MX5 NC - From can bus raw data to equation - a little help/ideas?


I'm in the processo of reverse engineering can-bus on my Mazda MX5 NC, my goal is to record data for RPM, Speed, % Accelerator, Brake Pedal position and Steering wheel from direct CAN-bus reading instead of OBD2. I'm almost there! I will show you what I found as soon as I manage to go on the road and do a full test.

In the meantime, I'm having a little bit of trouble with the steering anlge. I found the ID, it is 0x81, baud rate 500 kHz. It is a 4 byte long frame.


What I did is rotating the steering wheel from nearly neutral position to right, stopping for a while at 90°, 180°, 270°, 360° etc, then quickly back to 0 and repeating to left, then again quicky back to 0.

Byte 0 is not interesting and then not plotted
Byte 1 has 2 status: EF or 6F, depending on if the steering is at the 360° angle, in both direction of turning, starting from the neutral position (I've plotted the decimal value)
Byte 2 has 4 status: 00 or 01 or FE or FF depending on value of Byte 3 and direction of turning. If the angle is in the "positive" field, meaning from neutral position turning to right, Byte 2 equals 0 for angles up to 255°, then it swithches to 01 for angles above 255°. In the "negative" field, meaning from 0 turning to left, Byte 2 equals FF for angles from 0 to -255°, then it equals FE for anlges exceeding -255 (again I've plotted the decimal value)
Byte 3: the decimal value is ranging from 0 to 255, with a direct correspondance between decimal value and degree of rotation (0 = 0°, 90 = 90°, etc). Above 255°, it is resetting to 0. From the neutral postition 0° turning to left it starts from 255 and decrease to 0 while the steering wheel is reaching -255°, then it resets to 255

I came up with this equation, using the "less than/greater than" condition

I'm quite satisfied of the result, except for following consideration:
- I didn't use the Byte 1 information. Is there any way to use that information and smooth the equation?
- The values are not "symmetric". I got a -486° and +455°, it doesn't feel right

Any suggestion from a fresh set of eyes would be very much appreciated. Thank you!

Data collected using PiCan2


  • Sorry, the equation was incomplete:


  • edited February 2021
    Your equations look like unnecessarily complicated. It should be possible to use just one bytesToInt variant with correct parameters. Please provide the raw byte values (for all three bytes) for following positions:

    - Zero position
    - 90 degrees to left position
    - 90 degrees to right position
    - Max left position
    - Max right position

    With the data I can write up how to figure it out.
  • edited February 2021
    @aol thank you. Yes, I also think it is too much complicated, especially thinking that for the same input but with OBD2 data I was able to use this formula:

    To answer your question, position and full raw data including ID# and byte 0, I've also added 360° left and 360° right

    0° position: 081# 76 EF 00 00 90° left: 081# 8D 6F FF A6 90° right: 081# 17 6F 00 5A Max left (486): 081# 12 6F FE 1A Max right (455): 081# DE 6F 01 C7 360° left: 081# FE EF FE 98 360° right: 081# D3 EF 01 68
  • edited February 2021
    Not sure what the two first bytes are (and it's normal to have multiple channel values are on same CAN-Bus packet), but looks like the last two bytes are the steering angle in degrees.

    The negative values are being represented in two's complement:'s_complement When you know about two's complement it's rather easy to eyeball it from negative values starting from maximum unsigned integer value, in this case -1 is FF FF, and going down as the negative value grows.

    Byte order is Big-Endian: , it's obvious in this case that third byte is more significant than the fourth byte.

    The bytesToInt (notice not bytesToUint) can extract a signed integer, with two's complement and big-endian byte order. So just "bytesToInt(raw, 2, 2)" will solve your steering angle in degrees!

    Use the reference guide to pick the correct function:
  • @aol thank you! Now it is very elegant and simple :smile:
Sign In or Register to comment.