Bingo! Changing Q2 fixed the positive side.
Before changing Q2, the LEDs were brighter on that side and the output voltage was also higher.
Hopefully, if anyone else has the same symptoms here, they'll now know what to do.
Since Q2 and Q4 are easily damaged, is there a more rugged transistor(s) that could be substituted?
Something in say a TO-126 package? I know it wouldn't fit on Tea's boards, but I can easily change my boards.
Before changing Q2, the LEDs were brighter on that side and the output voltage was also higher.
Hopefully, if anyone else has the same symptoms here, they'll now know what to do.
Since Q2 and Q4 are easily damaged, is there a more rugged transistor(s) that could be substituted?
Something in say a TO-126 package? I know it wouldn't fit on Tea's boards, but I can easily change my boards.
OK, changed out the BC550 and BC560 and 4 diodes.
Unloaded voltages look better, but a couple of LED bars are not lit I'll order some more. Voltages are now -17.2, +16.74, +16.99, -17.45 (top to bottom related to pics)
Keith
Attachments
Last edited:
615498, you can easily check each LED segment with a 9V battery and a 1k ohm resistor in series with one of its terminals as Salas suggested to me in a earlier post.
I tried it and verified all my segments were working.
Changing those Kingbrights can be a real pain in the rear, especially if you don't need to.
I tried it and verified all my segments were working.
Changing those Kingbrights can be a real pain in the rear, especially if you don't need to.
Its in the mechanics of the Sziklai as a series pass element in this app that when the bigger transistor will see an output short, R3/R4 will be between Raw DC and zero potential, the beta booster small transistor will conduct much current through its emitter. Its not that the BC550/560 is weak for its normal operation, its the fault current will be very strong. Talking Watts of power. The small BJT really acts as a fast fuse and saves the MJE transistor.
And you want very high beta in the booster BJT for good performance. The bigger the transistor the lower the beta in general. You could fuse at the Q2,Q4 emitters with 50mA fast types if manipulation or connection mishaps are possible. When there is a umbilical cord and you may plug when energy is still stored maybe. But not confident they will act as fast as needed. Since your application runs the supplies cold use higher consumption bleeder resistors so it can a) perform better b) discharge faster at power off. 330 Ohm 2W.
Broken diodes say that was a big prolonged short. Your Mosfets got assembled without Silpads if I remember well. Check various Vbe and other points like in Ammel's measurement list, because the few dark segmnets can be just signs of other problems yet unfixed than broken themselves. To verify the segments alone try light them up individually with a voltage source as he did. Don't forget to measure for shorts between the the +/0/- DCG3 inputs before attempting to use the supplies again after fixed. We don't know if your DCG3 passed fully undamaged yet.
P.S. Why we see no R1, R2 on the top DCSTB in your picture?
I'm not using any bleeder resistors. I have a provision for them on the board but didn't think they were that important.
Perhaps I thought wrong(wouldn't be the first time).
Do you think not using bleeders could be why Q2 blew?
If so, I'll order some 330 ohm 2W as you suggested.
Could be. Its not a shunt reg, it won't self consume to ground. Can hold strong charge for >30mins if relying on internal losses only for the rail voltages to go down. A kickback can occur. You also want to run it >50mA for waking up the MJEs anyway. They need some idle to better lower the curve of output impedance.
I don't think 3 op-amps are going to draw 50mA.
So this supply won't work if the source doesn't draw at least 50mA?
Works fine with very light load, its just it works even better with more mA draw so those bleeders will be beneficial to its performance also in your case.
Thanks for the reply and for all the troubleshooting help.
I would hate to abandon this PS for op-amp circuits.
To my ears, it sounds better than the LM317/337 supply I was using.
There is a possibility you will also hear the technical performance difference after the mA consumption jump with those bleeders addition. If you will perceive anything subjectively let us know.
The DCG3's Class A case is already consuming more than enough, but your op-amps application (is it some preamp or an active crossover etc?) is loading much much lower the supply by comparison.
The DCG3's Class A case is already consuming more than enough, but your op-amps application (is it some preamp or an active crossover etc?) is loading much much lower the supply by comparison.
Yes, it is a simple preamp with only 3 op-amps.
I'll add some 330 ohm 2W Vishay resistors to my next order.
Most opamps can output 20mApk without getting near the short circuit output that typically is around 40mA continuous.
Now add on the quiescent current to that 20mApk and you have potentially 70mApk for three opamps sending transients concurrently.
Just received the AD823 and indeed they are much better sounding than the previous ones.