|In the mean time i tried to find out more about the problem and here are my results so far:|
1) The power supply is a "QTechnolgy Model NO: QT05360G" (who ever may be happy with that information) The specifications of it (as printed on a label on the outside of the cage) are the following:
+3.3+5V Combined 130W
2) I used a Voltmeter to measure the voltages on the cables which supply the peripherals (5V and 12V) These are the only ones that are accessable.
2.1) When working properly:
2.1.1) completely off (hardware switch): 0.0V and 0.0V (well that's not surprising)
2.1.2) stand-by: 0.35V and 0.0V
2.1.3) running: 11.96V and 5.2V
2.2) When NOT working properly:
2.2.1) completely off (hardware switch): 0.0V and 0.0V (again not surprising)
2.2.2) stand-by: 0.36V and 0.0V (not surprising either)
2.2.3) attempting to start: 11.4V and 5.12V
of course both values appear for just a very short time, so these are not stable values. Immediately after that the two voltages fall back to 0.
3) When the computer works (at least as far as needed to enter BIOS) there is a "Hardware Monitor" which also gives some information about internal voltages. The values shown oscillate constantly within a certain range therefore i give the limits:
VCORE: 1.280 .. 1.304
3.3V: 3.216V .. 3.408V
5V: 5.222V .. 5.273V
12V: 11.932V .. 12.038V
4) After off-hooking all peripherals and applying the above described "trick" the machine came back to life until i hooked the hard disk on again. Repeating that procedure a couple of times confirmed, the trouble returns as soon as i hook the hard disk. That seems to suggest, that the hard disk is the culprit.
BUT things are not quite as simple, as a couple of questions remain:
- If the hard disk is oververloading the power supply, why does it so only since saturday? It worked fine until friday and nobody changed it over night.
- After off hooking the hard disk the assumed overload should have gone and the machine should resume working, but it doesn't. It does so only after applying the "trick".
- It appears to remember a once experienced overload even across a total (hardware) power off. Only pressing the (soft) power button for a long time (longer than 20 sec) seems to make it "forget" its bad experience. Is this so? How does it work? Is that documented somewhere? The board's documentation doesn't tell anything about it.
- Is it really a hardware issue? Perhaps the power-up mechanics (north bridge???) are just a little bit too impatient? Is there a (software) time-constant somewhere that tells it how long to wait until it decides whether the power is good or bad?