oscillation

[merkur]

I running into oscillation with one amp that I have build. It is guitar oriented push pull 50W 6L6; Long tail driver. As soon as my signal is a bit over the bias supply (when there is grid current) oscillation start to occur. It's like a gost note, something in the back! If anyone could tell me advice or clue?

Maybe it's phase problem? but I not sure. Does someone know where I can get information about phase problem in the power amp?

Or maybe it's a layout problem?

helppp!

[Miles Prower]

Here's something you could try (see attached). Stabilizing the finals can be done with the plate stoppers consisting of a 100R, 2.0W carbon comp resistor in parallel with a coil of ten turns, #18 wire, ID= 7/16ths inches. Space wind (easy if you're making two: simply prepare as a single, ten turn, bifilar coil and pull the two sections apart). Insert the resistor inside the coil, and solder as one unit as close to the plate connection tabs as possible. Also, try to space at least one coil OD from any surrounding metal to reduce the tendency for xfmr action that'll essentailly "short out" the coil for AC. That'll render it ineffective for oscillation suppression.

A screen grid stopper of 1.5K, 0.5W, metal film or carbon comp will also help tame any sustained/damped oscillations. If the oscillation is simply xfmr self-resonant ringing, then a Zobel across the primary will suppress it.

[merkur]

The 100 resistor and the coil is to filter the power supply?
I'm already using 1.5k on the screen. It's not because I don't trust what you said but I need to understand why I will do this modification.

[EC8010]

If oscillation occurs when you go into grid current then it sounds more like a grid problem. Do you have carbon grid-stoppers butted up to the grid pins?

[Miles Prower]

Originally posted by merkur
The 100 resistor and the coil is to filter the power supply?

No. It's to break up parasitic resonances formed by stray capacitance, device capacitance, and inductance. That's what the coil does. The 100R resistor de-Q's the coil so that it doesn't form a parallel tuned circuit itself. The inductance is so tiny that it has no effect on the audio response, let alone PS ripple.

Without that, I'd've gotten a sustained oscillation of some 20MHz in my project. As it was, I was getting RF ringing on sine wave peaks. Adding screen stoppers put an end to that, and stabilized the final circuit nicely. As for grid current oscillations, that's not a problem either, probably due to the inclusion of low Z grid drivers.

[merkur]

Thank you Miles for the explanation. Do you apply that for both output tubes?

And Ec8010 Why are you suggesting carbon type resistor? Cause I already have grid resistor but they are metal film.

And what about phase problem that can cause oscillation?

Do you guys haves testing tricks to point the problem more precisely?

The high are not so good either, something effect those.

The power and output transformer are from a Bogen CHB 100
so I have done a voltage doubler to make 475V.

I'm also welling to post my schem and a picture of the amp showing the layout if someone is interresting to dig a little further.

[Miles Prower]

Originally posted by merkur
Thank you Miles for the explanation. Do you apply that for both output tubes?

Yes.

And Ec8010 Why are you suggesting carbon type resistor? Cause I already have grid resistor but they are metal film.

Carbon comps have less inductance since the film of metal film types, especially the higher values, is laid out in a spiral pattern that makes it a little inductor. In those parts of the circuit where you are dealing with RF, this inductance can be problematic. Since it doesn't amount to much, it doesn't have any affect at frequencies as low as audio. Otherwise, metal film is preferable due to being much quieter. Metal film stoppers worked well for me.

And what about phase problem that can cause oscillation?

Do you guys haves testing tricks to point the problem more precisely?

If you have a stability problem, it's caused by phase shift that turns negative feedback into positive feedback when the closed loop gain is greater than unity. You can either limit open loop frequency response, increase negative feedback at higher frequencies by means of a compensating capacitor across the gNFB resistor, or add a phase shift network to the plate of an early preamp stage to change the phase characteristic of the closed loop response. In order to cure the problem with stoppers, compensators, or phase networks, it's necessary to know what the frequency actually is. So you'll need to o'scope to determine this.

I'm also welling to post my schem and a picture of the amp showing the layout if someone is interresting to dig a little further.

Do that.

[merkur]

I got a scope but I not able to really see what frequency it is.
when I clip the output stage to the point of a square signal I see the oscillation on top of the wave. Maybe I don't probe correctly to see the real problem.

[Miles Prower]

Originally posted by merkur
I got a scope but I not able to really see what frequency it is.
when I clip the output stage to the point of a square signal I see the oscillation on top of the wave. Maybe I don't probe correctly to see the real problem.

Do you see it for any other waveform? If this occurs with square waves, but not other waveforms, then you have xfmr resonance. A Zobel will stop that if it's a problem. Usually, xfmr ringing occurs at frequencies above 30KHz.

[merkur]

It does'nt matter if the signal is a pure sinus or a square wave. It happens when the output stage clip. And when the output clip you have a square wave so my problem it's when the output start to clip(AB2 class kind of). Sorry if my english bad but i'm a french Canadian.