Flyback power supply repair help

[spencer2004]

Hello. I am currently trying to repair the power supply of a projector that I just bought (Sahara / Delta AV-3100), it worked great for about half an hour and it suddenly went dead, I am sure that it is the power supply that is at fault as it should be constantly on even when the projector is in standby. I would really like to be able to fix it as I don't think that there are any replacement power supplies for sale for such an old projector.

My knowledge of electronics has huge gaps in it so forgive me for using incorrect terminology, etc. Here is what I know about this power supply:

The power supply uses a flyback transformer which is controlled by a TOP234 IC. The output rails are -5v, 5v, 6.5v, 13v and 17v. Feedback is based on the output of the 5v rail and sent back to the TOP234 via an opto isolator, all other rails are controlled by the turn ratio on the transformer only except for the -5v rail which is regulated with a linear regulator.

Once it went dead there was no output on the power rails at all. I have spent the last week or 2 figuring out how the power supply works and testing all of the components and I came to the conclusion that the IC that controls the switching of the transformer had died to I replaced it and it now outputs something but the voltages are almost exactly half of what they should be (5v is 2.77v, 6.5v is 3.35v, 13v is 7.02v, 17v is 8.72v, -5v is 3.72v). Putting a 5 ohm load on the 5V rail causes it to drop to 2.22v.

Since I don't own an ESR meter I just replaced all of the capacitors and it made no difference. No components show signs of overheating and nothing gets hot when it is powered on. No safety components have been triggered (i.e. fuse and varistor).

The circuit is almost exactly like this except the voltage rails are different, feedback is based only on 1 rail (the 5v one) and the -5v rail comes from a linear regulator.

I am really out of ideas now so if someone could give me any guidance on what I could try next I would be very grateful. I know this amount of time and effort is not worth it but I'm mainly doing it for the learning experience.

 

[KrisBlueNZ]

Well, you've made a really good start! You've done your due Google diligence and saved us a lot of time.

So the thing is working, but all of the outputs are running at about half voltage?

One easy check is R1 in the above diagram. According to the data sheet, a 2M resistor from the M ("multi-function") pin to the positive input supply provides "undervoltage, overvoltage and duty cycle reduction". So if R1 is faulty, it could cause the device to limit its duty cycle, which in turn limits the output voltages.

There are two basic reasons why the output voltages are low. Either the controller has reached its maximum duty cycle and cannot put any more energy into the transformer (in which case an overload on one or more of the outputs could be the cause), or the feedback circuit is telling it that the output voltage is right even though it isn't.

So I would start by disconnecting the outputs from the rest of the unit. You'll probably need to put a resistor across one of the outputs to provide a load of, say, 5~10W, otherwise it may not be able to regulate and it may trip on overvoltage or minimum duty cycle reached.

If the rails come back to normal, connect the rest of the thing to the power supply, interrupt one rail and put an ammeter across the break, and power it up. Repeat for the other rails until you find the one with the overload.

It would be really useful to know the duty cycle that it's running at. If you have access to an oscilloscope, just test between any positive output tap on the secondary and the output 0V rail. The pulse you want to check is when the voltage goes slightly negative, just before it flies back and goes positive. The TOP234 can go up to 78% duty cycle. If it's already there, there's an overload or some similar problem downstream of the controller. If it's a lot less than 78%, something is limiting the controller's duty cycle.

The problem could be the feedback path. U3 in the diagram you posted monitors the 5V rail voltage and conducts, energising the optocoupler, when the voltage is high. This causes current from the reference winding (across C6) to flow into the C pin and this makes the controller back off its duty cycle. This is a bit more difficult to test because you have to be careful that the controller doesn't ramp its duty cycle to maximum and blow up the output capacitors!

Have you had a good look through the data sheet for the controller? It's pretty complicated and all of the pins have multiple functions.

 

[(*steve*)]

I would check a few things:

1) how did the original topswitch fail? If it failed with a short between the "drain" (D) and the "gate/control" (C), then the optocoupler and/or the 1N4148 diode could have been damaged.

2) You don't mention the voltage on the 3v3 rail. Both the 3v3 and the 5V rail are used in the regulation circuit. It's a bit weird, but maybe the 3v3 line is floating high?

3) The TL431 is probably also worth looking at. If it has failed it may have a much lower voltage across it.

edit: Aha! you say "6.5v is 3.35v". The schematic shows this is probably the 3.3V output. Try loading that one an see if the other rails come up.

 

[KrisBlueNZ]

3) The TL431 is probably also worth looking at. If it has failed it may have a much lower voltage across it.

Yes, that's a good thing to check too. Measure the voltage between the Adjust pin (the one that's fed from the voltage divider from the output rail) and the anode (0V on the output side). In other words, the voltage across R11. If it's less than the threshold voltage of 2.5V, the TL431 shouldn't be conducting any significant current, unless it has failed. Measure the voltage across the resistor that's in series with the optocoupler LED and use Ohm's Law (I = V / R) to calculate the current in the optocoupler LED. If it's significant (more than 100 µA or so), the TL431 is probably damaged.

edit: Aha! you say "6.5v is 3.35v". The schematic shows this is probably the 3.3V output. Try loading that one an see if the other rails come up.

That schematic isn't for his board. It's just a very similar design.

 

[(*steve*)]

The circuit is almost exactly like this except the voltage rails are different, feedback is based only on 1 rail (the 5v one) and the -5v rail comes from a linear regulator

I sure picked the wrong sentence to skip over in my eagerness to look at the schematic.

I'd probably first look for a voltage drop across R6 to determine if the optocoupler's LED is getting any current. If this is > 0 then the problem is more likely to be on the low voltage side.

If the voltage is 0 then I might suspect the output of the optocoupler and/or the rectifier

 

[spencer2004]

Hi and thanks very much for the replies.

I have ruled out any issues with the multi-function pin as I have tried shorting it to the source pin (which disables it) and it made no difference.

I have tested the power supply with it connected and disconnected from the main board, the datasheet of the TOP234 states that it can work with no load so there is no minimum load requirement. The voltages stay the same when connected to the main board and some power is getting to it as the LEDs light up (dimly) and the speaker makes a strange noise every 1/2 second or so as if it's trying to start up.

It seems unlikely that it's overloading as nothing gets hot or smells of burning and there are no short between any of the rails and ground but perhaps I could tap in to something just after the rectifier diode on the 5v rail and see what the current draw is.

Unfortunately I do not own an oscilloscope so I cannot check what the duty cycle is, my multimeter can test for duty cycle but I don't know if there is anything that I could probe that would work.

It does seem that the most likely culprit is the feedback path, I have ordered some replacement TL431s and I will try replacing it when they arrive. I haven't tried replacing the optocoupler yet, I have some that I pulled out of some old computer PSUs, will they work? I measured the voltages on the TL431; the cathode pin is at 1.846v and the ref / adjust pin is at 1.351v. I shall measure the voltage across R6 later.

 

[(*steve*)]

I measured the voltages on the TL431; the cathode pin is at 1.846v and the ref / adjust pin is at 1.351v. I shall measure the voltage across R6 later.

With the ref pin lower than 2.5V, the device should be turned off and the cathode pin should be at the supply voltage. (There should be almost 0V across R7 and even less across R6).

From your measurements it sounds like there is 2.77 - 1.86 = 0.91V across R7. Whilst this isn't enough to turn the LED on in the optocoupler, if there is any ripple on the output that might mean the peak voltage is somewhat higher than the average and the LED is actually turning on. The voltage reading across R6 should be instructive. Anything >0 will tend to suggest that the feedback loop is limiting the voltage.

 

[spencer2004]

R6 is actually 200 ohms in this circuit, there is 8mV across it so the optocoupler LED is drawing about 40µA. There is 0.932V across R7.

If it helps, the voltages at the optocoupler pins are 2.771 and 1.85 on the low voltage side and 7.79 and 5.82 on the high voltage side.

 

[KrisBlueNZ]

OK, well, the optocoupler isn't being energised. The voltage on its anode is low, but that's because all the output voltage rails are low. So now we need to find out whether the controller is being told that the output voltage is high. Cut the track to either the collector or the emitter terminal of the optocoupler (or desolder one of those pins, if it's a through-hole component) and connect a milliammeter across the break. Make sure that it can't slip off either connection, because that could cause the duty cycle to go to maximum and damage many components. Turn the power supply ON, record the reading on the milliammeter, and turn it OFF. According to Figure 4 on page 4 of the TOP234, current into the Control pin greater than about 2 mA causes the device to reduce its duty cycle. With the optocoupler LED not energised, the optocoupler transistor should not conduct any significant current. If the current you measure is more than, say, 1 mA, the optocoupler is damaged.

I also suggest that you check the soldering of the TOP234, anything else that you have soldered, and the optocoupler, for possible shorts.

I assume you have checked that D2 isn't leaky? You could just replace it on suspicion; the 1N914/4148 is very cheap. You could check the voltage across C6. I can't see anywhere in the data sheet where a voltage for this rail is specified, but according to Figure 32 on page 29 of the data sheet, it needs to be at least 7V, so with everything running at half voltage, I would expect at least 3.5V on it. Undervoltage on that rail wouldn't cause the controller to reduce its duty cycle, but if the rail isn't smooth (leaky D2 or faulty C6) it might have that effect. If you measure the voltage across C6, connect and disconnect the meter with power OFF; it's on the live side.

 

[spencer2004]

Guess what, it was the optocoupler that was at fault!

I just replaced it and all of the voltage rails are now within their 5% tolerance! Yay!

I'll now put it back together and see if the thing still works, thanks a bunch guys

 

[spencer2004]

Yup, it turns on and works just fine. It does now have an intermittant problem where it does not detect the bulb correctly and refuses to power on but I think the terminals need cleaning or something.

 

[(*steve*)]

Since the optocoupler was affected, I would tend to be a little suspicious of the 1N4148 as well. Just for safety, I'd replace it as well.

 

[spencer2004]

Yeah I might as well since I've already ordered some. I better not use it until I replace it then.

It's a good thing that the voltages didn't go the opposite way and blow everything really.

Thanks again

 

[KrisBlueNZ]

Congratulations and good work!