Stiletto wrote:

My gut feeling is that this is a hardware issue, not a software issue - and possibly power supply related.

I think your statement is right on the money. If the issue were software-related, you would most likely see more in the event logs. If it were thermal-related (outside the PSU itself), I imagine a lock-up would happen rather than a reboot.

And if it is the power supply, it needs to be repaired or replaced. Otherwise... one day that board isn't going to start back up.

Check the motherboard and the PSU (be careful) for bad capacitors. 2008 was toward the end of the capacitor plague era, but I still see OEM boards (Socket 775 and AM2) from that time once in a while with swollen or leaky caps.

I'd also test the power supply with a decent tester, even if it doesn't have visible bad caps. Thermaltake isn't known for using top of the line parts and that unit has mixed reviews. https://www.newegg.com/Product/Product.aspx?I … N82E16817153114 I'm also amazed that a 600W unit isn't even 80 Plus certified at all. Most likely it has a problem if they used such poor quality components that it is only 72% efficient.

It could also be related to P states or C states or something. I've seen some issues with certain hardware combinations (PSU,CPU, Motherboard, GPU) crashing, rebooting or not powering on properly sporadically due to power saving settings in the BIOS\UEFI. Its a fairly complicated thing to diagnose though and I've never really made sense of it.

I had an ECS nForce-6ma v3.0 AM2+ motherboard that lost a capacitor by the memory modules, would run for 20 seconds, and would shut off. I eventually got new guts for my main PC on my birthday in 2015.

I have a Dimension 4550 that has 4 swollen caps by the CPU and it still runs efficiently, my Riva TNT2 M64 AGP card has a bad SMD Electrolytic capacitor of the 2 that are on there (the one by the VGA port is dead, while the one by the GPU still lives), and a GeForce4 MX440 AGP needs either a solder reflow or swapping of the VGA ports between the MX4000 PCI and the MX440 (no video, but the computer detects the card without issues), even my GeForce4 MX4000 PCI has a missing capacitor, but it still runs (weird, and the heatsink broke off of it years ago and didn't run hot or overheat thankfully).

Not fun working on computers, especially my daily driver has the SMD Electrolytic capacitors that can leak at any time, even laptops use the SMD caps as well instead of the other caps (for space reasons).

Back around 2012, I replaced the original power supply in a Vaio Pentium 4 machine with one of those 600 watt ThermalTake PSUs from Best Buy. It died inside of 6 months. Thankfully the warranty replacement unit kept the system going a couple more years until I finally retired it. Interestingly, I've had an 850 watt ThermalTake PSU from Best Buy running my home theater PC 24/7 now for over 4 years without a hiccup.

I've never heard great things about Thermaltake TR2 series PSUs. A while back there was a 430w model with dual 80mm fans, which seemed to be very popular. It turned out to actually be a ~350w ATX v1.x unit, hiding behind its 430w label and ATX v2.x-style 6-pin PCIe connectors, and with terrible efficiency. That doesn't seem very promising, and makes me wonder how shady the 600w model in that PC could be.

Alright, we're replacing the PSU, hopefully that solves it. Just ordered a Seasonic SSR-550PX off Amazon, all of my power supply calculation knowledge says we shouldn't be exceeding that.
https://seasonic.com/product/focus-plus-550-platinum/

This is *roughly* how it's configured: https://outervision.com/b/LDT15D

I'd be surprised if it's bad caps on the motherboard, the motherboard was replaced by Dell in 2012. (Could have been "new" old stock, I suppose.) Once I have it out from under the desk again, I'll do a thorough visual check for swollen caps.

I didn't read through the whole thread -- did you try pulling out the IO Crest and Syba cards and using the motherboard controllers? I've had bad experiences with Syba anything, so my body shuddered involuntarily when I read that.

I built a new Xen host for someone based on the Dell T20. Used the approved Marvell SATA controller to connect 2 SATA3 SSDs. Right now they are still in my drawer as it was completely unstable. The server itself is running happily ever since with 4 HDDs. I'd at least try to exclude it.

clueless1 wrote:

I didn't read through the whole thread -- did you try pulling out the IO Crest and Syba cards and using the motherboard controllers? I've had bad experiences with Syba anything, so my body shuddered involuntarily when I read that.

I believe I unplugged the Syba card and the system still rebooted from time to time.

Syba wouldn't have been my first choice but there's not a ton of reliable manufacturers for USB 3.1 PCI-e x1 cards at any rate: USB 3.1/3.0 PCIe x1 cards
It was one of the few on Amazon that put "USB 3.1" in the description so I thought we were ordering a USB 3.1 Gen 2 card, FWIW.
Wrong! USB 3.0 had been rebranded to USB 3.1 Gen 1 by the USB consortium guys, and Syba was first to take advantage of that.

Can't remember if I tried removing the Marvell-based IO Crest card or not. (It benched faster than onboard in synthetic benchmarks IIRC, though I quickly learned to use Microsoft's drivers over Marvell's.)

To answer Jade Falcon, system has onboard Intel X58+ICH10R SATA 2.0 RAID, though it's a bit gimped by the Dell BIOS. The IO Crest card gives it "SATA III" with the Marvell 88SE9215 but it's gimped by coming over PCI-e 2.0 x1 slot, since the only slot with greater number of PCI-e lanes is in use by the video card. According to synthetic benchmarks, it was slightly faster to use the controller card versus the onboard controller. BIOS is "current" (2011).

Did a visual inspection of the motherboard: there were barely any capacitors on the motherboard which were barrel-type electrolytic with the "+" sign or "K" sign that can be easily checked for "dome-ing"/bursting, they were mostly seeming to be polymer capacitors - but those that were barrel-type passed a visual inspection. I don't have a ESR tester or ohmmeter of any kind to electrically test any capacitors in-circuit right now.

Received and installed the new Seasonic PSU on 11/19 and so far so good. Also replaced the CR2032 while we had it open. PSU was installed on 11/19 and even though the rate of rebooting had been increasing 11/11 - 11/18, they've dropped back down to zero over the last five days. I think we've solved it. I'll wait another week or so and if we can't find proof of purchase of the Thermaltake PSU, we'll take it apart to look for domed/burst capacitors. 😁