More SMPS repair (bad transistor)

[Rachie5272]

I have a broken SMPS from a commercial LED light fixture.  The primary culprit is obvious: an SMD transistor is fried.  Replacing it would be simple enough, if I had any idea what to replace it with.  I can just barely make out the markings "3D V1T1", but searching yields nothing.

The fuse and diode bridge seem fine.  There are scorch marks around the main FET and a diode, so I'll replace those as well.

Any ideas what the little transistor is?

[T3sl4co1l]

Not a clue. What's the chip say?

Stuff like this is usually a standard (mfg recommended) circuit, so you might be able to recognize other parts of the circuit based on that.

Tim

[Rachie5272]

The chip is an UC3845 from ST, which is a current mode PWM controller.  The datasheet says nothing about SMPS circuits.

[Chalcogenide]

MMBTA44 http://www.secosgmbh.com/datasheet/products/SSMPTransistor/SOT-23/MMBTA44.pdf (NPN 400V 0.2A) is marked 3D, but typically that is the only marking on the package, so I have no clue if it is the right part.

[T3sl4co1l]

Well, it's SOT-89, not -23, but the ratings seem about right.

Inspecting the board in a bit more detail,

Badly perspective corrected, flipped, top:


Bottom:


Overlaid:


The transistor looks like it's a supply pre-regulator, as you'd see in a one of these, for example:
http://powerelectronics.com/site-files/powerelectronics.com/files/archive/powerelectronics.com/mag/408pet07_web.pdf

Likely, one of these would work:
http://www.digikey.com/short/3tzdt2
(Doesn't look like any have the same marking; it could be some random Japanese (2SC or 2SD) transistor for all we know.)

But first you'll have to address why it died.

Check out the bottom.  There's a really crusty gray looking solder joint, just behind the transistor.  I can't see what's hiding behind there, but I'd guess it's a diode, possibly a UF4004 or P6KE200.  Looks dead in any case.  Transistor doesn't look much happier.

I might guess at a chain of events:
- Diode overheats, fails shorted (or overheats enough that leakage looks like it's a short; same thing to the circuit)
- Transistor can't handle the extra load current, dies
- Dead transistor becomes three-way short, zaps controller IC with >100V at lots-of-amps
- Dead controller IC draws mondo supply current, nuking the supply dropper transistor
- Dead everything toasts the input fusible resistor (which you might not've noticed having failed, as it's under heatshrink to prevent sparks flying)

I suppose there's an outside chance the transformer might have a short or something, too.

Basically, you can buy a brand new one of whatever this hunk-O'-junk was, for half the cost of even ordering the replacement parts, let alone the labor to fix it (and probably not succeed, then fix it again).  Let alone the time we've spent posting about it... oh by the way, that'll be $30...

Tim