I’ve watched through your video on the Argon One case: my immediate impression is that, given the problem was intermittent, it’s probably power-supply related. If there was an incompatibility it should never be able to successfully boot, while at 1:20 in the video it manages to get as far as gdm3 starting.
Your readings from the power supply are particularly interesting. The voltage reading as it’s entering gdm3 (around 17:36) dip to 4.81V which is getting perilously close to the brown-out level of 4.7V and makes me wonder if, with the Argon case (and its built-in fan), it’s just pushing the power supply into the 4.7V territory, so you’re sometimes getting away with it, and sometimes not. Note the later reading on the successful boot without the case; the voltage only dips to 4.86V. Even if the power-supply is happily capable of delivering 2A+, it matters that the voltage doesn’t drop (too much) while it does so. It also doesn’t surprise me it’s dying at the point it does: the startup of the windowing environment is quite stressful, power-wise, as is the early boot sequence when systemd starts firing up all the services in parallel. These are the two places we typically notice marginal supplies failing.
I’d be very interested to see the
/var/log/kern.log contents from one of the successful boots when the Pi is in the Argon One case. Even though the boot may be successful, if there are power-supply issues I’d expect to see some warnings appearing the kernel log (even if the lightning bolt doesn’t appear – and I can’t recall if it does when “full”
kms is in use – its appearance can be too brief to discern in some cases, but under-volt situations should nonetheless be reported in the kernel’s log). The definitive evidence would be to enable serial on the image and hook-up a serial console; even in the event of a brown-out that kills the boot that tends to manage to report under-volt warnings to the serial console first. However, it’s probably sufficient in this case to attempt a successful boot and have a look in
A quick note on the moving of SD cards (mentioned at ~10:00 and ~12:30 in the video): as you’ve noted there should be no issues moving SD cards between Pis (of any supported model). That also applies to the case of adding and removing HATs, so there shouldn’t be any issues with moving the card to/from a Pi within the Argon One case.
I haven’t currently got an Argon One case, but they do seem to be a fairly popular thing, so I’ll stick one on order and see if I can replicate this issue when it arrives.