Electronics > Repair
SMPS Troubleshooting. About to give up...
<< < (3/4) > >>
xavier60:
When VCC reaches the run threshold, it's being discharged very quickly back to UVLO.
Maybe a short on the 5V or on the drive output pin 6 or the IC is faulty.
Also during run state, some minimum voltage needs to develop on pin 1 either internally from the error amp or externally applied. I'm guessing about 1.5V for now.
Milmat1:

--- Quote from: fzabkar on August 16, 2022, 08:30:54 pm ---The way that these chips work is that they remain in "start mode" until the Vcc capacitor has charged up to a certain threshold level. The chip then switches to run mode and pulses the MOSFET chopper. The secondary winding of the transformer generates an auxiliary voltage which then augments the Vcc supply. If this secondary supply does not appear, such as when the DC output is overloaded, the Vcc supply decays with each successive pulse until it falls below the UVLO threshold. The chip then reverts to start mode and the process repeats. Some "smart" chips will give up after a certain number of attempts.

Typical faults with these circuits are the startup resistor and the small auxiliary filter capacitor.

--- End quote ---

Thank You !

Originally the startup resistor was open and the cap (82uf) had high ESR. I replaced both.
I'm fairly sure at this point that I have a failed chip. Because I cannot get the reference voltage on pin 8 to come at at all. And the chip appears to have everything it needs, just refuses to bring up the output. .
Anyone ever just pull the chip and breadboard it to see if you could get it to run ?
lowimpedance:
Why bother,  Digikey have the UC3842 in stock for a dollar each. If you're confident all other components are good and the original 3842 has failed,  just replace it.
Unless it's for educational purposes of course then have a go  :).
T3sl4co1l:
Note that failures generally cascade.  A typical failure starts with an overheating MOSFET, which fails as three-way short, dumping a fair fraction of the DC bus cap plus mains fault current into the current sense resistor, and gate circuit.  Both flow into the controller, which is subsequently damaged.  Resistors may also fail (shunt, or the resistor between it and ISENSE, or gate resistor and whatever else is with it (often a parallel diode for fast turn-off)).

The fault current blows the fuse as a final act, which might be tempting for its simplicity / obviousness -- it might be the first thing you check, but replacing it is the last thing you need to do in this case.  (Rarely, the fuse blows by itself, perhaps due to fault manufacture, thermal cycling, or mains surge; highly suspicious, but can happen.)

Startup resistor going open is relatively inexplicable, but perhaps an overvoltage condition occurred, or it was a thin-film type subject to environmental effects (corrosion).

A damaged chip may continue to work, in part.  A typical case is the output devices or ESD clamp diodes failing shorted, resulting in excessive current consumption, or excessive input loading, or insufficient output voltage.  This could be the case here, where the chip's UVLO/startup circuit is working fine, but when it enables power to the rest of the chip, it tanks the supply.  It could also be more direct, such as gate drive dumping into a shorted MOSFET -- you didn't mention if this was replaced or checked so as far as we know, it is still an option.

Also, because of ESD clamp diodes, it may be that the aux supply got overvolted, which could damage that capacitor for example.

More likely the aux cap just dries out, as it's a fairly stressed part, actually, despite the fairly low output current -- the controller typically needs a couple 10s of mA to run -- but it's also a small capacitor, and ratings in small sizes are rather modest.

Secondary side can end up toasted too; such a transient can dump excessive current into the rectifier, or maybe the failure was induced by a partial or complete failure of output filter cap or rectifier, or regulator (TL431?) for that matter.  And then the primary side tries to start up into a partial or complete short, and keeps chugging like that forever, or until something else blows up.

Have also seen one where the controller was just dead.  No idea how.  Everything else checked out, replaced it, worked fine. Go figure...

In short, replace everything active (transistor, controller, some diodes even), and check everything passive (resistors and capacitors).  Replacing electrolytics wholesale isn't a bad idea, but generally isn't necessary as long as they test OK.

Tim
xavier60:
A faulty start up resistor can lead to more serious problems if the PSU design doesn't include a Gate bleed resistor.
If VCC doesn't reach the UC3842's threshold needed for it to become active, the output pin remains in a Hi-Z state allowing the Gate to gradually get charged by leakage current.
It's possible that this problem has been fixed with current versions.
Navigation
Message Index
Next page
Previous page
There was an error while thanking
Thanking...

Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod