What happened to the diff amp in PSU ?

[lordvader88]

I made a PSU based on a Farnell L30 , with a 2n3055 on the output, and broke it for the 2nd time, while I was running it at 15V and a little over 1A into a MOSFET load tester I made. I think I had tried to adjust the voltage up on the PSU. I can't say if the current spiked after I noticed the output voltage going wacko and I shut it right off.

Farnell - L series Bench Power Supplies
the very last schematic on page 11
http://www.introni.it/pdf/Farnell%20-%20L%20series%20Bench%20Power%20Supplies

When it worked, the diff-amp worked and the feedback loop worked and it would balance. And so (VT3)/Q3 Vbe was held around 0.55V and barely moved a few milli-volts, never even goes over 0.6V. I thought maybe thats too low, but IDK. I could set the voltage from 0-25V or so with 30.5K total voltage control potentiometers P1 and P2

Now the output is stuck at max, which is the primary winding at 43V unloaded (it's a 36Vdc@1.5A winding).

When only the control circuit is turned on, everything seems about right. (VT2)/Q2 is off all the way (I think it used to be on? I don't remember). When the whole circuit is powered, (VT1)/Q1 BE is reverse baised, so off, and Q2 is fully turns on, so Q3 Vbe is 0.7V now, and saturated so Q3 Vce is under 0.1V , and so pulling the 10k collector resistor all the way low, and so the PSU output is stuck wide open.

I pulled the diff.amp BC558's, and I matched their beta and Vbe, and they both still read the same and match. The 2 1n4148 diodes seem ok in circuit, I matched them too.



When it's un-powered, everything else seems to checkout in circuit with a DMM, the voltage control resistors are all still working this time.

I should check the zener that fried last time, but it's at least working, last time it went short.

What would break the diff-amp ??

[lordvader88]

The output 2n3055 collector-emitter went short. IDK if it was the poor design of my load tester, previous damage, or maybe this is not a real 15A 2n3055

So it's working again, the Vbe of Q3 is back around 0.500mV

[xavier60]

Although I didn't make sense of all of it, some of the voltages were not consistent with having a shorted output transistor.
VT1 and VT2 are part of a constant current source for Z4 so that it will produce a stable reference voltage. The only thing that can affect this area is low rail voltage.

EDIT: I was looking at the wrong schematic.

[floobydust]

It's the last schematic page, all discretes?

The pass-transistor output stage is usually full on, due to R10 10k turning on VT5...VT6, VT7. VT3 (or VT4) shuts off the pass-transistor pre-driver VT5. So VT3 on, output should be zero. But if any VT7,VT6,VT5 shorted it would damage parts upstream like VT3 or worse.
It's a weird circuit, using 16 as common, measure voltage on midpoint Z2 (10V) Z4 (5.1V) voltage I would check too. R8 might have cooked or the V adj potentiometer if you tried dialing down output voltage to zero when it's stuck at full output voltage.
There are no current-limiting resistors on the triple EF so the driver VT6 can dump a lot of base current into the VT7.

[xavier60]

Am I seeing clearly? VT6 and VT7 appear to be two 2N3055's as a Darlington pair. That would be nuts!

[floobydust]

I checked and in 1976 2N3055 cost almost the same as a 2N3054 (TO-66):
Practical Wireless 1976
2N3055 55p, today $4.94
2N3054 50p, today $4.49
2N3053 25p, today $2.24

The British pound got trashed in 1975/76, so these figures might be off (used 1£=$1.78 USD).