Switchmode Power Supply Trouble - Help appreciated

[HandyMan]

Hi all,

I've got a switch mode power supply inside a satellite receiver which I am repairing for my father-in-law (he's into all things satellite...) but although I've got it working after changing many faulty components, it's still not right.

After blowing the fuse nastily, I found faulty and replaced the bridge rectifier, a couple of rectifier diodes in the secondary, two zener diodes in the primary and a capacitor in the secondary which blew after that before I discovered that the 8-pin optocoupler had one LED short circuit, causing the voltages to go too high.

Now the power supply is running but every secondary voltage is low by around 2V to 4V. No humming or squealing from the supply, it's quiet and nothing seems to be heating up excessively.

The secondary potentiometer (only one on PCB) used to adjust voltages will not allow adjustment to anything like the stated voltages on the panel (for instance, line marked 8V is reading 5.6V and maximum adjustment brings it to 5.7V).

I've checked just about everything on the board and all seems well, but I'm dealing with other variables which make this a difficult task - as I had no identical replacement for the optocoupler (LTV827), I replaced it with one that has identical pin layout (TLP621-2). I don't know for sure if it's a good enough replacement, maybe it's the cause of the low voltages?

As well as that, when I found that the 13V and 30V zener diodes were short in the primary (they are connected in series, 13V zener's cathode is connected to a large resistor which in turn is connected to the mains live through a normal diode. The anode of that zener is connected to other components and also to the cathode of the 30V zener. The anode of the 30V zener is connected to the mains neutral. Both these zeners are part of a small sub-circuit including a transistor or two but the circuit doesn't seem to go anywhere else, it seems to simply be connected to the live and neutral but the live and neutral go off to do their own thing with the main circuit without going through that lot. My lack of knowledge shows here, now. Anyway, the upshot is that I didn't have a 13V zener so I used a 10V and 3.3V connected in parallel, not realising they should have been in series. Apparently this means there was equivalent of only a 3.3V zener in place. The power supply still worked, giving the lower than stated voltages on all secondaries. After realising my mistake, I connected the zeners in series and put them back in. To my surprise the power supply was unaffected, still working and producing the exact same low voltages! Surely the value of the zeners is important so how come no difference, maybe this is a clue to the problem? I can't begin to guess while I don't know what the little sub-circuit is for. Just seems strange that it still worked the same whether a 3.3V or 13V zener was in place.

Anyway, as if that little lot isn't enough, there's yet another variable at play here - when I was given the power supply, someone had corrected some of the problems on the board and removed a component only marked as 'RT1' on the PCB. The mains live goes straight through this RT1 after the on/off switch and before going to the line filter and then rectifiers, etc. The white marking on the PCB is a largish oval shape, like seen in places where large ceramic capacitors go, so it was a slim device, whatever it was. Maybe a thermal fuse type of thing? Could this be affecting the circuit enough to severely reduce voltages?

Any advice would be appreciated. All electrolytics in the secondary and the few in the primary have been checked and read fine, although only using a normal capacitance meter as I don't have an ESR meter. All diodes read good as do all the resistors on the entire PCB.

I'm stuck now. Over to you!

Thanks,
James

 

[john1]

Hi James,

I am impressed by your optimism.
I find those switching power supplies to be very complex,
and i would not consider working on one without a diagram.

This is not an easy subject for a beginner,
and to get any worthwhile help you will have to give much
more information, so that people can find data on the
unit you are working on.

Maybe a photo would help too.

Frankly, i am surprised that you have got it to work
as well as you have done.

I wish you well with it, and i hope that someone here can
guide you to an effective repair.

Regards, John

 

[Nigel Goodwin]

I would suggest changing ALL electrolytics in the PSU, a very common cause of PSU problems is low ESR electrolytics. If you have an ESR meter you can check them, if not it's easier to change them all.

 

[Exo]

I've checked just about everything on the board and all seems well, but I'm dealing with other variables which make this a difficult task - as I had no identical replacement for the optocoupler (LTV827), I replaced it with one that has identical pin layout (TLP621-2). I don't know for sure if it's a good enough replacement, maybe it's the cause of the low voltages?

I've found that the optopcoupler is very critical in power supplies, mainly due to the huge diffirence in switching times between various types. I've checked the datasheet for both types you gave and there are large diffirences in the switching times between the two, this could very well be thr cause of your problems, replace the opto with the same type.

 

[ChrisP]

Your missing device labeled RT1 is more than likely a thermistor of the NTC type, used to limit inrush current when the unit is first powered up. Its purpose is to protect the remaining circuitry from excessive current during startup by limiting current until the thermistor heats. As the thermistor heats, its ressitance (if it is indeed NTC) decreases, allowing current through it to increase to normal operational levels.

 

[HandyMan]

John1 - Thanks for your reply. They are difficult to work on due to their complexity. Not much fun at all and too many bangs if something is slightly wrong for my liking!

Unfortunately I can't give much info on the item as it's a satellite receiver which I'm repairing for my father-in-law and he supplied the unit without the case so I must get hold of him for the case so I can see make and model.

I'm not really a beginner, I just have gaps in my knowledge mainly due to the way I learned about electronics - my Dad was a qualified electronics engineer for brown goods most of his life (he's 74 now). He taught me all about it as I was growing up, starting with electron theory when I was 8 or 9, I think. Of course, despite years of teachings, there are still gaps, some simply where I've forgotten what I was taught.

It didn't take too long to get it working as it is now... I've been repairing chopper supplies (usually in TVs and video recorders) since I was 13, earning a bit of pocket money while my Dad oversaw my work. I try to look at the PCB, get an image in my mind of what's happening on the board and then go from there.

It does seem that the culprit may be the optocoupler I used, it isn't the same as the original as Exo has kindly and helpfully pointed out. I am quite chuffed as I had a feeling it may be that the whole time, since it's responsible for regulation of the secondary voltages via the primary frequency.

Anyway, thanks for your kind words of encouragement, it's deeply appreciated. So many Internet forums lack decent people with manners, but this Electronics forum seems to have a good spread of good people.

Nigel - Thanks for the tip, unfortunately I have yet to get an ESR meter, I think I must do soon. I can't help but lack enthusiasm when I go to the trouble of removing a dozen capacitors on PSU boards only to check them with a normal capacitance meter which doesn't give any indication of the ESR, so I'm aware I may be putting a faulty capacitor straight back into the circuit. Getting one of those is my next priority.

Exo - Thank you for finding the datasheets for both items, I tried but only found a datasheet for one of the optocouplers which gave switching times, etc. The other was bare-bones and gave no such information and I couldn't find a better one... you must let me know your sources!
I'm now trying to source an LTV827 device which isn't proving easy here in the UK. All my usual sources don't have it.

ChrisP - Thanks for that info. I neglected to mention in my initial post that I'd used a loop of wire to bridge the contacts so the circuit had a power feed. Obviously the correct component really needs to be present to stop the circuit getting hit hard repeatedly. I'll have to see what I can find.


Regards,
James

 

[HandyMan]

While looking for an LTV827 using the term in Google, I came across this website: https://www.akaabel.ee/index.php?&PID=products&Code=904749&LANG=est

As you can see, it puts: "LTV827 dil8 dual = TLP521-2, TLP621-2"

The TLP621-2 is the replacement I've used in the circuit! Is this implying they are equivalents?

Regards,
James

 

[ChrisP]

The TLP621-2 is the replacement I've used in the circuit! Is this implying they are equivalents?

One of the problems with trusting obscure Internet cross-references (as opposed to manufacturer's cross-references) is that you don't really know how the cross was determined.

Often, a suitable replacement in one given circuit is not suitable in another, depending upon which particular parameter of the device is the most critical. In cases of speed differences, a faster device will usually work acceptably in place of a slower one, but the opposite is certainly not true. BUT... the person generating the cross-reference may not have taken that into consideration, usually because the device happened to work properly in his particular circuit, and was therefore deemed to be a "suitable" replacement device for that particular application.

When considering a suitable replacement, always consider all of the individual characteristics of the device to be replaced, paying close attention to the most critical characteristic(s). In the case of your optocoupler, speed is very critical and must therefore become a primary defining characteristic.

 

[Sebi]

Maybe You can to find a TL431 to drive the opto-LED. This circuit give the current-driving for opto-LED, depending from output voltage so the opto coupling-ratio spread (50...300%) no give a failure. The another opto - i think - overvoltage protector: find a zener and a resistor seriel with opto-LED, connected to output voltage. The zener voltage must be equal or 1V higher as the output voltage.
About the zener in primary side: i mean - without reverse-engineering - this is a start-delay circuit, so can work with almost any zener voltage, just the delay-time changed.

 

[Exo]

Exo - Thank you for finding the datasheets for both items, I tried but only found a datasheet for one of the optocouplers which gave switching times, etc. The other was bare-bones and gave no such information and I couldn't find a better one... you must let me know your sources!

The datasheet for the PC827 (=LTV827) can be found here

The TLP621-2 i've got from a toshiba datasheet cd
(rightclick on download and save to your harddrive first or it won't open)

 

[HandyMan]

Thanks for the info. I've now replaced that opto with two singles side by side, one is a PC 817 and the other is one that was listed as an exact equivalent - PC 123. This is until I receive some PC 817s I have ordered.

Incidentally, the power supply is from a satellite receiver, an Echostar AD-2000IP. Pictures of the supply can be put on here if wished, unfortunately I can't find a circuit diagram (schematic) anywhere on the net.

After fitting the two optocouplers side-by-side, every voltage but one on the secondary rose by around 2V to only 1V below the stated voltage on the PCB. Unfortunately, the line which has been lowest the whole time - an 8V line, reading 5.4V - didn't improve much, it only rose to 5.6V. It seems like this coupled with the other lines still being down by 1 or so volts is causing trouble still.

The 8V line measured right after the rectifying diode measures 5.6V, so it seems to be coming low from the transformer, although I am confused at that one being so much lower than the rest (percentage-wise and voltage-wise) when they all derive from the same transformer. It's a straight run from transformer to output line, only intercepted by a 3-pin device looking like a transistor with markings: SEC D288-Y to which the output from transformer rectifiers and smoothers goes. One of the other two pins of device then goes to the output wire marked 8V. The third pin seems to have voltage on it from another line, as it is higher than the 8V line (8V line measures 5.6V) at 6.6V.

Anyway, at this stage I got out the hair dryer. A blast on the PCB later and the voltages had all risen by a fair bit, with the 12V line almost hitting 12V. The 8V line was still far too low but it had risen from 5.6V to just over 6V.

While it was still hot and giving the higher voltages, I used freezer spray to find out which component brought the voltages back down. I eventually found that a quick blast on the optocouplers knocked the 8V line from it's record-breaking 6V down to 5.3V.

Assuming the optocouplers are fine, what problem may there be in the circuit if heat on the optos increases the voltage outputs and freezer does the opposite? And what of this 8V line being down by over 25% while the other lines are down by around 10%?

I've replaced every electrolytic on the board, incidentally. No improvement at all.

What a brain-teaser this one is. I could give up and be no less off, but I want to get this working now just for the huge amount I'll learn from the experience! And I'm stubborn.

Over to you

 

[Exo]

It's a straight run from transformer to output line, only intercepted by a 3-pin device looking like a transistor with markings: SEC D288-Y to which the output from transformer rectifiers and smoothers goes. One of the other two pins of device then goes to the output wire marked 8V. The third pin seems to have voltage on it from another line, as it is higher than the 8V line (8V line measures 5.6V) at 6.6V.

That would be a 2SD288 (NPN / 80V / 3A / 20W / 35MHz). As you discribe it i would assume it is connected as an emitter follower.

Assuming the optocouplers are fine, what problem may there be in the circuit if heat on the optos increases the voltage outputs and freezer does the opposite?

Hard to say, almost all electronic components characteristics change depending on temperature, so heating or cooling devices will cause them to behave diffirently, even in working equipment.

Is there any load connected to the psu as you are testing it?
If there isn't that could also be a cause. A SMPS requires load, if there is no load there are internal safeties that regulate the output down to safety...

 

[Nigel Goodwin]

Incidentally, the power supply is from a satellite receiver, an Echostar AD-2000IP. Pictures of the supply can be put on here if wished, unfortunately I can't find a circuit diagram (schematic) anywhere on the net.

You haven't filled your location in, so we don't know where you are?, but I seem to recall that Echostar service for Europe is based in Holland. I also seem to recall that they will supply a complete new PSU for the receiver.

 

[HandyMan]

I'm in the UK, I should have filled that out.

There is a load connected to the CPU - the satellite receiver itself. The PSU seems to have two permanent voltage lines (12V and 5V, measuring 11V and 4.5V respectively). The mainboard sends a voltage to a transistor on the PSU when the on/off button is pressed on the receiver front panel to bring the unit out of standby, which turns on all of the PSU voltage lines.

When turned on, a red LED on the mainboard lights up for a second and a fan aimed at the PSU spins for that brief moment too. Whether the voltage peaks for that second and is enough to do the job, I don't know, my meter isn't really responsive enough to notice a voltage increase which would last less than a second. If that isn't occurring, I don't know why the mainboard powers up for that second only.

Regards,
James

 

[Exo]

Sounds like a safety kicking in after that second...

 

[HandyMan]

Hmm, might be. I was just replacing some diodes around the transformer. Found that one leg of a disc capacitor marked 102M SWC was not soldered into the PCB. Although the pad had solder on it, the leg hadn't been pushed in enough so wasn't making contact. Solder looked original, so may have been like that since manufacture, the receiver was bought second-hand by father-in-law and passed to me for repair. The capacitor has one pin connected to the drain pin of 1M0880 (looks exactly like 2S0680 - KA2S0680 but don't know if it is equivalent) via a glass bead diode and the other pin connected to the DC line from the main smoothing block. I've soldered the leg in properly now. Haven't yet tried the power supply again because...

I checked some components in the primary and found the replacement 30V zener I'd put in had gone short like the original. The 13V one was still fine. Just replaced with a new 30V but wondering why it went short again. I had replaced it before taking out my loop of wire from RT1 and putting in an NTC thermistor, as someone helpfully informed me should be in that position, so maybe it was getting hit hard and got knocked out before NTC was in?

Anyway, I'm a little confused at the moment. I came across a component which I didn't recognise on this board. It's marked D32 with the D designation like all other diodes on the board, but it is marked differently and so is the board. It's a small glass cylindrical diode, red colour shining through the glass, very similar to a 1N4148. But the only marking on it is a thickish blue band just slightly off-center. That's all, no letters, no band completely at one end.

I made the mistake of taking it out assuming it a normal diode without noting whether the slightly off-center band was on one side or the other. I took it out to test as it was reading open-circuit both ways with meter on diode range. I then saw the PCB marking beneath where it sits - it's a rectangle with a thick line running through it on one side (like other diode markings on PCB) and a thin line running down the other side. That's got me stumped. I put it back in with the off-center blue band closer to the thick line side of the marking on PCB. Hope that was right.

 

[john1]

Hi James,

I am following this with interest.
Dunno what that component is, maybe a picture ?
I think some diodes were used as voltage dependant capacitors,
athough i don't know what the symbol is, its possible they
read differently to usual diodes.

Maybe someone will recognise it.

Best of luck with it, John

 

[Sebi]

Maybe a DIAC? But without reverse engineering can't see the exact function... The high resolution photo (with good focus, top and bottom wiew) can help for identification. Are You sure: no TL431 in the circuit?

 

[HandyMan]

Hi,

John1 and Sebi - thanks for your replies. I will get pictures done in the next 20 minutes and provide a link in a follow-up message where the pics will be on a server.

I had started to change components (mainly diodes) and had decided to do one at a time, test the equipment and then move on. Unfortunately, I got lazy and started changing a few things at a time before testing, my lazy idea had been I then would have only 3 or 4 components to put back in one at a time if the PSU then worked, to find out which had caused the trouble.

After changing quite a few, the PSU started working, with the 8V line, previously reading 5.6V now reading 9V. The machine powered up and switched on properly, with mainboard LED and fan spinning constantly but it wouldn't turn off each time, I then found every voltage was too high with the 21V line at 31V and the 31V line at 40V, etc. An improvement anyway, just the opposite problem!

Another strange thing is that after the voltages went up, the machine was operating in reverse - switching it into standby turned on all voltage lines and the fan and LED lit up constantly, turning it "on" did the opposite, shutting the supply down with only the standby voltages running. I put this down to the voltages all being too high.

Anyway, since my main reason for spending days on this being so that I could learn from it, I desperately wanted to replace the faulty component and have the fault back. Unfortunately, I'd replaced a few in one go and because the mainboard had been working in reverse, I'd been plugging it in and then immediately unplugging it when nothing happened, so maybe the PSU fault had been cleared earlier when I'd replaced other components. I had to try and remember the last time I plugged it in and also checked the voltages with my meter. I think I just about worked out when and it didn't leave many potential components.

Trouble is, I had replaced three identical 1N4148 diodes, two in the primary and one in the secondary. They were mixed up and I'd just thrown them to one side, foolishly. To make matters worse, one was and still is missing, so maybe that was a faulty one, because the other two made no difference when tried in the circuit! The two from the primary were okay, I'm sure, as after changing them I did do voltage checks and all were low still. So it leaves the one in the secondary, but I don't know if the one I removed from the secondary is the one missing and the two I have here are working ones from the primary hence why no fault returning when I tried them both in the spot in the secondary. I've scoured the carpet and bench and simply cannot find the third one! I'm extremely annoyed with myself for this. Looks like I'll never know what the cause was, so all this time wasted only to learn nothing.

To make matters worse, I now have all the voltages high except the 8V line which is reading 7.9V. So it looks like if I can somehow reduce the voltages, the 8V would then be a bit too low but perhaps that wouldn't be a problem.

I'm just so frustrated today, I should have been far more careful and kept the components safe and noted which came from which part of the circuit. It may not have been the diode and maybe a component which is no longer working in a faulty manner when replaced in the circuit, heat might have started it working, or maybe there was a dry-joint on a component I removed so component was good but removing it sorted the joint out. I couldn't see any dry-joints but I know that barely visible ones can cause problems in high-frequency applications.

I want to find that diode to at least rule it out

 

[HandyMan]

I've decided to put my foolsihness down to experience. I now believe that a imperceptible dry-joint was the cause of the low voltages and that it's unlikely the only component to go missing happened to be the cause of the fault. I think I changed a component with dry-joints, that got it working and then putting the original back in, of course, didn't bring the fault back. I also think it may have been the transistor I changed which is directly coupled to the opto. It is KA431, a programmable shunt regulator. Although I'd like someone to explain what that is and what it does in a circuit please, if someone doesn't mind! Also, if, knowing what it does, you also think that it could have been responsible for the low-voltage fault if it had dry-joints. It has one leg coupled to the cathode of one of the optocoupler's internal LEDs, one leg grounded, and the other on a 5V line fed through a 1K resistor.

Just to add to the previous post, I've inserted the original TLP621-2 8-pin opto in place of the two singles and the voltages have reduced by around 2V, except the 8V line which has gone from 7.9V to 7.8V. I still have the following voltages though, firstly I'll put the stated voltages as printed on the PCB followed by the actual measured voltages:

Printed:
31V, 21V, 12V, 8V, 5V.

Actual:
37.3V, 27.8V, 14.7V, 7.8V, 5.9V.

Bear in mind the unit is still working backwards and these voltages are when the mainboard LED is on and the fan is spinning even though the front panel says it is in standby. I still think the high voltage is responsible for the backward operation.