I've had only one experience where too slow of an output transistor caused a problem. It was on an SAE amp using 2N3773 transistors. The higher the test frequency the more current the amp would draw from the AC line. The amplifier output voltage would stay the same but draw more current. Popped in some newer MJ15024s and the problem was gone. Cross conduction was the problem, of the 100s if not 1000s of amps I've worked on this was the only time I've seen this.
You didn't mention what test frequency you were using but try starting at a lower frequency and monitor the AC current draw as you increase the freq.
I use the MJ21193/4 pairs all of time and never seem to have a problem. Since we're on the subject of "slow" ON semi TO-3 transistors, why do you think ON doesn't make faster TO-3s? They make fast TO-3P and TO-247s.
Craig
You didn't mention what test frequency you were using but try starting at a lower frequency and monitor the AC current draw as you increase the freq.
I use the MJ21193/4 pairs all of time and never seem to have a problem. Since we're on the subject of "slow" ON semi TO-3 transistors, why do you think ON doesn't make faster TO-3s? They make fast TO-3P and TO-247s.
Craig
I sourced my parts from Mouser and Digikey.
If I were to look at it "under a microscope" I would say that the negative peaks clip maybe .5 volts sooner than the positive rail. It's pretty darned close to symmetry.
My test frequency for these power output tests is 1KHz. I just performed a new test at 100Hz and 1KHz and both clipped at 36.5 vrms. I think frequency/capacitance loading is not an issue. There's something else going on.. driver has separate regulated supply, so now I'm going to check the sag on the high current supply. I hope it's not a bad transformer!
If I were to look at it "under a microscope" I would say that the negative peaks clip maybe .5 volts sooner than the positive rail. It's pretty darned close to symmetry.
My test frequency for these power output tests is 1KHz. I just performed a new test at 100Hz and 1KHz and both clipped at 36.5 vrms. I think frequency/capacitance loading is not an issue. There's something else going on.. driver has separate regulated supply, so now I'm going to check the sag on the high current supply. I hope it's not a bad transformer!
0.5 volts is nothing...
Can't see it being any problem with the tranny. Just scope all the rails in the power amp (DC coupling) and see how they hold up. I think you have to do that to get a handle on this. If the rails are good then it has to be the devices you have used.
Can't see it being any problem with the tranny. Just scope all the rails in the power amp (DC coupling) and see how they hold up. I think you have to do that to get a handle on this. If the rails are good then it has to be the devices you have used.
Repaired channel:
Supply rails to finals:
No signal = +77.5V / -77.5V
Just below clipping: +69.38V / -69.34V
Output at clipping: 34 vrms
Original channel:
Supply rails to finals:
No signal = +77.5V / -77.5V
Just below clipping: +68V / -67.76V
Output at clipping: 43.5 vrms
Supply rails to finals:
No signal = +77.5V / -77.5V
Just below clipping: +69.38V / -69.34V
Output at clipping: 34 vrms
Original channel:
Supply rails to finals:
No signal = +77.5V / -77.5V
Just below clipping: +68V / -67.76V
Output at clipping: 43.5 vrms
Just an idea and I can't just quickly trace the path on the circuit. It may be nothing at all...
Are these supplies OK under load. I can't see where they come. The block diagram shows them being the same supply as the output devices but are they ?
If they are good as measured on the driver and pre driver then I think you have to look finding alternative devices.
Although simulation doesn't always reflect real life I put the supply voltage and 34 vrms output into a standard class ab design using the MJL21194 devices and it easily works, even with just a single MJE340 driver and output. Base current of the outputs was around 150ma peak.
Are these supplies OK under load. I can't see where they come. The block diagram shows them being the same supply as the output devices but are they ?
If they are good as measured on the driver and pre driver then I think you have to look finding alternative devices.
Although simulation doesn't always reflect real life I put the supply voltage and 34 vrms output into a standard class ab design using the MJL21194 devices and it easily works, even with just a single MJE340 driver and output. Base current of the outputs was around 150ma peak.
Attachments
Have to leave it for today but this gives an idea of currents involved. Output is 37vrms into 8 ohms. You can see the base current of the output and driver (driver is around 2ma)
So your output triple should walk this. Something seems amiss in the devices, or their connection or connectivity.
So your output triple should walk this. Something seems amiss in the devices, or their connection or connectivity.
Attachments
Yes, they are the same rails, and I measured them and posted the no load and full load results above.
34 vrms output is no problem. The problem is reaching 42.5 vrms.
34 vrms output is no problem. The problem is reaching 42.5 vrms.
Does this amplifier have VI limiters? If so, make sure there isn't a burnt out resistor in the circuit. That could alter where it starts limiting, but at light loads or less than full power look perfectly fine.
EF3's almost always have enough current gain to drive any load off any VAS. Even with low gain parts. That's the whole reason to go to a triple.
Could also be a B-E short in the driver or predriver, or one of the collectors open. That would tend to be asymmetrical, affecting only that one rail. And with most loads it wouldn't be that far off because an EF2 would "almost" or "usually" have enough gain.
EF3's almost always have enough current gain to drive any load off any VAS. Even with low gain parts. That's the whole reason to go to a triple.
Could also be a B-E short in the driver or predriver, or one of the collectors open. That would tend to be asymmetrical, affecting only that one rail. And with most loads it wouldn't be that far off because an EF2 would "almost" or "usually" have enough gain.
Doesn't look like a VI limiter there in that driver board. I may have missed something, but it looks pretty straightforward downstream of the VBe multiplier.
All of the medium power semis were replaced. If any were open/shorted, the amplifier would not come out of protect mode.
I've checked every resistor, just in case one was off by a significant amount, but all are within 2% of marked values--impressive for their age.
I don't think there is VI limiting action because:
1. the driver supply rails don't sag under full signal/load
2. the high current rails sage even less than the original channel under load.
All of the medium power semis were replaced. If any were open/shorted, the amplifier would not come out of protect mode.
I've checked every resistor, just in case one was off by a significant amount, but all are within 2% of marked values--impressive for their age.
I don't think there is VI limiting action because:
1. the driver supply rails don't sag under full signal/load
2. the high current rails sage even less than the original channel under load.
You also mentioned that it biased funny - that you had to play with it to get the distortion to be normal. With a triple you usually don't get gross distortion even with zero bias in the outputs (though it will be higher than it's supposed to be). It really sounds to me like not all of the output stage is conducting. Even with brand new transistors you can get a solder bridge or an unhooked connection. Do the drivers or predrivers get really hot when testing with the dummy load? Is the Vbe stack running around 3.45V (or is it off by six tenths)?
OK 🙂 my mistake, but even so it gives a good idea of the currents involved. Base current of the output went to around 180ma at 45 vrms.
Imo you have to just try something else for those drivers and pre-drivers. I would do the pre-driver first. Something like a 2N5551/5401 might prove the point. Yes, they are small signal devices but the current and dissipation should be low.
Also... when you test at full power do either the pre-driver or the drivers get warm/hot ?
I still wonder whether there could be some problem with the devices. If either set (driver or pre-driver) 'wasn't present' and appeared just as a base-emitter junction in place of the transistor then I think it would give this symptom.
Could the transistors be fake or have an incorrect pin-out ?
That's the thing, after running the amp at full power for a few seconds, the drivers are still relatively cold. The output block gets warm though.
All parts came from Digikey and Mouser. If there are fake, then we've got a serious problem.
Next step: check the rails at the collector junctions of the output.
Good point about the relays. I'll check voltage at the Zobel network and compare with the output terminals.
I don't want to risk playing with low power devices in the pre drivers because a failure there could result in an expensive loss of parts.
The fact that it makes 45 vrms into a 250 ohm load makes me wonder if it's not a drive problem at all, but simply the outputs not turning on all the way under big signal conditions.
All parts came from Digikey and Mouser. If there are fake, then we've got a serious problem.
Next step: check the rails at the collector junctions of the output.
Good point about the relays. I'll check voltage at the Zobel network and compare with the output terminals.
I don't want to risk playing with low power devices in the pre drivers because a failure there could result in an expensive loss of parts.
The fact that it makes 45 vrms into a 250 ohm load makes me wonder if it's not a drive problem at all, but simply the outputs not turning on all the way under big signal conditions.
Results from testing rail voltages at the collector flanges of output transistors.. only sags to 68 volts at clipping.
Only measured a half a volt difference between Zobel network and output terminals.
Then I had a thought: I hadn't checked the base resistors on the output stage--only the emitter resistors. I was expected to find one or more open base resistors. Alas, all of them are right at their marked values. So were the 100 ohm resistors connected elsewhere on this stage. So it's back to square one.
Only measured a half a volt difference between Zobel network and output terminals.
Then I had a thought: I hadn't checked the base resistors on the output stage--only the emitter resistors. I was expected to find one or more open base resistors. Alas, all of them are right at their marked values. So were the 100 ohm resistors connected elsewhere on this stage. So it's back to square one.
What about those pre-drivers ? Did they get warm.
Yes, I can understand that. My thinking is that the current in the pre-drivers is going to be very small indeed.
Even at the levels it can achieve now, I would expect the drivers to heat up. Make sure its not the pre-drivers doing all the work as it stands now.
Are you getting full voltage to the bases of TR12 and TR13 or not? That would answer the question of whether it's an output stage problem or not once and for all.
That's a good idea. I will check that when I am feeling better. Had a rough day with severe back and joint pain.
My thought is: if the load on the output stage is affecting the base impedance drastically, then this could be the problem. I noted the base resistors are 4.7 ohms! That must mean the current drive is enormous. Perhaps these drivers can't deliver. I'm pondering playing with the values of R39 and R40 to increase the current sinking ability.
My thought is: if the load on the output stage is affecting the base impedance drastically, then this could be the problem. I noted the base resistors are 4.7 ohms! That must mean the current drive is enormous. Perhaps these drivers can't deliver. I'm pondering playing with the values of R39 and R40 to increase the current sinking ability.
Punch line first... I think its the MJE340/350 you are using as pre-drivers...
-----------------------------------------------------------------------------------------------
R39 and R40 are to aid fast turn off of the outputs. Altering the value of these isn't the answer.
The 4.7 ohms are a safeguard, they aid stability, they aid current sharing and they isolate each junction from the next. And they actually function as 'fuses' in the event of a catastrophic failure.
You need to deliver around 8 amps peak into the 8 ohm load for your aimed for output level. If it used a single output pair of gain of 40 then you would need to deliver around 200ma peak into the base. Multiple output pairs reduce that in proportion, so three pairs would see around 70ma in each base.
The MJ21194/3 are guaranteed to have a forward gain of 25 @ 8 amps collector current. Assuming one pair, that's 320ma drive needed. You have three pairs so lets say 100ma per pair.
wg_ski's suggestion of looking at the bases of TR12 and 13 is a good idea. Compare the voltage here with that at the other end of the two 390 ohm base feeds. There should be very little difference because the current in the 390 ohms should be minimal.
I've thrown this together to give you an idea of the currents involved. It shows one half of the output stage. Then we have the output current in the 8 ohm load. Next is the base current for the outputs (and these are lower gain than yours). Now we have the driver collector current and this has to supply the outputs and the 47 ohm. That's pretty substantial (and they should be getting hot). Next to the last and we have the collector current of the pre-driver. Again this has to supply the 560 ohm. Finally we have the base current of the pre-driver.
The 2N5550's are high gain devices. So when I said it was a walk in the park for the triple arrangement here... well I might have been over simplifying things. The currents in the pre-driver are higher than I thought. The MJE340/350 are very variable in 'quality' in my experience. Some of the best I have seen are from manufacturer CDIL.
-----------------------------------------------------------------------------------------------
R39 and R40 are to aid fast turn off of the outputs. Altering the value of these isn't the answer.
The 4.7 ohms are a safeguard, they aid stability, they aid current sharing and they isolate each junction from the next. And they actually function as 'fuses' in the event of a catastrophic failure.
You need to deliver around 8 amps peak into the 8 ohm load for your aimed for output level. If it used a single output pair of gain of 40 then you would need to deliver around 200ma peak into the base. Multiple output pairs reduce that in proportion, so three pairs would see around 70ma in each base.
The MJ21194/3 are guaranteed to have a forward gain of 25 @ 8 amps collector current. Assuming one pair, that's 320ma drive needed. You have three pairs so lets say 100ma per pair.
wg_ski's suggestion of looking at the bases of TR12 and 13 is a good idea. Compare the voltage here with that at the other end of the two 390 ohm base feeds. There should be very little difference because the current in the 390 ohms should be minimal.
I've thrown this together to give you an idea of the currents involved. It shows one half of the output stage. Then we have the output current in the 8 ohm load. Next is the base current for the outputs (and these are lower gain than yours). Now we have the driver collector current and this has to supply the outputs and the 47 ohm. That's pretty substantial (and they should be getting hot). Next to the last and we have the collector current of the pre-driver. Again this has to supply the 560 ohm. Finally we have the base current of the pre-driver.
The 2N5550's are high gain devices. So when I said it was a walk in the park for the triple arrangement here... well I might have been over simplifying things. The currents in the pre-driver are higher than I thought. The MJE340/350 are very variable in 'quality' in my experience. Some of the best I have seen are from manufacturer CDIL.
Attachments
Briefly back to the thermostat replacement... the original is a black plastic rectangular part with a tab off the long dimension for mounting.
Parts I've been able to find are round with two mounting tabs and very expensive!
This is the cheapest I could find:
Selco - CA-120 - Open at 90 F Close at 120 F Close on Rise Auto-Reset 1/2 in Disc Thermostat - Allied Electronics
I spent an hour fixing and recapping the meter board this afternoon. After recap, it would peg the left meter. Must have been a micro short between solder pads. X-acto knife between pads fixed it. It needs a DPST 24V reed relay. The meters work if I bypass the relay. The field coil is getting 27 VDC when the receiver comes out of protect mode. Oddly, both poles of the relay are permanently open.
Fixed the intermittent cutting out of the right and left channels earlier. Bad joints on the PCB that the pre in pwr amp out jacks are located at.
Progress is slow and steady. Will get back to the power amp measurements tomorrow.
I'm ordering batches of MJE350 transistors from three different manufacturers, in the hope that I find higher hfe devices I can pair with the few 'hot' MJE340s I have in my stock.
Parts I've been able to find are round with two mounting tabs and very expensive!
This is the cheapest I could find:
Selco - CA-120 - Open at 90 F Close at 120 F Close on Rise Auto-Reset 1/2 in Disc Thermostat - Allied Electronics
I spent an hour fixing and recapping the meter board this afternoon. After recap, it would peg the left meter. Must have been a micro short between solder pads. X-acto knife between pads fixed it. It needs a DPST 24V reed relay. The meters work if I bypass the relay. The field coil is getting 27 VDC when the receiver comes out of protect mode. Oddly, both poles of the relay are permanently open.
Fixed the intermittent cutting out of the right and left channels earlier. Bad joints on the PCB that the pre in pwr amp out jacks are located at.
Progress is slow and steady. Will get back to the power amp measurements tomorrow.
I'm ordering batches of MJE350 transistors from three different manufacturers, in the hope that I find higher hfe devices I can pair with the few 'hot' MJE340s I have in my stock.
- Status
- Not open for further replies.