little thinks, I worry about.....

htool / ESP32_Precision-9_compass

MPU9250 and ESP32 emulation of Precision-9 compass

Apache License 2.0

7 stars 1 forks source link

little thinks, I worry about..... #1

Open BerndCirotzki opened 2 years ago

BerndCirotzki commented 2 years ago

Hello, I think this is a good idea making this. But I found some thinks, They don't work.

The rate of turn never gives a negativ Value. But I think the reason is this : https://github.com/ttlappalainen/NMEA2000/issues/250#issuecomment-1040427529
It is not clear how the MPU9250 is mounted. in the Up case the direction of Rate of turn is the wrong way. change to : SetN2kRateOfTurn(N2kMsg, SID, -(gz * 0.017448352875489));

Why is there made a tumbling of the Compass, when the Values are never used ? It is overwriten immediately by "refresh_data()" "magScale" is never set to "Mag_x_scale" ... And in this case ... ? Why do you save "Mag_x_scale" in the EEPROM

In the function "calc_pitchRollYaw()" I thing the q[..] Values are a little confused. ... this Part also needs work. See exactly the code on this Side : http://en.wikipedia.org/wiki/Conversion_between_quaternions_and_Euler_angles

htool commented 2 years ago

The main focus so far was to understand how to communicate with the B&G MFD.

Rate of turn could very well be the case - I haven't tested it well yet.
MPU9250 was mounted upsidedown
Tumbling was part of the original code I based mine on. It can be removed.

I wasn't happy with the stability of the MPU9250 and ordered a CMPS14 to replace it. The other plan is to rework it using SensESP (https://github.com/SignalK/SensESP) to make over the air updates possible and have a clean intergration with SignalK. I'm not sure if you're using SignalK, but if you are it's worth looking into SensESP. We could work on a version together that has support for both MPU9250 (if you''re happy with it) and CMPS14.

BerndCirotzki commented 2 years ago

Ok, I understand now. I thought, I was tested out.

I am also not so happy with the MPU, but "rate of turn", roll, an Pitch works fine. ... with the little changes. The heading was not working correct , but I thought the reason is my "Testequipment" and the house around :-). the CMPS14 is a realy good Idea. ... but in the future ...

With SignalK ... I don't know for what ?? I don't know any Industrial Equipment woking with it. a new translation ... ?? Oh help, not now. And a new hardware connection ... why?

EduardNN commented 1 year ago

Hello htool. I'm trying to repeat your project, but the ESP32 does not see the MPU9250. How is your MPU9250 connected? By i2c or SPI? What is the best way to disable BMP280 in code (only heading pointer needed)? Is there a diagram? Thank you.

htool commented 1 year ago

@BerndCirotzki Did you manage to get the MPU9250 working well? Maybe you can assist Eduard?

HV40 commented 3 weeks ago

hi, I don't get how the MPU9250 has to be mounted, mine is vertical, I thought the sensors calibration would enable it to figure out its position but after calibration I still don't get correct pitch/roll/yaw behavior.. Would also be interested in CMPS14..either directly through SH-ESP32 which I prefer as much simpler and direct than having to fiddle around with SignalK..

EduardNN commented 3 weeks ago

Repeat the project on bno055 by BerndCirotzki - it works quite well https://github.com/BerndCirotzki/ESP32_Precision-9_compass_BNO055