• Welcome to PiBoSo Official Forum. Please login or sign up.
 
March 28, 2024, 10:09:53 PM

News:

GP Bikes beta21c available! :)


Time to improve ECU

Started by Manu, December 05, 2017, 08:39:29 AM

Previous topic - Next topic

Manu

Piboso, it's time to improve the ECU.

This simulator deserves a better ECU simulation. We can't continue with such a rudimentary ECU which gives more problems than benefits. I don't know if this is on your list but please keep it in mind.
It's Easier to Fool People Than It Is to Convince Them That They Have Been Fooled.

connorhall70

GanjaGod

Luponius

Can you please elaborate for that one moron who doesn't get it (asking for a friend  ::))?

Vini

December 05, 2017, 02:30:41 PM #3 Last Edit: December 05, 2017, 02:39:43 PM by vini97
Yes, big +1
For one it needs a much higher sampling rate.
At the moment you are much faster / closer to the limit without AW/TC because it just cuts way too much power too abruptly.
...I have to point out again that it's not the underlying cause of tankslappers, though.

MultiCOOLFRESH

+1
Quote from: vini97 on December 05, 2017, 02:30:41 PM
Yes, big +1
For one it needs a much higher sampling rate.
At the moment you are much faster / closer to the limit without AW/TC because it just cuts way too much power too abruptly.
...I have to point out again that it's not the underlying cause of tankslappers, though.

Yeah, the bike just cuts the ignition. It feels like my restricted 125ccm 2-stroke bike, when you rev it too much ;D
Heeeeeey Hoooooo <br /><br />

HornetMaX

Quote from: vini97 on December 05, 2017, 02:30:41 PM
Yes, big +1
For one it needs a much higher sampling rate.
At the moment you are much faster / closer to the limit without AW/TC because it just cuts way too much power too abruptly.
It probably needs a redesign more than higher sampling rate.
If the principle is not good, you can push it to 2000Hz and it would still be not good.

Vini

A higher sampling rate would be mean more, small steps instead of few, big ones. So it should certainly smooth it out, regardless of how the algorithm works exactly.

HornetMaX

Quote from: vini97 on December 05, 2017, 03:13:11 PM
A higher sampling rate would be mean more, small steps instead of few, big ones. So it should certainly smooth it out, regardless of how the algorithm works exactly.
Not necessarily, the "smoothing" may be totally irrelevant.
A low pass filter with a 10Hz cut frequency can be sampled at 100Hz, 1000Hz or 10000Hz and it wouldn't make any difference.
And if the low pass "is not the right filter", then even less.

Vini

I don't understand what you mean by low-pass filter in this context exactly.
And what is the "cut frequency", as opposed to "sampling rate"?

HornetMaX

Quote from: vini97 on December 05, 2017, 03:30:57 PM
I don't understand what you mean by low-pass filter in this context exactly.
And what is the "cut frequency", as opposed to "sampling rate"?
It was just an example to show you a situation in which, past a certain sampling frequencies, there's no difference at all. It's very basic control theory stuff.

Underdamped oscillations happen even in simple linear systems when the controller is badly tuned, even if the sampling frequency is 10, 100 or 1000 times the one actually needed.

In GPB case, I wouldn't be surprised if the ECU is in fact designed as a bunch analog filters (plus some logic) integrated over time just like the bike dynamics at a very high sample rate (like 500Hz or 1000Hz, can't remember if PiBoSo ever disclosed this). If that's true, then there's no way to increase the sampling rate (unless one wants to increase the integration frequency even more).

Your reasoning (higher sampling freq = better AW) would be eventually valid if the AW was something as rough as "if pitch angle > 10deg then cut throttle to zero" and the sampling freq was very low (let's say below 10Hz). But hopefully in GPB there's something a bit better than this (but very likely not as good as a good modern AW). AW control is not that trivial on a real bike.

P.S.
If I recall correctly modders have very little AW parameters to play with (and I'm not even sure we have a clear description of what they does).
It's way more likely that the bad tuning (or plain bad AW algorithm) is responsible of poor AW performance.

Vini

Ok, understood.


Quote from: HornetMaX on December 05, 2017, 04:31:36 PMYour reasoning (higher sampling freq = better AW) would be eventually valid if the AW was something as rough as "if pitch angle > 10deg then cut throttle to zero" and the sampling freq was very low (let's say below 10Hz). But hopefully in GPB there's something a bit better than this (but very likely not as good as a good modern AW).
Let's hope so...