This is a well-studied problem. 🙂
Both Marcel's and Edmond's circuits have fast sensing. Marcel's feedback loop is fast (audio frequency). Edmond's feedback loop is slow with one capacitor having the same rate of charge and discharge.
I would make the sensing as fast as possible. I expect that it will be inaccurate when the amplifier is slewing. At least Marcel's circuit will not remember its past.
Ed
Both Marcel's and Edmond's circuits have fast sensing. Marcel's feedback loop is fast (audio frequency). Edmond's feedback loop is slow with one capacitor having the same rate of charge and discharge.
I would make the sensing as fast as possible. I expect that it will be inaccurate when the amplifier is slewing. At least Marcel's circuit will not remember its past.
Ed
Krijg nou wat! (alleen gezonde gingen)
So im chatting to That Marcel. We're hosting humble gems here
🤣
Now my brain went click. Essentially its probably better to design a bias loop that can operate correctly in the audio-band, and then make it sub-sonic. Or make the loop truly common mode.
Its the peak/dip detector that bugs me the most.
Then there is the question of whether this method of interpreting the bias current is even valid.
The bias sensor in Edmonds autobias is exactly the same though, just same sex.
Does he use the opto coupler diodes to form a kind of peakdetect or does he avoid it altogether i ask myself but im yet to read the whole thing.
Ill be a while absorbing your new loop. Thank you. I tried googling it a while back. Here its handed to me like its Christmas.
Cheers!
Rupopulles,
I’m flattered by your comments about my auto bias circuit. Seems like you got the gist of it pretty well.
I have two general comments.
The detector circuit has to be fast but with slow decay, so the bias doesn’t change with signal frequency.
The circuit has to have very high common mode rejection because speakers are reactive loads. Current and voltage do not cross zero at the same time. Either sense the current at the supply rails (like I did) or build a detector that floats with the output voltage, like some commenters did.
Good luck!
Russell
I’m flattered by your comments about my auto bias circuit. Seems like you got the gist of it pretty well.
I have two general comments.
The detector circuit has to be fast but with slow decay, so the bias doesn’t change with signal frequency.
The circuit has to have very high common mode rejection because speakers are reactive loads. Current and voltage do not cross zero at the same time. Either sense the current at the supply rails (like I did) or build a detector that floats with the output voltage, like some commenters did.
Good luck!
Russell
Russell,
I also like your auto bias approach. I played with a "minimum peak Detector" + integrator circuit way back in 1990. I "floated" it on the output. I didn't have any stability issues but I was never happy with the performance. At the time, I chalked it up to imperfections in the peak detector. That may have been the main issue but I may have also been seeing some transient thermal behavior in the output transistors that I did not fully appreciate at the time.
If you want to do most of the bias processing at ground potential, you can reduce the common mode problem by floating just a gain stage (with limiter) on the amplifier output and using a packaged differential amp to translate to ground reference like Jan Didden did in his “Thermal transient variation of power amp quiescent current instrumentation and findings,” Linear Audio, vol. 9, pp. 171-188, April 2015. This works quite well.
Bruce