Hello all, my father picked up a rare amp/preamp combo, that amp is a Superphon Maxx 806 (or 80b) made right down the road from me in Eugene, Oregon by Stan Warren of PS Audio. Two of the N channel fets in the right channel had exploded prior to him buying it. I ended up replacing all of the mosfets with IRF520/IRFP520 devices and replaced all of the capacitors, increasing the filtering from 20,000 per rail to 30,000. I checked all of the balancing resisors and other transistors in the circuit prior to the outputs and all was good. Guessing maybe the output was shorted????
Powered it up and it was great for about 5 minutes and then I turned it off. I’m not super familiar with fet amps, BJT mostly. There is a small trimmer in each channel and looking at the paint they are in the factory positions. Should I leave it and run it as is? Should I make an attempt at setting bias? Unfortunately without any schematic or service manual I’m kind of lost as far as what I should do with it. Seems stable as is, but would love some guidance. Thank you.
Dan
Powered it up and it was great for about 5 minutes and then I turned it off. I’m not super familiar with fet amps, BJT mostly. There is a small trimmer in each channel and looking at the paint they are in the factory positions. Should I leave it and run it as is? Should I make an attempt at setting bias? Unfortunately without any schematic or service manual I’m kind of lost as far as what I should do with it. Seems stable as is, but would love some guidance. Thank you.
Dan
You can find out what bias you have simply by measuring voltage across those 0.22 ohm resistors.
Turning the pot one way will increase the value, in other direction it will decrease it. Its easy. Even without schematics.
Turning the pot one way will increase the value, in other direction it will decrease it. Its easy. Even without schematics.
Thank you, any idea what would be a good number to shoot for? 10 mA, 20, 50? I’ll measure both channels this morning and post my findings.
Dan
Last edited:
Given that you have replaced the FET's means the original bias setting is no longer valid. I would imagine something around 100ma per output pair... so around 22mv across each 0.22 ohm. You will have to just work to an average figure most likely.
Another line of thinking is that I can't see any thermal compensation on the output stage and so that could mean a low bias figure is called for. What is optimal and what is actually recommended can be two very different things.
Another line of thinking is that I can't see any thermal compensation on the output stage and so that could mean a low bias figure is called for. What is optimal and what is actually recommended can be two very different things.
I figured that the bias would need to be completely reset with the new outputs. With their current position I measured just above 13mV on the left channel (not exploded outputs) and just below 12mV on the right (exploded outputs). I checked after leaving the amp on for about 3 hours. At this point the heatsink was just slightly warm.
Really wish I had the thoughts of the designer. What do you think? Take them both to 13mV? Bump them to 15mV? Maybe even to 20-22mV as you suggested? It does have thermal monitoring in the form of a thermal switch on the heatsink to cut power if it gets too hot, but would prefer that isn’t ever utilized lol.
Dan
I do not think its thermal switch, just a thermistor feedback. How else would one utilize full power of the amplifier.
If it sounds good, I would leave it as it is. Doubt you will hear the difference.
If it sounds good, I would leave it as it is. Doubt you will hear the difference.
I believe it is, one side of the power cable comes in to the thermal switch and comes out directly into the power transformer. While the other side of the power cable goes into a fuse then into the transformer, no power switch.
That was one thing I was curious about, if there would be an audible difference. Well since the left channel is about 1-1/4 mV higher I may just bump them both to 15 and see how it goes. If anyone recommends something different please let me know.
Thank you,
Dan
Yeah I had the same thing on my HK Citation Nineteen amp, would trip all the time until I decided to check bias and see it was nearly double of what it should have been lol.
Dan
It really does depend on what the designer wanted. I can't answer that question.
To high and it runs hot. Is it too hot? we don't know.
How stable is the bias?
You don't want the bias to run away (thermal runaway) at high temperatures. If you run the amp at say 40% output power then that will give maximum heating in the FET's. Get the heatsink really hot and then cut the signal. Is the bias still where you set it or is it high or low?
If you have the bias to low and get the heatsink really hot then you might find the bias falls to near zero when you cut the signal. That happens if the stage is overcompensated thermally.
I don’t know what really hot would be, but I got it quite warm. Started out at about 13mV of bias in both channels. Injected a 1khz tone and drove the amp into a pair of 8 ohm dummy loads at about 7.5 watts each channel. At this level the bias was at 40mV in each channel. Left it for about an hour and it didn’t drift from 40mV at all. The section of the heatsink right at the mosfets was right around 110 degrees fahrenheit. I saw the actual devices reach about 140 and then cut the signal.
Bias immediately dropped to 2mV in each channel and took about 20-30 minutes to slowly come back up to the 13mV where it stayed.
Does that sound like it’s overcompensated thermally?
Dan
Last edited:
Either over compensated or it is designed to run at higher bias level or it is just the way it is. I'm afraid we just don't know.
The heatsink I can see in the picture looks quite small tbh and so I would say that 100ma per pair is going to be a safe value. A temperature like a really hot radiator is fine for the FET's, if the metal tabs of the FET's getting close to sizzling a drop of water, then its too hot,
The heatsink I can see in the picture looks quite small tbh and so I would say that 100ma per pair is going to be a safe value. A temperature like a really hot radiator is fine for the FET's, if the metal tabs of the FET's getting close to sizzling a drop of water, then its too hot,
Yeah the heatsink is relatively small. So with a 100mA per pair that would be 22mV measured across each .22 ohm resistor? You say each pair and I don’t know if you mean they share that current or each device. Each of the 8 outputs has a 0.22 ohm resistor at the source pin, so if I measure 22 mV across each of the resistors wouldn’t that mean there would be 100mA of current at each device?
I really appreciate your help with this.
Dan
I did mean per FET and so it would mean 100ma per device which for eight devices is 800ma. I can see the big caps are 35v and so if we say -/+30 volt rails as a maximum (and maybe its lower) then that gives 48 watts dissipation at idle. It is a lot.
Only you can make a call on where to set it. Why not try it at minimum and see if you hear any distortion at low levels. In practice anything over a milliamp or two will remove most distortion.
As to what is recommended.... we just don't know.
Only you can make a call on where to set it. Why not try it at minimum and see if you hear any distortion at low levels. In practice anything over a milliamp or two will remove most distortion.
As to what is recommended.... we just don't know.
And most probably asking too much from that particular heatsink for reliable operation, especially in the summer.
Okay, I’ll do some playing around. I’ll get it setup into a bedroom with a heater to set the room at 72-73 F as that is as hot of an ambient temp it’ll be in. Listen to it, run it fairly hard for a day and see how hot it gets. Maybe being at 13mV per channel it’s good. Rails are +/- 30v, they just measured 31 volts. So at their current bias setting I’m at about 29 watts dissipation at idle. It’s been sitting on for over 24 hours and in the current room the heat sink is just slightly above the ambient temp, can feel just slight warmth coming from it. It does dip with the fluctuating input voltage, as right now bias is at 11.8mV on each channel, but I can come back in an hour from now and likely see it at 13 or slightly higher.
Dan
Dan
Try it at minimum bias by listening to it.
Better still, have you got a scope? If so, then have a look and see if you can see any crossover distortion at 10kHz. See what bias you need to make it go away.
Better still, have you got a scope? If so, then have a look and see if you can see any crossover distortion at 10kHz. See what bias you need to make it go away.
I certainly do, so take bias all the way down, feed 10khz signal and adjust until crossover distortion goes away? I’ll give it a shot.
Dan
I took bias to its minimum which got me about 0.25 mV in each channel.
I got a 10 kHz signal put into it, is there a recommended amount of drive? Meaning a minimum wattage output from the amp I should be driving for?
I first took bias to 5mV
You can see the waveform looks pretty crappy. Slight crossover distortion and the peaks of the waveform are clearly distorted.
So bumped it to 7.5mV
Cleaned it up a bit, but still slight crossover distortion and the peaks look bad.
Bumped it to 9mV
Again, a little better, but the peaks look not so great and like 1/4 way down on the down slope (and 1/4 way up as it goes up) it looks like maybe some distortion???
So I just bumped it back up to 13mV
Much better overall with the exception of the peaks. At 1 kHz the waveform looks perfect
the distortion on the peaks of 10 kHz is there regardless of how much power it’s outputting, but that above is only about 14 watts per channel. It’s there at 2 watts.
Would continuing to drive up the bias eliminate that peak distortion? Maybe 22mV would be good lol. But if that distortion isn’t due to the bias then I don’t want to push this tiny heatsink.
Dan
I got a 10 kHz signal put into it, is there a recommended amount of drive? Meaning a minimum wattage output from the amp I should be driving for?
I first took bias to 5mV
You can see the waveform looks pretty crappy. Slight crossover distortion and the peaks of the waveform are clearly distorted.
So bumped it to 7.5mV
Cleaned it up a bit, but still slight crossover distortion and the peaks look bad.
Bumped it to 9mV
Again, a little better, but the peaks look not so great and like 1/4 way down on the down slope (and 1/4 way up as it goes up) it looks like maybe some distortion???
So I just bumped it back up to 13mV
Much better overall with the exception of the peaks. At 1 kHz the waveform looks perfect
the distortion on the peaks of 10 kHz is there regardless of how much power it’s outputting, but that above is only about 14 watts per channel. It’s there at 2 watts.
Would continuing to drive up the bias eliminate that peak distortion? Maybe 22mV would be good lol. But if that distortion isn’t due to the bias then I don’t want to push this tiny heatsink.
Dan