Is 11.93V close enough to 12?

[trents]

It is more than a general rule, it is the ATX specification (p. 22).

Also, just unplugging it and pressing the power button will not drain all the power in the caps. The (main?) cap holds a charge quite a while IIRC (I am regurgitating a post from BobN I believe).

Then why don't the fans keep spinning and the lights stay on? Caps store juice and that juice would seem to be drained away when the power button is pushed after the unit is disconnected from the wall and that's why the fans quit. Am I missing something?

 

[HankB]

Then why don't the fans keep spinning and the lights stay on? Caps store juice and that juice would seem to be drained away when the power button is pushed after the unit is disconnected from the wall and that's why the fans quit. Am I missing something?

Just guessing ...

1) The PSU requires a signal from the mobo to remain on. As voltage drops that signal could go away before the caps are drained.

2) The fans do not run off all available voltages and would not drain all of the caps.

3) The PSU is designed to provide specified voltage levels and will shut down the outputs before the outputs drop below specified levels.

 

[HankB]

Home stretch! I think I see the finish line.

After Chapter 2.

I replaced the Mobo with an ASRock Q1900-ITX One side benefit of the upgrade is that it had 4 SATA ports and allowed me to use an old 160GB laptop drive as the boot/system drive. The two big drives (2TB, 3TB) could be devoted 100% to data storage. This gave me an opportunity to deal with something that had been nagging me. The older 2TB drive is one of those that reports the physical sector as 512 bytes even though they are actually 4KB. I'm pretty sure I did not have proper alignment on 4KB boundaries and took this opportunity to fix that.

Before buttoning things up, I decided to run some benchmarks. Since this is shared via NFS I ran the benchmark (bonnie++) from another host - or tried anyway. Before the benchmark completed, the NAS locked up. I repeated this enough times to convince myself it was not a fluke.

Thus began chapter two. The first thing I did was to install LAN firmware that Linux was warning might be missing. No joy. Next I updated the BIOS. Joy! I've run the benchmark 5 times in a row w/out difficulty. With the previous BIOS it did not complete a single time. My guess is that there was some sort of issue between SATA and LAN. I was able to copy ~1TB contents from one drive to another without any difficulty (on the old BIOS) and that ran a lot longer than the benchmark.

Everything seems to be stable now. I still want to do a couple things before I sign this off. First I want to run Memtest86+ for a few hours just because. I'm not sure there are any other diagnostic tests available but this is a good starting point. The next thing is to do some cable management. I have some short SATA cables ordered and will start with that. I also need to bundle up the unused PSU cables since I was too cheap to spring for a modular PSU. Then if it looks good enough, I'll get a picture to post here. And I'll get wattage readings at the wall. Right now is is running about 30W running Memtest86+ which is about as high as I have seen it go. When OS is running and the drives are not active, I think it will be lower. That's a win for the new PSU as the older system ran about 50W.

 

[HankB]

OK... Not too fancy. I've seen a lot nicer cable layouts. The extent of my effort was to buy some short SATA cables so I didn't have a lot of extra cable to deal with and to bundle up the extra PSU cables to get them out of the way. Nor is it a terribly fancy case. It's about as plain as they come. But it does do the job and allows plenty of air space around the components. At the moment the drives are all at 33°C and the four cores at 34-35°C.

This is now done!

Thanks again for all of the suggestions.