lite mer info ifra V-tech
This is simple
There are two factors setting power (or regeneration) - position of pedal and speed of change (so first derivative in time of position of pedal).
All cars (so Leaf too) have a mechanism to improve pedal interpretation, basing on PID-R conversion. If we move pedal slowly, real power will almost perfectly folow pedal position. If we do that faster - power may go faster, than actual position (by prediction of future position of pedal). Same for regen.
All these algos are same and have R- part. It causes a non-reaction zone to exist, to avoid notorious algorythm reaction.
Using trick, that at the end of pedal changes last 10-14 kW in zero seconds (one step), prediction causes local jump of power demand with shortest possible time (one or two samples of AC converters) - in Leaf it is up to 85 kWh, then it releases it down immediately, but R-zone causes algorythm to stop at 82-83 kW (depends on luck at these probes (ac samples)).
This is not implemented in box now, but can be added in future. Until now I'm afraid of battery a bit, as with normal throttle using you NEVER make predicted power to be more than 80 kWh as leg of human is too slow. So factory did not want such behavior, but they did not expected, that pedal position would be translated and manipulated this way.
Same reaction is with regen (this is implemented in actual box). quick change in regen demand (releasing pedal to zero (to be honest - to physical zero, not zero of pedal, so this is about -5 % of opening, not possible position, but for PID-R algo it is normal value) - we cause regen to go deeper (again - this depends how system exactly interpretes that - sometimes AC converters catch 2 probes on "ultra - fast pedal release slope", sometimes one probe, so regen will be 2-4 kW higher, when algo stops intervention at R-zone.
So nothing special in it, just simple exploit of system weakness by using reactions, that were never tested (so never discovered) at factory.
Thomas
This is simple
There are two factors setting power (or regeneration) - position of pedal and speed of change (so first derivative in time of position of pedal).
All cars (so Leaf too) have a mechanism to improve pedal interpretation, basing on PID-R conversion. If we move pedal slowly, real power will almost perfectly folow pedal position. If we do that faster - power may go faster, than actual position (by prediction of future position of pedal). Same for regen.
All these algos are same and have R- part. It causes a non-reaction zone to exist, to avoid notorious algorythm reaction.
Using trick, that at the end of pedal changes last 10-14 kW in zero seconds (one step), prediction causes local jump of power demand with shortest possible time (one or two samples of AC converters) - in Leaf it is up to 85 kWh, then it releases it down immediately, but R-zone causes algorythm to stop at 82-83 kW (depends on luck at these probes (ac samples)).
This is not implemented in box now, but can be added in future. Until now I'm afraid of battery a bit, as with normal throttle using you NEVER make predicted power to be more than 80 kWh as leg of human is too slow. So factory did not want such behavior, but they did not expected, that pedal position would be translated and manipulated this way.
Same reaction is with regen (this is implemented in actual box). quick change in regen demand (releasing pedal to zero (to be honest - to physical zero, not zero of pedal, so this is about -5 % of opening, not possible position, but for PID-R algo it is normal value) - we cause regen to go deeper (again - this depends how system exactly interpretes that - sometimes AC converters catch 2 probes on "ultra - fast pedal release slope", sometimes one probe, so regen will be 2-4 kW higher, when algo stops intervention at R-zone.
So nothing special in it, just simple exploit of system weakness by using reactions, that were never tested (so never discovered) at factory.
Thomas
