Happy ending. But why?

[ehume]

I was downstairs earlier this week working on my testbed when a fan clip got away from me, bounced on the motherboard. My display went blank. Uh oh. Turned off the system, restarted. Nothing. Oops.

My testbed is an i7 860 mounted on a Gigabyte GA-P55A-UD3P. The heatsink is a Prolimatech Megahalems. The fan clips are 5-inch wires with another inch and a half side extensions. They have shorted out my video card before, but that is when the card was mounted in slot 2.

The video card is a Powercolor AX3450 256MB. It is passive (fanless) and it was mounted in slot 5.

When the clip has gone flying before and the display blanked, simply restarting the system fixed it. But that was when the card was in slot 2. So I moved the card to slot 2 and restarted. Nothing on the display. Oh, the little blue light on the monitor comes on for a few seconds, showing it is getting a signal from the video card. But then it turns yellow -- no signal.

And I'm noticing that when I power on, the fans on the Megahalems and the video card spin for a bit, then turn off. Power off, power on. Now the fans stay spinning. BTW -- I have the BIOS set to boot up with AC power ON so I won't have to deal with switches.

Hmm. The next step should be to go upstairs and pull a working video card from a rig and see if it will work, but with the odd behavior of the motherboard, I'm thinking the BIOS was corrupted or something shorted out. I know circuits can die: I shorted out my CPU Voltage control circuit once (the CPU pwm control continues to function). And I don't want to open up my main rig right now. So I jumpered the CLR CMOS prongs. BIOS is now reset to factory.

Oh, that's right. Now I need a switch. Hmm. I had taken the Phantom 410 out of its box to photograph the fans. So I pulled off its panels, plugged the switch wires into my motherboard, and pushed the button. No display. But at least I'm not getting the false start issue. So that's fixed.

Waited to Friday (last night). Pulled my Gigabyte GV-R467ZL-1GI HD 4670 1GB from my main rig (you can see I'm not a gamer) and put it in slot 2. Fans spin, but no display.

Cast my eye around. What is there left I can do? My downstairs monitor is an Asus VW193TR 19" that got good reviews. But I did not like it, so it became the downstairs monitor. It works though, and the colors are adequate to display my test software.

I decided I would unplug the monitor cable from the monitor and put it back. This I did. Turned to the Phantom 410, pushed the button. Glory be! The Gigabyte splash screen appears!

I let it boot into Windows. Open Office recovers the files I had open. Hit the Restart option in Shutdown. Delete key into Setup. Told BIOS to boot on AC power ON. Unplug the case, carry it away. Turn on the power. We're back. Delete key into Setup. Turn off the splash screen, restore my current OC settings. Back into Windows. Lovely. Shut down.

Move the 4670 to slot 5. Boot up. Works fine. Hmm. I wonder . . .

Pull the 4670. Put the original Powercolor card in slot 2, hooked it to the monitor. It works! It is not dead. And it works in slot 5. All back to normal. The question is why.

The crucial change seems to have been my unplugging and replugging the video cable from the monitor. But that doesn't make sense.

Happy ending. But why?

 

[Rvenger]

Its electronics. Some things are left unexplained, kind of like how my lawn mower was running with a disconnected plug wire. Normally when you completely disconnect a monitor cable it will reinitialize the driver for the display. Thats the only explanation I would have in that situation.

 

[T_Yamamoto]

Magic.

 

[ericloewe]

kind of like how my lawn mower was running with a disconnected plug wire.

Really? Well, that's a step beyond "graphics cards start working again randomly".

 

[ehume]

Its electronics. Some things are left unexplained, kind of like how my lawn mower was running with a disconnected plug wire. Normally when you completely disconnect a monitor cable it will reinitialize the driver for the display. Thats the only explanation I would have in that situation.

Interesting that the system somehow might distinguish between a disconnect at the monitor vs. the various disconnects and reconnects at the video card end.

I recall that on one of my boards -- Gigabyte or As.s -- I was getting on-of-restart early in the board's career. But it was so long ago I can't remember which board it was. This situation differed in that there was no automatic restart. I had to do it myself.

Another possibility is that it simply needed a certain number of starts to sort itself out, conflict by conflict, so that my unplugging/plugging made no difference.

One lesson that pounded itself in once again: the value of changing all the variables and trying all the possibilities. Restoring the video function after messing with the monitor may or may not be causally related. But trying out a "dead" video card and finding it worked was a definite win for the keep-trying-all-possibilities method.

Actually, I had a nice inexpensive passive upgrade picked out for my video card. But if the mb was dead, replacing it would have been a crapshoot -- maybe an expensive crapshoot.