Mullard 5-20 new chassis problems

[qualityten]

I have just moved one of my 5-20s from an old chassis to a new one. It was working perfectly before I moved it, but I have introduced a problem somewhere. Audio output is low and it is accompanied by an intermittent, not very rapid put-putting sound. Pics of the before and after attached.

I thought I might have misconnected the output transformer, but this is correct I think and disconnecting the NFB does not stop the oscillation.

Voltage measurements (Correct DC conditions in brackets):
Power transformer is fine for HT and heaters.
C15, C12, C5, C4 all just a volt or two under correct.

Both EL34s also measure close enough:
Anode: 423V (430)
Screen Grid: 416V (435)
Cathode: 30V (32)

But there are problems with EF86 and ECC83 voltages :
EF86
Anode 139V (85)
Screen grid 150V (100)
Cathode 4.2V (2.1)

ECC83
1st and 2nd Anodes: 187/250V (325). Not sure if these should be different?
1st and 2nd Grid: 140/100V (85) Not sure if these should be different?
the sound disappears when I measure the second anode and grid)
1st and 2nd Cathode: 140/140V (87.5)

Do these voltage differences suggest any obvious wiring error? I'm going to check my wiring again, as had problems with some of the brittle old wiring breaking off the tag board. There may be a hidden problem there, though I worked carefully and marked each wire I disconnected.

Thanks as ever for advice.
David

p.s. The large PIO caps are for later connection; I thought I'd get it working first before introducing any different components.

[Diabolical Artificer]

Hi David,it may be you have got the same problems I had with my 5-10,s.Your layout looks very much the same as mine;IE long wires all over the shop. Keep it neat and short and have a tidy up. Also your heater wiring crosses other wiring. Go back and have a look at how the old amp was layed out and compare,good luck,Andy.

[qualityten]

Thanks Andy. Point taken about tidiness. Most of the over-long wires are temporary, till I replace the can 8+8 can with the Dubilier PIO ones, and the two small electrolytics with the 8+8 can. But the wiring layout shouldn't affect the voltages.

[Herald1360]

It would if the amplifier is hooting at RF under the influence of the wiring layout!

Can you 'scope it?

Put-put noise sounds like motorboating which is LF instability which can be caused by poor power rail decoupling/layout.

[Diabolical Artificer]

Re wiring,thats what I thought David.The mismatch in ECC anode voltages turned out to be caused by a dodgy valve holder connection in my amp. As to the EF86 its so sensitive,especially if the earth aint right. Read what folks said re my 5-10 debarcle. Start at the beginning IE EF86,it dictates what happens next in line. Also REALLY read the instructions in the 5-10/20 bumf.,and the Mullard High quality amplifiction etc PDF.
Theres some good advice,which is easy to miss if like me your starting out.Andy.

[qualityten]

Thanks guys. It occurs to me that the only new components so far are new (NOS) valve sockets for the EF86 and ECC83. Will check and tidy as advised, and give the sockets a good cleaning.

I've checked the wiring of the whole circuit against the circuit diagram and checked for continuity. All is correct, except for a possibility that I might have made a mistake with the output transformer wiring. I did mark wiring carefully at disassembly, so I know which are anode and which are screen taps, but would it be a problem if anode tap A went to valve B by mistake? I stripped down the transformer to remove rust and respray the casing.

Chris, I have just bought a scope, but have not learned to use it yet. A friend will be able to give me help with this later this month. I assume I need a signal generator for this too? Where do I check my power rail decoupling?

Thanks, David

[PJL]

Is the feedback connection the right way around? Easiest way to test would be to swap over the audio coupling caps to the opposite output valve.

[GrimJosef]

Do be careful if you swap the feedback around. A 20W amp with a lot of positive feedback is capable of making a very loud shriek which may not be good for your speaker. I'd suggest at least keeping your finger on the off switch as the valves warm up and switching off immediately if it starts howling. It can be harder to tell if the amp decides to oscillate at supersonic frequencies. If you have a good-sized 8ohm resistor (ideally at least 10W or so) it might be wise to attach that in place of the speaker and to put a DMM across it set on the AC volts range before you switch the amp on. Again switch off quickly if you find there are lots of watts i.e many volts being delivered to the load.

Cheers,

GJ

[qualityten]

Swapping the feedback produced more oscillation and no audio. Disconnecting NFB removes the oscillation but still no audio.

Is there a way of measuring the output transformer to determine anode and screen taps? There are two pairs, a pair each coloured blue and brown. What I think I've confused is which pair goes to which EL34. Or, does this not matter?

[GrimJosef]

Resistance measurements will sort out the order of the OPT primary tappings. Find the pair of leads which have the highest resistance between them. These will be the two ends of the winding and they should go to the EL34 anodes. Now pick one of these anode leads and find the wire which has the lowest resistance to it. This will be the screen tapping for the same valve. Do the same for the other anode lead which will identify its associated screen lead. That should just leave the centre tap. As a check if you measure the resistances from one of the anodes (let's call it anode1) to the other four leads you should find they increase steadily in the order screen1, centre tap, screen2, anode2. It doesn't matter absolutely which anode lead you connect to which EL34 anode. But swapping the anode leads over (and the screen leads too, since each anode and screen do need to be kept together) will have the same effect as reversing the OPT secondary or reversing the capacitors from the phase splitter i.e. it will reverse the sign of the feedback and induce oscillation.

Cheers,

GJ

[qualityten]

Thanks GJ, just what I need to press on with this! I moved the amp to a decent chassis because these 5-20s are my favourite amps, and I want to get this one back in service.

[G8HQP Dave]

Once you have sorted out the first problem, you may have some buzz to deal with. The reservoir/smoothing cap ground seems to go to the middle of your ground bus, so it will be injecting charging pulses into the signal ground. Anyway, that is for later as it is unlikely to be creating the current problem.

When testing, always have a dummy load attached - you can connect a speaker too via an attenuating resistor. For now, work with the feedback disconnected but remember that this will increase the gain quite a lot.

[qualityten]

Thanks Dave. I've checked the OPT primary taps as GJ advised and they are correct. I found a dry joint on the 0.5uF C10 or C11 caps and repaired that.

I've also tried moving the pairs of G3+Cathode and G1 connections from each EL34, from their original pair of connections of 470ohm+50uF and C10/C11. Looking at the circuit while I type, I see that switching between C10 and C11 might not be good, but neither position works.

Oscillation continues, reduced of course with feedback disconnected.

I am puzzled that there is no audio signal at all now.

[GrimJosef]

Originally Posted by qualityten

I am puzzled that there is no audio signal at all now.

This is where your scope would come in handy. You'd inject the audio at the front end and use a scope probe to follow it through the system until it disappeared, which would tell you where the fault was. It is possible to build relatively simple audio frequency (AF) signal tracers which do the same job in a handier and cheaper if less informative way. There's a circuit for one suitable for valve equipment here http://www.pasttimesradio.co.uk/build/diyradio.html. Be careful if you decide to try another circuit. Many will be designed for transistor amps where the voltages (both DC and AC) may be much lower. Using one of those on a valve anode may destroy the tracer and/or give you a nasty shock. In any case make sure the probe lead is capable of withstanding the full HT voltage and that you're not holding any metal parts when you connect the probe to the circuit.

Cheers,

GJ

[qualityten]

Thanks again, GJ. Am I right in thinking that I can inject any audio signal from a CD for tracing purposes? If so, then I should be able to learn to use my 'new' scope for tracing the signal.

[DangerMan]

Hi David,
looking at your photo of the rewired amplifier, I think you might have the earth busbar connected to the wrong point near the EF86.
It looks to me as though you have connected it to the positive end of the cathode bypass electrolytic and the 2.2k resistor, and thus grounded the cathode. Should it not be connected to (what appears to be) the unused end of the 100 ohm resistor... second tag from the end?
I can't really make out the connections to the input signal screen but it looks like this may be included in the error. It looks like the feedback injection is around here, and so this may be the cause of the instability.
Pete

[qualityten]

Thanks for this suggestion Pete. I wish you were right, as this instability has got me baffled!

I've checked your query and my connections are correct, i.e. busbar is connected to 100ohm resistor and the EF86 cathode in not grounded, but does go to the positive end of the 50uf electrolytic and 2k2 resistor.

Would it make a difference if the input signal screen went only to the busbar and not also to the chassis at the tag mount (as at present)?

Thanks for your thoughts.

[GrimJosef]

Originally Posted by qualityten

Am I right in thinking that I can inject any audio signal from a CD for tracing purposes?

In principle yes. If you were using a simple tracer with a speaker or an earphone then music would be OK. If you're thinking of using a scope then a sine wave is better (it makes the triggering easier, for one thing). I don't know if anyone sells CDs of sine waves (I'll bet they do if you hunt for them !). Otherwise it's probably straightforward to 'burn' yourself one. I'd be the wrong person to ask for help with this though !

The most important thing if you're going to try your scope is to make sure you don't connect too many volts to its input. Good quality probes are helpful in this respect but it's possible to exceed even their voltage ratings with a high power amp. If in doubt, do ask first.

Cheers,

GJ

[DangerMan]

Originally Posted by qualityten

Thanks for this suggestion Pete. I wish you were right, as this instability has got me baffled!

I've checked your query and my connections are correct, i.e. busbar is connected to 100ohm resistor and the EF86 cathode in not grounded, but does go to the positive end of the 50uf electrolytic and 2k2 resistor.

Would it make a difference if the input signal screen went only to the busbar and not also to the chassis at the tag mount (as at present)?

Thanks for your thoughts.

Ah well, no cigar then

Do you mean that the amplifier ground (HT-) is isolated from the chassis? If yes, that can cause problems with stray oscillations.
If you don't want to directly connect the input shield to chassis, say to avoid earth loops, you can use a 220nF capacitor in parallel with 10 ohms to make the connection.
There should be only one such connection to chassis, which ideally should be at the bottom of the 100 ohm resistor.

[qualityten]

Thanks GJ. It looks like I need to bite the bullet and add a signal generator to my toolbox. I'll post a query in the relevant section here.