>Using different opamps result enorumous difference in simulations.
This statement doesn't fit my observations.
Perhaps this depends on topology, but with these LMK/Wiederhold style amps,
op-amp type doesn't make big difference.
Op-amp here is not being used as voltage amplifier.
I guess the bast way to proceed would be to try sim with different op-amps, and compare results.
I've done this numerous times for all amps in this thread, and my conclusion was to go with TL071
Some people claim they 'hear' the difference, but I can't dispute subjective opinions..
I don't hear the difference.
This statement doesn't fit my observations.
Perhaps this depends on topology, but with these LMK/Wiederhold style amps,
op-amp type doesn't make big difference.
Op-amp here is not being used as voltage amplifier.
I guess the bast way to proceed would be to try sim with different op-amps, and compare results.
I've done this numerous times for all amps in this thread, and my conclusion was to go with TL071
Some people claim they 'hear' the difference, but I can't dispute subjective opinions..
I don't hear the difference.
Last edited:
>Using the original rail voltage dissipations are: Q1 - 125mW, Q3 - 112mW. Opamp dissipates 152mW so it could be very hot.
Could be.
Power over 100mW is too much in my opinion for small plastic transistors (some people will disagree..)
Op-amp voltage can be lowered (E.g. to +/- 12V); it only swings below 1V at the output anyway... 15V is not needed.
Maxim uses 12V on his schematics, and I used to use 15V, but my last 2 amps have 13V for op-amp.
Had too many 13V Zeners siting useless
Could be.
Power over 100mW is too much in my opinion for small plastic transistors (some people will disagree..)
Op-amp voltage can be lowered (E.g. to +/- 12V); it only swings below 1V at the output anyway... 15V is not needed.
Maxim uses 12V on his schematics, and I used to use 15V, but my last 2 amps have 13V for op-amp.
Had too many 13V Zeners siting useless
>It is necessary to watch the modes in the clip , in the linear mode it can and will roll, and in the clip it will burn.
Yes, this quick sim was just a beginning. Square waves need improvement, FFT profile, clip simming. Doesn't look like this schematic was thoroughly simmed.
Even gate stoppers are at the same value, and no GS caps added to compensate for asymmetrical input capacitance of the fets..
====
Да, этот быстрый симулятор был только началом. Прямоугольные волны нуждаются в улучшении, профиль БПФ, симуляция клипа.
Не похоже, что эта схема была тщательно смоделирована.
Даже ограничители затвора имеют одинаковое значение, и для компенсации асимметричной входной емкости полевых транзисторов не добавлены конденсаторы GS.
Yes, this quick sim was just a beginning. Square waves need improvement, FFT profile, clip simming. Doesn't look like this schematic was thoroughly simmed.
Even gate stoppers are at the same value, and no GS caps added to compensate for asymmetrical input capacitance of the fets..
====
Да, этот быстрый симулятор был только началом. Прямоугольные волны нуждаются в улучшении, профиль БПФ, симуляция клипа.
Не похоже, что эта схема была тщательно смоделирована.
Даже ограничители затвора имеют одинаковое значение, и для компенсации асимметричной входной емкости полевых транзисторов не добавлены конденсаторы GS.
R8 (4k7) may need to be adjusted if changing op-amp voltage.
Also, the fact that Q1/Q2 draw more power than 2nd cascode is strange,
this should be investigated. Doesn't sound right/optimal.
Q1/Q2 definitely don't need to draw over 100mW.
Maybe original design was with higher rails (like in my sim, where power dissipation is more reasonable),
and someone just lowered the rails, without re-adjusting anything??
Also, the fact that Q1/Q2 draw more power than 2nd cascode is strange,
this should be investigated. Doesn't sound right/optimal.
Q1/Q2 definitely don't need to draw over 100mW.
Maybe original design was with higher rails (like in my sim, where power dissipation is more reasonable),
and someone just lowered the rails, without re-adjusting anything??
Last edited:
New Quasi Project
Here is what I'm going to build next. Should be good for honest 250W (rms).
Slowly starting to work on PCB now.
Taking advantage that I can work remotely from home now.
Once we are back at the office, there will be no time to play around
Commute will take all my time I can dedicate to amps..
Symmetrical HexFet Quasi, using Baxandall diode placement suggested by SteveU earlier in this thread.
Fet drivers seem to improve everything a little bit.
10kHz FFT could have been more linear..
See the sim, any suggestions for improvements?
Thd and FFT numbers shown at full power.
TTA/TTC transistors used (Q1/Q2), because I have too many of them, have to use them up. KSC/KSA can be used instead (idle current will need to be re-adjusted).
There is so many LEDs here, it will look like Christmas tree..
================
Вот что я собираюсь построить дальше. Должно быть хорошо для честных 250 Вт (среднеквадратичное значение).
Начинаю потихоньку работать над PCB.
Воспользовавшись тем, что теперь я могу работать удаленно из дома. Как только мы вернемся в офис, некогда будет играть.
Поездка займет все мое время, которое я могу посвятить усилителям ..
Симметричный HexFet Quasi, использующий размещение диодов Баксандала, предложенное SteveU ранее в этой теме.
Полевые драйверы вроде все немного улучшают.
БПФ 10 кГц могло быть более линейным.
Смотрите сим, есть предложения по улучшению?
Числа Thd и FFT показаны на полной мощности.
Используемые транзисторы TTA / TTC (Q1 / Q2), потому что у меня их слишком много, приходится их использовать. Вместо этого можно использовать KSC / KSA (ток холостого хода необходимо отрегулировать заново).
Здесь столько светодиодов, что оно будет похоже на рождественское дерево.
Here is what I'm going to build next. Should be good for honest 250W (rms).
Slowly starting to work on PCB now.
Taking advantage that I can work remotely from home now.
Once we are back at the office, there will be no time to play around
Commute will take all my time I can dedicate to amps..
Symmetrical HexFet Quasi, using Baxandall diode placement suggested by SteveU earlier in this thread.
Fet drivers seem to improve everything a little bit.
10kHz FFT could have been more linear..
See the sim, any suggestions for improvements?
Thd and FFT numbers shown at full power.
TTA/TTC transistors used (Q1/Q2), because I have too many of them, have to use them up. KSC/KSA can be used instead (idle current will need to be re-adjusted).
There is so many LEDs here, it will look like Christmas tree..
================
Вот что я собираюсь построить дальше. Должно быть хорошо для честных 250 Вт (среднеквадратичное значение).
Начинаю потихоньку работать над PCB.
Воспользовавшись тем, что теперь я могу работать удаленно из дома. Как только мы вернемся в офис, некогда будет играть.
Поездка займет все мое время, которое я могу посвятить усилителям ..
Симметричный HexFet Quasi, использующий размещение диодов Баксандала, предложенное SteveU ранее в этой теме.
Полевые драйверы вроде все немного улучшают.
БПФ 10 кГц могло быть более линейным.
Смотрите сим, есть предложения по улучшению?
Числа Thd и FFT показаны на полной мощности.
Используемые транзисторы TTA / TTC (Q1 / Q2), потому что у меня их слишком много, приходится их использовать. Вместо этого можно использовать KSC / KSA (ток холостого хода необходимо отрегулировать заново).
Здесь столько светодиодов, что оно будет похоже на рождественское дерево.
Attachments
Last edited:
Stuff
Cross coupling a Baxandal diode is not vey useful driving FETs because the FET vto is a lot more then the diode forward voltage. Looking at the op-amp output shows a burst of oscillation at the falling crossover and removing/shorting the Baxandal diode cleans it up somewhat.
Some suggestions are:
1. The total of two FET vto plus the VAS voltage drop is a lot of lost positive voltage swing. My favorite solution is a partial bootstrap for the driver and VAS.
2. The VAS transistors can be completely shut during clipping, which aggravates rail sticking... So I would reduce the current in Q1,Q3 so that Q2,Q4 never completely shut off.
~3. Stabilizing a quasi is always a compromise. Increasing C8 helps but not a great solution. I have altered the P channel driver load which seems to work but may be hard on the driver.
Cross coupling a Baxandal diode is not vey useful driving FETs because the FET vto is a lot more then the diode forward voltage. Looking at the op-amp output shows a burst of oscillation at the falling crossover and removing/shorting the Baxandal diode cleans it up somewhat.
Some suggestions are:
1. The total of two FET vto plus the VAS voltage drop is a lot of lost positive voltage swing. My favorite solution is a partial bootstrap for the driver and VAS.
2. The VAS transistors can be completely shut during clipping, which aggravates rail sticking... So I would reduce the current in Q1,Q3 so that Q2,Q4 never completely shut off.
~3. Stabilizing a quasi is always a compromise. Increasing C8 helps but not a great solution. I have altered the P channel driver load which seems to work but may be hard on the driver.
Attachments
Thanks Steve, I didn't notice this opamp oscillations at 10kHz...
BTW, in one of the quasi hexfet amps I built earlier in this thread, I did not use Baxandall diode,
and in the sim everything was looking good, but in real build I had to add it,
because of visible crossover distortion on the oscilloscope.. The diode fixed it very well..
BTW, in one of the quasi hexfet amps I built earlier in this thread, I did not use Baxandall diode,
and in the sim everything was looking good, but in real build I had to add it,
because of visible crossover distortion on the oscilloscope.. The diode fixed it very well..
Last edited:
This is the post where addition of Baxandall diode was discussed:
#314
Unusual amp from 1987
..
#326
Unusual amp from 1987
#314
Unusual amp from 1987
..
#326
Unusual amp from 1987
Something like this
If there is no overload protection , then you can do without them
Attachments
Last edited:
Here is quick update:
a) Hawksford cascode added
b) Maxim's clipping diodes added
c) current in Q1,Q3 decreased so that Q2,Q4 never completely shut off
I'm still playing with drivers and bootstrap, as suggested by Steve..
The oscillation at opamp out is now just a little kink, like in Steve's sim from last night.
a) Hawksford cascode added
b) Maxim's clipping diodes added
c) current in Q1,Q3 decreased so that Q2,Q4 never completely shut off
I'm still playing with drivers and bootstrap, as suggested by Steve..
The oscillation at opamp out is now just a little kink, like in Steve's sim from last night.
Attachments
>The total of two FET vto plus the VAS voltage drop is a lot of lost positive voltage swing. My favorite solution is a partial bootstrap for the driver and VAS.
So the only reason to add bootstrap, is to make amp more efficient due to lost swing in the VAS/drivers?
I tend to never worry about thing like this (I know, my bad), because I have 2 (external) PSUs I'm using for my amps:
50V rails, and 80V rails.
So all my amps are designed for these two approx. voltages, regardless of their expected power
I don't mind lost voltage and less efficient amp. BUT - if we can't make it more efficient at low cost, and without sacrificing quality, why not.
So the only reason to add bootstrap, is to make amp more efficient due to lost swing in the VAS/drivers?
I tend to never worry about thing like this (I know, my bad), because I have 2 (external) PSUs I'm using for my amps:
50V rails, and 80V rails.
So all my amps are designed for these two approx. voltages, regardless of their expected power
I don't mind lost voltage and less efficient amp. BUT - if we can't make it more efficient at low cost, and without sacrificing quality, why not.
Last edited: