TMOTOR F7V2: ERROR I2C MAGNETOMETER

[samsab1]

Hello everyone. I’m having trouble with a T-Motor F7 V2 flight controller; in iNav, I’m seeing I2C bus errors (the error count climbs rapidly to 55), and the magnetometer drifts significantly and won't work properly. I checked the board in Betaflight, and it works fine there. Is there a firmware fix available for the T-Motor F7 V2?
Best regards

[daijoubu]

Does it stop climbing? About 50 is normal at boot due to auto discovery of the different modules.
Did you follow the instructions on the wiki to set the compass alignment? Inav doesn't have auto alignment yet.

[samsab1]

I certainly don't have the expertise to fully grasp the software behavior of these controller boards—even though I've built about seven or eight drones and just as many aircraft (under INAV)—yet this is the first time I've run into this problem. While I understand that the I2C error count isn't necessarily a critical metric, I am unable to get the magnetometer working properly despite performing the usual calibration; it's a BE252i unit where the magnetometer is otherwise perfectly recognized under Inav.

[sensei-hacker]

You mentioned you did the calibration. That's great. Did you do the alignment?
I prefer the method that the wiki calls the "empirical method".

[samsab1]

Yes, absolutely—and that’s the major challenge with this GPS, or perhaps the firmware specific to this Motors F7 V2 board. I’m really struggling to find an alignment that yields a decent result; I’ve tried everything without success. With my other quads—even though the alignment process isn't always straightforward—things usually work out in the end, but here I’m stuck with a persistent issue. Maybe I should just bite the bullet and use a different GPS, like an HGLRC M100.

[daijoubu]

You said you had it working in Betaflight, they have automatic orientation which isn't in Inav yet, but will probably be in v10. Do you have a diff from your betaflight test? If so, search for align_mag or mag_align_[pitch|roll|yaw]. Do these settings differ from what you selected in Inav?

[samsab1]

Not being an expert myself, I asked Gemini; here are the answers:

Yes, in Betaflight, an I2C error count of 9 at startup is perfectly normal and does not indicate any malfunction during flight.

The fact that the packet error count is 0 confirms that communication is excellent once the board has started up.

Here is why you see such a clear difference between INAV and Betaflight on the same board:

1. The startup "Scan" phase
   When you power up the board, Betaflight doesn't know exactly what is connected to the I2C bus. The code therefore blindly tests several standard electronic addresses (such as those for the BMP280 barometer, various magnetometers, etc.) to see if a component responds.
   Each time Betaflight queries an "empty" address (where nothing is connected), the processor receives a rejection. This instantly generates an I2C error. Those 9 errors simply correspond to the few unsuccessful auto-detection attempts made before the board finds the barometer.

2. Why did INAV show 50 errors while Betaflight showed only 9?
   The difference is purely software-based:

Betaflight is highly optimized for racing or freestyle setups (which don't use GPS/compass by default). Its startup I2C scan routine is very brief and "gives up" quickly if there is no response.

INAV, being geared towards autonomous navigation, is much more persistent at startup. It sends multiple successive requests, adjusts its timing, and forces detection to ensure it doesn't miss any navigation sensors. It is this software persistence that caused the error count to climb to 50 in INAV with the same barometer.

What to look out for
Normal behavior in Betaflight is as follows:

You plug in the board: the counter immediately displays 9 (or a similar fixed value). Leave the drone stationary: this figure must never increase again.

As long as the value remains stuck at 9 and packet errors stay at 0, the I2C bus is perfectly stable in Betaflight.