Hello all,
I've recently built ESP / Rod Elliott's P101 Lateral Mosfet amp. The listening impression is simply amazing!
However, as I also want to quantify what's amazing, I decided to do a couple of square wave measurements using 1k, 10k, 20k signals, and then doubts arise.
Please see attached. It's one channel measured directly with the square wave at the input.
What worries me, and can't find yet an explanation is the overshoot at the bottom, and the fact that the amplitudes do not match between the upper and the bottom part.
The power supply was checked, and it's within 0.1V (+ve and -ve supply rails). Input signal is also to be trusted, it's generated by the scope, and I've checked before with wavegen linked to the input, and what comes out is perfect.
Output devices are Exicon 10N20/10P20 pairs, two of them (populated the whole board as in the high power P101).
I've also tried to simulate the schematic with models from Exicon, and it presents the exact same behavior, therefore I think that something needs to be adjusted.
Any ideas why is this happening?
I've recently built ESP / Rod Elliott's P101 Lateral Mosfet amp. The listening impression is simply amazing!
However, as I also want to quantify what's amazing, I decided to do a couple of square wave measurements using 1k, 10k, 20k signals, and then doubts arise.
Please see attached. It's one channel measured directly with the square wave at the input.
What worries me, and can't find yet an explanation is the overshoot at the bottom, and the fact that the amplitudes do not match between the upper and the bottom part.
The power supply was checked, and it's within 0.1V (+ve and -ve supply rails). Input signal is also to be trusted, it's generated by the scope, and I've checked before with wavegen linked to the input, and what comes out is perfect.
Output devices are Exicon 10N20/10P20 pairs, two of them (populated the whole board as in the high power P101).
I've also tried to simulate the schematic with models from Exicon, and it presents the exact same behavior, therefore I think that something needs to be adjusted.
Any ideas why is this happening?
Attachments
The asymmetrical clipping is to do with a mix of the use of a bootstrap current source, Vgs thresholds, and the diode D3. At normal listening levels you are not driving the amp anywhere near clipping (or you shouldnt be!) so it doesnt matter.
Once you drive an amp into slew-rate limiting you are likely to see somewhat untidy behaviour, the main thing is that ringing is kept well under control and recovery of the feedback loop is rapid.
Real audio signals should never provoke this regime, but its important the amp is stable under such input, else switching transients could lead to amp self-destruction by oscillation.
Slew-limited waveforms will vary with amplitude, typically higher amplitude response will have longer recovery time and be the most likely to misbehave - don't test with a loudspeaker load for this, that can fry tweeters and hurt ears. Test high-amplitude HF square waves only with short bursts to avoid overheating the Zobel network resistor(s), which are usually sized for less drastic signals.
Real audio signals should never provoke this regime, but its important the amp is stable under such input, else switching transients could lead to amp self-destruction by oscillation.
Slew-limited waveforms will vary with amplitude, typically higher amplitude response will have longer recovery time and be the most likely to misbehave - don't test with a loudspeaker load for this, that can fry tweeters and hurt ears. Test high-amplitude HF square waves only with short bursts to avoid overheating the Zobel network resistor(s), which are usually sized for less drastic signals.
You appear to be on 1V/div so the amp is not clipping. And of course the input is a square wave.
The overshoot is probably due to the bootstrap circuit having higher impedance at hf and unable to charge the gate cap quickly.
The dc offset could be: input dc offset (if the amp has no input blocking cap), output dc offset (easy to measure) or something wrong with the measurements.
The overshoot is probably due to the bootstrap circuit having higher impedance at hf and unable to charge the gate cap quickly.
The dc offset could be: input dc offset (if the amp has no input blocking cap), output dc offset (easy to measure) or something wrong with the measurements.
Indeed, the amp is nowhere near clipping 🙂 I had set it to be around 1-2W of output or so as I'm using a 1x only probe, and the scope's input is limited to 12V at most.
The amp is connected to a huge 8ohm dummy load, and it has an input cap of 1u/63V to block input DC.
However, the output DC offset I forgot to check, I need to do so.
Thanks for the tips!
Just measured the DC offset on the output of the amp, with the dummy load connected. I have around 8mV DC on one channel, 5mV on the other one. So this is reasonably good from my POV.
Must be a measurement error then.
I have never come across an amp that has different gain on +ive and -ive sides of the waveform.
I have never come across an amp that has different gain on +ive and -ive sides of the waveform.
