A1XE : cpu fan not running, no video output

[Floid]

Framiga wrote:

by lionstorm on 2003/7/15 16:44:55

@ Framiga :
Thanks I will try as soon I get home for the speaker. As for the fan, I tried several times to invert, revert etc this bloody cable, even to put it on the second fan connector but no , it does not work.

At this point, could be better, if you test the fan out of AOne mobo.

At this point, I'm going to play devil's advocate -- contact Eyetech.  You've been running the board without cooling, you apparently had a power supply flake out on you -- it may not be dead, but it seems you aren't having luck bringing it to life.

Perhaps they can put you in touch with a dealer (with the hardware on hand, like a spare processor card and fan) to debug the problem, or arrange for a board swap at the cost of some shipping.

[Floid]

Hm, dumb thought, maybe it's already implemented (I've not looked at the SourceForge U-Boot tree lately, let alone heard much new about the AmigaOne version) ... Maybe there should be a blind flash feature?

Hold down (Ctrl/Alt/Win) - 1, 2, 3, or 4 during boot, and the loader would ignore everything 'risky,' look for a properly-structured ISO9660 image on the IDE device picked (given that we don't really do floppies yet), and provide some feedback (speaker tones, keyboard LED flashes) as to whether it's flashed itself properly or not?

Be a bit troublesome for SCSI, but 1. 90% of everyone's using ATA right now, and 2. if it were worth doing for SCSI, you could, perhaps, have a (Ctrl/Alt/Win)-S to try selecting the first SCSI controller, followed by acknowledging/error response, and the user could then tap the ID of their SCSI optical to attempt the flash.

[Floid]

Tomas wrote:
Actually i have encountered some motherboards being picky about this.. One i had was even picky about what order the sdram was in. :O Not very usual but some few actually have a prob with this, and neither anything to loose from trying  ;-)  

At least one obvious thing that could cause such would be the "DIMM clock select" feature you might've noticed in some BIOSes.  I know not enough about SDRAM in general (ask me how much fun BX-era boards have caused me lately), but the clock-select feature apparently relates to how the board provides clock to the DIMMs - enabled, it decides if a module is present, only provides clock if it is (how it detects, how it switches, and how many traces all this uses, I've no idea); apparently this saves a small bit of power, maybe preserves signal integrity, and helps them pass FCC and other RFI certifications (I suppose it's worse to have the traces acting as an antenna, without a DIMM to sink the signal).

In any case, OEM BIOSes often lock down such settings, such that you might not be aware the design features are even there/decisions have been made... If that feature isn't perfect in some designs, I could imagine it favoring the use of sockets 'in order.'  Then you have odd designs like my old EPoX 8KTA3, which provided 'extra' slots wired single-sidedly?/single-bankedly?/with fewer lines, somehow, giving users a chance to recycle their piles of older, smaller modules while still taking advantage of the then-new 256 and 512MB DIMMs.

Plenty of design oddities out there; sometimes the manufacterers don't even know what their products support.