buba said:
Rafe said:
Not wanting to be a killjoy but I am going to have to mention it. Indicated human power (4.0) is now ridiculously high, I can easily get it past a Kilowatt...
Have not looked at that before and may look into it. Have made a pull request for the latest beta and I am simply dividing the value by half in that firmware. So it should represent human power a bit better. That will work for now. Thank you for your feedback!
Im curious on this as I have really been going over the power calculation code stuff quite heavily the last few days. At first I thought(human power calculation being off) might be something with the left/right pedals giving different torque readings but I verified they both gave the same readings (47) with 20# attached to each pedal with the back wheel blocked. This value matches what the code is using for torque sensor scaling. I even played with some c code today trying to characterize the reading of the torque sensor at 10Hz vs different cadence and the filtering of it by the display. No conclusions yet other than it can get complicated and we are dealing with 8 bit processors here
https://bicycles.stackexchange.com/questions/23405/analysis-of-stages-high-speed-data
0.17.x has no changes to the motor controller, only on the display.
The torque/power/assist level calculation is done in the controller software at 10Hz and sent to the display which applies a low pass filter so the display changes in 0.17.x shouldnt affect assist level multipliers, only the display of power.
So in summary, the controller power calculations seem spot on, basically measuring the torque sensor at 10Hz, using that and cadence to compute power. While there are fluctuations in the pedal pressure and computed power, I believe this is exactly how the assist should be working (i.e. amplifying your inputs throughout the pedal stroke).
How this works out on the display might/could be improved(i.e. computing average of all values from controller every second). Heck simply dividing by 2 on the display might be good enough.
I decided to abandon the 850c display today and go back to the LCD3 with the buttons from the 850C which are AWESOME. After I got it all wired up I noticed that what the LCD3 was showing for config settings in no way matched what the 850C was telling me. Although I had done 'restore to defaults' on the 850C, what was being shown on the LCD3 were the values I had set on the LCD3 prior to using the 850C. Just a note to casainho. This might be 'as expected' depending on what is stored in the display vs controller, just wanted to mention it