Z170 extreme6 latches off on suspend-to-RAM

bestsss wrote: Hello! I am experiencing some peculiar issue that's not always reproducible: Sometimes when the computer (Windows 8.1) is put in sleep mode [S3/suspend-to-ram], the PSU (EVGA 650 GQ) appears to 'latch off' and the computer is essentially down. The computer's power and reset buttons do nothing. The PSU has to be unplugged for about a minute in order for the computer to be able boot. It's important to stress the issue is rarely reproducible - i.e. often more than 10 consecutive sleep/awake cycles would be fine. The temperatures do not appear to have an affect i.e. under stress test multiple sleep/wake attempts may succeed, while sleep-on-idle (25min) may latch the system down. The 'latch off' happens upon sleep - likely 2-3sec after issuing the sleep command. Under normal operation I do not experience any other issues. The PSU provides very stable voltages under normal conditions as well. The CPU (i7 6700k) can overclock to 4.6GHz too. BIOS is updated to the latest firmware (P7.00). All drivers are the latest as well. No suspicions logs in the windows event log beside the fact the system has not shut down normally. I've attempted all possible settings of C sleep states to no avail (including the S5 deep sleep, just in case). Loading BIOS defaults, etc. Even tried connected HDMI TV instead of the monitor (to prevent power surge of DVI pin 14). Again the issue is related to S3 -suspend to RAM- that sometimes causes the computer to practically die off and the PSU has to be unplugged for a while (like 1 min for the capacitors to discharge) and plugged again. After plugging in the computer comes online (fans spinning, debug indicator on, etc.) for a short awhile, there appears to run some post test, then switches off -- by the moment power on button is available to boot up normally. That happens with power-off on boot setting in the bios. The next step would be trying a different power supply, however since the issue is not coherently reproducible - it may prove difficult to conclude. I kindly ask you if you can provide any insight what might be causing the issue and if there is any way to debug that. Other Hardware - GPU GTX1070, 32GB G.Skill 15/15/15/35@1.2v, 2666MHz, m2e SSD Thanks!

I had a vaguely similar problem with an older Intel P67 chipset board when waking from Sleep. In my case I would find an entry in the Windows Event Log about, to paraphrase, a driver not responding in the expected amount of time. It was related to the video card driver.

In my case the PC would otherwise start, fans would spin up, HDD could be heard, but no signal to the monitor. That allowed Windows to log the wake from Sleep failure event.

Your theory about the PSU latching off seems possible, given the symptoms you describe when the PC is unable to start at all when you try to come out of Sleep mode.

When in Sleep mode, the +5VSB (Stand By) power rail in the PSU is active, and keeps the PC "alive" and waiting/ready to respond to the signal to wake from Sleep. If the PSU +5VSB rail is intermittently failing, none of the UEFI settings changes or video cable changes.

I have an ASRock Z170 Extreme7+ board that is very similar to yours. I'm using the latest UEFI version (7.10), have an OC of the CPU and memory. I can use CPU power saving options, and enable the ASPM options (power saving settings for PCIe and chipset.) I have never had any wake from Sleep issues with this PC, it always wakes from Sleep fine, simply by clicking the mouse.

I'm using a Seasonic 650W Platinum PSU, FWIW. Your EVGA PSU's OEM is FSP. It has a 3Amp rating for the +5VSB rail, which is the usual maximum for that rail on any PSU. A good review of your PSU found no voltage drop on the +5VSB rail, and found it to be an excellent PSU. But that does not mean it can't have a failure in a component somewhere.

Your intermittent problem is the worst kind to diagnose, of course. One thing to check, normally during Sleep, the PC case power LED flashes on and off, all my PCs do that in Sleep mode. That of course requires power from the PSU.

If your PSU latching theory is true, you should not see the PC case power LED flashing when, and if, the PSU has that fault.

If that LED is always flashing, even during one of your wake failures, the +5VSB rail could still be alive, but unable to allow the board to recognize a signal from the power, reset button, etc.

Do you have Wake On LAN enabled? If it is, it is normal for the PC to start up temporarily and then shut off as you described, when power is removed from the board, and then restored again, IF the PC has NOT been shutdown normally first. That fits the scenario you described. The network chip is initializing itself to provide the WOL capability.

Regardless, testing with another PSU is really the simplest way to remove the PSU as a problem. Much easier than considering different PSU and PC state scenarios, that may lead you nowhere. PSU problems can produce strange symptoms that make no sense, except when the replacement PSU fixes the problem.


Edited by parsec - 08 Oct 2016 at 12:15pm

bestsss wrote: Great answer, a lot of useful info in there. Thanks, parsec! >>If your PSU latching theory is true, you should not see the PC case power LED flashing when, and if, the PSU has that fault.<< Indeed the led foes off. The very first time I experienced the issue it was a morning; the computer expectedly had gone in sleep mode... except the keyboard and mouse didn't wake it up. I assumed it must be some issue related to the USB (cases where the windows powers off the entire USB connector, so the mouse+kbd are off too). But then power off, reset did nothing at all. I unplugged, plugged -- nothing. Decided to open the case and check, keeping it unplugged. Obviously that was enough and soon the power on button actually started the boot process. Currently trying another PSU and luckily - everything is smooth so far (<knocks on wood>). Yet, since the issue is not easily and reliable reproducible, it's too early too call it off. Likely I will request a RMA and update the thread. (Hopefully this could be useful to someone else) >>Do you have Wake On LAN enabled? If it is, it is normal for the PC to start up temporarily and then shut off as you described, when power is removed from the board, and then restored again, IF the PC has NOT been shutdown normally first. That fits the scenario you described. The network chip is initializing itself to provide the WOL capability. I didn't know that, thanks a ton. Indeed Wake-on-Lan is enabled, only for the LED, if the system latches off the led goes offline too. (I am facing the back of the computer so the front LEDs are not easily visible). Cheers!

If the case's power LED is not flashing when your Sleep problem occurs, and you need to pull the AC cord, etc, to get the PSU to come on again, then an intermittent PSU problem seems more likely. I have never seen the case power LED not flash during Sleep, even if you have power to USB ports configured to be off during Shutdown.

You could configure that with the Goodnight LED option in the UEFI, but you haven't done that. Don't do that now, it will ruin your Sleep failure notification system!

I noticed a typo in my first post, or really I forgot to finish a sentence, I'll fix that one line here:

When in Sleep mode, the +5VSB (Stand By) power rail in the PSU is active, and keeps the PC "alive" and waiting/ready to respond to the signal to wake from Sleep. If the PSU +5VSB rail is intermittently failing, none of the UEFI settings changes or video cable changes will not do anything to fix that, as you know.

Besides the case power LED, you could also check this: When the PC is in a functioning Sleep state, if you use a wired Internet connection, look at the small LEDs next to the network input jack on the board's IO panel.

Do you see those LEDs lighting up in Sleep mode? That should happen if you have WOL enabled. The LEDs would also light up/flash when the PC is shutdown, with the PSU on.

Again, if you don't see those network jack LEDs flashing or on during a failed Sleep situation, that's another hint that the +5VSB rail seems to be failing.

Forgot to confirm this, you said the (wired?) mouse and keyboard will wake the PC successfully from Sleep normally, except during the failed Sleep event, right? Yet another clue, that they are not receiving power when that happens.

But it is still possible that the board itself has a component that is intermittent, and causes your problem.

Yes, please let us know how this turns out for you.

bestsss wrote: I've got a replacement PSU - EVGA 850 GQ...  It might feel like a steal (upgrade) but 850GQ is actually larger and the need for extra clearance caused quite a bit of issues fitting it in. Good new is: so far, there appears the cause for the unstable sleep was the original PSU and not the motherboard. Parsec, thanks again for your support! I've learned a bit more about the overlooked 5v stand-by rail. ---- I hope the posts in this very thread get properly indexed (by search engines) since the info Parsec provided (in here) is top notch, very detailed and on point. I do hope if anyone else has similar issue to be able to find this gem of a thread.

Great, glad that is working for you now. Thanks for your kind words. Your description of the situation allowed me to diagnose it, in theory at least, and that seemed to be correct. Unusual for the +5VSB rail to fail on its own, but it does happen.

That rail tends to be built separately from the rest of the PSU, and may be composed of only a few parts, with one or two chips doing most of the work. But unless you shut off the PSU itself, that rail runs all the time, which is usually not a problem for electronic components, but anything can fail.

Thanks for letting us know what happened!