My scheme on powering-up just the driver board is working. I needed to place a Vbe multiplier circuit across the bases of the driver transistors Q13/14, and to close the NFB loop I added two series-connected 100 ohm resistors between their emitters. Their center common point is the pickoff for the NFB. I used a separate +/-15V supply to bias up the collectors of Q13/14 because I was concerned that the power dissipation would be too high if I connected them to the +/-40V supplies (they normally run at +/-25V). I'm glad I have enough bench supplies to do this sort of thing.
And.....the darned thing just sits on the bench, happily running away. I left it on all day and it never misbehaved. I wiggled the transistors I suspected might be intermittent, no dice. A heat gun: no dice. Aargh.
But I had a thought on something else that might be going on. The collectors of Q13/14 are connected to the emitters of Q16/19 respectively. In addition to their function as "slave" transistors in the output circuit, they provide +/-25V (quiescent) to Q13 and 14. If there was an intermittent open in the slave circuitry, the amp might exhibit a similar failure signature if it caused the collector of Q13 to collapse.. Since I had the amp pretty much torn down anyway I started eyeballing the output board and lo and behold I found what looks to be a cold solder joint. It is the emitter pin of Q15, the top slave transistor. If that opens up Q16 goes into saturation. Q13 is still trying to source (about) 20mA into R38 so its collector current * the bias string resistance (R47 || (R49+R51)) pulls its collector down pretty close to ground. Which is what I see when the amp is misbehaving.
This is a theory that is relatively easy to check, the Intermittent-Failure gods willing. The biggest hassle will be removing the stuff I did so for debugging the driver board.
I have a little USB pencil-sized microscope that I'm going to use to get a close-up photo of the problematic solder joint. It's perfect for this kind of thing because it has a built-in adjustable LED light source. I'll post the photo later.
And.....the darned thing just sits on the bench, happily running away. I left it on all day and it never misbehaved. I wiggled the transistors I suspected might be intermittent, no dice. A heat gun: no dice. Aargh.
But I had a thought on something else that might be going on. The collectors of Q13/14 are connected to the emitters of Q16/19 respectively. In addition to their function as "slave" transistors in the output circuit, they provide +/-25V (quiescent) to Q13 and 14. If there was an intermittent open in the slave circuitry, the amp might exhibit a similar failure signature if it caused the collector of Q13 to collapse.. Since I had the amp pretty much torn down anyway I started eyeballing the output board and lo and behold I found what looks to be a cold solder joint. It is the emitter pin of Q15, the top slave transistor. If that opens up Q16 goes into saturation. Q13 is still trying to source (about) 20mA into R38 so its collector current * the bias string resistance (R47 || (R49+R51)) pulls its collector down pretty close to ground. Which is what I see when the amp is misbehaving.
This is a theory that is relatively easy to check, the Intermittent-Failure gods willing. The biggest hassle will be removing the stuff I did so for debugging the driver board.
I have a little USB pencil-sized microscope that I'm going to use to get a close-up photo of the problematic solder joint. It's perfect for this kind of thing because it has a built-in adjustable LED light source. I'll post the photo later.
Lots of stuff going on on that mother board, parts on both sides of the board and fairly crowded with large parts plus it's a mother to work on. You'll end up having to pull all of the output transistors anyway to fix the solder joint. Make sure the transistors fit tight in the sockets, the last III I rebuilt and the last Ampzilla II I repaired both needed new sockets. I have a stash of original RN sockets but Keystones will work also.
I'm curious why would I need to pull the transistors. The only scenario I can see is the possibility of soldering the transistor pin to the socket pin. Is that the concern?
BTW I see that Mouser indicates the MJE11022/11021 devices are EOL status. Ebay has quite a few but one always has to worry about counterfeits.
BTW I see that Mouser indicates the MJE11022/11021 devices are EOL status. Ebay has quite a few but one always has to worry about counterfeits.
Correction, the EOL devices in question are MJ11022/11021. They are not MJE's. But I'm sure you know that 🙂
I measured the resistance between the PCB pad and socket pin. 177 ohms. So definitely a problem. Is it THE problem? I dunno yet. I feel like I'm still peeling the layers of this onion. Proof will be an extended run with some decent power output to get things nicely warmed up. I have a 100W 8 ohm resistor bolted to a heat sink that will come in handy for that. It also has an attenuator so I can feed the signal into my USB sound card for doing some REW-based spectrum measurements. Its baseline noise level is around -110dB so should be OK for checking this guy. More modern amps, not so much.....
Oh, and thank you for the info on alternative replacement devices!
Oh, and thank you for the info on alternative replacement devices!
Apologies for the somewhat-blurry image. But it was taken, hand-held. Magnification may be in the 25X range:
It actually doesn't look like a cold solder joint to me. It looks like some kind of long-term creep problem. That's a fracture line, no question about it.
This particular pin is the closest to one of the board mount points, and it is quite possible that the board was subjected to bending stress due to some mount-point oddities -- due to the absence of a spacer under one of the mount points underneath the board, it could have been bent. This problem predates my ownership of the amp. The stress at the socket pin could have been magnified due to lever action, and I will fix that by inserting a space underneath the boardr. I have a machine shop. I have micrometers. I can size that sucker to a few thousandths if I want to...
It actually doesn't look like a cold solder joint to me. It looks like some kind of long-term creep problem. That's a fracture line, no question about it.
This particular pin is the closest to one of the board mount points, and it is quite possible that the board was subjected to bending stress due to some mount-point oddities -- due to the absence of a spacer under one of the mount points underneath the board, it could have been bent. This problem predates my ownership of the amp. The stress at the socket pin could have been magnified due to lever action, and I will fix that by inserting a space underneath the boardr. I have a machine shop. I have micrometers. I can size that sucker to a few thousandths if I want to...
The bias IC usually sticks up a bit higher than the surface of the xsistor sockets so a little board bending stress there.
Yeah that's what my eyeballs were saying, too. The spacer I'm adding will be sized to avoid adding any more bending moment to the PCB. Then it will be on to the reassembly/test phase (yet again....).
So some (possible) closure on my amp's problems. I put everything back together and, yet again, once the amp had been running awhile it misbehaved in the same way, clipping off the positive side of the waveform. But some poking around with some DVM probes on the driver transistors caused the amp to recover. And, despite my efforts to cause it to fail (by power-cycling it a lot), it hasn't malfunctioned. But the key phrase is: not yet. I fully expect it to, but , based on what I've seen I strongly suspect those TO-66 transistors. As @llwhtt mentioned, the aluminum cased TO-66 devices can develop problems, probably (in my estimation) due to aluminum oxide formation. It might be possible to address the problem for quite awhile by cleaning the devices' collector contact points.
Time will tell.....
Time will tell.....
If you mean cold solder joints or similar, I did perform a fairly complete visual inspection of the driver board. The output board is more difficult because there are a LOT of parts mounted on the copper side of the board. I didn't see any indications that any of the other output transistor socket pins were suspect.
The output board is pretty simple compared to the driver board. For the most part it just uses much beefier components.
A true cracked trace could be a PITA to track down, no doubt about that....
The fact that the board gave you a clue while physically probing with a spuger or such has me on this path. While working telecommunications cables in outside plant, we had what was referred to as a high open. It wouldn't always present itself but depended on humidity and temperature. In rare cases you might accidentally find one by the physical bending of the cable itself, not that this was the intent of the bend, which was basically to get into the cable pack itself. That is why I say here that luck might have given you a clue.
The tricky part about the problem is that the cracked solder joint I found was on the output board -- but the most-recent intermittent problem just about has to be related to the driver board. It's hard to believe that the amp has more than one problem but maybe it's an exception to the rule that "one fault rules them all" (so to speak).
One bothersome feature of this amp's design is the use of some pretty funky connectors that join the two boards. The words "connector" and "intermittent" are closely related in my mind. I used contact cleaner on them but, still, they remain on my list of suspects.
One bothersome feature of this amp's design is the use of some pretty funky connectors that join the two boards. The words "connector" and "intermittent" are closely related in my mind. I used contact cleaner on them but, still, they remain on my list of suspects.
Maybe it is like the soles of my shoes. First the right one cracks a tread and I am able to glue it back in place and keep wearing the shoes. Today, the left one cracked a tread same as the right. Now they are not connected, but made of the same material. I am thinking this way about the circuit boards in your amp.