Mackie SWA1501

[DB Techs]

Initially, R17 smoked, R16 went open circuit and Q5 and Q17 shorted.

Detailed examination with an in-circuit curve tracer (Huntron Tracker) 'health' test caused me to replace Q7, Q14, Q18, Q19 .

Every diode and semiconductor has been tested.

Further testing revealed that the component legs of Q4 were intermittently open-circuit inside the body of the To-220 package. Definitely difficult to diagnose. This may have been caused by the stress of clamping the component to the heat sink with inaccurate component leg-lengths during the manufacturing process, (there is some evidence that the input PCB heat sink mounting technique broke the legs of T6...but...later for that one...)

Applying low voltage with a variac reveals only one of D15 or D16 illuminating, depending on whether or not the input board is connected.

Increasing the applied VAC with the variac reveals the amp still wants to blow up...

I am open to advice from anyone as to how to proceed.

I'll attach a summary document from many threads I have read on this forum as an FYI.

[km6xz]

The dynamic ground system used on this and several other sub-amps use is confusing unless you really understand what it supposed to do. In what way does the amp want to burn itself again? Does that mean both main output transistors are conducting at the same time? Or that short exists full time and you don't see excess current until mains volts gets higher? The only parts of the circuit that can pull high current are the complentary outputs and the ground reference steering MOSFETS Q12 and Q15. If it is the outputs, suspect a part[s] in the bias/level shifting network associated with Q7. I assume you are testing without a load on the amp. If you are using a load.....don't. What is getting hot, how much current is being pulled by measuring the drop across the emitter resistors of Q14 and Q17? Without a load, if one of those outputs is shorted, you will get DC offset but it should not be pulling high current, if you are using a load, it will pull a lot of current with DC offset. Why did the Huntron cause you to replace those other components, did they measure bad independently from the Huntron? In-circuit testing can cause probably in the circuit when sweeps with negative exciting voltage which the Huntron does part of each cycle. If it was pulling current from the start and Q5 is shorted, I would immediately suspect Q18, both power outputs, R64, R63 and D29 and D27. If you suspect the driver and bias circuit, pull the Q14 and Q17 transistors out and see if the level at the voltage drop across R31 and R32 are the same. If they are, the drive circuit is probably OK. See if a signal on both resistors are the same but with different offset of less than 1 volt references to ground.
That focuses on the amp portion but the dynamic, active ground system can also be suspect. Check for equal voltage drop across Q12 and Q15, with no signal. That will tell you whether the artificial ground reference is centering.

[DB Techs]

Thank you so much!

You have given me much to think about.

The Huntron tests were on components out-of-circuit.

I'm running the AC input with a variac (auto-transformer) at ~40-50 VAC to obtain the following readings, (please let me know that you can view this screenshot):



As you can see I'm missing the positive voltage from FWB D35. I trust this would explain why I can blow Q5 & Q17 and burn R16 & 17 if were to apply full power - which I have done once already, (so much for my confidence).

Now to understand why...lots of circuitry attached to #1 power supply path...

The other supplies +/- 15 & 25 seem OK.

One thing that is interesting is the way only D15 or D16 is on. They alternate in illumination when I toggle the power switch on and off under 40 VAC input. This machine is quite the balancing act...

[km6xz]

Yes, D35 not showing a positive voltage tells you a lot. Remember it has no fixed ground reference so the supply is not seeing a short to ground when the positive terminal is at Zero potential above ground. That means something is referencing that rail as ground. The dynamic ground system allows higher voltage swings across the load than is across the output transistors by moving the ground reference away from Zero in the opposite direction to the power amp output. So AF ground is being shifted towards the positive rail even without a signal. That could...would...happen if Q15 was turned on. Is should be barely turned on, the same amount that Q12 is until there is signal. A negative going signal at the output of the amp stage would be accompanied by a positive going AF ground reference. The opposite would involve Q15 with a negative going signal at the amp output so the difference between the more negative ground and more positive power amp output would be a greater potential difference across the load than either the ground shift or the power amp swing....mucho power from few devices and low heat. This only works well with low rise time signals unless small filters and faster shifting can be done. The time constant is intentionally kept long to prevent ripple modulation of a higher frequency signal. Below 100hz or so it is great, otherwise a real bridging configuration works better like Crown has has for decades.

Back to your problem. What is turning on Q15 and if it is not, what is leaking so the ground reference is being brought up to the positive rail? Have you tested Q15 passively or with actually driving it's gate to see if it conducts or cuts off as expected? If the gate is being driven hard on, look for the drive circuit at the junction of R7 and R42. What is the voltage at pin 7 of U2?