Sleep and 2nd HD breaks warm reboot

[jbb1003]

EDIT - changed PSU and everything worked fine.

Not really sure where to put this, so trying here...

Basic system is:

Gigabyte Z77-DS3H rev 1.0 with F9 BIOS
Intel i3570K
BeQuiet 430W PSU
320GB SATA 2 drive (Chimera 1.11 plus Mountain Lion on here)
TP-Link WDN4800 WiFi
Onboard HD4000 graphics using DVI
8GB Corsair Vengeance RAM, but currently running as Normal profile.

BIOS settings - cleared CMOS then loaded Optimized Defaults. Set AHCI.

With this setup, everything is fine. Sleep (hibernatemode 3), multiple sleeps, autosleep, and shutdown/restart all work.

If I add any of a SATA 3 drive (tried 2) or a SATA 2 drive, then the everything works fine except for the following sequence:

Boot, sleep, wake, restart (either sw restart, or Reset button). The system shuts down, the screen goes black, and I never get a POST beep. The HD keeps spinning up and shutting down.

If I remove the additional hard drive, and add a SATA 2 SSD, then the above sequence works.

If I use just the SSD and a SATA 3 drive as a fusion drive, I get the same problem.

Any suggestions? I'm stumped. Have tried every BIOS setting there is, moving SATA ports around, and mucho googling.

 

[jbb1003]

I had a working theory that I need to test a bit more and suffers from a lack of knowledge

When the system goes to sleep it puts all drives to sleep.
When the system wakes up, it only wakes up drives it needs. Drives it doesn't need remain in the sleep state.
When I restart, drives which are in the sleep state will cause a POST failure, with the exception of SSDs (what could possibly cause this?)

This theory fitted the facts for the following use cases:

Two HDs. I woke up, and then forced an access to the second drive. It POSTed on restart.

But failed for this case

Fusion drive. I woke up, and then mounted partition 3 on the HD. It failed to POST on restart.

So something about the theory is wrong, or mounting a partition doesn't properly wake up the drive, which I find hard to believe.

It feels like I've reduced this to either a motherboard or a software problem. The only other thing that just occurred to me is I installed the Marvel eSATA kext - I'll try the SATA one.

 

[PIETROMETER]

How curious?

do you have s3 acpi mode sleep enabled in bios? i would go with a hardware issue on this one but its still very strange?

good luck

 

[jbb1003]

How curious?

do you have s3 acpi mode sleep enabled in bios? i would go with a hardware issue on this one but its still very strange?

good luck

No BIOS setting for ACPI mode.

Also tried manually fixing the DRAM voltage at 1.5V - and also fixing VTT at 1.2V (I'm not overclocking). Neither fixed the problem, though the DRAM voltage may have improved it (reboot does occasionally work).

Very occasionally a power switch boot fails too, when I power off following a restart failure and then power back on again. However, I suspect that's because I'm immediately powering back on, so it's not a true cold boot.

 

[jbb1003]

I can reliably reproduce this issue with no hard drives in at all if I boot, wait for the POST beep, power off, and power on as quickly as possible. It then fails to POST.

That said, it feels like electrical abuse - does it actually demonstrate anything useful? i.e. is this something that should work? If I wait 5 seconds between power off and power on, it POSTs just fine