Traction Control
...having a big engine with huge turbo and rear-wheel-drive is really fun, but we don't want this fun always (especially at high speed cornering on narrow roads).
- So I have the idea that I need some traction control implemented...
For the first time I don't speak about fancy stabilizing system with angle and G sensors utilizing brakes, and so on.. It's just a simple idea to control wheelspin when boost is suddenly too much.
For a simple installation we already have:
- Stock wheel angle/speed sensors (VR at all wheels from ABS system)
- Control over excess power by playing with ign retard and fuel-cut
What we need:
- Some extra electronics to process the four VR signal
- we are sending in an ARM board (just hours before layout completion) that is capable of this (with very good trigger and AD inputs, power and analog output capabilities, and networks, of course)
- I'm curious!
- we are sending in an ARM board (just hours before layout completion) that is capable of this (with very good trigger and AD inputs, power and analog output capabilities, and networks, of course)
- Some logic -probably the good-old PID algorythm- to reach the limited slip
Tell me your thoughts!
Simple is good! Have a look at MembersPage/DavidBlades/TCSInputs I don't think it's very up to date though.
I have a couple of promising algorithms in development based on just wheel speed sensors for working out what is slipping. Corners have a big impact on expected speeds... Really must get that Matlab finished...
- Yes I know, if we want something serious, we need to model the whole car, but that needs extra inputs also (f.e. steering wheel angle sensor, G-sensors...)
- Extra sensors are "nice to have" but I don't think they'd be essential if you have a good set of algorithms working behind the scenes.
1. Straight line acceleration. No problem to detect slip as wheels on the same axle are spinning at the same speed.
2. Powerslide. Not too difficult again, front wheels have a very different speed, but rear ones has same speed (LSD). Definitely spinning faster than the front ones.
My first thought was it only needs a defined threshold and some very simple algebra to determine. Of course this is not universal for a car with AWD, FWD without LSD...-But those cars rarely need this kind of control either..
Yes, the defined threshold works - say 10% slip between front and rear if you have an RWD with LSD... (My car is RWD with LSD too) You could probably average the pairs of wheels for non-LSD. FWD is more complex, and I don't have one of those so I'll stop here...
After that it gets tricky since you don't want false alarms at any cost, hence my 1st idea for a simple curve fitting system across the 4 wheels to see if there's any slip. IE if you can fit a pair of 'matching' arcs through the wheels then you're not slipping. If you can't then something is wrong - the data should indicate what action to take. No other sensors than the 4 wheel speeds, so there's the obvious limitation of it saying you're stopped while in a 4 wheel skid... (say on ice)
Then you have to think of what action you can take - and can it do anything to counter the slip you're monitoring. If not, then it's probably pointless detecting it.
Have a look at Megasquirt, those quys implemented quite simple traction control using engine accleration table values and few other parameters. I have an MS ecu in tuned Peugeot 205 GTI, tc quite works flawlessy. One could develop it further by adding gear sensor or vehicle speed sensor and mapping speed vs. engine speed vs. gear vs traction limit.
Links:
- Is there any progress in TC like that one on this link? I have some AVR programming knowledge and I was going to do it myself anyway, but if you have some algorithms for testing, a would be happy to make my contribution to the development of TC...