MA6100 output bias creep issues

[CVElectronic]

I took in a MA6100 that has turned out to be a major headache. It came in with the right channel blown, so I fixed that. The right channel is now stable. Then, I noticed that the left channel would get warmer and warmer the longer it was left on. I was about to do some diagnostic troubleshooting measurements on the left channel when it suddenly blew both 5A rail fuses at the same time. Output transistors were shorted.

I've replaced the outputs with MJ15003G and the sentry transistors Q407/Q409. Because I got odd readings from one of the drivers, I replaced Q411/Q413. I have not yet replaced Q405 (seems it can only be ordered from McIntosh) or the differential pair. I have also replaced out of spec resistors, including both 22R.

At idle, the right channel is stable with ~2mV across the .33 emitter resistors. The left channel starts out at ~3mV across these resistors and stays that way for a while, but then it gradually starts to rise the longer it's left turned on at idle. The left channel outputs start to get warm while the right channel stays cool. The voltage drop across both left channel emitter resistors remains roughly equal across both as the voltage increases. Last night, it got up to 17mV before I shut it off.

My problem, as a tube amp tech, working with this direct-coupled circuit is that I can understand how it works, but I need to know what component is mostly likely to be causing this bias drift, assuming that it's not normal.

I am immensely grateful for any help that points me in the right direction. I'm chalking up the hours I've sunk into this as my education on this circuit.

 

[dewickt]

Time for a can of freeze spray and a chilling of individual parts to find the most unstable one.

 

[c_dk]

What devices did you use for the drivers? Sentry monitor?

Since I had a few left I have not yet had to figure out what the replacement is for the Class A amp.

I assume you have swapped the L and R boards to confirm that it follows the board.

 

[CVElectronic]

Good question about the board. I'd forgotten that one channel got warmer despite a board swap before this channel melted down, which would eliminate the PCB components. That would tend to point me in the direction of D403, the diode on the heatsink. It appears to have some kind of plastic ring around it, though someone else suggested it should be covered in thermal paste. I'm also under the impression that it's some sort of NLA part.

I used 2N4920/2N4923 for the drivers, which is what someone said McIntosh ships for these, and I used MPSA06/MPSA56 for the sentry transistors, which is what McIntosh shipped me for the other channel.

 

[c_dk]

So you are saying the overheating is not traveling with the board?

Did you replace all 4 outputs, that is both channels, with MJ15003Gs?

 

[c_dk]

so I do not have to look them up later, D409,411,410 and 412 are Mac #070-031 a 1n4820.

 

[CVElectronic]

D403/D404 are 070-046, which I believe is the hard-to-source part. One is mounted on each heat-sink on a terminal strip with a band of some type around the middle in contact with the heat-sink.

There are also two 070-046 on each PCB.

I did replaced all 4 outputs in both channels with MJ15003G from ON Semiconductor. Two I got from McIntosh for the first channel, and I got the second two from Mouser. The two from Mouser measured pretty tightly matched.

 

[CVElectronic]

Another finding. Taking a fresh look, I turned it on, and found bias high from the outset on the left channel. Looking closely with a flashlight, I didn't like the way the wires to the output transistors looked against the edge of the chassis cutout, so I pushed them away a bit with a chopstick. Voltage across the emitter resistors dropped to 2.3mV and holding steady. Time to inspect those wires very carefully.

Owner told me that this MA6100 was damaged by rough handling in shipping when he first bought it.

 

[CVElectronic]

I think I've got it. The wires look good, but I think when I moved them, I moved the PCB slot contacts they're hooked to just slightly. The contacts were a bit uneven, so I got out my tube socket dental tools and retensioned them for a tighter fit, along with a thorough cleaning.

Bias is holding steady in both channels. DC offset is 9mV in one channel, 13.2mV in the other with no load. Close enough?

Many thanks for pointing out to me that it must be something NOT on the PC boards.

 

[c_dk]

070-046 is listed as a zener, commercial number MZ2361, 1.34 volts with a 2% tolerance at 10mA. They are selected stabistors......

I think Mac in a later MC2205 design replaced them with multiple diodes in series.......