I miss ambient temperature in the article, and testing methodology too. Hard to compare then. If they wrote how many degrees over ambient temp, I could compare it to CooliPi 4B.
Also note that the USB3 bridge firmware version plays a role. The latest ones lower idle temperature of the USB3 bridge more than the former, and probably limit bandwidth of small, fragmented transfers over the USB3 bus.
Any case with the bottom metallic part continuous will attentuate WiFi signal. It's much better to make it two piece, the upper being metallic, connected to the two most power hungry chips - USB3 and BCM and the lower part plastic, to let microwaves go through. Using an external antenna might void microwave certification if it leads to more concentrated beam in any one of the directions.
The PMIC circuitry is the third most heating element on the board. The temperature of the PMIC limits current that can be supplied to the BCM chip, essentially limiting total power before it reboot spontaneously. If you load all 4 cores and overload the PMIC (which happens only with overclocking), it heats up (the PMIC) and drops its output voltage for a short while. It causes Pi reboot.
Some nice thermal images of the PMIC are here:
https://www.coolipi.com/Overclocking.html
The rest of the chips is directly cooled, so nicely shows how the PMIC heats being nearly thermally isolated.
To wrap it ip, why remove heat - because it not only limits the max frequency (leading to downclocking - thermal throttling, also not mentioned in the article if it ever happened during their test) but also shortens the lifespan of all the chips - approximately in half for every 10°C hotter. Note that your SD card is at this temperature too, being next to a BCM chip and the PMIC circuitry. So, having the temperature at 67°C or 34°C results in about 8 times worse lifespan using this rough method. It may not be that high for SD cards, because they tend to write with less current at elevated temperatures, but anyway... capacitors suffer the most. The electrolytic ones definitely do, hopefully arent' on the Pi at all. And with overclocking, you want the lowest temperatures because it may hang up just because it's hot.
There are norms for surface temperatures to prevent finger burns. See BS EN 13202:2000 or UL 60950-1. A case that does not contact the chips will not likely reach extremely high temperatures, because thermal throttling will have prevented it sooner. Some cases with a pillar over the hot chips might theoretically have a hotspot on the outer side in a place where the pillar meets outer shell, if the case itself is thin (perhaps FLIRC, but it covers this hotspot with a rubber, because of aesthetics). And then there are cases like CooliPi which distribute heats using massive metallic heatsink, essentially disallowing these thermal hotspots to show from the outside.
You can see the thermal hotspots in this review made by Gareth Halfacree for the MagPi magazine:
https://magpi.raspberrypi.org/artic...aspberry-pi-4-thermal-cases-tested-and-ranked
I get about 11°C higher temperature of the BCM chip over ambient temp at idle, RPI4 overclocked at 2GHz. Attached HATs don't block airflow, because they're from a side ;-)
Facebook
Twitter
Instagram