500W PSU MOSFET 20N60 replacement

[Buriedcode]

I am repairing a 500W powersupply that has a blown input fuse, and a couple of burnt SMD parts in the PFC.   I have the part numbers for the missing SMD caps and resistors in the snubber on the PFC switch from the manufacturer, but one of the PFC switches also looks like it has some charring around one of its legs. Given they are driven by a totem pole transistor driver with reverse diodes, I'm assuming these are MOSFETs.

The power supply has two TO-3PN switches in parallel for the PFC, marking 20N60 (with F e P above it).  I will desolder the one nearest the snubber to check it hasn't shorted, but a quick check across the switches shows no short.

In my haste I ordered the only 20N60 parts I could get in that TO-3PN package, FCA20N60 but I haven't replaced them yet.  Whilst double checking I downloaded several datasheets for MOSFETs with the same package, with 20N60 in their part numbers to compare specs.  Whilst the VDS max and Ids max are all the same (the 60 for 600V and 20 for 20A) there seems to be a lot of variation in total gate charge, input capacitance, turn on/off times etc..  What doesn't help is these datasheets give different conditions for each spec so its not an apples-apples comparison.

My question is, as these two MOSFETs are in parallel, should I replace both rather than just one? On the one hand, if there are differences then it could mean one switch takes more load than the other, or turns on quicker/later than the other.  On the other... if the turn/on times or gate charge are very different from the originals I'm essentially doubling this effect by using two new ones.

I'll get a photo of one of the switches to see if anyone recognises the manufacturer symbol. The F looks like fairchild, but not quite.

Cheers!

Update:  Got a photo, I'm thinking Fuji Electric FMP20N60S1: http://www.farnell.com/datasheets/1810676.pdf  although that datasheet is for the TO220

[Buriedcode]

Well, I might as well ask about this 500W PSU again, but this time, a general question.

Since the PFC switches weren't short, I decided to replace the burnt snubber SMD components on the board, and the fuse, and test the unit with a 60W bulb in series with the input.

With no load attached, the bulb flashes at about 4Hz, a few seconds later, the relay in the PSU clicks - which is the relay across the inrush limiting resistors - the output LED lights, and I get 30V on the output. Its rated for 27V, but given theres no load, 30V is fine.

However, the bulb continues to flash, the unit then switches itself off for about 5 seconds, then back on for 5 seconds, and continues.  I'm assuming that it requires a minimum load, and without it the output goes out of spec and it shuts down.

What do people recommend as a reasonable load for a 500W PSU? 1A?

[wraper]

Bulb flashes because its resistance is too high to run this PSU. You need higher wattage bulb or a few bulbs connected in parallel. Also as bulb does not stay constantly on, it should be pretty safe to just connect PSU directly to mains voltage.

[Buriedcode]

Bulb flashes because its resistance is too high to run this PSU. You need higher wattage bulb or a few bulbs connected in parallel. Also as bulb does not stay constantly on, it should be pretty safe to just connect PSU directly to mains voltage.

Thanks for the reply!
As it's a pretty large power supply, I was hesitant to just plug it in, so wanted some kind of limiting. Especially as it looks like the fault destroyed the snubber network on the PFC, and then the main input fuse - and I'm still unsure as to why they failed (I just assumed it was one of the PFC switches, or main input recitifier, but none show any shorts).  I'll try with 2 100W in parallel, and if it still shows the same behaviour, just go direct into a breaker.

The user manual has no mention of a minimum load, but the efficiency graphs stop at 10% (18.6A max out, so ~2A).  At 27V thats pushing 60W load, but I have some 20W 50 ohm resistors to test it with, at least to get the output voltage into spec just incase it trips the OVP.

Thanks!

[T3sl4co1l]

Sounds like it's maybe working fine.  Anyone's guess how it blew a fuse without nuking transistors, I guess...

FWIW, the input characteristic is negative resistance, i.e. drawing more current at lower input voltage.  Probably, under whatever load this counts as, it happens to be more negative than the light bulb's positive resistance (or indeed fighting with it, as light bulb resistance varies considerably with current as well), probably combined with some burst-mode behavior given the light load.  So, probably that goes away with stable input and optionally a modest load.

Supplies these days are usually fine without a minimum load, but anyway the interesting stuff is near full load, where everything's operating continuously and drawing real current.

Tim

[Buriedcode]

Well, seems it handled two 100W bulbs in parallel (in series with the input) just fine.  That gave me enough confidence to plug it in directly. I now have a quite handy test box with 3 bulb holders, switches to switch them in and out fuse etc.. for testing SMPS.

Also, it didn't seem to mind no load on the output, altohugh the voltage crept up to 31V because of it.

I've sent the unit back to customer, but still unsure exactly what caused the fault (vapourised SMD parts). Apparently it was powered by a large generator on site that failed, so perhaps over voltage on the input?

Cheers