Very Unstable Amp - Oscillation

[Robert M. Martinelli]

Hi Enzo,
I suspected ripple myself and asked Martin to measure the oscillation' s frequency, but he says he's from the UK, and mains voltage there is 240VAC 50Hz, so hum/ripple should be 100Hz.....and I get "dazzled and confused"
( more than I usually am ) by the fact that opening the feedback loop the oscillation problem gets better.....quite the contrary of what should be in theory...well, I guess that' s why they call this world the real world

BTW - Do you think my hypothesis about the unused anodes is worth investigating? I've always thought grounding unused tube' s pins was a sound practice to keep stray signals away and avoid such oscillation problems.

Best regards

Bob

[Wildcard]

6av6

Pin 5&6 are grounded OK. If I pull V1 it still oscillates, if I pull V2 it stops. If I ground the pin of V2 it stops it. No pins are touching each other. I'm reading 2VDC and 2.6VAC on the grid of V2.....

(see correction below....actual voltages are: -0.135mvDC
0.025mvAC)

[Wildcard]

correction!!!

Sorry, I need to correct those voltages!

The grid of V2 - 6av6 - reads:

-0.135mvDC
0.025mvAC

[Wildcard]

creeping bias notch!

Another observation if this helps diagnose the problem...

When i look at a wave form on my scope, you can see a "bias notch" appear as i turn up the volume. Now, as I increase the volume very gradually the notch on the left hand side leaves it's parallel with the right hand notch, and creeps up towards the top of the waveform, when it reaches the top and goes over the top of the wave the amp goes into oscillation. The right hand notch pretty much stays where it is until this point.

What am i looking at here? Some sort of imbalance or phase problem??

[Robert M. Martinelli]

Does your o-scope have two channels? If so it would be great to ( properly scaled ) compare the waveforms, if the problem is located at V2, pick the signal before it and after, or pull out V2 and inject the signal after V2 and look at the waveform on the output.....I don' t remember so pardon me if I ask again, have you checked / changed the coupling caps as well ? The change in the bias could be the result of some leaky capacitor. Oh, one last thing - have you checked the tube sockets or cleaned them ? Oxidization on some terminals could be the culprit, as it rises the contact resistance. To do this I usually use a good quality contact cleaner ( Philips 390 CCS ) and I push and pull the tube in and out of the sockets several times, letting the tube pins do the job.
Hope this helps
Best regards
Bob

[Wildcard]

OT

Hi Bob,

Thanks for the input. I've got a 2-channel scope so i'll do that test and report back. I switched on the amp this morning and with all the controls on zero the amp started a squealing high-pitched oscillation that was coming from the OT (the amp is linked up to a dummy load not a speaker but I could still hear it). If I turned up the volume it just got worse, i once again set about moving leads to see if it would make any difference but it didn't.

The only way to stop it was by removing the feedback loop! The amp will still go crazy when turned up but the high-pitched noise from the OT stopped.

I checked the resistance readings on the OT from the B+ lead to the plates (with tubes removed), from every lead to the chassis and from the primary to the secondary and all looks fine.

Cheers
Martin

[Robert M. Martinelli]

Hi Martin,
measuring DC resistance to check the OT only gives you a rough indication about the OT' s status, but I don't think your OT to be defective....
It definitely seems you' ll have to do comparative measurements with your 2-trace o-scope, maybe you' ll be able to locate the problems, and identifying the area would make finding the culprit easier.

The behavior you described in your last post is quite different from before, it seems the problem is getting worse....it' s very difficult to understand where the loop closes and the oscillation starts....again....I suspect phase problems given the behavior with/without the NFB loop ( which probably behaves like a PFB now ).

As to the "shifting bias" problem, double check all coupling capacitors for leakage, some DC could pass through one of those.

BTW, what resistance /power rating your dummy load has? Is it properly matched with the amp' s output? Dummy loads usually warm up a lot, so there is a chance the resistors' value drifted off spec mismatching the output stage and altering the NFB behavior.

[Wildcard]

Thanks Bob,

OK, I've rechecked for any DC on the coupling caps and there's hardly any if at all...0.001mv at most. I put two traces on the scope, once the oscillation starts (turn up the volume to full), I can see it everywhere, from the input of V1 (very low level), to it's output, to the input of V2 to it's output and so on right through the amp, getting louder and louder.

Incidentally, I moved the ac cables to the switch, the twisted cables ran on the floor of the chassis between the two output tubes and close to their grids, they now avoid that area and I "think" the oscillation seems slightly harder to trigger and not quite so pronounced, I plugged a guitar in and it seemed a bit more stable, that is until any bass is added by clicking the rotary tone control.

I think that if I can tame down the drive on the first stages and reduce some bass it could be useable. With the feedback loop removed there's absolutely no headroom....great sounding drive but waaayyy too much.

Any suggestions as to the best place to start with taming the stages and reducing the bass?....the 5-way rotary switch seems to affect the oscillation when in the "in-between" stages. The switch is marked up as 1. bass, 2., 3.flat, 4., 5. treble. I'm almost guaranteed to get the oscillation on positions 2 & 4, the other 3 postions aren't so bad. When in positions 2 & 4 you can really hear a big increase in bass and hum and that's causing the freak-out.

Dummy Load
The dummy load I'm using at the moment is 8ohms / 100w, and that's fed from the 8ohm tap on the amp, it seems handle the heat OK...that and the amp gets much hotter during these oscillations.

Cheers
Martin

[Robert M. Martinelli]

Hi Martin,
a simple way to reduce gain at low frequencies would be to tinker with the cathode bias cap of the first triode, right before the PI section ( the first half of that ECC83/12AX7 ), but I don' t like the idea of having to modify something to sneak around the problem, I' d rather like to have the problem fixed for good.... I don' t know much about that particular amp's make/model, but the schematic ain't nothing terribly complicated, and I don't think there are design flaws, maybe it' s something related to the layout and/or some wire' s wrong path....I can remember one of my VOXes was humming, and I realized a signal-carrying wire was catching noise, so I substituted that wire with a shielded cable and changed its path and the hum disappeared....sometimes we are so concentrated and frustrated we forget the basics....at least this is what happens to me from time to time.

Oh, I' m still thinking about it and I just took another look at the schematic....
The anode voltages at the first tubes are 75VDC, but I can' t read the grid voltages, can you tell me how much the grid voltages are supposed to be ?

Regards

Bob

[Wildcard]

voltages!

Thanks Bob,

Yeah, I'd much rather fix this thing properly but maybe it's something to do with the bass response thats pushing it over the edge?

I've just re-checked the pre-amp voltages and they seem to have changed! The plate of input 1 seems low at 37.5v. I subbed another 6sc7 and got pretty much the same readings. The Grid leak resistors are 4.7meg and read good.

The plate of the 6av6 seems to have crept up to 153.5v....previously i had read 121v on that pin.

What gives?

On the schematic (actual voltages in red) the votages are:

V1 - 6SC7
GRID 1 = -.5v (0v)
GRID 2 = -.5v (-.6v)
PLATE 1 = 75v (37.3v!)
PLATE 2 = 75v (70v)


V2 - 6AV6
GRID - 0v (0v)
PLATE - 110v (153.5v!)

[Robert M. Martinelli]

Hi Martin,
I assume your "1" and "2" on the 6SC7 electrodes stands for "a" and "b" on the schematic diagram.....uhmmmm....ok, on the plate of V1A you are getting half of the voltage that should be there, so you should check the voltage drop across R6, comparing it with the other half, which seem to be pretty close to the specs - Both resistors are connected to a 350 VDC +B, both resistors have the same value, and the double triode is expected to run under the same conditions.... I would say that R6 has drifted to the high side....well, maybe R8 has drifted a little too, but 70 vs 75 V still seems pretty close.

Voltage on the first and the second grid of V1 should also be the same, and I can see V1 is biased using the "grid leak" method, this practice is not that good because the amount of bias can vary greatly with different valve samples, as the bias amount relies upon electrons "leaking" - another thing I don' t like is that this method requires a very high grid leak resistor, and high value resistors are noisy.....

About V2, the plate voltage seems high, so either R12 has drifted on the low side, or the 2700 Ohm cathode resistor ( R13 ) has drifted on the high side,
Measure the voltage across R12 and R13, across R12 you should read about 240 VDC ( 350-110 ) and across R13 you should have about 1 VDC -

The given values and the voltage drops also allow you to calculate quiescent currents on V1a, V1b and V2 - on V1a and V1b you should have (350-75)V/470K=0.585 milliAmps, on V2 you should have (350-110)/470K=0.51 mAmps.

As to the measurements, always remember that the reference point of a valve is always the cathode, so the 0 V on the grid of V2 actually is -1 looking at it from the cathode; you should also keep in mind that sometimes bad resistors look good if measured statically ( with the amp off ) but they can drift with temperature so measuring the voltage drops across them with the amp on is a better method IMHO.

Hopefully we' re almost there....

Best regards

Bob