I bought this NUC6i5SYK a little over 2 years ago, and it's been working great. However, yesterday I unplugged it for a few minutes, and afterward it refused to power on (no light or fan).
After enough times unplugging/replugging it and hitting the power button, I finally got it to power on, and I see the message about the CMOS battery being dead. While it was plugged in, I was able to reliably get it to power on, but as soon as I unplug/replug it, it goes back into the state of refusing to power on.
After finding some other forum answers recommending to unplug the CMOS battery for 30 seconds, I was able to open it up, disassemble it (which is a PITA, btw), and unplug the CMOS battery. After putting it back together, I was able to power the NUC on again. Of course, once I unplugged AC power and plugged it back in, it goes back to refusing to boot.
Just to confuse the issue further, the NUC will readily boot (even after unplugging and replugging the AC adapter multiple times) if I completely remove the CMOS battery. If I reinstall it, it will boot the first time that I plug it in, and then it goes back to refusing to boot afterward.
Knowing what I know now, I'm fairly sure that I've needlessly replaced another NUC unit (also a 6th gen) that had this same issue (after even less time), which is quite annoying.
So I'm now left with a few questions:
- how/why does the CMOS battery die so quickly, especially when the NUC has been plugged into AC power for most of the time?
- why does the NUC refuse to power on when the CMOS data (apparently) gets in a weird state?
- what can I do to prevent this situation?
- why would I ever want to buy another NUC knowing that I'll likely run into this same issue?
I have worked with literally *hundreds* of NUC units and only once (testing with a DE3815TYKE that had been sitting on a shelf, in a sealed box, for a couple of years) have I seen a dead battery. In that case, the dead battery did not affect the unit's ability to boot, however. Regardless, I would certainly not say that you have enough of a correlation to suggest that this could be the cause of other folks issues.
Ok, that said, let's answer your questions...
- As I said, I have worked with literally *hundreds* of NUC units and I have only seen one dead battery. In theory, while plugged into A/C power, the battery is not being accessed at all (standby current takes care of maintaining the CMOS state). It's an alkaline battery, so it's not being recharged. So, why did your die so quickly? Well, first of all, it wasn't so quickly; you said that you've had the NUC for over two years. Secondly, let me ask: how many A, AA, AAA, C, D, 9V or plethora of watch batteries have you purchased that ran out of power faster than you expected? I have seen plenty. Shiite happens.
- I have worked with or on BIOS for a lot of years, yet have never understood why the AMI BIOS cores have had a problem with CMOS state. In theory, once they had made the switch over to flash-based firmware hubs, their use of the the CMOS memory could have been eliminated. Like other BIOS cores, they continue to use the overall state (set or reset) of the CMOS to determine whether the BIOS should be initialized from scratch (and thereby getting around the effects of BIOS Configuration corruption, for example). Still, there is something more going on -- and it isn't isolated to the NUCs; I have helped people with similar issues who are using motherboards from many of the other ODMs/OEMs.
- If I only knew this.
- If you want to blame the NUCs, I can't stop you. As I said, I have helped folks with the same issues with other ODM/OEM motherboards as well. If you want to make what I think is a ridiculously stupid, completely rash decision, that's still your prerogative. If I had a dollar for every time I have run into someone who finds that, gee, the grass really isn't greener on the other side, I would be a rich man indeed.
That's my say - and no, I am not interested in your rebuttal.
It's a fair point that this problem may not be exclusive to the NUCs. However, out of the 4 NUCs that I have owned, all 3 6th-gen models have had issues. The first had a bad ethernet port (and was replaced with the 2nd one), and the 2nd (the one that this post was about) and 3rd (the one I mentioned that I had replaced) both ended up in a state where they would not power on after a loss of power. I do not know for sure that the 3rd one suffered from the same issue that I describe above, but I have my suspicions. In general, I'm just not really impressed based on my experiences.
This issue in general was quite weird. Through many rounds of testing both with and without the CMOS battery plugged in, I can fairly confidently say that the state of the CMOS battery (perhaps putting out less than the design voltage) is somehow causing these issues. I've now completely removed the CMOS battery, and it's happily running again (with the date reset to 1/1/2015). I've ordered a replacement battery, but since it's a bit of a pain to replace and ntpd is a thing, I'm probably not going to bother to replace it when it arrives.
I'm not entirely sure what answers I was really looking for. I realize now that most of my questions were rhetorical and more meant for me to vent than anything.
I was careful in how I worded my response because there are certainly things that would easily turn into *very* long conversations. I am a former (now retired) member of the Intel NUC and Compute Stick/Card team and was also a longstanding member of the Intel Desktop Boards team from which the NUC team derived. I am a software engineer, but during the course of my career, I have been involved in board and chip design, thermals and acoustics (yes, they did call me "fan boy"), BIOS and ME firmware development, etc. When it comes to BIOS update processes, I am an expert. None of the issues that you describe (well, other than things like your dead Ethernet PHY) have ever happened to me. The only time that I have run into a corrupted BIOS and an inability to boot has been when working on preproduction BIOS releases. Once they achieve production, I have never had an issue. So why do so many NUC customers run into these issues? If it doesn't happen to me, someone that knows what he is doing (give or take a few, um, "senior" moments), then it must be the result of user inexperience, right? Since I know for a fact that folks "in the business" (namely folks involved in the Integrator business) have also seen these issues, I can't blame it on inexperience. What I have to conclude is that either I am just awfully darned lucky (I'm not, as folks who know me will attest) or it is the result of a non-robust update process. This is what I think Intel needs to be working on improving...