jukebox amp oscillation conundrum

[Alex R]

WE have here a schematic you might have to pay money to get elsewhere. It's one of the stereo power amps in an AMI Continental. Early 60s jukebox amp. They run 6973 outputs. These tubes are expensive and modern replacements tend to be unreliable, and down the years I have converted maybe half a dozen of these to EL84 operation, simply by rewiring the bases to suit and rebiasing, without problems. With this one however, I do have a problem. It oscillates. I have tried a number of things to stop this without success (I'll list the things I've tried & can recall below).

Here's the thing: the oscillation stops when I remove the NFB. Yes the plates are the right way round - tried both ways and it howls like a demon the other way round, also I checked the phase with a 2-channel strobe.

It has had the OT replaced this side with some standard Hammond type job.

I tried:
Grid stoppers on all grids, soldered to tags on the base
Extra filter capacitance, including adding a cap on the first B+ node where the 5500R 10W inrush resistor meets the 150R 10W dropping resistor.
Screen resistors soldered to tags on the base
Separating plates from screens using the node between the two 1K 1W dropping resistors for the screens. (Why are the plates connected to the screens btw?)
It has been fully recapped.
I checked the circuit node by node and it looks good to me.
Grounding the inputs doesn't stop the osc.
I can't even stop it fully by touching a big cal to various grids with the other end grounded. It's very determined to oscillate, makes me think I've hard-wired it in somewhere but durned if I can see where.

So yeah the thing oscillates somewhere in the 50-100KHz range, at full output, pretty clean sine wave visible on scope. Remove the NFB and it disappears. Any ideas?


Some other things to note:

The other amp, which is on the same chassis, oscillates at about 3Hz, visible as the speaker cone pumps in and out, but otherwise works ok.

The OT is unusually connected - the 4 ohm tap is grounded. The replacement OT fitted to this side doesn't have the 70v taps but they aren't needed. It's connected the same way, with the 4 ohm tap grounded and the NFB taken from the 16 ohm tap.

I'd try disconnecting NFB from both amps and trying to drop gain but the other amps hums loudly if I remove the NFB, also it's not a simple matter to see where you'd drop the gain unless before the first stage, which would leave a lot of noisy gain running through. Anyway that's not a solution really.

[nosaj]

On the antiques radio site the other day someone had an old zenith amp I think. It would oscillate on him. Hie solution turned out to be the tone controls were on a separate board and not hooked up. Once he hooked the tone controls up the oscillation disappeared. He also was pondering the NFB.
nosaj

[g1]

I'll wager that if you swap the OT's, the problem will stay with the replacement OT.
I'm wondering if it has to do with the 4ohm tap being grounded, and maybe some extra shield/ground in the replacement OT?
Or maybe someone went by the colour coding on the OT wiring and it doesn't match up with the schematic?

[Alex R]

I'll wager that if you swap the OT's, the problem will stay with the replacement OT.
I'm wondering if it has to do with the 4ohm tap being grounded, and maybe some extra shield/ground in the replacement OT?
Or maybe someone went by the colour coding on the OT wiring and it doesn't match up with the schematic?

Many thanks for this. I did check the windings and I think reconstructed the tap order correctly. But the grounded 4 ohm tap - yes that could be an issue couldn't it? - I will try it. Cleared the thing off the bench for a few days to put other jobs through but I will get back to it and report back.

[Alex R]

oh one other thing - screens connected to plates? Isn't that triode operation? Wossatabout?

[Enzo]

I am looking at schematic and I do not see plates wired to screens.. I do see the screens wired to the same B+ node as the center tap of the OT primary, but that is not the plates.

[nickb]

I'll wager that if you swap the OT's, the problem will stay with the replacement OT.
I'm wondering if it has to do with the 4ohm tap being grounded, and maybe some extra shield/ground in the replacement OT?
Or maybe someone went by the colour coding on the OT wiring and it doesn't match up with the schematic?

^^^^I'll wager that too. Stray capacitance in the new transformer is the likely cause. That said, there are a couple of HF caps that should be checked viz, the 300pF to pin 6 of V2 and the 400pF (?) to the plate of V3. If these are OK you'll have to try some compensation to see if you can make it stable.

[Helmholtz]

Please note that the impedance between the 8 Ohm and the 4 Ohm taps is not 4 Ohm (probably lower than 2 Ohm). OT output impedances are not additive.

[Alex R]

I am looking at schematic and I do not see plates wired to screens.. I do see the screens wired to the same B+ node as the center tap of the OT primary, but that is not the plates.

Dur. Thanks Enzo. Currently on the bench is an AMI AA, I guess abut 1950. Blessedly simple and mono. This one too lacks screen resistors and uses the same node for screens and centre tap. I saw it and thought, , better hurry back to the forum and delete my stoopid question before Enzo sees it. Talk about esprit d'escalier. Thanks again.

[Alex R]

Please note that the impedance between the 8 Ohm and the 4 Ohm taps is not 4 Ohm (probably lower than 2 Ohm). OT output impedances are not additive.

Thank you. This is a promising line of enquiry. When I get back to it I'll connect it normally and see what happens.

[Alex R]

OK I came back to this amp and moved the speaker ground to the normal place, then tried tapping the NFB off each point in turn. The lower the impedance the less HF oscillation. As it lessened a hum crept in, that looked like ripple on the scope. NO NFB - no HF oscillation but quite a loud ripple buzz.

A reminder that the NFB phase has been checked and is ok.

Took some voltages around v1 and noticed that in stage 2, the cathodyne splitter, grid voltage was slightly higher than cathode voltage. The cathode resistor had drifted up to 130K so I changed it for a good 100K to match the one I'd already put on the plate (all caps and many resistors already replaced in this amp). Still grid positive but the HF osc went away and was replaced by this ripple.

I thought I'd take v1 voltages in both the problem amp and it's good stereo sister, with and also without power valves fitted (oscillation stops with power valves removed, or with either 12AU7 removed). Here are the voltages (vDC):

PROBLEM AMP
power tubes in/out:
1st stage
pin1 (plate) 97/145
pin 2 (grid) 0/0
pin 3 (cathode) 2.6/4.3

2nd stage (cathodyne)
pin 6 (plate) 104/158
pin 7 (grid) 97/145
pin 8 (cathode) 96/144

GOOD AMP
power tubes in/out:
1st stage
pin1 (plate) 75/120
pin 2 (grid) 0/0
pin 3 (cathode) 3.7/5.5

2nd stage (cathodyne)
pin 6 (plate) 126/188
pin 7 (grid) 75/120
pin 8 (cathode) 83/123

So the problem amp's cathodyne stage still has the grid positive relative to its cathode. Any further ideas?

[Helmholtz]

I still suspect the OT. HF oscillation may change DC levels and may give false meter readings.

In tube amps HF oscillation is typically produced by the power amp having strong over-all NFB. By action of parasitic capacitances and inductances, the signal phase at HF inverts and the negative FB turns into a positive one.
(In amps with a presence control the oscillation often stops at higher presence settings, because more presence means less NFB.)

A replacement OT having a different frequency response and/or different impedances may turn a stable amp into an oscillating one.

So first make sure that the OT is connected properly. The second step, if necessary, would be decreasing the HF gain within the power amp and/or reduce feedback at high frequencies.

[Alex R]

Many thanks for the responses. You guys were right about the transformer - thanks for the guidance. It led me to clip in a spare transformer - and hey presto, no oscillation. The quick way with this is to fit the RS £35 2xel84 transformer. I've tried various suppression techniques including networks across the primaries and secondaries. The suppression caps in there have all been replaced, and I have tried different values. Something about this OT is wrong, and of course I can't be entirely sure what the windings actually are. So we'll fit a new one. Thanks again everyone.