Mullard 5-20/Beam Echo DL7-35 Strong Hum

[frankmcvey]

This is the same amp with which I had distortion problems some months ago; it's a 1950s machine that owes a lot to the basic Mullard 5-20 circuit. It's been on the back burner while I got on with other stuff and has bubbled its way to the top again.

Circuit diagram here.

With the amplifier running on full mains voltage and the audio input shorted with a 4K resistor, there is a strong hum from the speaker, appearing as the valves warm up. It's at 100Hz and does not vary with the volume control.

Removal of V1, the EF86 driver stage, gets rid of the hum almost completely, leaving just the faintest of 100Hz hum that I can only just hear with my ear against the speaker, and about what I would expect if the amp was serviceable. This would lead me to suspect that my problem lies in the V1 (EF86) driver stage or its supplies.

Replacement of V1 resulted in a reduced hum (still unacceptable), but I think that may be down to the fact that the original EF86 tests as very strong according to the HSVT, but the replacement, while still serviceable, is considerably weaker on test.

Since the hum is at 100Hz, I'd expect it to be HT-related, rather than 50Hz heater break-through.

My first thought was that new reservoir/filter caps in the HT chain had broken down, however, on scoping the HT chain, I found that hum on the HT line at the driver stage (junction of R2/R3) was very low, even down to the 5mV/div setting on the scope. See Pic 1 upper trace

Moving the scope down on to the anode of V1, there was a fairly strong hum signal, above 20mV – see Pic 1, lower trace

Taking a slightly different tack, I compared the HT chain on the unserviceable amp against its serviceable stable-mate. The AC hum figures using an AVO 8 were as follows:

Top of C14 reservoir Good Amp 112VAC ripple Bad Amp 125V AC ripple

Top of C11 Smoother Good Amp 52VAC ripple Bad Amp 55VAC ripple

Top of C6 Smoother Good Amp 37VAC ripple Bad Amp 40VAC ripple

Top of C5 Smoother Good Amp 0V AC ripple Bad Amp - see next para*

* A slight deflection on 250VAC range on the AVO; slightly larger on the 100VAC range, audible hum decreases momentarily then comes back; virtually no deflection at all on the 25V range and the audible hum decreases significantly and stays low.

All in all, there isn't a great deal of difference between the good amp and the bad one; I expect the anomalies on the 25VAC range are down to the AVO shunting the HT line.

Components changed during previous work:

C11, C6 and C5. During test on this fault, I substituted C14 for a 4uF 450V electrolytic, which made no difference to the hum during the 30 seconds or so that it lasted before expiring with a pfffft and a wisp of smoke (HT at this point is around 465 DC)

R9, R10, R2, R3 and R4.

I then scoped around the other V1 electrodes with V1 fitted, audio input socket shorted and strong hum from the speakers

Pic 2 shows HT on the upper trace and V1 control grid on the lower.

Pic 3 shows HT on the upper trace and V1 cathode/suppressor grid on the lower.

Pic 4 shows HT on the upper trace and V1 screen grid on the lower.

So I'm mystified at the moment and, having gone through the checks several times, I find myself in need of inspiration.

Any ideas, please?

Cheers,

Frank

[bikerhifinut]

I wonder Frank, if there's any dry joint or component failure around the first stage? There's R1, the grid stopper and VR1 to look at. And then the cathode resistors and decoupling caps?
I'd be wondering about that area but you say its 100hz which does suggest coupling through from the HT?
All the same I would have a good look around that area. It might even be just a bad earth connection at the ground end of the pot?
Turning the volume pot down to minimum will effectively short the grid stopper to earth, which kind of negates the resistor on the phonos. I usually test with a shorted phono plug to eliminate any externally induced noise.
It also looks like Avantic built this amp without a grid leak resistor on V1, I have never liked that idea as it relies on the wiper of the volume pot always making a good contact, and the track of the pot itself being in good nick to give that route to earth.
If you have already investigated this area then i apologise.

Andy

[G8HQP Dave]

The hum in Pic 1 looks rather asymmetrical, more like full-wave rectified 50Hz than pure 100Hz sine wave. Is it possible that there is coupling from the PSU to the first stage? Could be magnetic or a common ground. Old amps often had ground arrangements which would be considered poor by modern standards.

The slightly strange behaviour when you measure hum at C5 might be a clue - but I don't quite understand what it means. Loading C5 will shift the DC bias for the LTP, and this Mullard LTP is quite sensitive to bias, being easily pushed into grid current on one side.

Always remember that strange hum problems can be caused by oscillation.

[barrymagrec]

Heater cathode leaky on the phase splitter?

[PJL]

Pulling V1 tells you nothing as it is DC coupled to the phase splitter which will then not work at all.

I guess you have compared the valves DC conditions of the good and bad amps?

Have you already tried swapping the full set of valves in the good amp with those in the bad? If so its not a valve.

V1 has -ve feedback from the output transformer so any hum will end up there. You could try disconnecting the feedback but I would use a dummy load and a capacitor from input to earth to stop the thing oscillating. You should then be able to locate the origin using your scope.

[ronbryan]

I don't think that checking the hum level with the EF86 pulled is a very good idea, because it is DC coupled to the phase splitter. With no anode and screen current to the EF86, there could be around 400V applied to the grids of the phase splitter, saturating both triodes and reducing the gain to zero and as a side effect, exceeding the maximum permitted heater-cathode voltage. A better technique to eliminate the effect of hum generated by the EF86 would be to decouple the anode with a reasonably large capacitor.

Ron

[Boater Sam]

Does changing the output (speaker) impedance throw up any clues?

[julie_m]

Excessive ripple on the screen grid of a pentode can manifest as hum.

[ukcol]

Originally Posted by frankmcvey

......it's a 1950s machine that owes a lot to the basic Mullard 5-20 circuit.

Yes, a Mullard 5-20 with speaker impedance switching and a volume control added.

Originally Posted by PJL

Pulling V1 tells you nothing as it is DC coupled to the phase splitter which will then not work at all.

Originally Posted by ronbryan

I don't think that checking the hum level with the EF86 pulled is a very good idea, because it is DC coupled to the phase splitter.

That direct coupling means that DC conditions are critical. Do make those voltage measurements as suggested.

[frankmcvey]

Thanks for all your comments and observations.

Taking a look at Peter's suggestion of attaching a dummy load, placing a cap across the input to prevent oscillation and disconnecting the feedback loop, I broke the FB loop at the junction of R13 and R14. On connecting the scope to the flying ends of the resistors (where they would normally connect to R5/R6 junction), I got the results shown in Pic 1.

This shows just a millivolt or two of hash on the anode of V1 at 50Hz (where before we had upwards of 20mV at 100Hz) but upwards of 1 volt at 100Hz coming into to the V1 cathode circuit via the feedback loop, so it seems we're on the right track and this is where the 100Hz is entering the signal path.

Just had a visitor arrive, so I'll have to be quit investigating and be sociable. I'll get back to you tomorrow with the progress.

Cheers,

Frank

[PJL]

You have tried swapping all valves with the good amp?
Also you should check DC voltages to make sure everything is taking the right current, particularly the phase splitter anodes and the output valve cathodes.
Hum problems can occur if there is a fault with the output transformer primary as the two primary halves cancel hum introduced by HT ripple but faults here should show up with your voltage measurements.

[Sideband]

Maybe leakage across the valveholder?? Just suggesting something that hasn't been mentioned yet

EF86's are pretty sensitive so even slight leakage across adjacent pins might cause a problem.

[frankmcvey]

Hi, All,

Swapped out all valves with the serviceable amp - no change.

I've also checked every voltage on every valve electrode in both amps - not long since I finished However there are a number of anomalies - with the serviceable amp as well as the humming one - so I'll do some double-checking again tomorrow before I publish the results.

Note - this is the first time I've used my valve breakout boxes in anger since I made them last year. I made one octal and one B9A. Pic of the octal one below. Makes it a snip to check valve voltages and (to a lesser extent) currents when you can't easily get to the back of the valve base. Chocolate block terminal, valve base, small enclosure, matching octal or B9A plug ( I found mine on Ebay), some fairly flexible wire and a handful of ferrite beads (just in case) and you're away.

The plug with the label sticking out of the amplifier is a standard phono plug with a 1K resistor across the input (not 4K as I said in an earlier post ) used for shorting the input. A bit redundant on this amp, since it has a volume control across the input as someone has observed.

Dummy load in the background 4 x 8 Ohm @ 100W resistors mounted on a redundant heatsink and attached with croc clips to the amp output terminals. It'll be switchable 4-8-16 Ohms in a nice box with proper connectors when I get around to it...

Cheers,

Frank

[Boater Sam]

Only problem with a breakout box Frank is that it will induce hum and/or instability on its own.
I have a B9A socket mounted on top of a B9A plug with external links from pin to pin, except the heaters, that I can prod for voltage checks. Being only about 30mm tall, it doesn't upset the valve running conditions too much.

[frankmcvey]

Hi, Sam,

Well, all those unscreened wires running in parallel certainly doesn't help, and it might give problems with more touchy amps, but so far it seems to be happy with this amp. I've put ferrite beads on each pin in an effort to calm things down. It's certainly handy while it behaves - it might be an idea to put a scope connection on the dummy load so at least I'll get some warning if it does break into song!

Cheers,

Frank

[Diabolical Artificer]

Sometimes it's possible to take readings with the valve not quite pushed all the way in so you can get at the pins, you have to be carefull though.

My dummy loads (stereo) have 3 BNC in series, in parallel with the dummy load which means you can put a scope, meter and analyser on easily with out leads flying everywhere. The only problem with this is it means the grounds of the two channels are tied to mains earth at the scope which is ok for most tranny amps but can be tricky when testing valve amps with imaginetive OPT/ground/FB layouts.

Andy.

[cathoderay57]

If you haven't done so already I would try replacing C1. It is evident from what you say that the hum seems to originate from V1 area and is HT related.

[PJL]

The most important observation on your scope image of V1 anode and the output is that the anode is 50Hz and the output is 100Hz so we can eliminate V1.

This also tells us the noise is not pickup or leakage from the heater circuit as that would be 50Hz.

Keep the feedback disconnected and the capacitor across the input in place and scope the grids of each output valve. Any 100Hz noise here from the phase splitter should be of equal amplitude and in phase and should be cancelled in the PP stage. Then scope the cathodes of each output valve which are decoupled by C12 and C13 and should be of an equal low amplitude. I guess you replaced C12 & C13 as a fault in one of these could cause your symptoms?

Is the output transformer original and untouched?

[frankmcvey]

Thanks for the further replies, gentlemen. Peter asked that I check on the DC conditions. These are attached below in tabular from and as cct diagrams with the voltages annotated, which may be more convenient.

The reading for the cathode of V3 on amp #H10002 looks like a misreading ad was probably down to my forgetting to swtich the meter back to the DCV range after having set it to AC to read the heater voltage on the previous pin (it was getting late!) I'll confirm that tomorrow.

In amplifier #H10002 there is also an anomaly between the anode voltage of V1 (76V) and the g1 voltage of V2 (89V) - these are directly connected and should read the same. I'll check on that again tomorrow.

In answer to other questions, C1, C12 and c13 are all new items.

The ouput transformers are original and don't look to have been disturbed from the original build.

Peter, I'll do scope readings you requested and come back tomorrow.

Cheers,

Frank

[PJL]

The grid voltages look odd and I guess are measured wrong? At 11ma/V, the cathode voltages don't reflect the anode current that would be drawn if the voltages were correct.

The higher voltages are probably due to the amplifier being designed to power a pre-amplifier from the same HT/LT.

I have a pair of these in dark blue. They are very slightly different with one having 3 feedback networks and the other 2 as in the schematics. They are on the original reservoir and smoothing electrolytics and hum is very very low even on speakers with decent bass. Are all your cathode bias standee resistors still original?