My days of thinking I'm clever are long over, so nobody need worry about that. I just noticed that all my power amp simulations go potty whenever I leave class A, and everything I could find on sliding bias seems to have quietly scurried away with time except new products branded as such. Pioneer solved all of class AB's problems in '75 with the 8800, the Levinson 33 sounds good. I get it, but there are no .asc files showing it works at all?
I recently obtained some Exicon fets and determined I would try to use them as best I can. A friend told me class A is dead so I tried to avoid it. The following happened and I can't come up with a good excuse except I started with a "current feedback" circuit and wanted to keep the pulse performance and have less distortion. Please critique or I might be doomed to make a terribly expensive oscillatorsparksmoke. I'll probably do it anyway.
There are some magic current and voltage sources here to make things easier to simulate but the idea I think is pretty obvious. The bipolars are thrown in only because I noticed the latfets didn't like low voltage in heavy saturation so much. They really needed at least 10 volts on them, which is more heat. The "third loop" differential input stage is sensitive to the input filter shift so that the exact values have to be tuned to get the error signal to "null" at 20kHz. Of course the overall feedback resistor must be as well, but you could probably watch the current on the output coupling resistor while adjusting. The differential output coupling resistor value can be tuned for sound, damping factor, or oscillations, whatever you're into. The current on that can come down to some hundred nA of pretty much just the error in simulation. Being off by couple 10uA doesn't seem to wreck it though. I haven't come up with a great way to match the first two loops without a distortion analyzer but haven't thought about it much. Other than that anything else is or isn't real either. Sorry for the scrubby drawing but at least it actually ran the last time I tried. Thanks for looking.
A
I recently obtained some Exicon fets and determined I would try to use them as best I can. A friend told me class A is dead so I tried to avoid it. The following happened and I can't come up with a good excuse except I started with a "current feedback" circuit and wanted to keep the pulse performance and have less distortion. Please critique or I might be doomed to make a terribly expensive oscillatorsparksmoke. I'll probably do it anyway.
There are some magic current and voltage sources here to make things easier to simulate but the idea I think is pretty obvious. The bipolars are thrown in only because I noticed the latfets didn't like low voltage in heavy saturation so much. They really needed at least 10 volts on them, which is more heat. The "third loop" differential input stage is sensitive to the input filter shift so that the exact values have to be tuned to get the error signal to "null" at 20kHz. Of course the overall feedback resistor must be as well, but you could probably watch the current on the output coupling resistor while adjusting. The differential output coupling resistor value can be tuned for sound, damping factor, or oscillations, whatever you're into. The current on that can come down to some hundred nA of pretty much just the error in simulation. Being off by couple 10uA doesn't seem to wreck it though. I haven't come up with a great way to match the first two loops without a distortion analyzer but haven't thought about it much. Other than that anything else is or isn't real either. Sorry for the scrubby drawing but at least it actually ran the last time I tried. Thanks for looking.
A
I don’t thing Class A is completely worthless in the modern world. Lateral mosfets have very low distortion biased into class A.
I found Nelson Pass’s patented cascode output design, highlighted in John Broskie’s post Post 554 but using lateral mosfets for the class A stage gives very high performance.
I found Nelson Pass’s patented cascode output design, highlighted in John Broskie’s post Post 554 but using lateral mosfets for the class A stage gives very high performance.
Ah yes you can nearly get rid of one whole amplifier that way. I'll have to see if it works better. Thanks!
I will use Class A for tweeter in a future 3-way active speaker. Maybe it sounds better if I use it for midds as well. -I don't have enough experience in critically comparing HiFi components subjectively to judge this. so I will settle with using my Cambridge poweramps that supposedly doesn't have crossover distortion.
I would have moral issues with running powerful Class A amplifiers.
I won't swear ill never get a Vox AC50 Class A tube amp for guitar again though.
Cheers!
I would have moral issues with running powerful Class A amplifiers.
I won't swear ill never get a Vox AC50 Class A tube amp for guitar again though.
Cheers!
You can have ClassA, let say about 200W/4ohm with a bit higher efficiency of a ClassB, look here: https://www.diyaudio.com/community/threads/100w-cfa-class-h-classa-classb.394923/#post-7545142
The same principle can be done with ClassD as the helper amp.
The same principle can be done with ClassD as the helper amp.
Most of modern loudspeakers are not very efficient and 100W on 8 ohm is not to much. Angtyfatcat have you looked the schematic in the link I possted?
I was talking about classic ClassA with ClassD as a low voltage power supply modulator.
I worried about that. I'm No Pro when it comes to models. The best I can do is use what's already floating around on the net. One thing is for sure, no matter what I use on the final output amp they make more distortion when the bias drops below the signal. 🙂
A class D amp that can match the pulse response of the whacky setup I got going on here would be something else indeed! I hope to go up to 80V rails with this and a Lot more power transistors in the near future but my budget doesn't cut it for that right now. Plus, I'd just like to see this work first..
ALrighty. I got around to trying the Pass bootstrap idea and it works awesome for getting rid of the power supply driver at audio frequencies. It doesn't seem to handle the step response of two identical amps running parallel. I don't know if it really matters if you can go faster than 100V/uS, but I certainly like seeing that happen. Also, the "helper" transistors on the primary input stage are bogus far as I can tell once the differential is tuned, unless you don't plan on any differential, which I didn't at first. When I get to the first set of boards I'm not sure that will still be there.