After receiving my Pocket Reform, I noticed some heat-related issues. The main annoyances were CPU throttling and a sudden drop in WiFi speed at times. I suspected that the primary cause was the PCB lid. Although it has two copper layers, the insulating base material (FR4?) in between likely contributed to the heat retention.
My idea was to replace the lid with a simple aluminum plate. After discussing this idea with others on IRC, I felt a bit discouraged due to concerns about potential WiFi shielding. Nevertheless, I decided to give it a try.
To my surprise, it was a complete success. The temperature dropped significantly, and the WiFi remained functional. I decided to do some measurements to quantify the improvements.
I tested the WiFi speed with the different lids using iperf from the same position in the room (my desk). The temperature was tested by streaming a 1080p video on YouTube with chromium and hardware acceleration turned off.
The results were quite good. The temperature dropped by 30 degrees and the WiFi seemed to be even more stable than before. It appears that the plastic cover of the ports on the side is enough to allow the WiFi signal to propagate into the laptop.
I customized the lid by engraving it and spray painting it with UV reactive neon colors. The first prototype was a pure aluminum plate while the second one was a black powder coated one. Notably the coating had no impact on the temperature.
For the sake of completeness here are my testing conditions:
Test Video Norway - Youtube
Kernel: 6.10.9
Thermal pads applied to CPU and WiFi chip: Amazon
Aluminum plate thickness: 1.5 mm
Link to the CNC files: On github
A big thanks goes to my boyfriend since he did most of the work and milled all the parts!
Wow, I tip my head to you and your boyfriend! I love that you collected and plotted the data to have some empirical data to draw a conclusion of your project. And I love to see all the photos you took and shared with us. Amazing project. It really stands out!
This was my redneck solution to the problem. I thought temps were pretty good with the banana pi cm4 but when I upgraded to the rk3588 I found the lid getting very hot. I had a bunch of heatsinks laying around from classic reform so I reused the heatsink from the cm4 kit and double sided taped it to the lid with some thermal compound underneath. I don’t have any pre or post graphs but system temp is around 45-50C at idle and seems to max out around 70-72C after playing Half-life 2.
I had a solid aluminum lid cnc’d by jlcpcb’s cnc machining service, cost me about $25-30 shipped, and I’m anxiously awaiting my pocket reform with the rk chipset to try it out on. I also got it engraved. I’ll be happy to provide the files once I actually get a test fit in, but the specs I had it machined to should (hopefully) fit. I’m hopeful of the performance given your benchmarks, thanks for the data points!
That sounds very exciting! Will JLCCNC countersink holes at that thickness? (I’ve looked at a few other services and it doesn’t seem to be an option for 1.2mm.)
PCBWay, Xometry, and some others can do milling from CAD, with custom finishes. (If you make a plate in CAD please post the step files, I’d love to do this upgrade as well.)
A plate with full-width fins would probably reduce temps even more.
Yes, they countersunk the holes for me. I just hope that a chamfer is enough to get the job done, part of why I’m hesitant to post the files just yet is I can’t test for fit, but I will as soon as I receive my pocket reform.
I’ve been satisfied with PCBWay. Messing with the files on KiCAD, I found a 0.9mm chamfer works well for the screws. 1mm left the heads a little recessed, which I didn’t like.
