When I compare the distances in my CSV file and the ones displayed on the original panel screen, they do not match and I am not very surprised.

Currently I just take for granted that the treadmill runs at the speed it is supposed to, and that acceleration/deceleration is instantaneous, I know it is not true.

On the  6 pins connector between the panel and the treadmill, I strongly suspect there are some rotary encoder signals. This is probably how they read the speed of the engine.

To get the real speed of the treadmill and improve the accuracy, | see 3 solutions :

• add a PCB with a LS7083N and a 12V-to-5V converter
• read the speed at the LCD controller level
• not using any other signal but using a less naive modelisation of the treadmill speed in my arduino code, including the time ittakes to accelerate

I will probably try the 3rd approach first since it doesn't involve any hardware modifications. It won't be as accurate as the 2 other approaches but it's not a big deal if I don't know the precise number of calories I burn each day. After all, the number of calories is already an estimation, and in my case I mostly use those values to track how much exercise I do during the week

On another topic, I added a git repository to the project, and I improved the charts by removing the year from the dates and limiting the display to the last 10 days.