old standby box won't boot

[kortec]

I built myself an AMD64 box a couple years back. I plugged it in the other day, and it won't boot up. I've been able to fix things of this nature before (bad ram, busted disks, etc.), but this one has me and mine puzzled.

The hardware involved
■ AMD Athlon 64 3800+
■ Rosewill nVidia 5900 GPU
■ 1GB of pqi DDR 3200 (one stick)
■ assorted SATA hard drives, optical drives
■ antec True 430 PSU
■ ASUS KV7 motherboard (iirc)

When I boot it up, I can feel the PSU spinning, and a small, green, LED on the mother board is lit up. I can also feel the drives spin up, but they don't sound like they're being accessed (I've heard this box boot up enough to be pretty familiar with that sound). The fan on the CPU and GPU heat sinks are both visibly spinning.

There is, however, no video output on either the VGA or DVI port (with known to be working monitors). Also, I had it plugged into a switch, and the indicator light on its port was not active. I tested both the cable and the port by plugging it in to a known working box, and I'm convinced they're fine. I'm getting no beep codes, even if I remove the RAM before I power on, and I can't boot to a memtest86 CD.

That's about all I know. I'm not sure what to think of all this any more, so if anyone has some insight, I'd be much obliged.

 

[ghmage]

The motherboard might have decided to be 'at rest'. (Aged capacitors, or dead CMOS battery.) Can you narrow it down by removing all components except what is required to boot? Graphics card, mem, cpu, nothing else. Do you have another system to test/swap components? Is the motherboard mounted safely on risers, or is something possibly shorting out? I had an old Asus a7v which morphed into a brick in storage, and did exactly what you are describing. I even had the same little green light next to the AGP slot. I ended up using the motherboard for Origami.

 

[kortec]

The mobo is indeed on risers -- my nice Antec case saw to that.

I have another system that I can swap things into to check them, but I'm not sure I have to any more. I unplugged everything except the GPU, RAM, and CPU, and lo and behold the old girl booted! This is the sequence of events that I get to see:

■ I hit the power button, things start spinning
■ Monitor flickers
■ I see the motherboard splash screen
■ The serial drives get scanned
■ The IDE drives get scanned
■ I'm told that the BIOS is not installed for the FastTrak controller (I'm pretty sure it never was; its a hardware RAID thing I don't use)
■ Text displays: CMOS checksum bad / NO Hard Disk Drive Detected ! / Press f1 to run SETUP / Press f2 to load default values and continue

I'm going to start putting things back in one at a time see if I can't figure out what's breaking the system.

 

[ghmage]

Could be the PSU choking when it has too much load, too. Nice diagnostic work so far!

 

[kortec]

I didn't have anything terribly special set in the CMOS, so I reset it, and started adding things back in one at a time. It booted just fine with the sound card back in, so I added my optical drives, and it booted to memtest! Unfortunately, it ran memtest for about three seconds, and then the top half of the screen went black.

I'm *reasonably* sure that the contents of the lower half of the screen mean that my RAM is roasted, but just to be sure, I'm posting an image here:

So if you could verify that, I'd be much obliged. Also, if that's the case, why wouldn't it boot before? The only difference now is the presence of my hard drives and my floppy -- everything else has been plugged back in.

 

[ghmage]

Have you tried reseating and retesting the single RAM stick? You could try a different RAM stick or a different PSU.

It might have not have booted before due to a PSU (or other unit) that's borderline. Out of spec power can show up in various ways. It is also possible that when you were in the case pulling stuff out to test, you jiggled something and that made the difference.

 

[kortec]

I have a friend with compatible RAM that I'm going to mooch off. I'll post back about that soon.

How can I test out the PSU? I've got a multimeter that I could pull out, and just start testing leads. That strikes me as being more deterministic than plugging in a different one and "seeing if it gets better" (which could easily be "seeing if the issue manifests itself in a different odd way).

 

[ghmage]

I agree wholeheartedly, but not everyone has a multimeter, or knows what one is, or could use one without electrocution.

The 12v rail should be in range of +/- 10% from 12v. (Yellow wire is +12v, black wire is ground.)

My multimeter probes fit right into a molex connector, so that's easy. You can also slide the probes into the backside of the big power connector going to the motherboard. Be sure to test under load, such as during a memtest86 cycle.

Even with voltage in spec, a PSU could have wattage/amperage problems, or intermittent burps. So, another PSU could be a good test if required.

If you want to do some Ohm's Law calculations -
http://www.anderson-bolds.com/calculator.htm

Let I = Current (measured in amperage)
then let E = Energy (measured in voltage)
then let R = Resistance (measured in ohms)
I = E/R

 

[kortec]

Sorry for the long lag in response -- I hope you're still watching this thread. It's finals week, in some parts of the world, though, so you know how that goes.

Anyway. I put the RAM from the Problem Child box into my known box, and ran the memtest disc I had on it, and it still was full of errors. To double check that, I ran the memtest with just that box's RAM -- lo and behold, it apparently had the exact same errors! As did three other known working boxes. *sigh*

So once I burned a new memtest disk, and checked it against the ISO md5, I put the questionable RAM back in my testing box and memtest ran for 6 hours without a single error. I put the problem RAM back in the problem box, too, and ran memtest and it looks okay. (This was with just GPU, optical drives, RAM, and CPU.)

I'm going to start testing leads with the multimeter in a short while. I'm now reasonably convinced that the RAM is fine. The only oddity I had was when I booted up my testing box with this RAM. For some reason, it booted fine, but X crashed and rebooted the computer as soon as I launched it -- and some of the display was garbled. No idea what to think of *that*.

 

[Zorg]

... or could use one without electrocution.

You know you can't get electrocuted with 12V.

 

[ghmage]

You know you can't get electrocuted with 12V.

What fun is that!?

 

[kortec]

Oh, no worries there. I've spent enough time inhaling solder fumes to be okay with multimeters and the like.