Mullard ECL82 stereo - where has the voltage gone?

[qualityten]

Chris, I don't know. There is a black wire linking the HT centre tap to the screen tag (bottom right in the pic). That tag is in turn connected to the 0V tag of the heater supply via a thin bare wire. Both were there on the radiogram originally if I recall correctly, but I guess I can't damage the transformer by temporarily disconnecting either or both of them?

Al, yes, I did see your mention of the bulb holder. The terminals are isolated from the chassis, and at present one measures 6.3V and the other 0V.

[PJL]

Originally Posted by qualityten

With all the above attended to, I’m pleased to report that the amplifier is working well.

But there is a softer but still very audible 50Hz hum. I’ll try and get a large 6v battery today to see if it disappears with a DC heater supply. A humbucking arrangement is ruled out because one side of the 6.3VAC is at 0V. I wish I knew how Dynatron did it!

Not sure I follow this as you were using 2x 220 ohms so that neither of the 6.3V AC connections were directly tied to 0V. A humdinger is a wirewound pot wired across the heater with the wiper connected to 0V. The theory is any hum introduced by coupling from the heater connection can be balanced out by fiddling with the relative amplitude of the opposing phases. Your 220/220 split is a fixed 'humdinger'.

[qualityten]

I've now disconnected the two wires connecting the HT centre tap to the screen tag. This gives 3.1V and 3.1V on either side, which is progress, following Chris and Al's questions.

On this basis I've reinstalled the 220/220 fixed humdinger. The 50Hz is still there, but reduced so I'll now experiment with a 100/100 pot.

[qualityten]

I have installed a 250ohm pot, which makes no difference to the hum.

All the voltages are spot on, except those for the ECC83. The anode voltages there are 20V too high (110, rather than 90V) and the cathodes too low (0.9, rather than 1.5V). The C2 capacitors 2200pF are very old. My LCR meter measures them at 1800pF each, but I wonder whether a fault/leakage there would raise the voltage? (Running out of ideas here!)

Failing that, should I look for inadequate power supply filtering? I have an oscilloscope and have just got a high voltage probe.

[snowman_al]

Okay,
I assume you have tested it with both inputs connected? Remove either left or right input and try it.
HT filtering is 100hz hum. Ground loops and heater wiring can result in 50hz 'buzz'.

[qualityten]

Removing either input has no effect, with or without a shorting plug added. It's remarkably stable, even with both inputs disconnected. It's definitely 50Hz.

[PJL]

Originally Posted by qualityten

I've now disconnected the two wires connecting the HT centre tap to the screen tag. This gives 3.1V and 3.1V on either side, which is progress, following Chris and Al's questions.

On this basis I've reinstalled the 220/220 fixed humdinger. The 50Hz is still there, but reduced so I'll now experiment with a 100/100 pot.

Are you saying that the connection on the transformer we have been calling 'screen' is the centre tap for the 6.3V? A screen is normally just that, a metal screening applied to the transformer between the primary and secondary windings and should not show continuity to any winding.

[qualityten]

Originally Posted by PJL

Are you saying that the connection on the transformer we have been calling 'screen' is the centre tap for the 6.3V?

No. I'm saying that the 6.3V winding is separate, but has no centre tap.

[PJL]

My understanding is the circuit 0V, the transformer screen, the HT centre tap and the wiper of your LT humdinger (or centre of the 220/220 LT divider chain) should all be wired separately to a single point where it is connected to the chassis.

[qualityten]

Yes. What I don't understand is why the transformer originally had a thin uninsulated wire also connecting the transformer screen tag with one side of the 6.3V winding.

And, can the single point be the C15 ground, with a connection from there to the chassis near the RCA inputs?

[Diabolical Artificer]

"And, can the single point be the C15 ground" Yes, but remember what I said about big currents.

Another approach is to try and find where the 50hz is getting into the audio stream, do you have 50hz on the IP V1 grid, if so you can fit another humdinger with reference to grid.

It could also be a badly placed bit of wiring, try poking about with a stick, gently pushing heater wiring from being so close to decoupling caps, cathode resistors etc. Htr wiring if it crosses other wiring, should do so at 90 degress & shouldn't lie along side.

A pic of the underside of your amp may help, someone may spot a potential issue.

Another potential improvement can be to up any cathode and g2 decoupling (by in/decreasing C), tweaking the 3dB point, same goes for any other coupling, which can be tweaked to attenuate 50hz, but the grounding layout is the first place to look.

Andy.

[Sideband]

Originally Posted by Diabolical Artificer

this side connect is exposed when you take the cap off, if you have live connected to it, therefore you have live that can be touched. Usually the L is connected to the bottom lug & can't be accessed.

Not if you have an approved fuseholder. In safety testing, live and neutral are both considered as 'live' so an approved fuseholder both cantacts should not be accessible when the cap is removed. I can't remember the distance for the side contact off-hand but 3mm seems to ring a bell somewhere so with the cap removed there should be 3mm distance from the screw top to the side contact. There are many fuseholders out there that do not meet this spec.

[peter_sol]

Just to clarify.
Your transformer screen, Ht winding centre tap and one side of the heater wires should be earthed to chassis. Make that point the centre of the chassis. No need for resistors or humdinger pot.
Make sure Neither heater side goes to chassis elsewhere. Heaters should be with a twisted pair run close to the chassis. Don't let the input socket earthy side touch the chassis. Earth it to the buss bar, which in turn should be earthed to the same central point.
Hope that helps I've recently had all the same problems
and have made 5 amps, 4 single ended and 1 push pull, and the hum is very low now.
I am aware that the Mullard instructions say different but all the above cured my problems.
Good luck with it.

[vidjoman]

The fuse holders in the photo look like old reclaimed bits. Today, if new, they would need to be of the correct standard.

[snowman_al]

I do not know of the Dynatron the transformers came from, but as we learn more about the original wiring I suspect the heaters were fed one side through the chassis, that is the side that linked the HT centre tap, the screen, and the bare wire to the 6,3 volt winding. (That also means the HT -ve was originally also through the chassis.) So I speculate each of the valve heaters were wired one side to the chassis and the other via a single wire looping round valve to valve from the other side of the 6.3 volt winding.

Thinking out loud, to prove that it is a heater problem without using a battery, why not put a temporary switch in the heater line. So with the amp operating, warmed up fully etc, you can switch the heaters off. If the buzz stops immediately it is a heater / wiring / cathode heater fault. If it fades away as the heaters cool, we need to look elsewhere...

[qualityten]

Al, you are a genius! I still have the Dynatron chassis and that is how the heaters are wired up. As I’d prefer not to have to buy another power transformer. Does this mean a heater rewire lies in my future? And further ground requiring?

Following Andy’s question, I used my new high voltage probe to look for hum and found it on the grids and anodes of the tetrode of V3 and V5, but not earlier and not on V2 and V4. (I think. I’m still not confident in using an oscilloscope.)

Off to try the heater switch test now.

Ps re the fuses, they are both recent, one new and the other used, but recently bought.

[Herald1360]

It would seem to me that the hum problems here are caused by the wiring of the signal ground, 0V supplies and heaters being neither as per original Dynatron (rather poor grounded one side of heaters and everything else just taken to chassis though this can work well enough if done carefully and has assembly cheapness as a benefit) nor as per Mullard design nor even as per any other one reasonable layout. The problem will likely be sorted by choosing one scheme and sticking to it. Starting from where it is now, stripping out all the heater and 0V wiring to revert to Dynatron would seem to be more work than going with the centre tapped (humdinger) heater circuit with twisted pair heater wiring and simply tidying up the PSU filter capacitor 0V and signal grounding along either single point ground at the Filter cap -ve or busbar ground as described, lines. The TX screen can simply be taken independently to chassis and the incoming safety earth connected to chassis near the mains inlet using its own dedicated fixing not shared with any other function.


It can be surprising just how much can be "gotten away with" but also intensely frustrating to sort out hum problems arising from poor layout!

[qualityten]

I connected a switch to the heaters of all the valves, except the EZ81 (the latter left in for convenience of access). When those heaters were switched off there was no change to the hum.

[PJL]

No to rewiring, your twisted pair and humdinger is a much better arrangement. If you say that you fitted a humdinger (a wirewound pot wired across the LT with the wiper connected to the common ground) and rotating the pot made no difference then it is reasonable to assume that the heater is not responsible.

Tracking down hum with a scope is difficult particularly in an amplifier with -ve feedback that will be trying to counter any hum. The scope can be used to check the hum frequency, post a picture of the scope trace for the output and HT here.

You also need to keep in mind that a PP output transformer is very good at cancelling hum and output pentodes are close to constant current and will only translate HT ripple voltage into a small AC current in the transformer.

[Diabolical Artificer]

"Not if you have an approved fuseholder." Good point, but it's good practice anyway.

"Just to clarify.
Your transformer screen, Ht winding centre tap and one side of the heater wires should be earthed to chassis. " It may be better to use "earth" with regards to mains earth, and ground for all other grounds, else it's easy for beginners to get confused.

"Following Andy’s question, I used my new high voltage probe to look for hum and found it on the grids and anodes of the tetrode of V3 and V5" looking at the schematic via the link you posted, i couldn't find V5, but I did notice V3a's grid isn't decoupled. I wonder if popping a 10u cap in parallel with R12 might improve things?

When you say HV probe, is that with your scope? Can you confirm it's 50hz and not 100hz? 100hz ripple is a common problem on long tailed pairs, at least I've found that to be the case. You could try increasing C8 to 22u if it's 100hz ripple. To find the frequency, get a steady, stable image on the scope, use your horizontal shift to line up a peak with a graticule line, then count how many divisions there are until the next peak. EG if your TB is set to 5mS/div and there are two divs from P- P, do this sum - 1 divided by 2 x 0.005 = 100, so that'll be 100hz.

Try the cap across R12 first, hope that works, Andy.