Hartke 7000 output bad with load

[rf7]

The output of this amp looks unstable on a scope when it's feeding a load resistor. I feed a sine wave in, and the output has discontinuities in it. The output looks really jumpy on the scope too. If I remove the load the output looks fine.

I can feed the power amps directly and they drive the load all the way to clipping with no problems. If I stick a dummy plug in the return, the output of the send looks ok. When the problem is occurring, it seems to be throughout the whole amp. I have been occasionally able to get it to work for short periods, but usually it won't do that. And no amount of chop sticking anywhere inside will change the output. Power supply rails look ok.

Any ideas?

[Enzo]

Did you scope the HV rails while this was happening? Weak filters - or ones with cracked solder, this IS a Hartke after all - might hold the HV rails up OK under no load, but ask them to hold up while pushing a load and they can collapse.

Also look at the LV rails for the op amps while this occurs.

Make sure the grounds all have integrity. Is there continuity from the PT CT to chassis./ From the PA input connector ground to chassis?

By feeding the PA direct, do you mean clipping to its input connector pins or through the jacks on th panels?

The power tranny sends a number of wires to the main board. They are in several Molex connectors. FLip the board over and check the solder along all those pins along the rear edge of the board. See if any are cracked - or just resolder them all.

[rf7]

Did you scope the HV rails while this was happening? Weak filters - or ones with cracked solder, this IS a Hartke after all - might hold the HV rails up OK under no load, but ask them to hold up while pushing a load and they can collapse.

Yes, they don't collapse, but a smaller version of the output signal rides on the rails. I think that's normal.

Also look at the LV rails for the op amps while this occurs.

I also see a version of the output signal on the 15V rails. That does not seem right to me.

Make sure the grounds all have integrity. Is there continuity from the PT CT to chassis./ From the PA input connector ground to chassis?

Yes

By feeding the PA direct, do you mean clipping to its input connector pins or through the jacks on th panels?

I clipped to the input connector pins on the PA.

The power tranny sends a number of wires to the main board. They are in several Molex connectors. FLip the board over and check the solder along all those pins along the rear edge of the board. See if any are cracked - or just resolder them all.

I resoldered the whole power supply section on the front board, along with those connectors.

The signal on the 15V rails is puzzling, I was not expecting that. Could that be bad 15V filter caps or a shorting out IC?

[Enzo]

Probably neither.

The input connector is important. The ground there must be connected to the ground of the rest of the amp - a clip to chassis will do fine. SO if you unplug the PA input cable to clip on your test signal, the only ground it sees is the signal generator ground back through the mains ground to the Hartke. That might well explain the funny scoping.

You can clip a signal in there, but you still need to connect that ground pin to the chassis.

If I see signal just everywhere, I start to think the ground I am using for a reference is spongey. Note some of the screws that mount the main bottom board stick through metal grounding tabs. Put screws through those holes when doing these tests.

Where is your scope probe grounded? FInd some place close to the power ground - that is where the filter caps join ground.

And for that matter there might be some missing ground connection, even a burnt trace.

[rf7]

Finally got it figured out. The Bridge/Mono switch on the rear panel was flaky, and no amount of cleaning would cure the problem, so I replaced it. They run the output from the Pre-amp into the EQ section and then through that switch, before it hits the power amps. I was lifting coupling caps in the EQ section path and feeding signal in through the cap, trying to find the culprit. Took me long enough.