If the fuse is intact then perhaps the switch is faulty? Check that first.

I'll give you a few hints but this advice assumes you know how to safely work on mains connected devices. If not then perhaps at least watch/read some safety tips. One such tip is to use isolation transformer but I'll also assume you don't have one. So at least know for sure which wire coming from the socket is the hot one and put ~50W lightbulb on it, in series with the PSU. This will protect the PSU from most kinds of massive damage due to shorts (including accidental, by you), and also provide some personal protection. Not much but better than nothing.

Anyway, check if there is rectified mains voltage on the input caps. Note these discharge slowly with no load (possibly very slowly if the PSU has no bleeding resistors) and the shock from those is fully capable of inducing cardiac arrest. Also quite unplesant to experience in general. If the voltage is missing or low the culprit would be the rectifier - possibly one of the diodes died open. I also have some vague recollection of a series of PSUs that had a silly design issue, the bridge rectifier got so hot it actually unsoldered itself. Was actually fine otherwise. I think it was an IBM PSU, but can't quite remember. Check that too since it's a simple job.

If the rectified voltage is there then perhaps the PWM chip is dead, or more likely not getting a proper starting voltage due to dried out electrolytic cap. There's usually a 10-100 uF at 50 to 63V somewhere near it, bigger than the smallest caps but smaller than the input and output filters. These ones like to die quietly by drying out, that would be the next thing to check. And the voltage on that cap - try to figure out that PWM chip is, unless IBM branded it with their own custom code/number. Then find the datasheet and the typical application schematic in it should at least explain what pin is what and what voltages to expect.

Again speaking of voltages, do note these kind of PSUs have separate input and output side grounds. DO NOT connect any grounded equipment unless you know what you are doing. Stick to battery operated meters and/or scopes (and in general always double or even triple check every connection you make when using a scope of any kind).

Hello Deunan,
Thanks for your replay,
I have a bulb in serie when I work on a PSU, and I have observed the folllowing:

- The primary circuit: 2 Massive capacitors are charging to 160V (which is normal for a 240V input AC voltage). The diode bridge seems to work fine. The thermoresistance is also working.
- The auxiliary circuit with the smaller transformer is also working, and delivering around 20V.

My guess is the circuitry to switch on the PWM is not working properly:
- There are 3 boards on this PSU: Card 0 with the Primary circuit and 12V, 5V -12V secondary, Card 1 Startup circuit card, Card 2 the PWM card.
- The connector P2 (connected to the Motherboard and linked to the power button), Pin4 grounded to Pin6 is going to CARD1 on a voltage comparator TM339N,
- The output of the TM339N is going to 2 successive series of transistor (C1213), and the output of the transistor are drive the PWM VCC on the CARD 2
- On Card2, there is a PWM chip TL494CN.

What I observe:
-When shorting the button (Pin4 grounded to Pin6), the transistor are receiving 11V but the last Emitter stays at 0.5V and not powering the PWM circuit.
- I have checked on a working PSU the output (Emitter of the C1213) should give at least 11V.

There are multiple Capacitor in Card 1 driving the Card2 PWM, but noting big:
- 470 uF 16V
- 100 uF 16V
- 2.2 uF 50V
- 10 uF 16V
- 10 uF 50V

There are all in a good visual condition, no leakage,

Hereafter some image to describe more easily:

What is the best approach ?
- Should I replace the Diode and the Capacitor in CARD 1 Board ?
- Is there a way to test without desoldering each component ?

Thanks
Vincent

Well, the PWM chip not being powered would certainly explain why the PSU doesn't start. So that transistor you mentioned is suspect. If you put 11V on the base of NPN transistor then the emitter should not stay at 0.5V - unless it's output is shorted.

Check if the collector of that transistor is powered, because if that part of circuit is somehow shorted then the base-collector diode might be "stealing" the current going into the base. If the base is getting 11V then the collector should be at least this voltage, preferably even higher. If so and the emitter is 0.5V then I would replace the transistor.

Thanks,
I am currently travelling and I will be able to test the transistor at the end of the week,
I have order a few capacitor and transistor in case
I will let you know
Vincent

Hello,
I received the order from LCSC (transistor, caps, ICs)
I have been working and the PSU and I have written a recap of my work with photos, I hope it will help others, as well,

https://www.r3tr0.net/index.php/2025/12/10/re … n-ibm-8540-psu/

Thanks for your support in this journey
Vincent