Hi,

Does anybody know what U5 component is (bottom of the PSU board, roughly in the middle of the board) - see photo? Mine blew up and can't really tell what it is. Also - the values of the resistors nearby (R27 and R39 - see the photo) which alse seem to have blown up... And finally - that diode nearby the resistors...
Perhaps somebody has a disassembled PSU and could make a photo...?

No idea but what did you do to cause that?

Which one - CRS328-4C-20S-4S+RM or CRS328-24P-4S+RM?
There may be still more failed components (failure is not always visible to the naked eye), it is important to find the root cause (otherwise trying to repair might cause even more damage), and remember about safety working with power circuits.

It's CRS328-24P-4S+RM
All valid points and there are more failed components on the board. But (so far) I need to identify those mentioned in my previous post.
I've no idea what caused it but it would be a shame to throw away a working device (ie. Everything alse works perfectly and it's juat PSU that failed).

The burned resistors and diode seem to be part of the gate drive circuit of a power MOSFET or IGBT which got shorted too. The unknown chip could be a PWM or PFC controller, if pin 1 is upper right then quite a few PFC controllers (L6562 for example) have gate drive output on pin 7 (between ground on pin 6, and power on pin 8 ).

Also check out viewtopic.php?p=856559 for instructions how to replace the failed power supply with an external one, just +24V DC is sufficient if you don't need 48V PoE. Or try to order a replacement https://www.i4wifi.eu/www/en/291935-mikrotik-g1070 - not cheap but could save you a few hours of work.

If I guessed right and it's an active PFC, you could try to bypass it (remove the shorted MOSFET) and the next PWM stage might work at slightly lower voltage (not boosted by PFC, just peak rectified from the mains - PFC is mostly there for compliance with harmonics regulations to make the power utility company happy) if you don't try to load it to the max power. It may or may not work, it really depends on the design though.

Thanks for that. I was thinking of using 3rd party 24v power supply but unfortunately PoE is a key requirement for this box.
As for replacing powr er supply - sure, that is an option. But I enjoy fixing stuff (I know - I'm a masochist ).
And yes - that chip seems to be driving a mosfet (which is indeed shorted and I think the whole catastrophy started at that mosfet and it's the root cause). But I do t want to try and guess what that chip was (or try to replace ot with something else) because it's a nice power supply and I dont want to cause more damage...
Anyway - I'll keep looking...

I have a working CRS328-24P-4S+RM but it's in service, so I can't take it down to look inside the PSU right now. I plan to migrate to netpower 16P (closer to the many powered devices on a 12m high mast, and then only run DC power and fiber from the bottom to the switch, replacing a bunch of aging outdoor cat5e starting to fail after 10-15 years), then the CRS328 will be freed but it may take a month or two.

You can have PoE with external DC power - just need to supply both +24V and +48V if you need selectable 24V/48V PoE output.

The cause of the MOSFET short could be AC overvoltage (lightning strike, loose neutral wire, wrong settings of nearby PV inverter to sell more solar power to the grid, etc.), but also another failed component that could open the feedback loop so it was trying to boost the voltage too high. Low power resisitors working at high voltage (like in a feedback loop voltage divider) sometimes fail open without overheating or any visible damage.