It is capable to run low impedance. But demanding 400W out of 3 output pairs is well optimistic.
Yeap! I guess 250-300W will do the trick!
Any suggestion about output transistors? I would like to bias it as much class A as possible - modu 5U 500mm case
Sounds like a recipe for secondary breakdown failure. I'd suggest biasing as designed given the HoneyBadger doesn't have VI limiting protection circuitry and the distortion is at a minimum if correctly biased.
NJW3281 / 1302 are easy to get output devices.
Yes I agree with the others. Don't try to overdo the basic design. I will let you know the outputs I used after I take a look. Most people were telling me that I was going to high with supply voltage with a 4 ohm load. I have had no problems though. Now that is with forced air cooling when the output heatsinks felt warmer than comfortable. But cooling was 24 volt fans running at 12 volt. You cant hear them at all. I'm very happy with the sound for a few years now. bias is right around the middle if I remember correctly. Details that I have to check on.
I was wondering the same thing. Aren't high speed output devices more prone to oscillation?
Slow switch-off leads to heavier dissipation at higher frequencies and more distortion as the device that's supposed to be off stays on for a significant fraction of the half-cycle. The 18/19kHz intermodulation measurement is a good one to spot high frequency issues like this,
as is a plot of THD v. frequency.
as is a plot of THD v. frequency.
Not necessarily - you'd have to get your hearing measured and the distortion measured and figure out if there's an issue...
You can try an ultrasonic I.M. test (say 22kHz + 23kHz) to see if any audible 1kHz breaks through. This may distress nearby pets though, and is testing your speakers for distortion too.
Another check you can do is measure current consumption at moderate power for 1kHz v. 20kHz (into a dummy load) - ideally the current consumption should be the same. Any large difference likely indicates an issue with switch off speed. You need to check the output amplitude is the same for the test to be meaningful (not all amps have dead flat frequency response upto 20kHz).
Don't drive an amp at full power at high frequencies for very long, this is high stress for the driving and output stages and the Zobel network. Certainly don't do this into a real loadspeaker.
You can try an ultrasonic I.M. test (say 22kHz + 23kHz) to see if any audible 1kHz breaks through. This may distress nearby pets though, and is testing your speakers for distortion too.
Another check you can do is measure current consumption at moderate power for 1kHz v. 20kHz (into a dummy load) - ideally the current consumption should be the same. Any large difference likely indicates an issue with switch off speed. You need to check the output amplitude is the same for the test to be meaningful (not all amps have dead flat frequency response upto 20kHz).
Don't drive an amp at full power at high frequencies for very long, this is high stress for the driving and output stages and the Zobel network. Certainly don't do this into a real loadspeaker.
By comparing the transistion frequencies (4 MHz for MJL21193 vs. 30 MHz for MJL1302A or even 35 MHz for MJL4302A) I assume some demand for another compensation regime. I'm not that of an expert here, though, so Ostripper, the designer himself, or others might help with a competent answer?
Btw, MJL1302A/3281A are much cheaper @Mouser than MJL21193/94, so why these old devices?
Best regards!
From what I can remember the MJL21193/94 were fine in an EF2 design amplifier. They didn't become an issue until used in an EF3 like the Slewmaster. The slower devices would suffer shoot through / cross conduction and be destroyed. Bob Cordell say the cause of the issues was too low bias current in the drivers, they simply didn't have enough power to switch the output off fast enough.
Compensation brings the loop gain down to 1 at some frequency before the phase margin is too small, its the gain of the devices at that frequency that affects compensation, not really how high their transition frequency is (although they are related). However the output devices have current gain only, no voltage gain, so are less relevant to compensation than the VAS.
OS, which drivers would you recommend with the above? I've sourced a few of them!