Lionel RUN

Welcome to the forum.
Can you give the exact model number, please? If it's a rebranded Axpert (they can have many names), it is possible to re-calibrate the battery voltage reading using special commands. But you may not need to.
With higher loads, there is a higher voltage drop across the battery cables.
Are you measuring the battery voltage directly at the battery terminals?
Is your multimeter trustworthy?
If the reading drifts with time, temperature, or moisture, then there are specific tiny resistors inside that can be replaced to stabilise the reading. Then you can perform battery voltage measurement calibration. But if it's a fixed error percentage (e.g. always reads 7% high or 4% low), then this can be adjusted by calibration and you likely don't need to replace the resistors. All the above assumes that it's an Axpert; if not, you'd need to find a schematic diagram or service manual for your brand and model of inverter.

Hi Green Bum, Hi everyone.
Sorry for huge delay because of emergencies (+ mild COVID) recently.
Yes, the replacement worked great.
I changed only one of the two fans : i.e. only the "dead" one (wich had a true "speed feed back failure"). Even if it was still running marvelously. https://eu.mouser.com/datasheet/2/6/AD0812XB-A73GL(N)-A-1155697.pdf
I replaced it with a 9.2 cm diameter fan (because I coudn't find 8cm anymore). The important thing to be respected here  is the nominal air flow only (not the RPM indeed). Here, 68CFM for the spare one VERSUS 57CFM for original fans. (spare one : https://www.delta-fan.com/Download/Spec/AUB0912VH-CX09.pdf  ).
Please note that here, I prefered boring 4 new holes in the plastic of the single spare fan, than modifying 4 small holes for screws + 1 big hole for air in the inverter casing.
I agree that the too small original inlet hole of thet 8cm fans could have been a problem here though, but it might compensated with the better airflow of the 9.2cm spare fan.
Also note that I finally decided to be flipping both the two fans upside down (in order them to be extracting the inside air instead of blowing outside moist and dust directly onto the boards and connectors, without inlet air filters).
Indeed, I also discovered that boards and connectors were already oxydized (in front of the fans flow) after 2 years only (!) Here, the filters are mounted on top of the inverter, and to me, filters means inlet of course, not air outlets.
As I said, check the air flow on every fans datasheet and it's wires colours (because you might be using a 4 wires fan with no problem, after knowing which wire to be plugged to where).
As I planned, once I was totally sure that the machine was fixed, I also tested with very slow fans, in order to make sure that the software of this device was intelligent enough to be accurateley checking the speed of the two fans separately.
I also tried manually while the fan was turning between 0 and 100 rpm only : the results are as followed.
 
- The system only checks if the fans are emmitting few pulses (no matter the frequency of the pulses).
- You can block one fan completly for few seconds with no error message at all (for approximately 15 seconds continuously, if I remember correctly) : And there will be no "fan fault" message at all since the fan can perform at least few turns avery 15 seconds.
- If you exceed 15 seconds with one fan blocked, the system then consider that you have a fan failure of course (even if temperature is still very low).
- Once the fan fault (ERROR 27) is triggered, the current switched to OFF, and the beep sound has been warning you for approximately one minute, the system might try to recover from this event by itself... Switching the main relay ON again, but the fault might resume several times, even if the fan is no more blocked. It seems that the error stays in memory even if there is no more fault actually. 🤔
Scary, but I needed to check it several times (in order to know that blocking a fan was not dangerous). I'm happy because now I know.
This several reseting procedure of the inverter could be dangerous though : especially for computers, old refrigerators, congelators, or TVs because no appliances like beeing turned ON and OFF so many times. Not dangerous for the inverter of course. Because he must be designed for voltage failures of course.
 
If the Error 27 fault is occuring while you're "playing" with faulty fans, or manually blocking or slowing down the fans (I had fun to play with) I would recommend to switch the infinisolar OFF right after the fan fault is  triggered ON. The fault would be then reset and the inverter will then start again with no more fault.
 
Hoping that these informations will help the maximum of users. Special thanks again to Coulomb who helped us a lot about informing that the standard codes only check for lack of pulse from fans (instead of searching for fan unexpected decelerations).
 
Have a nice Week-End everybody.
 
Lionel
P.S. : All you needed to know about how to connect a 4 pins fan to these original 3 wires fans... Always check colours of wires correspondance on each fan datasheet (because pin out positions may vary from the main standard computer boards.
https://www.quora.com/How-do-I-connect-a-4-pin-fan-to-an-old-3-pin-motherboard/answer/Janis-Baumanis?ch=10&oid=258771781&share=42265742&srid=uoWx47&target_type=answer
 

It's been a long time since I looked at Infini firmware, and I'd rather not revisit them, but that's what Axperts do (just check that the fan isn't locked rotor), and I expect the Infinis to be similar.