Working on a Sony T-N86B power amplifier.
It arrived from the customer, non working. I recapped the power supply and the entire amp and it came up and worked great. Met or surpassed all factory specs on measurements on the bench. Then it went to a listening test.
I set it up with a pair of Theil CS3.5's and let it play music at background levels for several hours. After five hours, it shut down.
Initial troubleshooting indicated a failure in the switching supply, five transistors within had shorted.
After repairing and testing the supply, I decided to check and see if there might be an external reason for the supply failure. There is. Both output transistors on the left channel are shorted, C-E.
This is a curious failure as the amp was operating in class B at the time and program levels were low.
At any rate, these output devices seem to be unobtanium except in Hong Kong. There are NTE equivalents. Does anyone have a better source within the US for original 2SA1028 and 2SC2398 parts?
It arrived from the customer, non working. I recapped the power supply and the entire amp and it came up and worked great. Met or surpassed all factory specs on measurements on the bench. Then it went to a listening test.
I set it up with a pair of Theil CS3.5's and let it play music at background levels for several hours. After five hours, it shut down.
Initial troubleshooting indicated a failure in the switching supply, five transistors within had shorted.
After repairing and testing the supply, I decided to check and see if there might be an external reason for the supply failure. There is. Both output transistors on the left channel are shorted, C-E.
This is a curious failure as the amp was operating in class B at the time and program levels were low.
At any rate, these output devices seem to be unobtanium except in Hong Kong. There are NTE equivalents. Does anyone have a better source within the US for original 2SA1028 and 2SC2398 parts?
I did see an offer for the 2SC part from Florida, nothing for the pnp though. Good luck finding high-fT power transistors like that in TO-3 in this day and age.
I am guessing old age and secondary breakdown got you on this one. When you measured its performance, I am assuming you did check bias current and all?
Class B operation is not necessarily less risky, as Class A operation also reduces rail voltages considerably.
That's one weird biasing circuit, btw. Took me a while to wrap my head around this one - it all starts to make sense if you look at Class A first. It certainly is not failsafe - if one of the diode strings goes intermittent or the 10k pot for Class B bias develops bad contact, say bye-bye to your output stage.
I am guessing old age and secondary breakdown got you on this one. When you measured its performance, I am assuming you did check bias current and all?
Class B operation is not necessarily less risky, as Class A operation also reduces rail voltages considerably.
That's one weird biasing circuit, btw. Took me a while to wrap my head around this one - it all starts to make sense if you look at Class A first. It certainly is not failsafe - if one of the diode strings goes intermittent or the 10k pot for Class B bias develops bad contact, say bye-bye to your output stage.
The bias was fine at the time of bench testing and both channels were running nice and cool.
Inspection now suggests that the bias circuit suffered an abrubt and sudden failure, some five hours into music testing.
I've rebuilt the switching power supply, and tested it with a current limiting resistor at it's input and it produces appropriate output rails. But I'm reluctant to test the amplifier will full power supply rails, so I'm using a bench supply, with the current limit set to under an amp and supply 30V rails instead of 41.
I've installed MJ21193/96 output transistors and both channels pass signal as far as the speaker relays (measureable at bias current test points). In class B, the left channel is idling at 216ma instead of 20ma. The right idles at about 7ma. The right channel bias can be adjusted. The left cannot. In Class A, there is SOME variability in current, but not much, when adjusting the class A bias pot. The class B pot tests good, so the problem is elsewhere. D106 tests good, so Q123 and associated circuits must be the problem area. I wonder what the modern replacements for D104 and D105 would be? A string of 1N4148s perhaps?
Agreed, it is a funky bias circuit. This is one of those "in for a penny, in for a pound" situations, where what started out as a dead amplifier, come to life after a recap, suffered a failure after five hours of testing and is now into both major PSU rework (done) and now amplifier work. So many hours sunk into this, but it's too late to turn back.
Inspection now suggests that the bias circuit suffered an abrubt and sudden failure, some five hours into music testing.
I've rebuilt the switching power supply, and tested it with a current limiting resistor at it's input and it produces appropriate output rails. But I'm reluctant to test the amplifier will full power supply rails, so I'm using a bench supply, with the current limit set to under an amp and supply 30V rails instead of 41.
I've installed MJ21193/96 output transistors and both channels pass signal as far as the speaker relays (measureable at bias current test points). In class B, the left channel is idling at 216ma instead of 20ma. The right idles at about 7ma. The right channel bias can be adjusted. The left cannot. In Class A, there is SOME variability in current, but not much, when adjusting the class A bias pot. The class B pot tests good, so the problem is elsewhere. D106 tests good, so Q123 and associated circuits must be the problem area. I wonder what the modern replacements for D104 and D105 would be? A string of 1N4148s perhaps?
Agreed, it is a funky bias circuit. This is one of those "in for a penny, in for a pound" situations, where what started out as a dead amplifier, come to life after a recap, suffered a failure after five hours of testing and is now into both major PSU rework (done) and now amplifier work. So many hours sunk into this, but it's too late to turn back.
Well, had it failed shortly after being back at the (then presumably rather angry) customer you would probably have had it back on the bench soon enough anyway, so that saved it one trip back and forth... *shrug*
Bias current is in the right ballpark, so stupid question, have you checked the 10k pot and all the connections around it? Are both switching transistors and associated relay contacts operating as they should? In Class B operation the Class A (470R) pot is supposed to be shorted out and the 10k pot connected. In any case I would guess that the bias circuit developed this problem a while before the failure, and then heating, aging and secondary breakdown took care of the rest.
It's definitely not oscillating either?
As far as potential VD series replacements go, I read somewhere that UF4002 are supposed to be a better match than 1N4148. No idea why they used two different types here (I can only assume tempco?), I can't even find a datasheet on this MV104V because there's also a tuning diodes called MV104. In any case the diodes used here appear to be generally reliable, as opposed to some.
Using a double diode to drop down from 8 to 6 B-E drops seems a clever idea at first (if beta isn't too high, it'll limit current), but it's pretty daft from an AC perspective given the low current the double diode would be running at. That's not what symmetrical low impedance drive looks like - I imagine even-order harmonics may be higher than they would have to be. The whole affair might appreciate a bit of an electrolytic across either the double diode or the bases of the first EF stage.
BTW, MJ21195/96 are quite the opposite of the originals (which I'd say are closer to something in between 2SC5298/2SA1941 and 2SC5200/2SA1943 or the like) - a good bit bigger, presumably much better SOA, but much slower as well. I sort of doubt you're going to get away with a measly 20 mA on those, that seems barely adequate even for the originals (is the unit that thermally challenged?).
Bias current is in the right ballpark, so stupid question, have you checked the 10k pot and all the connections around it? Are both switching transistors and associated relay contacts operating as they should? In Class B operation the Class A (470R) pot is supposed to be shorted out and the 10k pot connected. In any case I would guess that the bias circuit developed this problem a while before the failure, and then heating, aging and secondary breakdown took care of the rest.
It's definitely not oscillating either?
As far as potential VD series replacements go, I read somewhere that UF4002 are supposed to be a better match than 1N4148. No idea why they used two different types here (I can only assume tempco?), I can't even find a datasheet on this MV104V because there's also a tuning diodes called MV104. In any case the diodes used here appear to be generally reliable, as opposed to some.
Using a double diode to drop down from 8 to 6 B-E drops seems a clever idea at first (if beta isn't too high, it'll limit current), but it's pretty daft from an AC perspective given the low current the double diode would be running at. That's not what symmetrical low impedance drive looks like - I imagine even-order harmonics may be higher than they would have to be. The whole affair might appreciate a bit of an electrolytic across either the double diode or the bases of the first EF stage.
BTW, MJ21195/96 are quite the opposite of the originals (which I'd say are closer to something in between 2SC5298/2SA1941 and 2SC5200/2SA1943 or the like) - a good bit bigger, presumably much better SOA, but much slower as well. I sort of doubt you're going to get away with a measly 20 mA on those, that seems barely adequate even for the originals (is the unit that thermally challenged?).
I wonder how is possible an "amplifier expert " is seeking this type of advice here or even think an NTE replacement ....
As it turns out, the inability to adjust bias was due to a failing driver transistor. That transistor went full failure during troubleshooting with a low current supply source. I subbed an MJE340 to see if the channel would function again, and it does, and the bias current can be properly adjusted now.
No offsets at the output. Just need to find out why the speaker relay doesn't pull in. But definite progress. I'm testing with MJ21119x output transistors, which seem to be passing an undistorted signal in class B operation. Not sure what will happen under load, but since the parts are unobtainium, this is a foray into uncharted territory.
No offsets at the output. Just need to find out why the speaker relay doesn't pull in. But definite progress. I'm testing with MJ21119x output transistors, which seem to be passing an undistorted signal in class B operation. Not sure what will happen under load, but since the parts are unobtainium, this is a foray into uncharted territory.
The MJ21193/4 transistors are good parts and will perform well. However MJE340/350 are lousy drivers - if you must use a TO126 part, use MJE243/253 or KSC2690/KSA1220.
The PLPS supply in these Sony amp's is a pain in the ***, as it is fussy about the switching transistors used on the transformer primary. Make sure Q609-Q611 are gain matched.
The PLPS supply in these Sony amp's is a pain in the ***, as it is fussy about the switching transistors used on the transformer primary. Make sure Q609-Q611 are gain matched.
I'm amazed. The amp performs VERY well. Distortion is measuring at 0.0019% and the amplifier can make 100 watts per channel in Class B before clipping now. Before it broke (but just after recapping, it clipped at about 85 watts per channel.)
The 340/350 might not be my favorite choice, but voltage and current ratings are overkill for this application and the amplifier's performance seems to bear out that it's a synergistic match.
I'm burning it in this week. I've switched it on and off on the bench many times, seeing if I can trigger a fault, but no glitch so far. Now it's connected to a pair of 4 ohm Thiel CS 3.5's and playing music all day into the late evening.
The 340/350 might not be my favorite choice, but voltage and current ratings are overkill for this application and the amplifier's performance seems to bear out that it's a synergistic match.
I'm burning it in this week. I've switched it on and off on the bench many times, seeing if I can trigger a fault, but no glitch so far. Now it's connected to a pair of 4 ohm Thiel CS 3.5's and playing music all day into the late evening.
- Status
- Not open for further replies.