Hi Guys

Faced with a broken power supply...no volts out:-( i set about trying to repair.
I sketched the circuit to the best of my ability and i think this is how it goes.[see attached]
After staring at the circuit for a few days and doing some comparisons with various textbooks on the subject it became more intriqing
Looking at the circuit it does not seem to be a forward converter as there is no diode from collector of the BUT11A to the HV rail.
C3 SHOULD BE 22 U 400V oops sorry i mis labelled,but it just has the rectified mains across it . The interesting parts are r2/c4 in parallel with the primary and hanging on the collector of Q1. AAAH is this then a parallel resonant circuit that is self starting and self oscillating?
I think it works as follows ...on switch on R3 TURNS Q1 hard on which charges C6/R6 and starts to remove the base drive from Q1 via Q2 collector.
The base of Q1 has just received a pulse from the secondary of T1 tertiary winding via C7...about 2V to enforce the turn on of Q1 and save some power.
Q1 continues to be turned off by Q2.and finally C4 R2 charge in the opposite direction as the collector of Q1 swings positive.
This is the inner loop. The outer loop is from the o/p and is fed back by the opto isolator to control the amount of drive to Q1.
If O/P volts go up U1 sinks more current setting the dc point of Q2 a little higher and throttling back Q1 slightly.
Anyway it turns out that there was no oscillation and hence no output as C7 had reduced in value to 50u, so i removed and fitted a new one and bingo we are in biz. C7 is fitted very close to the heatsink of Q1.....too close and it dries up.
If anyone has another idea of how this circuit works please let me know. What is bugging me though is the frequency it is working at and how you would set about designing this circuit. Putting the scope on the collector of Q1 revealed pulses running at about 67Khz which does not hold water with the jolly old "one over two pi root LC equals the resonant frequency". If i put in the numbers[i measured the inductance of all windings] and used the parallel resonance of T1 primary and R2C4 i end up with a resonance of 6.25KHZ a factor of 10 in error.
Can anyone offer an explanation of what i am doing wrong in my calcs and what topology this circuit is?
Can anyone point me in the direction of some design guides for this type of circuit as i think it should be a good circuit to use for 5v as well.
Can you leave the inductances as is and just change the reference on the output. maybe not.
Any information gratefully received.

 

Did you intend to attach a schematic?

 

[
here is the schematic
sorry just getting used to this forum and it does not do what i ask it)

 

here is some images of the power supply out of his container. Note the 100u cap close to the heatsink on the left which i replaced.Did not have a radial to hand still it works and proves the point. Careful on the top oscilloscope pic it relates to the collector of Q1 not as shown as the secondary. I think the secondary has peaks of about 20v.pk pk
careful also for the diode shown as 1n4001 its actually a fast diode.suspiciously like a 1n4148
Bear in mind i also hand traced this circuit so might have some errors but i'm sure the collector of Q1 and the resistor and cap are correct. cap says 331K...330n?

what does r4 do is it only preserving Q2 collector emitter max volts? or clever feedback temperature related?

 

It is a feed forward converter. The dots on the winding tell you that (assuming they are correct). It is self-oscillating with feedback through your red line and hence no oscillation with fauty C7. R5, R6, C6 provide current limiting via Q2. The TL431 and optocoupler via Q2 provide the voltage regulation.

 

Thank you for your kind reply and explanation on the control.
There is no diode for c4 r2 in the collector of Q1 and every forward converter seems to have one where r2 c4 is a snubber.
what governs the frequency of oscillation? c4/r2?. as i said the frequency measured bears no relation to the calcs.
Do you know of any design article that covers this circuit. The billings one has a diode in the collector.

 

Are the dots on the transformer winding at the correct places?

 

I think you have drawn it wrong, the Tl431 has 3 pins, and i don't see the point of an opto-couper if the Negative mains is joined to the Negative output.!!

 

I think you have drawn it wrong, the Tl431 has 3 pins, and i don't see the point of an opto-couper if the Negative mains is joined to the Negative output.!!

The negative output is on ground 2 not ground which the input mains refers to. I did label them to differentiate.
TL431 is indeed a three pin device and is not drawn strictly correct as in the circuit, i was pushed for time and tried to simulate the effect of the reference voltage and where it was applied with a zener, so apologies for that. Nevertheless there is no refernce to the mains as they are physically not connected on the board, the 9v floats as it is galvanically separate.

 

Are the dots on the transformer winding at the correct places?

Not sure as i'm using a stock symbol. Can it be any other way given the orientation of the diodes and the positive supply o/p?

 

Just to clarify the opto diode transmitter is on the floating 9v side and the rx is referenced to the negative on the fully rectified incoming mains.

 

If the dot on the primary was at the other end then it becomes a flyback converter. That's more compatible with the snubber circuit.

 

If the dot on the primary was at the other end then it becomes a flyback converter. That's more compatible with the snubber circuit.

apart from a feed forward converter has a tertiary winding a flyback does not?

 

I think it is likely that the primary dot is at the wrong end of the winding and it is a flyback converter.

 

oops i take that back it looks like it can but the second inductor is in the lead of the collector of Q1.

 

i think you are right, it must be a flyback which as you say would make the snubber correct, however does it not need a diode in series with the snubber with the cathode pointing away from the collector of Q1?

 

if the snubber is just a snubber i wonder what governs the oscillation frequency of the circuit.......maybe my old friend C7.

 

no frequency does not match scope frequency for either of the inductances...the plot thickens.

 

It works something like: Q1 starts to conduct and the feedback drives Q1 hard on. The current through the primary rises exponentially. This will continue until the current reaches the limit set by R5 or the transformer saturates. At this point the feedback voltage falls and Q2 turns off. The feedback now drives Q1 fully off. The cycle repeats.
So the timing for this depends on many components, and probably on the load on the supply. The L/R time constant of the primary is a big part of how long Q1 stays on.

 

Hi Guys

Faced with a broken power supply...no volts out:-( i set about trying to repair.
I sketched the circuit to the best of my ability and i think this is how it goes.[see attached]
After staring at the circuit for a few days and doing some comparisons with various textbooks on the subject it became more intriqing
Looking at the circuit it does not seem to be a forward converter as there is no diode from collector of the BUT11A to the HV rail.
C3 SHOULD BE 22 U 400V oops sorry i mis labelled,but it just has the rectified mains across it . The interesting parts are r2/c4 in parallel with the primary and hanging on the collector of Q1. AAAH is this then a parallel resonant circuit that is self starting and self oscillating?
I think it works as follows ...on switch on R3 TURNS Q1 hard on which charges C6/R6 and starts to remove the base drive from Q1 via Q2 collector.
The base of Q1 has just received a pulse from the secondary of T1 tertiary winding via C7...about 2V to enforce the turn on of Q1 and save some power.
Q1 continues to be turned off by Q2.and finally C4 R2 charge in the opposite direction as the collector of Q1 swings positive.
This is the inner loop. The outer loop is from the o/p and is fed back by the opto isolator to control the amount of drive to Q1.
If O/P volts go up U1 sinks more current setting the dc point of Q2 a little higher and throttling back Q1 slightly.
Anyway it turns out that there was no oscillation and hence no output as C7 had reduced in value to 50u, so i removed and fitted a new one and bingo we are in biz. C7 is fitted very close to the heatsink of Q1.....too close and it dries up.
If anyone has another idea of how this circuit works please let me know. What is bugging me though is the frequency it is working at and how you would set about designing this circuit. Putting the scope on the collector of Q1 revealed pulses running at about 67Khz which does not hold water with the jolly old "one over two pi root LC equals the resonant frequency". If i put in the numbers[i measured the inductance of all windings] and used the parallel resonance of T1 primary and R2C4 i end up with a resonance of 6.25KHZ a factor of 10 in error.
Can anyone offer an explanation of what i am doing wrong in my calcs and what topology this circuit is?
Can anyone point me in the direction of some design guides for this type of circuit as i think it should be a good circuit to use for 5v as well.
Can you leave the inductances as is and just change the reference on the output. maybe not.
Any information gratefully received.

Its a pretty typical flyback converter, more or less a blocking oscillator.

Start with the power transistor - they often fail short all 3 ways and take out the emitter resistor. You have to test everything and search the solder side for dry joints or it'll probably just go bang again.

The reservoir electrolytic for the opto coupler supply can go bad - no regulation = saturated core and bang!

Q2 does both current and voltage control, if that circuit doesn't work; it'll go bang - if Q2 is shorted; it wont work.

Simple failure to start is probably R3 high or open. Shorted secondary side diodes can also hang it up.

Live testing is a very good chance of making it go bang - lift one end of each component and test it cold.