More info on predictor algorithms?

[Philip Lochner]

Hi Guys
 
Where can I find more info on the merits as uses of the different timing redictor algorithms offered my MS2?  I searched the site but did I miss it somehow?
 
Kind regards
Philip Lochner
 

Posted by email.

[grippo]

The reason there is not much said about the predictor algorithms is because nobody has done any testing on this. There are only 3 choices plus a Kalman filter gain, so it wouldn't be hard to do this. Inside the code there is a spare output (outpc.spare[0]) that grabs the difference between the present measured delt_t between tach inputs and the last predicted value. Ideally these should be the same. The error difference is in microseconds and should be very accurate. This error is directly proportional to the error in spark timing - just need to convert it to degrees, which is very easy to do. So what would be needed would be to datalog this output for the different algorithms. I believe this can be easily done by putting the variable in the .ini file.

You could start off on the bench with the stim by datalogging while turning the rpm pot very quickly. Plot out the resulting curves of rpm and outpc.spare[0] vs time. Do for no preditcion, 1st deriv and 2nd deriv. Also fool with the Kalman gain. Then repeat while driving a real car.

[Philip Lochner]

"The reason there is not much said about the predictor algorithms is because nobody has done any testing on this."
PL: If you are asking if I would do this testing, then certainly but I would need an "misinformed person"'s guide to "putting the variable in the .ini file" and any other requirements.  My knowledge at this level is non-existent, but I can learn.
 
"start off on the bench "
Don't have a stim yet but I have a spare VR dissy and intend getting a signal gen to simulate VR.
 
Kind regards
Philip Lochner
 

Posted by email.