Mosfet amp goes crazy at clipping

[Zero Cool]

I am working on a power amp that uses the 240/9240 mosfets. the amp works fine at all levels until you get to clipping then it goes nuts.

I couldnt figure out why this thing was burning up output relays. the relays are rated for 10 amps.

I replaced the relay and the amp ran fine. In fact it ran into a dead short until i figured out that the output relay had disintegrated and shorted all the contacts together (one being ground). I removed the relay and the amp ran fin on the bench. nice clean signal with no load up to clip.

So i replaced the relay, put it back together again and tested it. sounded great. so i cranked it up a little but and the power light flickered and then dimmed and then no sound. power light still on.

Back down to the bench i had no output. after a bit of checking, output relay melted again. this time i removed the relay and re tested the amp. works great. hooked up a 8 ohm resistive load and cranked it up. looks good until you get close to clip, then a little bit of what i would describe as noise on the wave form. Hook up a 4 ohm load and NOW the problem becomes clear!!!

Just before clip with a 4 ohm load, lots of noise appears on the waveform. then it starts to look like ripple, then the amp goes massivly into clipping and the scope screen is a scrambled mess of massivly clipped waverform. the amp starts mechanically screaming a very high pitched squeal.

I think it was ultrasonically welding the relay contacts together!

back the input signal down and the amp works fine. its only when the amp gets near full power does the problem happen.

So now i am wondering if it is something in the IV limiter, or if the mosfets are bad???


Any ideas???


Zc

[CeramicMan]

Polarity inversion in the feedback loop? Some op-amps do that at clipping too.

[Zero Cool]

Originally posted by CeramicMan
Polarity inversion in the feedback loop? Some op-amps do that at clipping too.

What would cause that? and what would i look for?



Zc

[HDTVman]

It sounds to me like the amp is starting to oscillate as it goes into clipping. Try adding some bypass caps from the PS rails to grd. @ the outputs. .1uf might to the trick or try .1 ohm series resistors in the power supply lines to the outputs.

My guess is that the clipping is causing the power supply to ring like a bell and that the ringing noise is getting into the drive circuits causing the oscillation.

Just a thought, what's it cost to give it a try?

BZ

PS what are you using for a scope?

[Leolabs]

Output zobel network???

[Workhorse]

Post your schematic first and then ask the other members for help.

[Zero Cool]

Originally posted by HDTVman
It sounds to me like the amp is starting to oscillate as it goes into clipping. Try adding some bypass caps from the PS rails to grd. @ the outputs. .1uf might to the trick or try .1 ohm series resistors in the power supply lines to the outputs.

My guess is that the clipping is causing the power supply to ring like a bell and that the ringing noise is getting into the drive circuits causing the oscillation.

Just a thought, what's it cost to give it a try?

BZ

PS what are you using for a scope?

I will try that. I do notice that when i bring the amp up on a variac slow, it oscillates until the rails get closer to there rated voltage.

I have a sencor SC-61 for a scope. 60Mhz usable to 100Mhz it says. not a great scope but i like the built in DVM and Freq Counter plus the delta T functions. makes some calibrations really easy.

[Zero Cool]

Originally posted by Workhorse
Post your schematic first and then ask the other members for help.

Thanks, unfortunatly i have the sams on paper and not in a PDF or any easily postable schematic....Let me see if i can dig my old scanner out and see if it still works. havent used it in years.



Zc

[Zero Cool]

[Workhorse]

Also show us your opamp frontend, and check for series connected back to back 2 zeners of value around 6.8V from inverting pin to output pin of opamp. if not there then connect them. They will solve your unstable clipping behaviour.