FTWWTK (For Those Who Want To Know) 
I saw at http://nov55.com/amr/amf2.htm#Top
that the driver transistors had small caps over the B-C junction.
I tried that. At least the excessive amount of quiescent current had disappeared.
I'll do some measurements today !
grtz
Simon
I saw at http://nov55.com/amr/amf2.htm#Top
that the driver transistors had small caps over the B-C junction.
I tried that. At least the excessive amount of quiescent current had disappeared.
I'll do some measurements today !
grtz
Simon
Thanks Upupa Epops.
The amp is oscillating, and this is caused by too much gain at the pole frequency. If you know how high this oscillation is, it will give a clue to the origin of the problem, but I'd suggest it will be the voltage amplifiers.
You can reduce this gain by adding 15pF silver micas across the base/collector of Q33 and Q34. This will improve phase margin.
You could probably go as high as 33pF; and all the advice about grounding the heatsink and installing base stoppers is very important too of course.
Cheers,
Hugh
You can reduce this gain by adding 15pF silver micas across the base/collector of Q33 and Q34. This will improve phase margin.
You could probably go as high as 33pF; and all the advice about grounding the heatsink and installing base stoppers is very important too of course.
Cheers,
Hugh
I placed 150p styro's over the B-C junctions of Q32,Q36, Q37 and Q39. This helped to stop the oscillations.
Hugh : i did not try your proposal yet.
I'll let you know.
Upupa Epops : This additional resistor will be in the next (updated) design.
MikeB: i did simulations to get a conclusion on feedback compensation. The phase does never get close to 180 degrees a unity gain. 20 degrees of, at least in this setup.
I used
http://www.sumuller.de/audiotester/
To sweep my amp. No FFT and AC seeps yet. I listened to it only !
The amp is silent with no source connected but it hums when i do connect one. Next challenge
grtz
Simon
Hugh : i did not try your proposal yet.
I'll let you know.
Upupa Epops : This additional resistor will be in the next (updated) design.
MikeB: i did simulations to get a conclusion on feedback compensation. The phase does never get close to 180 degrees a unity gain. 20 degrees of, at least in this setup.
I used
http://www.sumuller.de/audiotester/
To sweep my amp. No FFT and AC seeps yet. I listened to it only !
The amp is silent with no source connected but it hums when i do connect one. Next challenge
grtz
Simon
Hi Blu_line !
Ok, i tried a bit in sims for feedbackcompensation, check attached
schematic. I had not the models for your transistors, so i chose
something "similar".
The changes there are mainly:
c26,r49,r50,c25,r45,c11
c21,c22,c23,c24,
r52
r52 is a need to try, if it sounds less bright with this, it's necessary.
It's purpose is to enhance openloopbandwidth.
Your circuit was a bit tricky, it does not show a dominant pole, means
it does not oscillate in sims. But it shows overshoots when fed with
a 20khz squarewave. With my changes these should be gone.
Ok, have fun with my suggestions, if it still oscillates you might have
to deal with local oscillations. In this case add 47ohm basestoppers
to q7,q3,q11,q12 and so on...
Mike
Ok, i tried a bit in sims for feedbackcompensation, check attached
schematic. I had not the models for your transistors, so i chose
something "similar".
The changes there are mainly:
c26,r49,r50,c25,r45,c11
c21,c22,c23,c24,
r52
r52 is a need to try, if it sounds less bright with this, it's necessary.
It's purpose is to enhance openloopbandwidth.
Your circuit was a bit tricky, it does not show a dominant pole, means
it does not oscillate in sims. But it shows overshoots when fed with
a 20khz squarewave. With my changes these should be gone.
Ok, have fun with my suggestions, if it still oscillates you might have
to deal with local oscillations. In this case add 47ohm basestoppers
to q7,q3,q11,q12 and so on...
Mike
Attachments
Why don't you try the classic 5 to 10 ohm and 100nF snubber network from output to ground?
Without this network the output stage is unloaded at high frequencies and this makes loop gain unpredictable [actually it changes when you touch the heatsink]
Try this network, it will fix the open loop gain and will make most of the extra compensation you've added unnecesary
Without this network the output stage is unloaded at high frequencies and this makes loop gain unpredictable [actually it changes when you touch the heatsink]
Try this network, it will fix the open loop gain and will make most of the extra compensation you've added unnecesary
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.