search

Digital1O1 / MPU_9250_Calibration

The intention of this repo was to provide a general guideline in order to calibrate the MPU_9250 since there seems to be a lack of documentation on how it should/could be done.
6 stars 2 forks source link

No issue, just a note of thanks! #2

Open steveincolo opened 3 years ago

steveincolo commented 3 years ago

Your calibration routine is extremely well documented and implemented! I just used it to calibrate an MPU9250 that I'm using for a "push-to" telescope control via WiFi to the Sky Safari iOS app. After calibration, the Madgwick sensor fusion algorithm seems to be working very well, with negligible drift. I'll be trying the scope out tonight!

Digital1O1 commented 3 years ago

@steveincolo my bad for just NOW getting back to you since I never really check my GitHub.

But that honestly made my day and I'm freakin stoked that I could help you out with your project!!!

I do plan on revising the MPU calibration code sometime this month (with hopes that it'll be more useful), but I'd love to hear the progress you make down the road with your project!

steveincolo commented 3 years ago

No problem, Christopher, I'm just getting back to this myself. I'm using an MPU9250 with an ESP32. The ESP32 creates a WiFi access point that I can connect to using an astronomy app (SkySafari) on my iPad or iPhone. The box with the MPU9250, ESP32, and an SSD1306 OLED are mounted to the scope. The scope is mounted to the tripod by an "alt-az" (altitude azimuth) unmotorized mount, which moves in yaw and pitch. You start by aligning the scope on a bright star or planet, and then telling the app that the scope is aligned to that object. You can do a two-star align if the scope tripod is not perfectly leveled. Then in theory you can select another object in the app and select "push to." The app will show crosshairs as you move the scope and zero in on your target.

I'm using the "SensorFusion" library, which allows use of either the Mahony or Madgwick algorithms. If I understand them correctly, they use quaternions to transform the sensor data into yaw, pitch, and roll (which I don't need). I'm having trouble getting the Madgwick algorithm (which uses the magnetometer data) to work. The Mahony algorithm seems stable but I haven't tested it on the real sky yet.

Here's a post I did on an astronomy website with a link to your calibration code:

https://www.cloudynights.com/topic/589521-37-dobsonian-dsc-for-diy-makers/page-11#entry10964763

At the time Madgwick seemed to be working but not now.

The goal is to have a "push to" box that doesn't need to be physically linked (e.g. with optical encoders) to the mount to sense the movement of the scope. For this application a sufficient accuracy would be about 0.5 degrees, within the field of view of a low-power eyepiece. The altitude calculations are stable and more accurate, azimuth is the main problem.

Steve

On Sat, Mar 27, 2021 at 11:04 AM Christopher Ta @.***> wrote:

@steveincolo https://github.com/steveincolo my bad for just NOW getting back to you since I never really check my GitHub.

But that honestly made my day and I'm freakin stoked that I could help you out with your project!!!

I do plan on revising the MPU calibration code sometime this month (with hopes that it'll be more useful), but I'd love to hear the progress you make down the road with your project!

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/Digital1O1/MPU_9250_Calibration/issues/2#issuecomment-808762706, or unsubscribe https://github.com/notifications/unsubscribe-auth/ATI3ZXOX7HMV3GXTA2Y6E3DTFYFZ7ANCNFSM4ZLIQYAQ .