Best way to cure output osscilation

[catalin gramada]

Hello, I just recorded some pretty high amplitude 300 kcps osscilation at the OT output with speaker connected. This are not present with a resistor, and also are suprimed with speaker and resistor in parallel. Sometimes are suprimed just connected one of my voltmeter over outputs, sometimes just disconnected one of preamp stages. It is not a well trusted osscilation but when appear is pretty consistent. Does it worth to bother to found its source into circuits or just to suprimed snubbing it over the OT winding, please?

[J M Fahey]

Please post the actual schematic involved

Unless you want the classic "it must be some thingie in there" generic answer, that is.

[catalin gramada]

No,no, I like more that one


I.ll post one, Thanks JMF

[Chuck H]

Just to be clear, it's some thingie in there causing it.

[catalin gramada]

Hi, this is what I did

[alexradium]

i told you,that was it

[Helmholtz]

First all I am not surprized that the oscillation stops when you wire a load resistor in parallel with the speaker. Amp stability is influenced by open loop frequency/phase response, NFB ratio and open loop gain. All three of these influences depend on load impedance. A speaker has an impedance of over 100 Ohm above 100kHz and turns the phase because it is inductive.

Also your PI circuit has a much wider HF bandwidth than typical designs, which makes things worse. The main reason is the low impedance cathode follower driving the PI input. As a consequence the Miller effect limiting HF response at the PI input grid in standard high impedance designs is essentially eliminated.

To reduce HF gain of the PI, you can try to wire a capacitor of 100p or more between the PI plates.
To compensate for the reactive speaker impedance, a suitable Zobel network across the OT primary is recommendable.

[pdf64]

Are the power tube 1k and 10k grid resistor bodies mounted at the tube socket terminals?
Is the feedback resistor 470 ohms?
That seems a lot of NFB (16ohm output, 470 ohm/100 ohm feedback network).

[catalin gramada]

Thank You gents,
Yes 1k and 10k resistors are over the tubes socket, 470 ohm resistor on nfb ( ca 15 db) Weird thing I had before the pi, biased by cathode bias resistor and did not noticed any problems. And just now is osscilate by a margin, sometimes did sometimes not, but when appear is pretty consistent and around 2v pp level.
Also, when osscilation appear get some buzzy ,sort of hum noise in speaker but not found any components to justify those noise ,other than this 300khz osc. I can suspect a bad component like 12at7 tubes, which from my curiosity I toasted hk insulation at 150v for a while, but it measured very good so I think are out of question.
This osscilation appear not suddenly at startup. It takes 10 sec around after the stb switch is engaged, of course with tubes all ready heated

[pdf64]

Is it stable when open loop?
OT plate wires, and LTP output wires, twisted? As these wire pairs carry balanced signals.

With using that tap and those feedback network resistor values, you are applying a lot more NFB than a regular BF Fender.

[Helmholtz]

Sometimes it helps to ground speaker chassis.

[catalin gramada]

I will try, thank you

[catalin gramada]

Is it stable when open loop?
OT plate wires, and LTP output wires, twisted? As these wire pairs carry balanced signals.

With using that tap and those feedback network resistor values, you are applying a lot more NFB than a regular BF Fender.

Plates wires are not, there are a lot of building issues I have to cover but just for now it is a platform for experiment. It was excellent stable before to try this c.f. direct coupled with pi stage. Maybe is a building issue and becomes more vulnerable to osscilate due to rised hf sensitivity as Helmholtz states

[J M Fahey]

+1 on too much feedback and check whether problems disappear with NO NFB applied.
Just try it.

Then use something like 1k5 to 2k2 instead of the 470 ohm one.

And above suggested 100pF plate to plate at the PI

I cleaned it a little:



the original drawing made my eyes bleed.

[catalin gramada]

+1 on too much feedback and check whether problems disappear with NO NFB applied.
Just try it.

Then use something like 1k5 to 2k2 instead of the 470 ohm one.

And above suggested 100pF plate to plate at the PI

I cleaned it a little:

[ATTACH=CONFIG]52528[/ATTACH]

the original drawing made my eyes bleed.

Thanks JMF. I found the source of osscilation in front of c.f. so I need to have care about that. Anyhow followed you advice and changed nfb resistor to 1.5k for 8db nfb around. Found crossover distorsions appears as amplification goes up.this was not present with 15db nfb before. There is a way to get rid of it? I tried to bias power tubes as hot can be around 75percent at iddle. PI outputs are pretty well balanced. I get 495 plate, 490 screens with -44 bias at iddle and 433 plates with 396 screens -42.3 bias below onset clip for 64 vpp into 4.2ohm load for 1.9k OT. Thanks