STK2MV64CC locks up during picture manipulation

[GLong1]

Hi

We are testing a STK2MV64CC stick for digital display application using Linux Mint. The OS is installed and the bios to 0058. We can play a movie trailer from storage but the system totally locks up if we resize a static picture or move it within a document the device locks up totally. Needs the power pulled or 5 sec hold on the switch.

Will do more testing of other functions but to date seems to be just graphics.

[n_scott_pearson]

My gut reaction would be that the processor is overheating. Use AIDA64 or something similar and monitor the processor temperatures across these operations. Since the processor cannot represent temperatures above its Maximum Junction (Tjmax) temperature, what you want to watch for is readings reaching and staying at Tjmax for some period of time. Since the Tjmax for mobile/embedded processors is typically 100c, this is relatively easy to watch for (though the time that it takes to rise from that temperature to the thermal shutdown temperature can vary significantly).

Presuming that you do see this occur, the next step is looking at causes. For starters, I would suggest that you verify that the fan is continuing to run in the unit (Aside: my recommendation is to ensure that the fan is always on; do not leave it in the mode where it can be turned off).

...S

[GLong1]

Hi Scott

Looks like this is close to the mark. Thank you. Did some hardinfo tests. Before setting the fan to M2, did complete set of tests, including a graphic stress test, but the stick locked up trying to save that report.

After changing M1 to M2 got the following temperature readings following each test run. What will be interesting , if we deploy this unit, is it will be operating in the tropics and how much dust before the cooling is ineffective. It may not be durable as a display board system. Maybe needs to be mounted on a heat sink to carry the heat away. Also found it MUST have its own PSU. Other sources via a USB C cable are not enough to install an OS!

SensorRun 1Run 2Run 3Run 4acpitz/temp1 27.80°C27.80°C27.80°C27.80°Cacpitz/temp229.80°C29.80°C29.80°C29.80°Cpch_skylake/temp156.50°C46.50°C45.50°C63.00°Ccoretemp/temp168.00°C52.00°C55.00°C68.00°Ccoretemp/temp2 58.00°C51.00°C53.00°C67.00°Ccoretemp/temp368.00°C52.00°C56.00°C68.00°Ciwlwifi/temp161.00°C50.00°C49.00°C65.00°Cthermal/thermal_zone256.50°C46.50°C46.00°C63.00°Cthermal/thermal_zone027.80°C27.80°C27.80°C27.80°Cthermal/thermal_zone363.00°C53.00°C56.00°C68.00°Cthermal/thermal_zone129.80°C29.80°C29.80°C29.80°Cthermal/thermal_zone461.00°C51.00°C49.00°C65.00°C

[n_scott_pearson]

Well, these kinds of units are, in general, not recommended for locations that are very hot, very humid, poorly ventilated or very dusty. I was thinking that maybe you could add some sort of dust filter, but the reduction this causes in airflow may be a killer itself. Ideally, if your solution is going into tough environmental locations, you need to use a base product that is designed for these kinds of environments.

OMG, don't do that! The cable into the CC Compute Sticks is NOT a USB-C cable! It is a custom cable that just happens to use the (physical) USB-C connector. Connecting any other cable is potentially going to fry the stick. If you decide you really want to use the Compute Stick for your product, you may want to get an NDA with Intel, so you can get the connector spec, and then design your own power supply.

...S

[GLong1]

Thanks Scott

It would probably work really well in Alaska but not the real world in the tropics. Think I will switch back a a basic NUC unit for the job.

[n_scott_pearson]

I would look into some of the fanless chassis that companies like Akasa offer for the NUCs.

...S