Are there any problems when shutting down a Mini PC?
Are there any problems when shutting down a Mini PC?
OP already indicated that he monitored the system under various conditions and that it never exceeded thermal compliance specifications. Obviously, that was the first thing I indicated as being the mostly likely cause but if the OP says it hasn't broken the T spec then we have to assume it hasn't unless and until there is evidence it has.
I had a thought and not sure why I didn't think of this sooner. When the screen went blank, I thought the pc was still running(could still hear fans running) but lost signal to monitor for whatever reason. So I had music playing on youtube and ran OCCT again and when screen went blank, music stopped playing as well. Why is it the fans continue to run but also can't hold power button to shut down?
I have a new power adapter coming for it. But don't have a good feeling that will fix it. Also waiting on paste and will replace thermal paste and see what happens.
EDIT: I decided to open it up and take a look. Now I am no expert but something doesn't look right. In the pic, the round thing in black, that is the cmos battery correct? This is below where the nvme sits. If that is the cmos battery, why is it covered like that and looks like that black cover has melted onto the motherboard. Also looks like there is a burnt mark on top of the wifi card.
Well I'd say that was factory and probably with a device like this, they had no intention for Joe Blow to be taking it apart or working on it internally. Just like many devices these days, if something is disturbed, regardless of laws allowing repairs outside the company's control, they still do things to check if anyone has accessed it or made changes. This helps them stay ahead in avoiding costs. Not too confident about that mess on the wifi card either—it seems like some internal work was done poorly.
Yes, that setup involves a CR2032 battery wrapped in a PVC heat shrink insulation and secured onto the motherboard using black adhesive to prevent it from moving and possibly detaching the wires. If I needed to replace it after five years and feared damaging the board, I’d keep it in place and secure a new one nearby with tape or glue.
https://www.amazon.com/ZAHARA-Batte...3-...Y-fGLDQS3T
The same principle applies to the WiFi card. The black adhesive is there to keep the tiny coaxial cables (WiFi antenna) from coming loose. These cables can be tricky to reinsert if not handled carefully and are prone to breaking if you’re careless. I haven’t seen WiFi cables permanently glued to a board, but it’s a practical solution to avoid future troubles.
https://www.globalwellpcba.com/wifi-pcb-design/
Summary: Everything seems fine to me; focus on finding the real issue elsewhere.
Moving forward from that point, both memory sticks were taken out and only one was reinstalled. The covers weren’t reattached properly; they were just placed back without being secured correctly. I began the process again, executed OCCT as before, and it operated smoothly for the entire hour without any interruptions or errors. Temperatures stayed below 78 degrees. This suggests an issue with the other memory stick that was removed.
I reinserted the second memory stick, but faced some difficulties getting it to activate. It reminded me of previous problems when it crashed. Eventually, I removed the second stick again and continued trying. After disconnecting the power brick and clearing the CMOS a few times, it started working once more. Once it was running, I ran OCCT once more, this time with both memory sticks installed, and it functioned flawlessly for the full hour.
Regarding the recurring activation problems, it appears I encounter difficulties when unplugging the power cable. It seems I need to wait a certain period before turning it back on. However, other times plugging it in immediately results in it powering up without me needing to press the power button.
Beyond replacing the NVMe drive and memory sticks, the only notable change is that the inside cover isn’t secured and the outside cover remains off. This might improve airflow, but based on temperature readings, overheating hasn’t occurred unless there’s an issue not visible in hwmonitor or hwinfo.
At this stage, I haven’t replaced the CPU thermal paste and will wait for it to settle before proceeding. The next step will be reattaching all covers and running OCCT once more.
While testing OCCT, I observed something unusual according to hwmonitor: during its operation, the two 2-core processors ran at around 900 MHz, while the 4-core cores operated near 2700 MHz. After OCCT stopped, the 4-core cores dropped back down to about 900 MHz. My Windows power plan was set to Power Saver at the time, which I initially thought might be related to heat or power delivery issues—though it still crashed under that setting. However, during OCCT, the 4-core speeds were significantly higher than the 2-core speeds. I wondered if this discrepancy would explain the instability, but it still caused a crash even when using Power Saver.
I suspect companies sometimes make adjustments to prevent customers from incurring repair costs. I don’t expect the company to cover these issues after just three years and a one-year warranty.
EDIT: I switched to a high-performance power plan, reran OCCT, and it crashed within minutes. The highest temperature recorded was 88°C—still well below thermal limits.
EDIT2: After removing one memory stick and switching to a high-performance plan, the system still crashed shortly after a few minutes.
I have experimented with another power cable that matches the original one. It didn’t resolve the issue, the device still crashed. I cleaned the old thermal paste and applied Arctic mx-6 paste, but it failed again. I removed the board completely, took out the Wi-Fi card and used a different memory stick. I powered it on with the board removed from the case. Running OCCT for an hour without any crashes was successful. During this time, temperatures stayed lower, around 66 consistently, whereas previously they reached up to 88 but remained steady in the mid to upper 70s. The lowest reading was in the low 70s. Once testing finished, I restarted and let it run for several minutes before stopping it. It would crash after about a couple of minutes. After connecting the Ethernet, it operated smoothly for several minutes without any crashes. This was still with the board out of the case.
Some people recommend running more than one complete Memtest86 pass, with each run usually taking several hours. I leave the machine running MemTest86 overnight, or start it at breakfast and end the test after supper. Even one error means you should investigate further.
There's also the possibility it's a bad IMC channel in your CPU, but I suspect a bad DIMM or an unsuccesful XMP overclock, if the fault continues after repasting.
If CPU World is to be believed, the i5-1235U should be capable of running at up to 100°C, at which point it should throttle.
https://www.cpu-world.com/CPUs/Core_i5/Intel-Core i5 i5-1235U.html
Electrical / Thermal parameters
Minimum/Maximum operating temperature
?
0°C - 100°C (junction)
Thermal Design Power
?
15 Watt
On the other hand, my i7-13700H laptop shows a tendency to hang when the CPU is pushed up to +100°C, running intensive tasks in a Hyper-V VM. The underlying OS remains stable, but the VM stops running.
Memory shouldn't be assumed to have passed Memtest86 unless you've achieved FOUR COMPLETE PASSES across all included tests. I've experienced memory kits that performed well during the initial three passes but failed the fourth when the memory was warmed up. Any incomplete test or error should count as a failure. A failure might stem from faulty memory, incorrect settings or timing, damaged CPU pins, defective motherboard, or an unstable power supply.
I also reinserted both sticks and performed memtest86+ for more than 10 hours, with 15 passes and no errors detected. Could there be an issue with the memory that isn't showing up in the test? I might try swapping the sticks and run the test again.
Additionally, when attempting to shut down normally, the computer enters the start and shutdown screens without closing properly. The screen turns black, but the power button remains illuminated and the fans continue to run, just like during a crash. However, I can hold the power button down to force a shutdown.