Well I said I'd been wrong before. 11.5 volts on video card.
 

I finally got ticked off with the clumpiness of the video. Slow to paint, and even with pre-viewed photos in memory, they painted on the screen a segment at a time.

I measured the voltages on the dedicated supply to the card. 5v solid. and 11.5v with minute fluctuations. Viewing Good photography showed no dipping of voltage on either line.

I still thought Bow-lox to the notion that a 250 w PSU was the culprit, because the computer seemed stable in all other respects. I had to know.

I took an old AT PSU out of the junk box and compared ground voltages with the computer. I got 50v or so AC between them, but when I dropped the meter setting, (AVO8) it showed scarcely any more needle deflection. A good sign that the stray v was of very high impedance. I nailed the chassis together with a wire that could cope with 50 amps at least. Checked the voltages at all points as common.

Took a deep breath, and fired it up.

The effect was astonishing. The whole system had seemed to take on a new vibrancy, sluggish painting of the raster, now in a blink of the eye. I can still hardly believe it.

Disturbed a duff connection? We'll see when it's normalized again, but I don't think so.

I know that some of the dips would be of very short duration, an not show on a meter, but I had thought there would be a general lowering during heavy video processing. If there was, it was in milli-second bursts.

11.5v steady, just too low?

That's the only other notion. Time will tell.

I often wonder when browsing through here, what the next problem Loose rivets will be having with his PC... :ok:

It makes my tinkering look very tame in comparison.

ATX spec is 12V +/- 5%. If you can see wobbles on an AVO, the supply's borked. Max ripple @12V is 120mV; if you've a cheeeep PSU (say a Bestec) the 5V line may have little or no regulation, but be relying on the 12V components. There's a "power_OK" signal from the PSU on pin mumblemumble (it was all a long time ago) but, of course, the cheaper PSUs can lie or be unable to work out their own problems. Again, on a cheap PSU, perceived problems on the 12V line could actually indicate problems on the 5V line: current capacity required at 5V is increasing more rapidly than current required at 12V these days, so that side of things tends to be the weakest link.

I see the specs of your card specifies a 350-watt supply. I bet it's the 5V giving you problems.

...but comfortingly successful :)

Although Rivets is an inveterate tinkerer...:) it is nice to see an intractable problem solved. Most of us can re install software, update drivers, swap cards about etc. but when you get to the nub of one of those 'tried everything I can think of' problems, it is very satisfying indeed. Well played LR.

I am just chuffed you fixed it.

Way to go.

The moment you take a voltmeter out most IT pro's start running a mile. Obviously not part of the system engineers exam.

I am suprise the graphics card was the only indication though. I would have thought you would have had other errors as well.

All the best for the new year of tinkering

14
Green :)

(Says he having also been tinkering with PSUs this week per earlier post!) :)

AHA!

8 grey :):)

14 green is PS_ON

Sorry, I read it as PS_ON that you meant!

This is the pin-out I used when troubleshooting my PSU

Well if you insist on using antique pinouts. 18 is now deprecated, pins 21 - 24 are missing and, thanks to you, I'm suddenly remembering lots of other pinouts from the past which will no doubt push useful, current info out of my brain.


Wait, I can't tie my shoe.