I use 4x Huawei R4875G1 and in my opinion they get too hot when delivering high power.

The garage ambient is between 27°C-29°C (light blue curve). From 14:30h they delivered 50A per unit (out of 75A max.) which is about 2750W at around 54V. The rectifier fans were spinning at 6700RPM at this time. The highest peak was when the garage door was open for 30 minutes with an outside temperature of 38°C (which increased the garage ambient by 2°C).

All 4 units are close or at 80°C and I don't know if this is healthy for longer duration. I've noticed that the fans are only speed up to higher RPM if the current limit will be increased.

I found the following default (auto) RPM for different AMP limits (per unit):

AMP RPM
===========
20A 3000
40A 5100
50A 6700
60A 12000
75A 15000

I did NOT notice an automatic RPM fan speed increasing while the units getting hotter. So I'm not sure if the "fan automatic" mode really taking the current temperatures into account.
Because I'm using a DIY ESPHome program to control the rectifiers, it would be easy for me to set RPM speeds by myself regarding their temperatures - but why is the "automatic" fan mode not working like expected.

Any ideas?
=========
answer:
The temperatures of my R4875s also reach 80 °C before the units bother spinning up their fans further, which really only happens when they are under fairly heavy load. The maximum ambient operating temperature for these units (at full output, not derated) is 55 °C. If they were not running at 80 °C internally, they would not shed any meaningful heat at a 55 °C ambient.

@fmeili1 I do however agree that the inbuilt control (internal firmware, not esp32 firmware) is a little on the crude side:
Set current to near max, or drop the voltage to something silly, the fan goes nuts,
but as long as the internal temperture stays under 85C the fans dont really change speed.
As you said, it wouldnt be hard to code something more sophisticated without the need for extra sensors even)


I’ve come to the conclusion that the R4875G units are simply designed to operate under these conditions and at those temperatures. What we might consider “running hot” is what we expect from consumer-grade supplies, not from industrial units such as these. They use expensive, heat-resistant components but reduce operating costs by lowering the need for additional cooling from cell-tower-cabinet supplied feed air.


Many people have designed “cooling solutions” for these devices, but if you look at how they are deployed in the wild—boxed tightly into racks with the only real cooling provided by their internal fans—you’ll see the design intention. Mine already have over 45,000 operating hours, and I’m no longer concerned. I’ll report back if and when they reach 60,000 hours. (once in a while, I do run them for hours at full output but mostly at around 50% load at ~25C ambient)


source:
https://diysolarforum.com/threads/diy-chargenectifier.56329/post-1560146