Shuttle Mini PC power supply repair

[gildasd]

Old SN27P20 that I need to keep going for a few more months... Typing this with a Celeron laptop with a broken screen.

Chain of events:
After one year of use, the PC would need a few minutes to start up after a an unplug.
Nothing else to complain about since, solid as rock appart from that, best PC ever.
On Sunday, there was a lighting strike in a another neighbourhood, but it cut us off too.
And the Shuttle wont start at all,. No fans, no leds, not even in the supply.

The supply is a Shuttle PC4213502II 400W.
No bulging caps or any other obvious problems on the mainboard or supply.
No magic smoke smell.
Seems to be a common problem (did a search), but the obvious "get another one" solution is not my current option, repair if possible is.

Here are my checks in chronological order:
Cord works.
On/off button works.
Fuse works
Green to black does nothing
... more to come

[gildasd]

Sorry, no photo software, images "as is" and Kicad is asking too much of my temp Celeron beast.

The filtering/protection on the mains input is nice, I'll copy it for my future projects.
So far, I've found no bad caps or blown stuff, only thing is that I'm getting "nothing" on the secondary side of the coil.

Mystery still thick.

[gildasd]

Bloody hell this thing is dangerous!

Even after discharging with a resistor, the primary 15V filtering still retains sufficient energy to spark a short significantly.
I've tested most of the passives on the primary side, and they all seem to work, I'm starting to think it's the IC before the transformer that is gone.

In any case, this does not seem to be a "quick fix" reapair...

[mich41]

How about replacing the whole PSU? Isn't it ATX compatible? Isn't it by any chance a normal SFX PSU?

[gildasd]

I've ordered a new boxed PSU.
However, I'll still try to fault detect and repair this one as a learning experience.

First step is to test the "regulators" on the primary side, there are 6 of the bastards.

[wraper]

Bloody hell this thing is dangerous!

Even after discharging with a resistor, the primary 15V filtering still retains sufficient energy to spark a short significantly.

Just like almost any other SMPS. 15V remaining means your resistor has too high resistance to discharge it reasonably fast.

[SeanB]

I see a lot of SMD resistors in series on the input side. Most likely is that there are high value or open ones there in the start up supply to the controller, which would explain the behaviour after power cut of the resistor slowly going higher in value till it was essentially open. Simple to check in circuit as well, just go along the input side resistors and look on resistance range for any that are either at marked value or higher value, though you might have to wait 10 seconds for capacitors to change. Easiest to only check where there are 2 or more in series, as this will be high voltage withstand. Replace with a standard leaded part as temporary test ( and put on a thick insulating sheet to test out of chassis) and then replace all of the same value on the primary side when you find which 47k plus resistor is open circuit.

The second hand refurbished ACER I am using to type this on had an annoying power supply whistle after a while. Rather than get a warranty replacement I simply went and got a cheap ( not the cheapest, a little better than the cheapest PSU I have had reasonable results with before) replacement PSU and removed it from the standard footprint case and shoehorned it into the custom ACER case. Took the fan as well, just did a quick mod to have it run all the time at full speed, rather than the default "run to lightly brown" that the thermally controlled fans are set to do out of the box. Even though the case is a third smaller the board did fit, and aside from having to solder on the mains leads to the socket it was easy enough. As a plus the new board did come with the optional input filtering and EMC chokes and capacitors, along with a wired in fuse, something ACER omitted from the original. Bonus was the extra 12V leads and the extra SATA power leads as well, as opposed to the original having exactly 1 power lead with SATA, IDE and FDD sockets along it. Also added 2 extra case fans to shift air around inside, seeing as it now has 2 HDD inside instead of 1, and they do benefit from extra cooling.

[gildasd]

After a watch worker's loupe check, there were no obviously damaged SMD parts.
But I'll do your check anyhow, replacing a resistor beats changing a regulator!
As for using the guts of another supply, difficult as this one is tiny yet still allows 400W (450 peak).
It's a neat design, but suffers from "dying" from main cuts, maybe the main capacitor discharging into the IC?

[SeanB]

Change any low voltage electrolytics on the primary side. They will be from 10-220uF at between 16 to 63V, used as a startup supply. If they are failing high ESR or leaky then they will also cause the slow no start issue. Easier to unsolder and bulk replace a few small caps than rebuild a power supply.

[gildasd]

Change any low voltage electrolytics on the primary side. They will be from 10-220uF at between 16 to 63V, used as a startup supply. If they are failing high ESR or leaky then they will also cause the slow no start issue. Easier to unsolder and bulk replace a few small caps than rebuild a power supply.

MAGIC
It was that startup cap you speak of!
My analysis was off the mark.
It looks perfect, no traces of leaks of bulging, but tests at 60uF instead of 100uF...
All leads test out good, all voltages are spot on (or 1 to2% high) and fan turns. 0.2€ repair.