[Ltsp-discuss] EPIA mobo power

[GraftonCottage]

Can we chuck in our EPIA5000 tupence worth in case it gels something for some one?  We've experimented extensively with a tiny 18 watt supply for the fanless CPU version.  We see two distinct intermittent problems: 1/. mobo won't start at all, 2/. mobo starts but won't even try to boot off a lan via a hub.    All of our mobos came from the same VIA shipment, possibly the same batch.  Most are faultless and never fail to start or boot, just a few are very troublesome.  Once started, all our mobo's will always boot off an ethernet *switch* but not a *hub*, so there's one clue.   We have 4 wires going to the ATX plug:  +3.3v, +4.9v, +11v, (don't ask....), and earth.  On the ATX plug we generate the PwrOK signal with an RC timeconstant off the 4.9v.  We've kind-of concluded it won't start at all if the +11v drops below +10.6v - otherwise voltage stability and accuracy is not an issue per-se.  We don't believe powersupply *wattage* is the issue or the solution.   We've tried to determine the cause of both problems, and the only thing we've found that makes a difference is the *rate* at which the +12v comes up against the other two. If the 12v comes up slightly slower, it ain't gonna boot regardless of whether it gets to 10.5v or 12.5v by the time the PwrOk signal kicks in.  If the 12v comes up significantly slower, the mobo probably won't soft-start either.   Altho the CPU doesn't "start" until the start button is pressed, we suspect some bit of circuitry looks around to see what's connected the instant the power is applied, decides what devices are connected, and then goes to sleep.  If +12v comes up too slowly, maybe it doesn't see anything that needs the 12v, so the power-good delay and all the rest might be academic.   Any ideas, anyone?   Cheers Mark Lochore