I have been in the process of resurrecting my old SAE 3CM amplifier. It's been more problematic than I expected. It hadn't been fired up for 30+ years so I bought a re-cap kit and swapped out all the electrolytics before I punched the "on" switch. The re-cap job went OK, although I first had to un-do some "improvements" the previous owner had done. Like remodeling old houses, you never know what you're going to find once the studs have been exposed.
Initial testing after the re-cap job looked OK -- the driver board's idle-current measurements were good and the output voltage offsets were reasonable and could be adjusted to a few mV.
But longer-term operation revealed a problem in one of the channels. The output would limit on the negative portion of a test waveform (1KHz sine). And I also observed that the bias voltage went down, from the normal ~2.1V to about 1.85V. I suspected a problem in the Darlington VASes so started my troubleshooting by measuring the emitter resistors on the Darlington emitters (R24 & R25, see below). The 3CM uses symmetrical NPN/PNP LTPs that drive two Darlington pairs, one for the + side and one for the - side. So there were two emitter resistors to measure: and BOTH measured quite high. The schematic indicates that the resistors should be 62 ohms. The one on the + side measured over TWICE that, and the one on the - side measured about 90 ohms! That is far beyond the 5% tolerance the SM specifies for those guys.....
For a sanity check I also measured the 10 ohm resistors that are used to isolate the driver board's supply busses from the output board supplies. They were spot on.
I also measured the emitter resistors in the "good" channel and they also measured high, although not as bad as the failing side. The SM indicates that the R's are carbon film types. Visual examination didn't reveal any indication that the resistors were subjected to excessive current, and I didn't see any obvious mechanical damage.
In normal operation those 62 ohm resistors should be dissipating far less than 1/2 watt. I think they are that size so they will survive if/when a short circuit condition occurs & Q11 or Q12 turn on.
I'm puzzled why the resistor on the plus side is the worst, but the output waveform clips on the negative side. That channel initially works -- more or less -- but then starts clipping after it's warmed up. So there may be a thermal intermittent situation. My next test will be to monitor the resistances while I heat them with a hot-air gun.
The main point of this post, however, is to ask folks if they have seen a similar problem with old carbon film resistors spontaneously failing like this. If not, there's a good chance something else is going on that is over-stressing those resistors. The other thing is: if those went bad simply due to their age, it doesn't bode well for the other CF resistors in there. And there are a lot of them. Replacing all of them is not my idea of a fun time.....
Mark
PS driver board schematic below:
Initial testing after the re-cap job looked OK -- the driver board's idle-current measurements were good and the output voltage offsets were reasonable and could be adjusted to a few mV.
But longer-term operation revealed a problem in one of the channels. The output would limit on the negative portion of a test waveform (1KHz sine). And I also observed that the bias voltage went down, from the normal ~2.1V to about 1.85V. I suspected a problem in the Darlington VASes so started my troubleshooting by measuring the emitter resistors on the Darlington emitters (R24 & R25, see below). The 3CM uses symmetrical NPN/PNP LTPs that drive two Darlington pairs, one for the + side and one for the - side. So there were two emitter resistors to measure: and BOTH measured quite high. The schematic indicates that the resistors should be 62 ohms. The one on the + side measured over TWICE that, and the one on the - side measured about 90 ohms! That is far beyond the 5% tolerance the SM specifies for those guys.....
For a sanity check I also measured the 10 ohm resistors that are used to isolate the driver board's supply busses from the output board supplies. They were spot on.
I also measured the emitter resistors in the "good" channel and they also measured high, although not as bad as the failing side. The SM indicates that the R's are carbon film types. Visual examination didn't reveal any indication that the resistors were subjected to excessive current, and I didn't see any obvious mechanical damage.
In normal operation those 62 ohm resistors should be dissipating far less than 1/2 watt. I think they are that size so they will survive if/when a short circuit condition occurs & Q11 or Q12 turn on.
I'm puzzled why the resistor on the plus side is the worst, but the output waveform clips on the negative side. That channel initially works -- more or less -- but then starts clipping after it's warmed up. So there may be a thermal intermittent situation. My next test will be to monitor the resistances while I heat them with a hot-air gun.
The main point of this post, however, is to ask folks if they have seen a similar problem with old carbon film resistors spontaneously failing like this. If not, there's a good chance something else is going on that is over-stressing those resistors. The other thing is: if those went bad simply due to their age, it doesn't bode well for the other CF resistors in there. And there are a lot of them. Replacing all of them is not my idea of a fun time.....
Mark
PS driver board schematic below:
My hot-air gun testing showed that the 62 ohm resistor on the positive side has a very high negative temperature coefficient compared to the one on the negative side. In the absence of anything else that's a very good reason to replace it -- but it still doesn't explain the amp's failure signature. The negative tempco would suggest that the amp starts out in a bad way and gets better as everything heats up -- but that's not what happens.
So, while the 62 ohm resistors need to be replaced they don't appear to be the source of the problem. This is how failure analysis goes -- you form a hypothesis then figure out some way to test it and evaluate the result. From there, devise some other test(s) to perform. And so on.
I'm starting to think that I need to come up with an LTSpice version of this driver board. The transistor models don't need to be all that accurate, I just want to see how something like the driver circuit behaves when I change things. The "fun" part is that the driver is inside a feedback loop so a fault will propagate throughout the circuit.
So, while the 62 ohm resistors need to be replaced they don't appear to be the source of the problem. This is how failure analysis goes -- you form a hypothesis then figure out some way to test it and evaluate the result. From there, devise some other test(s) to perform. And so on.
I'm starting to think that I need to come up with an LTSpice version of this driver board. The transistor models don't need to be all that accurate, I just want to see how something like the driver circuit behaves when I change things. The "fun" part is that the driver is inside a feedback loop so a fault will propagate throughout the circuit.
Amp still has the original aluminum cased TO-66 and TO-3 transistors? Just repaired an Ampzilla I, circuit is closely related, and the TO-66s were the problem. This is not the first time I've experienced the old aluminum case transistors causing problems, both TO-66 and TO-3.
Craig
Craig
Yes they look like originals, they have the SAE p/n's on them. Along with the owner manual and service manual, the amp came with an invoice from a repair shop, dated 1983. The invoice indicates that the 4558 opamp was replaced. That actually is useful information because I'd been wondering what that IC was. I knew it was a dual something-or-other but in that application a dual comparator OR dual opamp could do the job.
Nothing to do with the current problem though.
The repair job cost $59.40. Those were the days.....
Thank you for the information! Based on your comment regarding its similarity to the Ampzilla 1 (and it sure does look similar) I found a parts list indicating that the PNP devices in question may be similar to the MJ3584 and the NPN's are similar to 2N3584. The 2N3584 still can be found here and there but so far no luck on the MJ3584. I found one cross reference indicating that a 2N5345 might work and did find some sources on ebay. Its Ft is higher, 60MHz vs 10 for the MJ3584.
Nothing to do with the current problem though.
The repair job cost $59.40. Those were the days.....
Thank you for the information! Based on your comment regarding its similarity to the Ampzilla 1 (and it sure does look similar) I found a parts list indicating that the PNP devices in question may be similar to the MJ3584 and the NPN's are similar to 2N3584. The 2N3584 still can be found here and there but so far no luck on the MJ3584. I found one cross reference indicating that a 2N5345 might work and did find some sources on ebay. Its Ft is higher, 60MHz vs 10 for the MJ3584.
More debug info. One important item, due to the design of my USB ADC. My initial observation of the problem occurred when I was using REW to check the amp's distortion specifications, and the output waveform appeared to be clipping/limiting on the negative side of the waveform. It turns out that my sound card INVERTS the input signal!!!! Given that, the ~2X too-high value for the resistor on the plus side now is consistent with the waveform behavior and the low voltage across the "thermal block" -- the current is way too low. I'm glad I have a separate signal generator and scope, otherwise I would have been barking up the wrong tree.
I found a schematic for my ADC and sure enough its input amplifier design is inverting. Jeez....
Note to self: when using REW's scope app, invert the waveforms.
I checked the LTP outputs (actually the collectors of the common-base transistors) and they looked reasonable. So the tail currents are alright.
Just for another sanity check, I also measured the voltage across the 51V zener diode. It is OK. It could go majorly-bad and still not affect the offset voltage adjustment much, because the voltage across the adjustment pot is clamped to +/- .6V by a couple of diodes. However, the LTP tail currents definitely would be way off in that situation.
It's looking more and more like the problem is that 62 ohm resistor. I'm thinking I may go ahead and order a set of driver transistors at the same time I order the resistors. If nothing else, I will have them when/if the old aluminum-cased TO-66 transistors go TU.
I have seen comments that replacement resistors also should be carbon film because they contribute to the amp's sound character. So I guess I will go with them, even though I'm unhappy about their apparently poor reliability. Hopefully they're manufactured better now than they were back when.
I found a schematic for my ADC and sure enough its input amplifier design is inverting. Jeez....
Note to self: when using REW's scope app, invert the waveforms.
I checked the LTP outputs (actually the collectors of the common-base transistors) and they looked reasonable. So the tail currents are alright.
Just for another sanity check, I also measured the voltage across the 51V zener diode. It is OK. It could go majorly-bad and still not affect the offset voltage adjustment much, because the voltage across the adjustment pot is clamped to +/- .6V by a couple of diodes. However, the LTP tail currents definitely would be way off in that situation.
It's looking more and more like the problem is that 62 ohm resistor. I'm thinking I may go ahead and order a set of driver transistors at the same time I order the resistors. If nothing else, I will have them when/if the old aluminum-cased TO-66 transistors go TU.
I have seen comments that replacement resistors also should be carbon film because they contribute to the amp's sound character. So I guess I will go with them, even though I'm unhappy about their apparently poor reliability. Hopefully they're manufactured better now than they were back when.
The output signal of my Sound Technology 1700B going to the scope is also inverted, really screws up your troubleshooting if you don't know this. I just turned around the dual banana going into the 1700B, the input of the 1700B is balanced so it's easy to do.
MJ3584 is 2N6421
MJ3584 is 2N6421
I pulled the driver board and removed the 62 ohm resistors. My "silver" soldapult came through again -- for several pins that's all that was needed, no solder wick required. Re-measuring the resistors confirmed that all of them were significantly out of spec, ranging from 75 ohms to 112 ohms. The highest value resistor was in the obviously-bad channel.
65 ohms would be the worst-case tolerance excursion for a 5% 62 ohm resistor so these guys are pretty far out of spec. My in-situ heat gun test also showed that the 112 ohm resistor was REALLY unstable w/respect to temperature.
A couple of the resistor bodies looked a little discolored but nothing really jumped out. No cracks, no pins were loose. Weird.
I'm hoping that replacing the resistors is all that's needed to restore this guy. Replacement resistors are on the way.
The fact that all 4 of the resistors have failed in lesser or greater degree has me wondering if the amplifier has some sort of turn-on transient that momentarily activates the short-circuit protection circuitry. In normal circumstances that can only happen when there's enough current flowing in the sense resistor, which is a .125 ohm 10-watt resistor. But I'm not seeing any scenario where that can occur in the absence of an external short.
65 ohms would be the worst-case tolerance excursion for a 5% 62 ohm resistor so these guys are pretty far out of spec. My in-situ heat gun test also showed that the 112 ohm resistor was REALLY unstable w/respect to temperature.
A couple of the resistor bodies looked a little discolored but nothing really jumped out. No cracks, no pins were loose. Weird.
I'm hoping that replacing the resistors is all that's needed to restore this guy. Replacement resistors are on the way.
The fact that all 4 of the resistors have failed in lesser or greater degree has me wondering if the amplifier has some sort of turn-on transient that momentarily activates the short-circuit protection circuitry. In normal circumstances that can only happen when there's enough current flowing in the sense resistor, which is a .125 ohm 10-watt resistor. But I'm not seeing any scenario where that can occur in the absence of an external short.
Maybe the quiescent current was set too high, R92. OR maybe spent some time with the VI limiters in action.
My amplifier has no "R92", the quiescent currents were set at the factory by adding the right-value resistors. I'm leaning toward the VI limiter circuit as the culprit. My main concern is that the limiter is somehow being triggered during normal operation, perhaps at turn-on. It isn't obvious to me how that could be, but.....call me a worry-wart 🙂.
To answer that concern, I'm planning on temporarily connecting an LED with a series current-limiting resistor across one of those 62 ohm resistors. In normal no-load operation the voltage drop across the resistor will be approximately 1.2V. If the resistor experiences a condition where it temporarily dissipates 1/2 watt or more, the voltage across it will be much higher -- enough to turn the LED on. A red LED's turn-on voltage is around 2V so it won't affect the circuit if it's operating correctly....at least, not when it's in a no-load condition.
I have some flea clips so removing the monitoring circuit would easy, just pull the clips off the resistor leads.
On my amp, one channel has no added resistor bridging R21/25. The other channel has a 1.5M resistor tacked on.
I won't even THINK of messing with that until I have some good 62 ohm resistors in there.
I won't even THINK of messing with that until I have some good 62 ohm resistors in there.
Replacing the 62 ohm resistors didn't fix the problem. But while measuring some circuit nodes I discovered that Q1 is very sensitive to being mechanically disturbed. Q1 is the common-base transistor connected to the positive-side LTP. Even lightly touching it would cause the amp's output to clip, outputting zero voltage above ground.
Information I have suggests that SAE801 is an MPSU-06 device. Radwell.com's website indicates they have them for not a lot of money, but I'm wondering if it would be best to also replace Q2, the corresponding device on the negative-side LTP. Radwell has the MPSU56 but they are MUCH more expensive. It would be cheaper to replace both with more-modern equivalents like MJE243/253. Unfortunately the pinout is different but I can manage that. Shifting the BCE-style device over by one position will match 2/3 of the pins.
Information I have suggests that SAE801 is an MPSU-06 device. Radwell.com's website indicates they have them for not a lot of money, but I'm wondering if it would be best to also replace Q2, the corresponding device on the negative-side LTP. Radwell has the MPSU56 but they are MUCH more expensive. It would be cheaper to replace both with more-modern equivalents like MJE243/253. Unfortunately the pinout is different but I can manage that. Shifting the BCE-style device over by one position will match 2/3 of the pins.
Had an I61 Mullard fail in V1 - damaged in shipping. Installed, start up, a few seconds later... pop. Replaced it and all is almost well. Since then, now and again, right after warm up, an occasional "spit" or quick crackle, once or twice, then disappears. Is that a symptom of a plate resistor, which could have been damaged with the failed tube? Thanks
Try using the MPSW06/56 pair, if you can't find any I have a good supply of them. The MPSUxx series have very delicate legs and break if you stare at them too long. Let me know.
Craig
Craig
I'm sorry but I don't have much experience w/regard to tube amps. The last time I used one was sometime around 1979 or thereabouts.
My home-made thermal IR camera says that the MPSU's are running at about 35C, and they have a heat-dissipation tab. It could be higher due to the actual emissivity of the tab. That's on the side that works. The MPSW series are in TO-39 packages so they will be running a lot hotter than that, a reliability concern for me. I figure that the worst-case PD would be about 150mW.
It's clear that I need to pull the driver board out of the amp no matter what, so that transistor and its solder points on the PCB will be VERY carefully examined. I also have a scheme to power up the driver portion by itself, although it won't be running at the full +/- 75V. My bench supply's max output is +/-40V but that should be enough to generate the needed +/-25V from the 51V zener, and everything else looks like it shouldn't care as long as the LTP tail currents are right.
Have you used the MPSW pairs in this particular portion of the 3CM? If so, I'm wondering if there is something else going on....particularly since I have already found some resistors that were seriously out of spec! Man, just like remodelling an old house, you never know what you're going to find.....
Mark
I use them everywhere SAE 801/902 are used. MPSWxx series are giant TO-92 style and were the replacement for the MPSUxx series within limits. Also there are slip-on heatsinks if needed. There is very little current going thru Q1/2, less than 2ma. Approx. 2.5v across the R13,15,17 combo (1.5K total) for 1.7ma.
Have you checked the Zener diode? If that's open the voltages would be a bit high in the front end.
Craig
Have you checked the Zener diode? If that's open the voltages would be a bit high in the front end.
Craig
It measured OK, close to 51V. It's sounding more and more like something else is going on.
Thank you for your quick responses. More later!