Best way to cure output osscilation

[trobbins]

What is your circuit schematic for the scope waveforms?

[catalin gramada]

What is your circuit schematic for the scope waveforms?

This is a sketch, preamp circuit was omited as is not part of discussion. Thanks

[trobbins]

I get an invalid attachment message.

[catalin gramada]

I get an invalid attachment message.

[trobbins]

So if the waveshape at 40Hz is not to your liking, and if it was being degraded at high signal level due to the drop in forward gain (and hence feedback level), you could measure that if you had a spectrum analyser and took some measurements, and then changed the value of the coupling cap and repeated. Imho, your appear to want to investigate distortion but don't really have any instrumentation to support that incentive.

Are the OPT primary half-winding idle currents matched? Have you tube-rolled the output stage, or somehow matched the valves for high-signal gain, to match the high signal current levels in each half-winding? I recall there is a hi-fi technique for matching at high signal levels, based on the signal current residual being drawn from B+ supply. OPT and output stage distortion can be very influenced by imbalance - one of the Heathkit Williamson construction manuals graphed an example of the influence of imbalance.

[pdf64]

If a 1 ohm cathode resistor is fitted to each power tube, then the waveforms across them can be compared for magnitude, hence weak / lower gain power tubes can be identified and replaced, good tubes can be swapped around to get the best balance.

[catalin gramada]

You're right trobbins.I work with the valves, but different sort of. No spectrum analysers laying around . But at least pdf64 idea sound interesting to me and is something I can do. Thank You guys.

[Chuck H]

Well I'm out of my depth on this thread technically at this point (when am I not ) but...

The distortion on the waveform is shifted at the P/P crossover point, but both shifts are in the same, not opposite directions. If I interpret this correctly this represents a phase shift and is probably related to the OT's ability to fully shift magnetization at this lower frequency. The current demand is outrunning the time constant of the OT's limitations at that frequency. I think this would be consistent with what Helmholtz has been saying all along. The fix would be an OT with higher inductance. Whether that comes with any HF response consequence I don't know and whether correction is even necessary has already come into question.

[Helmholtz]

As explained above, the distorted 40Hz waveshape is typical of partial core saturation. The shape distortion is maximal around voltage zero crossing, as the current of an inductor is maximal when its voltage is zero (90° phase shift). Core saturation always starts at low frequencies as low frequency means high parasitic magnetizing current. And it's high magnetizing current which drives the core into saturation. As signal frequency approaches zero, any iron core transformer loses its ability to transfer voltages.

Imbalance of large signal plate currents will make the situation worse, because it means DC-premagnetization. But that would show as asymmetrical waveshape/distortion.

As the PI output signals show no signs of grid current distortion and it is a bass amp, I would go with the 0.1µ coupling caps. Especially with the low value 100k grid load resistors. Compared to 0.047µ this will result in a little more forward (open loop) gain at low frequencies, making NFB somewhat more effective. Also increasing the power tube grid resistors from 100k to 220k could help.


But as the OP reported better bass performance with a Hammond OT, I would go with the Hammond in this bass amp and save the Partridge for a guitar amp project.

[Helmholtz]

As the PI output signals show no signs of grid current distortion and it is a bass amp, I would go with the 0.1µ coupling caps. Especially with the low value 100k grid load resistors. Compared to 0.047µ this will result in a little more forward (open loop) gain at low frequencies, making NFB somewhat more effective. Also increasing the power tube grid resistors from 100k to 220k could help.

Doing both, increasing coupling caps to 0.1µ and increasing grid resistors to 220k will give you between 2dB and 3dB more forward gain and effective feedback at 40Hz.

[catalin gramada]

shared 220k meant 440k single equivalent. Is not a bit too much for EL34 into fixed bias mode,please?

[Helmholtz]

shared 220k meant 440k single equivalent. Is not a bit too much for EL34 into fixed bias mode,please?

Datasheet allows for 500k per tube with fixed bias.

[catalin gramada]

And think mistery was solved. Was today to buy a trashed Sound City ST 50, and guess what? Found the same Partridge trafos kit like in 100W PA amp which I used for my project. What the heck they did in '70? Judging by actual facts it seems my tranny is a 50W one, they just added two tubes and changed OT tap for 100W aplication.....% @$&#*%

[trobbins]

That may not be the case. It was not uncommon for a manufacturer to get a batch of higher powered OT's and use them for lower powered models as well. The cost of a mixed batch of OT's was more than the cost of just one type at the highest power rating spec.

[catalin gramada]

One more question,please. Can I shift the point where core saturation start changing the plate voltage,please? Thinking to drop the supply from 450v around to some like 400v....?