Blog

[ do default stuff if no widgets ]

Archive for July, 2019

Cases Shipping

Wednesday, July 24th, 2019

Cases came in and we’ve started shipping them Monday. Orders are flying out, and customers should start getting email notifications with tracking.

Kodi cases are shipping to us this Friday, and will go out as soon as they arrive. Because both cases are so different, their manufacturing processes are also distinctly different. One had to be done before the other.

Orders were supposed to start shipping out about a week ago. However, as soon as I got the first units back, I did a lot of thermal testing.

This new pi runs hot. I started testing with various benchmark scripts, python, etc. But results were generally inconsistent across tests. Much depended on the initial starting conditions. Should the pi have started cooler, I’d see much different results than if I had it running idle. If I just finished a test, results were different as well.

I tried other tests and settled on cpuburn-arm.

Results were consistent. I felt like I saw the most consistent and highest power draw. Which means we could theoretically get temp as quickly as possible.

Without any case, in 26-27C ambient, I got a naked raspberry pi to nearly 90C in 5-10 minutes. Okay great. Now with the Flirc case, I get it to 80C in 25 minutes. I saw the temp hang around 80C +/- 2C for 30 minutes. Seems like we hit steady state, but I think we can do better.

So we started studying the design, the gap, and the thermal material, and simulated. I ordered some new thermal pads and did some more tests that looked great. I got some arctic cool pads off amazon with a much higher thermal conductivity, and started testing.

In my lab/warehouse 26-27C which is really hot, it took 90 minutes of cpu burn before it got to 80C. Ambient matters. I ran another test on Monday, where it was 24-25C, and in 3 hours, it never went above 77C.

But let’s talk about something important. cpuburn is meant to burn as much power as possible, it’s not realistic of any situation. Yes, cpu utilization could get up to 100%, but that doesn’t mean the processor is going to burn the same power. Cpuburn is meant to try and draw the most power. Depending on the software that’s running, power will not only be different, but it will be dynamic. Even running cpuburn, on stopping, I saw a 1C drop every second. So in actuality, a real load alleviates the stress on the cpu, allowing it breathing time to cool.

I’m confident in the case and the new pad. Confident in saying that any strenuous situation will be fine without the need for active cooling or worrying about overheating.

I got in touch with arctic directly, and were shipping with their pads.