Well that didn't turn out well!!!
Substituted 38494s with 2N3055s........just went into massive oscillation - the bulb limiter was ramping up and down. Reinstated 38494s
Had a look for collateral damage:
Turns out TR105 has gone - checked TR106.. its fine and checked TR1L...its fine
Replaced TR105 and tried again - getting 0.5mV across emitter resistors but I cannot alter it with VR101. With an output voltage of 6.5v the trace flattens off at the bottom - the R channel is ok at this point. TR105 gets VERY hot very quick.
Checked the Miller capacitors ( before substituting the output transistors) - they measured out ok but changed them both just to see......no difference, so the originals are now back in place.
seem to be going backwards here!
Substituted 38494s with 2N3055s........just went into massive oscillation - the bulb limiter was ramping up and down. Reinstated 38494s
Had a look for collateral damage:
Turns out TR105 has gone - checked TR106.. its fine and checked TR1L...its fine
Replaced TR105 and tried again - getting 0.5mV across emitter resistors but I cannot alter it with VR101. With an output voltage of 6.5v the trace flattens off at the bottom - the R channel is ok at this point. TR105 gets VERY hot very quick.
Checked the Miller capacitors ( before substituting the output transistors) - they measured out ok but changed them both just to see......no difference, so the originals are now back in place.
seem to be going backwards here!
Are you talking about R120 and R122? - just want be sure before I measure!!
Oscillation or just the bias current altering wildly and pulling the rail down... bulb lights... current falls... rail comes back... and repeat.
I can't understand TR105 keep failing. The output device should be the one doing all the work up to failing (or something else failing).
This flattening off at 6.5 volts output. Is that with or without a load?
(I assume the midpoint volts is correct)
I can't understand TR105 keep failing. The output device should be the one doing all the work up to failing (or something else failing).
This flattening off at 6.5 volts output. Is that with or without a load?
(I assume the midpoint volts is correct)
TBH- didn't really look at the scope, was more bothered about causing any further damage!!
So this is with 8 Ohm dummy load and the supply and mid rail are spot on.
Definitely got PNP-NPN in the right places.
I'm going to explore Johns suggestion about measuring the driver currents separately - just need to clarify across which resistors.
The driver currents are really what I was mentioning back in post #24.
If you work forward from the output transistors you know you must have at least 500mv give or take for each output to conduct. I've quoted a low figure there because the quiescent current is so low and the output just on the verge of conducting.
So 500mv means each 68 ohm needs around 7 milliamps flowing. As long as you are happy with the resistors then the currents are easily calculated from the volt drops.
Virtually the same current also flows in each 10 ohm.
TR105 and TR106 in turn also need around 500mv across those 1k resistors. So current in these is around 0.5 milliamps. Same current is flowing the 100 ohms.
Any chance there is a problem around MR105 and MR105, the overcurrent protection circuitry? You can snip those out for test purposes (keep using the bulb though).
The lower 10 ohm (R126) or a problem with any of the transistors in that lower output triple would cause problems with the lower part of the sine levelling off.
If you work forward from the output transistors you know you must have at least 500mv give or take for each output to conduct. I've quoted a low figure there because the quiescent current is so low and the output just on the verge of conducting.
So 500mv means each 68 ohm needs around 7 milliamps flowing. As long as you are happy with the resistors then the currents are easily calculated from the volt drops.
Virtually the same current also flows in each 10 ohm.
TR105 and TR106 in turn also need around 500mv across those 1k resistors. So current in these is around 0.5 milliamps. Same current is flowing the 100 ohms.
Any chance there is a problem around MR105 and MR105, the overcurrent protection circuitry? You can snip those out for test purposes (keep using the bulb though).
The lower 10 ohm (R126) or a problem with any of the transistors in that lower output triple would cause problems with the lower part of the sine levelling off.
I'm wondering if one of your output transistors have expired. I have seen reports that replacing old 3055's with new (epi) can lead to oscillations - especially with triples the way Quad have them (NPN-PNP-NPN likely to be worse). I've had to use additional caps and adjustments to use modern 3055's. But also long leads need decoupling back to ground too, which would be another important point with faster (epi) transistors.
Perhaps your Tr105 was having to do all the work of the output transistor. That might explain a loss of output power (6 V clipping?) not got the scope probe on 10X? (but that would have been 60V, unlikely). Did you check the devices when you disconnected the original leads?
We need to be certain whether your oscillations are low frequency (PSU voltage related, for example) or high frequency.
Perhaps your Tr105 was having to do all the work of the output transistor. That might explain a loss of output power (6 V clipping?) not got the scope probe on 10X? (but that would have been 60V, unlikely). Did you check the devices when you disconnected the original leads?
We need to be certain whether your oscillations are low frequency (PSU voltage related, for example) or high frequency.
Did a few prelim measurements:
across R119 1.2V
across R121 153mV
across R127 650mV
There was a few mV difference between these and their opposites ( on the same board)
but more interestingly they were pretty similar to the other board.
Checked out TR103/104, TR105/106, TR1L/R and all measure out fine.
what I will say is that TR106 gets screaming hot.
Do you think that I am looking at cumulative problem here - as in all the semiconductors check out individually, but they start to go out of spec when put together under load?
The amplifier is passing audio - so nothing is dead dead short anywhere.
across R119 1.2V
across R121 153mV
across R127 650mV
There was a few mV difference between these and their opposites ( on the same board)
but more interestingly they were pretty similar to the other board.
Checked out TR103/104, TR105/106, TR1L/R and all measure out fine.
what I will say is that TR106 gets screaming hot.
Do you think that I am looking at cumulative problem here - as in all the semiconductors check out individually, but they start to go out of spec when put together under load?
The amplifier is passing audio - so nothing is dead dead short anywhere.
Last edited:
If Tr106 gets a bit steamy perhaps Tr108 (lower O/P tranny) is dead. Even with a base looking sensible the collector may have gone OC. Or possibly R127 (lower 68 ohms) is shorted or low - esp. if it shows signs of being damaged.
What is the voltage across R126?
I've not seen whole circuits fail because they don't work "all together". It's always been one or two transistors taht have failed in one or more ways that causes the problem.
I think someone above suggested shorting out one of the bias diodes to see if that cuts off the quiescent current. Or bypassing it with a resistor temporarily (or permanently if it fixes the problem).
If you have a two channel scope, and an oscillator (50Hz sinewave is about ideal) and a working amp and transformer, you could try making a transistor curve tracer to check your transistor operation with voltage and current. (Not recommended for measuring breakdown voltages, but you can at least check that the devices operate within spec.)
I have to say, I do not think the early version of the circuit was very good. Adjusting the current to set a bias voltage in a diode or transistor is actually a good idea, but not when the current is also the VAS driver, since the Miller capacitor needs current to drive it. Does your control pot (22k) actually reach 22k (for minimum current) or is it low?
I would recommend replacing the diode scheme with the updated transistor scheme and cut the pot out. You might be able to put the additional components on a small "daughter board" and glue or screw it to the main pcb.
What is the voltage across R126?
I've not seen whole circuits fail because they don't work "all together". It's always been one or two transistors taht have failed in one or more ways that causes the problem.
I think someone above suggested shorting out one of the bias diodes to see if that cuts off the quiescent current. Or bypassing it with a resistor temporarily (or permanently if it fixes the problem).
If you have a two channel scope, and an oscillator (50Hz sinewave is about ideal) and a working amp and transformer, you could try making a transistor curve tracer to check your transistor operation with voltage and current. (Not recommended for measuring breakdown voltages, but you can at least check that the devices operate within spec.)
I have to say, I do not think the early version of the circuit was very good. Adjusting the current to set a bias voltage in a diode or transistor is actually a good idea, but not when the current is also the VAS driver, since the Miller capacitor needs current to drive it. Does your control pot (22k) actually reach 22k (for minimum current) or is it low?
I would recommend replacing the diode scheme with the updated transistor scheme and cut the pot out. You might be able to put the additional components on a small "daughter board" and glue or screw it to the main pcb.
Last edited:
171mV is high. I take it that is the one on the left hand channel that is getting hot. It should be about 100mV.
Not 104, I think that is looking like the lower output transistor (Tr2L) is leaky or low gain or something.
But next time the amp is off, check the value of R126 on the left channel. Should be 10 ohms, of course.
Not 104, I think that is looking like the lower output transistor (Tr2L) is leaky or low gain or something.
But next time the amp is off, check the value of R126 on the left channel. Should be 10 ohms, of course.
No, he means the lower output transistor, is it Tr2?
And 2N3055 is not an adequate replacement for the output transistors. You need MJ15003. What you have used is a backward step to something pretty marginal in this circuit.
Last edited:
Completely off track - I'm curious. Does anyone know what devices the original "U17219/U17229" driver transistors were based on? Seemingly selected TO-18 can devices. Not very important, as many options for alternatives (ZTX304/504, MPSA06 A56, 2N5551/5401) but I've not worked out what the originals were. Perhaps the PNP was a BC477? The NPN a selected BC107? Even BC546/BC556 might work (most of the time).
And yes it should be Tr2L. Just carried on with the numbering instead of looking at the circuit!
And yes it should be Tr2L. Just carried on with the numbering instead of looking at the circuit!
Last edited:
Not sure about your comment that a 2N3055 is "not adequate". As I mentioned I've got a version running with epi 3055's at double the bandwidth of the original 303. What I would say is that the epi devices need some compensation tuning.
The current 2N3055 from ON Semi have second breakdown levels approaching the original RCA device (38494 etc). (60V 0.9A) That was not the case until recently. Also capacitor coupling tends to limit the power under short circuit conditions. That would be my main concern rather than the audio performance.
But an MJ15003 is a good choice and matches the original (60V ~2A).
The current 2N3055 from ON Semi have second breakdown levels approaching the original RCA device (38494 etc). (60V 0.9A) That was not the case until recently. Also capacitor coupling tends to limit the power under short circuit conditions. That would be my main concern rather than the audio performance.
But an MJ15003 is a good choice and matches the original (60V ~2A).
Seemingly OK but gain is low compared with TR1L. 6mA is ratehr low to test the devices. As you can access the device thorough the leads and it is on a heatsink, you could try testing at higher currents. Put a meter in series with the collector and try a few base resistors to see what it needs at 1A. You could use a 12V supply for a short while. Start with 10k and work down (slowly). With 10k I'd expect 100mA or so.
(Check out first that the device isn't leaky at 12V - or rely on your meter having short circuit protection? If you have a 10 ohm power resistor or similar you could use taht to prevent damage as a first test))
Is the 68 ohm still 68 ohms when Tr2 is disconnected (and you might check by lifting a lead to remove it from any other component). There does seem to be something amiss in this part of the circuit.
(Check out first that the device isn't leaky at 12V - or rely on your meter having short circuit protection? If you have a 10 ohm power resistor or similar you could use taht to prevent damage as a first test))
Is the 68 ohm still 68 ohms when Tr2 is disconnected (and you might check by lifting a lead to remove it from any other component). There does seem to be something amiss in this part of the circuit.
Last edited:
R119
It is in the expected range. Two Vbe drops of TR106 and the output transistor.
R121
That sounds reasonable.R127
Again, reasonable.TR106 should definitely NOT be hot and should only be barely even detectably warm.
Given the other voltages sound OK then this points to TR106 conducting heavily into the output transistor which in turn should then be conduting itself.
Do we have a voltage reading across R126 while it is in this exact condition... I see we do...
John has picked up from this now leads... that the lower NPN appears not to be conducting for some reason.
Have you got continuity from its collector to the 0.3 ohm and from there to the upper 0.3 ohm and upper output device?
