Quad II test run.

[GrimJosef]

It might well. There is no such thing as 'near enough' with the Quad II output transformer (or most of the rest of the wiring, to be honest). Everything has to be exactly as designed. Anything else invites disaster.

EDIT: One of the effects of the negative (or supposed to be negative) feedback is that it spreads any problem out around the whole amplifier. So when you have trouble you'll see it everywhere - both in the location which is the source of it and in all the locations which aren't. When you're fault-finding on an audio amp one of the first things to do if you safely can is to disconnect the feedback. You'll lose all the good things that it does, but you might be able to track down where the trouble is arising.

Cheers,

GJ

[sparkymike]

Hi GJ, removed the feedback link and the multi waveforms have gone.
What is the next check ?
Mike.

[Radio Wrangler]

Originally Posted by GrimJosef

When you're fault-finding on an audio amp one of the first things to do if you safely can is to disconnect the feedback. You'll lose all the good things that it does, but you might be able to track down where the trouble is arising.

Actually, losing 'all those good things' allows you to see what the base amplifier is like... what its distortion, frequency response etc are like without the feedback trying to level the playing field.

What could be a slight effect with the loop closed, can be a lot more visible with the loop open, and an awful lot easier to track down.

Valve amplifiers with AC coupling between most stages give a lot more freedom to open loops

Transistor amplifiers with DC coupled stages and lots of gain matched with lots of feedback are critically reliant on their overall feedback loop to establish correct DC conditions. You have to get very inventive to do anything similar with these.

Some instabilities involve the feedback path, so going open loop can make the fault go away. At first glance, it's difficult to track a fault that isn't manifesting, but it allows you to check gain and phase versus frequency, looking for anything unexpected in the results.

David

[GrimJosef]

The wiring connections. Make sure each turret on the output transformer is connected as per the circuit and in particular that they go to the correct KT66. Then check that the right EF86 is driving the right KT66.

When you say that you built this amp out of parts, are you using an actual Quad II output transformer and an actual Quad II tagboard for the components ? If so then it should be quite unlikely that the feedback has somehow been applied to the wrong EF86.

Cheers,

GJ

[sparkymike]

In laymans terms,to me it seems that the feedback is going round in circles and multiplying itself, as shown by the multi waveform on the scope. A bit like microphones on stages when they feedback through a speaker not being in correct position. It does worry me if one of the windings on the output transformer needs reversing, as there are six windings and one of those is centre tapped
Mike. !!

[sparkymike]

I made notes of the KT 66 voltages.
Anodes 332/333 (pins 3.)
Grids 359/359 (pins 4)
Cathodes 33/34 (pins 8) (In line with the same voltage on R12)
Grids ? (pins 5) What should I expect to find there ?

While I am at it I will also list the EF86's.
At least I will be able to refer back to it here, long after I have lost that vital bit of paper !!

Mike.

[sparkymike]

Now checking the EF86 voltages.
Anodes 129 118 (this is one furthest away )from Jones plug. (pin6)
Grid 2 111 98 note: while checking this I checked R2 and R3 and could not get a constant reading on either and resistance was gradually reducing.
I have changed C1 (Hunts)110 v. working and replaced with a 400v. working cap. That original cap must have spent all its life on the edge of disaster !!
R2 and R3 are reading a little high, so if I can find similar ones, I will replace.
Mike.

[John Caswell]

Hi Mike,
Have you changed the coupling capacitors to the output valves? If so what type did you use. If you used anything other than a metal cased type bolted to chassis then to compensate for the slight capacitance reduction of the elements to chassis in metal cased types you will need to add 27pF from each EF86 anode (pin6) It will probably have to be a Silver Mica type as 400V polystyrene caps are getting hard to find.
C1 has got very little voltage across it as it goes from V1 to V2 screen grids (g2) which should both be at approximately the same potential so 100V rating is more than adequate.
There really should not be any volts on the control grids (g1 pin5) of the KT66s, but in practice I have always found a little. If you have any significant voltage ie above 1-2 volts either the coupling capacitors are leaky or the valves are poor.

John

[sparkymike]

Hi John,
they are metal case bolted to chassis, STC make.
Next step is to inject 50hz into the output transformer windings with GJ's kind assistance.
Mike.

[GrimJosef]

For the record I did the phasing check on a Quad II output transformer this morning. The results are in the attached pictures.

I took all of the valves out of the amplifier (we really don't want the rectifier making 370V DC on the transformer primary while doing this test !). I then plugged a lead into the amp's upper speaker socket (the ungrounded one) and croc-clipped the other end of that lead to one of the 6.3V heater turrets on the mains transformer. It doesn't matter which turret we use. This 6.3V winding is centre-tapped and the centre is connected to the mains transformer's Earth turret, so it's grounded. This arrangement will put a nominal 3.15V RMS (actually slightly more since the mains transformer is off load) across the output transformer's speaker secondary - turrets T (signal) and P (ground). In this case the secondary was jumpered for a 15ohm load (turret R wired to turret S).

I also connected the scope Channel 1 across the secondary - probe ground to P and probe tip to T - and I used the signal on this channel to trigger the scope. The result is displayed as the upper trace in the pictures, revealing the nasty clipped nature of my mains supply waveform .

In the first picture I've connected the other scope channel, slightly confusingly numbered Channel 4, with the probe ground to turret V and the tip to turret W. This allows us to check the phasing of the KT66 cathode winding relative to the speaker winding. We can see from the picture that WV is in phase with TP i.e. when one goes up, so does the other. The RMS voltages are 3.32V across TP and 2.56V across WV.

In the second picture I've repeated the exercise but now Channel 4 is connected to the output transformer primary - probe ground to turret Y and probe tip to turret Z. The voltage here is sufficiently large that I've had to set the probe to X10. Again we can see that ZY is in phase with TP and the voltages are now 3.30V on TP and 23.18V on ZY.

If we think this through then we ought to be able to convince ourselves that these phases are correct. Let's consider the KT66 V3. This has ZY connected to its anode and UV to its cathode. We've measured the phase of WV and found it to be in phase with TP and therefore with ZY. So since UVW is a centre-tapped winding we know that UV will be out of phase with ZY. If V3's grid voltage rises, causing the valve to pass more anode current, this will pull Z more negative of Y. The out of phase relationship means that U will go more positive of V and will push V3's cathode positive, lowering the V3 current and so providing the negative feedback that it's supposed to.

In the meantime as Z goes more negative of Y it will cause T to go more negative of P and, to a lesser extent, Q to go more negative of P as well since Q is effectively a tapping on the TP winding. This negative going signal will be fed back to the cathode of V1, causing the current in that valve to increase and its anode voltage, therefore, to go negative too. This negative excursion at the anode is coupled to V3's grid, counteracting the positive-going grid voltage which started the whole process off. So once again the feedback is negative, as we would wish .

Cheers,

GJ

[Radio Wrangler]

Funny how your past comes back to haunt you.... I was part of the team which re-packaged one of that scope family into a plug-in ATE module (HP 70703)

There are only two 20MS/s ADCs shared amongst the channels, so you don't get the full sampling rate with all channels firing. However, it does a nice trick of randomly jittering tha sampling times, and then measuring that jitter so that samples get accurately positioned in X. The jitter, however, scrambles improper aliases.

So we now need the same tests done on the problem amplifier.

G-J's mains looks pretty clean, but with flat tops and bottoms due probably to the prevalence of rectifier loads on the power grid. the key in the picture is the alignment of the zero-crossing times.

David

[sparkymike]

I have had to order another SO239 UHF BNC adaptor so I can run the two probes, so that will be a couple of days I expect, before that arrives.
The output transformer, from distant memory is not potted. At the time I decided to wait until the amp was working ok, so if any problems, the core could be removed ok.
Mike.

[sparkymike]

I decided to change R2 and R3 (both 1meg), as one was 1.33meg and the other was 1.29 meg. This could not have helped matters. After checking the rest of the resistors on the board ,they are all mostly way out, so probably best to change the lot.
R4 762 (680) /R5 1202K (180K) /R6 209K (180K) /R7 714K (680K) /R8 3K (2.7K
R9 711K (680K) R10 106 (100) R11 484 (470) R12, (off the board) 181 (180)/
R1 across pins 1 &2 on the Jones socket. 1.67meg (1.5meg)
Figures in brackets are correct value.

Mike.

[GrimJosef]

R5 looks very odd but if the rest are the original Eries then they look pretty much as I'd expect for 60 year old carbon composition resistors (R4 and R8 are sometimes closer to their original values than yours are).

Cheers,

GJ

[sparkymike]

Hi GJ, typo on R5, that should read 202K
Mike.

[sparkymike]

After changing R2 and R3 I expected the EF86 voltages to even up as those resister form a potential divider and both now are same value, however voltage on the anodes which was 129 and 118 has lowered to 107 and 96. (The first figure being the valve next to the Jones plug. However I forgot about R5 and R6.
They are 202 and 209 volts. I am waiting for new board resistors to arrive but I do have a decade box and it would be interesting to adjust one of the resisters to bring it to same value as the other and check volts at the same time.
Would the EF86's alter the voltage slightly as well if not a matched pair ?
One thing puzzles me. Why did Quad bring the mains plug to the opposite end or the chassis to the mains transformer ? No chance of mains induction there in the wiring loom ?
Mike.

[PJL]

Reducing R2/R3 will increase the grid 2 current which will increase the anode current. The imbalance may be from a mismatch in the EF86's.

[GrimJosef]

R2 and R3 aren't a potential dvider. Each one interacts with the screen grid characteristic of its particular EF86 to set the voltage on and the current flowing into that screen grid. The voltages are different because the EF86s are different. Perhaps the most useful way of looking at it is that if the HT was at 360V one resistor now has 253V across it (360-107) and the other has 264V. Those two voltages are within a few percent of one another, which is not a bad match, all things considered.

I don't know for sure but I think Quad wanted all the connectors - mains, audio input and audio output, on one end of the amp. This way in an industrial application you might slide the amp onto a shelf and only need access to one side of the shelf for your wiring rack. It's a bit like most hi-fi where all of the wires come out of the back. Since the speaker connections come from one transformer and the mains connections from the other, and there isn't room for both transformers at the same end of the amp, the wires to one are going to have to run the length of the amp. They chose to make that the mains transformer. The most sensitive (input) audio wiring is very short and doesn't run in the loom at all. The interstage wiring between the EF86s and the KT66s is in the loom, but on the other side of the amp from the mains.

Cheers,

GJ

[sparkymike]

Those particular EF 86's I am using are the Russian variant, millitary grade and by various accounts are said to be a quite good and robust valve. (6X32n with a large C in a circle.) I was wondering about the proximity of mains to the other DC wiring as I worked on outside broadcast trucks for several years and the effects of induction could be quite noticeable on long runs of maybe 20 feet or more. Those we built for the Germans specified different trunking and conduits for ac or dc wiring.
Mike.

[sparkymike]

Well the postman delivered (at a 2 metre distance) the other UHF/BNC converter today and all would have been fine, but Val decided she would like to paint the spare bedroom.It soon dawned on me that the "she" was in fact me.!! So the morning came and went and it was around 8pm that I got back to the big test.
Both probes were connected and valves removed and the scope was switched on to warm up. I was about ready to go and there was a "pop" and a small cloud of smoke inside the scope, which amazingly was still functioning ok. I then decided that I had best check the scope out and found that it was another of the myriad of capacitors that had reached its sell by date. I have a spare of the correct rating so will fix that tomorrow and hopefully resume.
Mike.