I broke out the FLIR camera this morning to see how the RPi5 amp/PSU performed under load. The short answer is “fine"; the longer answer is below.
All of these tests were conducted in still air at about 22°C. The addition of a small fan would make a big difference, and I’ll re-run these tests later with one.
First thing, of course, is to get a cup of tea:
Next was to add a 2A load to the PSU side, in the form of 2 x 5Ω, 50W (when bolted to a massive heatsink) resistors in parallel. Running a kernel compilation (make -j 8 Image.gz) across all cores resulted in this:
The CPU was being throttled at 85°C; the board around it was at about 70°C, and the hottest part of the PSU (the switch-mode chip) about 65°C.
Finally, this is where I got to when running 5W per channel into load resistors with the additional 2A load still present on the 5V rail - it’s interesting to note that the hottest thing (at about 95°C) is the load resistor at the bottom of the image.
The RPi5 CPU looks cool. It isn’t; this is a result of its having a metal casing which has a lower emissivity than the PCB around it, so, for a given temperature, it emits less IR radiation than the board.
Now, time to measure its performance: