QSC USA850 problem

[bkahuna]

I've got a USA850 that is behaving very strangely. The +15 volt supply on both channels will not stay up. In fact they never really get up to +15 volts. They hover around 12 or 13 volts. Definitely not regulated. Both sides will produce output and seem to function with no load, however, when a load is added and signal applied, the +15 volt supplies on both channels will crash to about +5 volts and the output will mostly go away, leaving only part of the positive half cycle. The crash to 5 volts does not happen simultaneously in both channels. In other words, each channel crashes when driven, but the other channel will stay up until IT is driven.

I found and replaced a shorted zener (Z1) in channel A. This got me to the point I'm at. Prior to that, channel A had no output at all, while channel B was already exhibiting this symptom.

I thought it was likely the muting circuit, but I have removed Q12 and the symptoms remain. R20 is a little discolored (it's the one that was driving the shorted zener), but I've measured it out of circuit and it measures fine. I can't imagine it would be affecting the other channel.

I've replaced both 5532s(they're in sockets) as well, thinking they could be dragging down the supplies. Thinking that my stock of 5532s might draw more current than the originals, I replaced them with 4565s. The result was that I had to load the amps down to 4 ohms to get them to fail, but they did still fail.

I've talked to several techs at QSC, and while all very helpful and knowledgeable, they couldn't really produce any new ideas beyond what I'd already done.

Any suggestions before I pull the rest of my hair out?

[g1]

In fact they never really get up to +15 volts. They hover around 12 or 13 volts.

This would be a good place to start (may require putting 5532's back?). How are the other supplies under this condition (-15, + and -78)?
You are getting more voltage drop across R20 than R21? (measured across resistor, not to ground)

[bkahuna]

This would be a good place to start (may require putting 5532's back?). How are the other supplies under this condition (-15, + and -78)?
You are getting more voltage drop across R20 than R21? (measured across resistor, not to ground)

Thanks for the response.
The 5532s are already back in. Voltage drop across R20 and R21 are stable at about 60vdc when the unit is operating. When I "goose" the input signal to make it fail, R20 jumps to about 85v then drops slowly back to 60. R21 goes the opposite way: It drops to 45 and starts climbing back toward 60 when input is removed. When those two voltages stabilize around 60vdc, the clip light goes out and that channel becomes functional again.

The -15 volt supplies drop to around -8v in failed condition.
+73 jumps to around 100v in failed condition.
-73 drops to about -45 in failed condition.
These last 3 measurements are all referenced to ground.

[bkahuna]

Here's an update: After my last measurements I started thinking it might be a grounding issue since the rails were actually changing with respect to the ground I was using. Turns out that was a correct assumption. This amp has the collectors of all output devices grounded through the heatsink. I found two outputs that did not read continuity to ground. The screws holding them to the heatsink were loose enough that they were no longer electrically grounded. In fact, a lot of the other output xistor screws were loose as well. After tightening the xistor screws one channel started working. After tightening the bolts that actually ground the heatsink, the other channel started working.

Then I checked the +15v supplies. They're still flaky as hell! They just don't crash anymore. In fact, they don't actually hit +15v until you're driving the amp pretty hard. That's when the diodes on the output are rectifying the output signal and adding it back into the low voltage supplies. Anyone know if this is how these amps work, or should there be a solid +15v there all the time?

I've also had one instance where one channel powered up with the clip light on and wouldn't come out of it until I turned it off and back on again. Guess I'm not out of the woods yet!

[Enzo]

This is a "flying rail" or "grounded output" type amp circuit. The HV rails will stay at their voltage with repsect to each other, and indeed across their own filter caps. They are stable to their common, but their common isn't ground. It is the speaker connection. SO if the amp "output" swings one way, it does so by swinging the whole power supply that way. So from 70v, one rail goes up 30, and so does the other, in your example.

DO yourself a favor, before going further, look left center for the RIPPLE NULL pot TR2. and associated with it are R7, R8 47k 1w. See if any of those are open. A problem ther can be maddening to diagnose if you have never seen it, and it can cause all sorts of odd problems.

Those 15v zeners at the 5532 are there to limit the 15v to 15v, they are not there to ensure 15v all the time like a comon zener supply. Certainly a defective zener will cause problems, but if the thing idles at close to 15v, especially if you have replace them, then collapsing 15v rails is not their fault. More likely the collapsing voltage is a symptom, not a problem. WHen the 5532 output tells the V+ side to conduct harder, that pulls the V+ rail down, and down far enough it starts to reduce the +15v to the IC, this serves to limit how hard it can turn on the outputs beyond that point.