output - top trace the louder left channel, the bottom trace the quieter right. But the scope would only give a sensible trace on the channel it was set to trigger from, the left channel in this case.
.

20240204_160738.jpg (107.42 KiB) Viewed 8493 times

.
The peak to peak amplitude is around 1.25V. But the frequency looks to be 1M Hz (1,000,000 Hz).

Triggering on the quieter right channel on Ch 2 of the scope gave a peak to peak amplitude of nearly 3V at a similar frequency.

So this does look like oscillation somewhere?

[simon]

Whilst the bottom was off I disconnected the coupling caps between phase splitter and output valves in order to isolate the output stage, and with the AC wiring to the side and rear of the amp there was little detectable hum. This is what the scope picked up.
.

20240204_160827.jpg (84.08 KiB) Viewed 8489 times

.
Not much really in comparison, around 10mV peak to peak at a frequency of 50kHz.

Connecting the AC wiring back to the switch I could hear a little hum. Reconnecting the coupling caps and the was a little louder than when I started looking at the amp today.

I might need another trip round to Dave's again just to check what's happening in his system. And as Nick suggested yesterday take the grounding plugs too, in case Dave's kit is causing earth loops somehow.

Does oscillation look to be correct? If so it looks to be coming from one of the ECC83s...

[Nick]

If it is oscillating, I would have thought the output pentodes were more likely to be the cause, maybe grid and screen stoppers could be tried if not there already. Getting a ecc83 to do that seems unlikely at first sight.

[bostod]

So I have got the amp back home. Simon plugged in his phone as a source and the amp was quiet.
I was keen to put the amp back in my setup but when I did the audible noise came straight back. God dam it. Tried swapping out tube pre to solid state, then passive. All produced the same hum. Put grounding plugs on preamp. Unplugged cd player no hum. So I figured the power to pre was ground loop free. Tried plugging in cd player on a different electrical ring. Hum returned. Finally I've tried a RCA group loop isolator between rca cable and cd player. No hum. Fingers crossed.
I think the way the amp is grounded is causing the issue but it would need a redesign to solve that. No wonder Simon was pulling his hair out. I think there must have been multiple noise issues of which are now fixed/solved.
I say fingers crossed as with Simon, this amp has been quite temperamental to different noise issues.

[Nick]

I think from the above description, you might also point the finger at your CD player. What is the CD player? Is it Type II isolated or does it have a ground connection?

You could isolate the chassis and safety ground on the amp from the signal ground with a small resistor, but I dislike that idea in a valve power amp as I have seen cases where a fault would burn out the resistor without anyone noticing leaving the fault in the amp to find ground via the rest of the system, or via the user. A safer higher current way of doing the same thing is to use the Vf of a couple of diodes in a 25amp bridge rect to isolate the grounds as long as there is no fault current/voltage.

[simon]

My gut feeling is lifting the amp ground might well solve some problems, but it's a bit of an involved job.

That looks like it's a Tubecad dwg? Any suggestions for values of the cap and resistor? I think I might use this in my amps rather than a simple resistor.

[simon]

Found the Tubecad blog which has values

https://www.tubecad.com/2019/11/blog0482.htm

[simon]

I'm not sure I really understand the full circuit - there appears to be two circuit ground connections to mains supply ground. One through the bridge, R and C - the other from the bridge via a cap. Is the single cap C14 to allow ripple currents in the bridge to have an easy path to mains supply earth?
.

PS-3 HV Schematic and Part Values.png (121.61 KiB) Viewed 8166 times

[Nick]

Is the single cap C14 to allow ripple currents in the bridge to have an easy path to mains supply earth?

Maybe, sounds like could be correct. Not ripple though, its 10nf, so will be to pass switching noise from the diodes I would have thought. HF signals will see the wire to the other end of the 0v line to be a series of inductors, so the direct path to ground via the cap will be a lower impedance route to get noise out of the system.

[simon]

Hey, I was nearly right. Ish.

[simon]

The amp has been pretty well behaved at Dave's for three months now. It's still had its odd moment and Dave's had to use ground breakers at times, which isn't great as he can hear the device. But mostly it's been okay. Phew.

Dave wanted to try a volume pot in the amp itself rather than use a pre amp. So I put together a pot with short leads to connect to the amp's input phonos which Dave really liked. But there's just too much gain in the amp for Dave's speakers and listening and just a gnats on the volume pot is loud enough. I tried adding a resistor divider with two 4k7 resistors in front of the pot, a c8dB loss I guess, and it's now loud enough at 9 o'clock, so still far too insensitive.

It perhaps needs as much as 24dB loss, but that's just a guess.

I don't really like the idea of throwing away so much signal because there's too much gain in the amp. I don't want to mess with the phase splitter but I wonder if it's worth changing the ECC83 SRPP pre stage to something with less gain.

[simon]

A reminder of the circuit:
.

WAD K5881 mk 2 small 2.png (215.18 KiB) Viewed 4544 times

.
Consulting the bible (thank you Mr Jones) the gain of the SRPP is around 82. So I'm guessing really that a gain more like 30 might be better. Playing with different op points for the existing ECC83 pre valve as a simple common cathode stage, even with an unbypassed cathode resistor is still a bit high really. A Va of 250V is around 39, but the load resistor is only c2.3x the ra, so a bit low really.

ECC81 might be a better option, or perhaps an ECC82, but I don't have any of either. I think I have some 6CG7 somewhere which I could try if I rewired the heaters. They draw a bit more current too, hopefully the transformer is up to it...

Anyway, here are some numbers. I can't really balance up the best compromise of current draw on the HT, gain, ratio of RL to ra, and heater current. I'm thinking the simplest and probably best is to parallel the two halves of the ECC83 with Va=200V and don't bypass the cathode resistor. I might still need some gain throwing away though before the pot.

Am I missing anything, or does anyone have better suggestions?
.

WAD K5881 mk 2 small 2.png (215.18 KiB) Viewed 4544 times

.
Don't ask me why the circuit has been displayed twice...

[Mike H]

It might be much simpler - I've just clocked the NFB resistors ratio - 201:1 ?! (20k : 100R)

Cripes!

100R for R4 is fine, but I'd look at reducing R18, say to one tenth its current value - easy to do, just bung another resistor in parallel, say 2.2k. That should get us about 20:1, which I would have thought was plenty for most practical purposes. (My amps mostly tend to be 10:1.)

I wonder if '20k' could be a typo?

It might fix the erratic noise issue as well!