Power Supply Failure (with Ultraviolet Radiation)

[metrologist]

I really wanted to use this power supply, but it was dead. I opened the case and found F1 an open circuit, so I attached a jumper and powered up the supply.

R3 in this diagram began to hiss glow, like it was welding...

https://us.tdk-lambda.com/ftp/manuals/17133.pdf

I'm concerned because U3 looks like two distinct ICs, not one. There is a clear divide between. Is it a lost cause?

[barry14]

F1 is meant to be a protective device and is designed to go open circuit if the current through it exceeds its rating.  Since F1 was open in your power supply, something is faulty.  You should not place a short across F1 in order to run the supply.  As you found out, this can cause unsafe currents to flow and parts to burn out. You will need a full schematic to do proper troubleshooting but you can look for burned components and other damage.  Replace these parts, especially F1, and then try again.

[metrologist]

Thanks! I noticed a burned spot on the cover that I removed and it was next to one of the two secondary outputs that plug into J2 and J3 (16V, 8A each). I was hoping the fault would have been in one of those because I was just interested in the primary output. It was a small jumper 

The smoking device looks like a typical brown disc capacitor, but is marked R3. I'm not able to find a schematic, but my main concern is U3, which may not be a part than I can get.

[metrologist]

The last pic shows the relative  location of the burned section, input line on the left. The burnt item is marked R3 on the parts layout. Any ideas what this could be? Thanks.

P.S., the part (U3) that I was worried about seems to be an opto...
http://www.vishay.com/docs/83652/ild615.pdf

[aargee]

Sorry, but I find it hilarious that someone with a user name of "metrologist" would use such a fault finding technique. 
The danger is that bypassing the fuse may make the faults worse and more numerous.

[metrologist]

I removed the burned part and the traces seem OK. The PCB silkscreen is marked R3. I was thinking it's a thermistor. Anyway, any ideas from the numbers? One leg goes to the induct or, not sure yet about the other. I have, I think, a 250 ohm thermistor that came out of another supply.

[Gyro]

You really really do need to work out where it is in the circuit - you can't just do it by guesswork substitution. You need to sit down and trace out where it sits in the circuit.

It looks to me like a 2??V MOV, especially as I can see what looks like an NTC termistor in the center of your third photo in the reply above (the black disc part). 250 ohm would be a really high value to be in a power supply, very unlikely.

As I say, trace it out properly or you will do even more damage.

[wraper]

This is certainly NTC for inrush current limiting. By carelessly shortening the fuse, you made more of the fault as this isn't the part which was originally faulty.

[metrologist]

I recall long ago from technicianing that thermistors were a not totally uncommon failure. Anyway, this part was already glowing before I got in there.

I'll try to figure out this part of the schematic. I think the big black part is labeled as R1 and the burned one R3.

[Gyro]

This is certainly NTC for inrush current limiting. By carelessly shortening the fuse, you made more of the fault as this isn't the part which was originally faulty.

Not possible that a shorted MOV blew the fuse and then melted down when the fuse was bypassed then??? 

One thing's for sure, if you replace a MOV with a NTC thermistor then you're going to know about it in a hurry!

[kxenos]

If it's a mov:
it's not needed. Remove it, install new fuse and power on. If no fire then check voltage in big cap. If voltage ok then install a resistor (like 330R or something similar) in output and check for pulses in mosfet's gate (the pin with the thin trace) and capacitor's -.
If it's an ntc:
You can measure for shortcircuit in the bridge and all mosfets / transistors in the primary side. Probably the mosfet is dead and you alive, ok. If there is no short then remove mosfet, install fuse, replace broken ntc with 10R resistor and power up the PSU. If there is no fire then measure voltage on big capacitor. If voltage ~ 325V and you alive, replace controller ic. May the holy spirit be with you.   

[Gyro]

May the holy spirit be with you.   

I'm really not sure the OP is capable of testing the primary side of a SMPS while powered without doing himself serious harm (Edit: You just need to look at how the mains input is wired). Best to be cautious and trace it out and check for shorts unpowered.  If in doubt, play safe!

[metrologist]

What wrong with the input wiring? That's a good solder joint - I'm sure up to code  once I goob it up with hot melt

It looks like a single strand 35 ga. copper wire will work as a 5A fuse.

[Gyro]

Err, I can't tell if you're joking now 

- wires just tack soldered onto the connector pins, no mechanical joint.
- No sleeving of connections (in case one of them actually does fall off!).
- No strain relief.
- No colour coding, just three black wires from a flat mains cord.
- ...and being in the USA, no upstream fuse! 

If you start turning that board upside down and back for measurements, one of those connections is going to fail, either from solder joint failure or from the stranded wires fracturing where the solder has wicked up them.

It looks like a single strand 35 ga. copper wire will work as a 5A fuse

No it won't a fuse comprises a fusible wire in a suitable housing (even with rewireable types). You're not doing anything to reduce my concern. 

[Fraser]

Whoa, this thread is scary.

Basic safety procedures apply

1. If not competent to work on live mains power supplies...don't ....The Darwin effect will potentially cost you your life
2. Never, ever just bypass a fuse. It is in circuit for a very good reason. If it fails, it failed for a reason that should be investigated.
3. Always check the input to the SMPSU for low resistance.If such is present, connecting it to the mains with a new fuse serves no purpose !
4. If a low resistance exists across the input to a SMPSU you should start checking the input components for failure with the multimeter until the cause is found and rectified. Be careful with the high voltage primary side capacitor as it may be charged to almost 400V (on 240V supplies). It hurts if you get across it !
5. If the input does not appear to have a low resistance across it, fit a new fuse of the correct currret rating and connect a 240V 25W incandescent light bulb in series with the live supply line feeding the SMPSU. The lamp provides some protection if excess current is drawn. Fit some dummy loads on the outputs. Suitable resistors or light bulbs will do.
6. If the SMPSU is switched on and does not illuminate the light bulb the outlook is good. Check for correct output voltages on the SMPSU. If present and correct, remove the light bulb from the supply circuit and see if the SMPSU functions normally with some dummy loads on the outputs. If not, you need to investigate the output stages of the SMPSU for faults. The smoothing capacitors should always be considered suspect unless proven otherwise.
7. If the SMPSU works correctly during testing, there is the possibility of a fault causing high current draw from the SMPSU when it is powering its intended load. This should be investigated.

This is a generic and somewhat simplified decription of how to approach a faulty power supply. It is not specific to your case. In your case, high current was drawn when the fuse was by-passed. That cold be a failed (low resistance) MOV across the supply rails, a failed rectifier, failed switching transistor or significant low resistance failure in the seconday circuit drawing excess current.

You should always do as many quick checks on a SMPSU as possible without power connected. It tends to avoid electrocution and burn-ups !

Fraser

[wraper]

What wrong with the input wiring? That's a good solder joint - I'm sure up to code  once I goob it up with hot melt

It looks like a single strand 35 ga. copper wire will work as a 5A fuse.

If NTC blows up, you get open circuit with no current consumption, not a blown fuse. There is 99% chance that some semiconductors are blown up, especially after your doings. Most likely at least MOSFET and probably driver circuty failed. But after what you've done, much more additional parts might fail too. NTC was glowing because the circuit after it was shorted. By shorting the fuse you made huge current to pass through without possibility to interrupt it, resulting in NTC glowing and blowing up.

[wraper]

It looks like a single strand 35 ga. copper wire will work as a 5A fuse.

[metrologist]

OK, peel your hands off your faces for a sec. I removed the burnt component and I also removed the two 16V secondary supply sections, then I powered the PS up again.

I'm getting 6.75V and 24V on the main supply outputs. I was scrounging around my salvaged parts and found a Vishay varistor with I think the same markings 275V 594-xxx, so I think this device was the same kind of varistor. The salvaged part also has the ceramic beads around the leads.

I'll get the full part number of the replacement for a final sanity check before whacking it in there...

 

[LektroiD]

It looks like a single strand 35 ga. copper wire will work as a 5A fuse.

[metrologist]

I started drawing the circuit over a parts layout, but did not try and convert that to a schematic form that I know you all will want to see instead.

I also found this replacement part and soldered it in, replaced the fuse, and the supply seems to work fine now. Since this is on the input side, I suspect it was a power surge that took it out, or it just failed with age?

[Ice-Tea]

Ageing in MOVs is a thing, actually: characteristics vary over time and with surges applied and it will degrade so that at some point it will die because of a surge it handles a thousand times before...

[metrologist]

The part I salvaged came out of a DOA power supply unit that I stripped.

[Ice-Tea]

Ageing in MOVs is a thing, actually: characteristics vary over time and with surges applied and it will degrade so that at some point it will die because of a surge it handles a thousand times before...

It ages with each surge*, but I don't think time is a bug issue for MOVs.

*: Which means it is not a permanent device. Just like flash memory, after you use it a certain amount of times, it will deteriorate.

True. But installed and working it is likely to be submitted to a surge from time to time. So age and #surges are probably correlated.