Conbsidering the attached schematic it is a simple push-pull arrangement of which I have built and been using one.
Now I made another copy and there are a few problems. One of the power transistors, the Q4 MJE15029 in the picture, gets hotter than the others and eventually runs away.
All six power transistors are mounted on two heatsinks, three and three and they all maintain same temperature, except that odd one. Before placing I checked the HFEs of all transistors to be similar.
Therefore I can not see what could be different between Q4 and Q9 or Q13, or on the other side the Q5, Q8 and Q12, they all maintain same temperature, no problems.
So I did some debugging and placed the oscilloscope, not grounded, running off of battery, onto both sides of R7, to see how the transistor opens and closes. Of course I did the same for all other five transistors.
All five transistors open and close in the same way without cross conductance. If you remember sometime ago we discussed cross conductance, which is when the transistors stay open for far longer than they should allowing cross currents to flow into the other side, thus pulling a lot of unnecessary power, generating heat, and eventually running away.
So, all other five transistors open and close perfectly. But this odd one, Q4, it has cross conductance from the word go, when it is still at 40C, all other transistors heat up evenly and reach around 50C but this one goes on to over 80 at which point the cross conductance is so severe it simply shorts.
I cannot determine what factor could be causing this transistor to fail like this, being that they were all measured before soldering. Now my suspicions
are:
1) even though measured before soldering, it was done at around 200mA, and the currents here are about 2A RMS for each transistor. Maybe at those currents something fails in that transistor ?
2) the driver transistor Q3 may have a problem, even though it measures OK, with hfe=400.
3) the R6, nominally 33R, could be broken and say be at 100R, highly unlikely, and then we get cross conductance.
4) it seems the 0.22R resistors are not doing very much in terms of balancing, at least they cannot stop this odd transistor from running away.
I am lost for ideas. My next step is to replace both Q3 (driver) and Q4 (power) and see if anything improves.
If you have done similar debugging before please help.
PS: I forgot to say, whereas all the other curves on the scope are sines, the odd transistor's curve is more triangular, as if it cannot keep up with the frequency. I have noticed on general amplifiers that when you increase the frequency two things happen: (a) you get amplitude attenuation and (b) the sine becomes a triangle, a sign that the amplifier cannot keep up. Could it be that this transistor is defective and cannot keep up?
Now I made another copy and there are a few problems. One of the power transistors, the Q4 MJE15029 in the picture, gets hotter than the others and eventually runs away.
All six power transistors are mounted on two heatsinks, three and three and they all maintain same temperature, except that odd one. Before placing I checked the HFEs of all transistors to be similar.
Therefore I can not see what could be different between Q4 and Q9 or Q13, or on the other side the Q5, Q8 and Q12, they all maintain same temperature, no problems.
So I did some debugging and placed the oscilloscope, not grounded, running off of battery, onto both sides of R7, to see how the transistor opens and closes. Of course I did the same for all other five transistors.
All five transistors open and close in the same way without cross conductance. If you remember sometime ago we discussed cross conductance, which is when the transistors stay open for far longer than they should allowing cross currents to flow into the other side, thus pulling a lot of unnecessary power, generating heat, and eventually running away.
So, all other five transistors open and close perfectly. But this odd one, Q4, it has cross conductance from the word go, when it is still at 40C, all other transistors heat up evenly and reach around 50C but this one goes on to over 80 at which point the cross conductance is so severe it simply shorts.
I cannot determine what factor could be causing this transistor to fail like this, being that they were all measured before soldering. Now my suspicions
are:
1) even though measured before soldering, it was done at around 200mA, and the currents here are about 2A RMS for each transistor. Maybe at those currents something fails in that transistor ?
2) the driver transistor Q3 may have a problem, even though it measures OK, with hfe=400.
3) the R6, nominally 33R, could be broken and say be at 100R, highly unlikely, and then we get cross conductance.
4) it seems the 0.22R resistors are not doing very much in terms of balancing, at least they cannot stop this odd transistor from running away.
I am lost for ideas. My next step is to replace both Q3 (driver) and Q4 (power) and see if anything improves.
If you have done similar debugging before please help.
PS: I forgot to say, whereas all the other curves on the scope are sines, the odd transistor's curve is more triangular, as if it cannot keep up with the frequency. I have noticed on general amplifiers that when you increase the frequency two things happen: (a) you get amplitude attenuation and (b) the sine becomes a triangle, a sign that the amplifier cannot keep up. Could it be that this transistor is defective and cannot keep up?
Attachments
Last edited:
I have found an error on the PCB, a whole track was missing, connecting C4 and Q3 also leaving the collector of Q3 and base of Q4 disconnected from the Vcc !
Sorry for having wasted anyone's time.
Sorry for having wasted anyone's time.
You said it was a copy of your first build, but obviously not. Any change, like a new pcb, must be carefully verified before trusting it.
It takes very little to upset an amplifier.
I once designed and built a valve amplifier.
One of the valves red plated all the time.
Turned out I had missed inserting a grid stopper resistor.
I once designed and built a valve amplifier.
One of the valves red plated all the time.
Turned out I had missed inserting a grid stopper resistor.
- Status
- Not open for further replies.