Hi everyone,
It's an oscillation thread
I am experiencing oscillation with a P-P amplifier VERY similar to the symptoms described by Wavebourne in one of his old threads: http://www.diyaudio.com/forums/tubes-valves/128992-weird-oscillation.html
The bursts (seemingly around 200Khz, one period being around 1 division on 5µS/div timebase) only seem to appear on the negative SLOPE of a sine wave, and appear when you feed the amplifier 40Hz or lower, and near full power. It seems to correlate to the transformer being near or at saturation.
Annoyingly, it only seems to do this on one channel. The amp uses negative grid bias, and tweaking it seems to move the position in which the burst happens. It seems to be the one valve of the pair causing it - if you bias the suspect valve nearly off and leave the other to do its best to reproduce a low frequency sine through an ungapped TX - these bursts don't occur. I say valve - it seems to be any output valve in that particular socket.
The amp uses negative feedback, has an ultralinear EL34 output stage with 1K screen stoppers. Interestingly, despite the amp having a dominant pole on the input stage at around 10k, the oscillations DO appear on the anode of the first stage, at the same relative amplitude as the output. This strikes me as slightly odd. If the oscillation was entirely output stage bound (snivets?) - wouldn't they be naturally attenuated by the first stage?
The amp seems otherwise stable and already has a zobel network across the speaker terminals, and the frequency response doesn't have any strange anomalies. Oddly, the channel that has the issue has a slightly smoother HF response than the stable channel.
Any pointers would be MASSIVELY appreciated
It's an oscillation thread
I am experiencing oscillation with a P-P amplifier VERY similar to the symptoms described by Wavebourne in one of his old threads: http://www.diyaudio.com/forums/tubes-valves/128992-weird-oscillation.html
The bursts (seemingly around 200Khz, one period being around 1 division on 5µS/div timebase) only seem to appear on the negative SLOPE of a sine wave, and appear when you feed the amplifier 40Hz or lower, and near full power. It seems to correlate to the transformer being near or at saturation.
Annoyingly, it only seems to do this on one channel. The amp uses negative grid bias, and tweaking it seems to move the position in which the burst happens. It seems to be the one valve of the pair causing it - if you bias the suspect valve nearly off and leave the other to do its best to reproduce a low frequency sine through an ungapped TX - these bursts don't occur. I say valve - it seems to be any output valve in that particular socket.
The amp uses negative feedback, has an ultralinear EL34 output stage with 1K screen stoppers. Interestingly, despite the amp having a dominant pole on the input stage at around 10k, the oscillations DO appear on the anode of the first stage, at the same relative amplitude as the output. This strikes me as slightly odd. If the oscillation was entirely output stage bound (snivets?) - wouldn't they be naturally attenuated by the first stage?
The amp seems otherwise stable and already has a zobel network across the speaker terminals, and the frequency response doesn't have any strange anomalies. Oddly, the channel that has the issue has a slightly smoother HF response than the stable channel.
Any pointers would be MASSIVELY appreciated
Last edited:
Capacitive coupling back to an earlier stage? One of the output valves is the right phase to give positive feedback. If the OPT nears saturation then it might begin to decouple the two output anodes from each other. Or maybe one output valve is approaching cutoff (trying to drive the OPT) so leaving the other free to oscillate.
I do this as standard practise, this is the only time due to a number of reasons when I don't have one right at the socket, and look what happened
Don't worry - I always use a 10:1 probe on valve circuits. The exception being a loaded and known to be stable OPT secondary. I am willing to be wrong if this is generally a bad thing to do.
Before I scoped output grids, I put the scope onto the cathode resistors, and the oscillation seemed to correlate with the highest peaks of anode (+screen if you want to be pedantic
) current on one of the phases - so it is probably something similar to what you suggest. When I say peak, the actual current waveform was very distorted at this frequency / power so it was more of a 'nipple' of high current (obviously due to the OPT). The global feedback was doing a shockingly good job of keeping the output waveform clean, however!
At low frequencies, where the OPT is running out of inductance, the global feedback will push the output valves nearer cutoff in an attempt to maintain the correct output voltage. Of course, the other valve will be pushed harder into conductance. Taken together, this means that the output stage is quite unbalanced for a while. Output valve gain, and hence Miller effect, will also change.
I once had to put an electrostatic screen in to stop a LTP phase splitter from oscillating due to the rather large coupling caps to the output grids being too near the input stage.
I once had to put an electrostatic screen in to stop a LTP phase splitter from oscillating due to the rather large coupling caps to the output grids being too near the input stage.
It is OT guilty (mine was ringing right above 35 kHz 10K P-P 100W transformers). I cured it by increasing local feedback around output tubes; as the result global one that goes from secondary decreased.
You probably have better tubes and as the result slightly higher open loop amplification factor in that channel that is ringing.
You probably have better tubes and as the result slightly higher open loop amplification factor in that channel that is ringing.
Last edited:
Actually for tube gear testing 100:1 1.5kV rated probe is more appropriate. Standard 10:1 probes are usually rated for 300V only and thus they are not safe enough.
For some reason, this problem seemed to go away but has since come back again. I have installed some grid stopper resistors, which helped, but didn't cure it completely. Luckily it only happens at extreme overload conditions so I'm not going to worry too much. It's practically impossible for it to happen with a music signal. It is irritating though.
My future designs will incorporate HF plate to grid feedback around the output stage so it isn't left wide open to excite OPT resonances. You don't need gain at these frequencies.
My future designs will incorporate HF plate to grid feedback around the output stage so it isn't left wide open to excite OPT resonances. You don't need gain at these frequencies.
This would probably be something I would have to experiment with. I was thinking perhaps if I rolled each valve itself off at around 30-40khz with a tight feedback network around each valve's socket. It seems silly that you should have a device with gain up to RF directly coupled to a load that's so complicated at high frequencies.
I thought of another culprit - since this is a conventional P-P circuit, perhaps when one tube goes into hard cut off, its winding has an open resonance that is coupling back into the circuit - my anode/screen leads are longer than I would ideally like. Maybe some braided shielding around my OPT feed wires would cure it.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.