Traction control with 4 wheel speed inputs?

[seb209726]

Not much online these days and a bit late to the party. Sorry.

It's a shame for this discussion that the one guy who I know to actually work with real traction and active diff control on real rally cars using several different ecus in actual real racing series was told his ideas were BS and bowed out...
If you want constructive information about how rally cars, 2wd or 4wd, tackle this issue to get this further, and possibly even implemented and tested, it might well be worth shooting him a PM. My take is more data/theory oriented than the information he could provide, and therefore much less valuable.

As a side note, a good TC strategy (i.e. one that lets you corner more easily, not only accelerate in a straight line) can be one of the decisive factors on which ECU to run for some people. As MS is also officially meant to be a racing product, this is one of the missing features to broaden its use in other series than drag racing, along with the ability to properly ramp in/out the torque output of the engine for sequential gearboxes use and a good DBW implementation... slushboxes are quite rare on the circuits, much like corners are on a drag strip. - This is however a digression.

From my limited experience, you want in the most basic implementation to allow a certain slip (driven wheel vs undriven ones ) based on TPS and RPM, with possible refinements adding speed, gear, lateral-g or yaw-sensor and steering input sensor modifiers. This allowable slip can in the latter cases be mapped by calculating the cos/cus (coefficient of over/understeer) which of course require some input about physical dimensions of the car for the user and a good grasp of vehicle dynamics for the developer writing the equations. Quite often, this part of the modelling is not part of the ECU/Tuning software but left to the engineer to figure out. If somebody was to change this, it would get many people interested I think.
Cars required by the regs to use an open diff should obviously also compare both driven wheels for it to be useful.
In the implementation I've seen, the difference of average speed per axle is indeed the main input under power (with a tolerance range on the allowable speed differential on a single axle). You then have a single "cut-curve" (be it spark or timing retard) which is function of the surplus slip for any situation.
Brake input possibly important as well for left foot braking action, I never had the chance to do any rallying to see this in action.

For circuit/hillclimb use, in an ideal world, the diver should be able to mash the throttle right at the apex and not worry a bit about what the car will do, as its response will always be exactly the same. No biting oversteer in RWD cars, no ditch plowing understeer in F/4WD applications. Look up videos of the european Hillclimb championship in the last 4-5 years, especially the larger E2-SS / E2-SH cars and notice how on every corner you hear the TC working (cut cylinders) just like some V10 F1 cars back in the days. I do agree that this could just as well be called "dynamics control" than "traction control", but that is what people want and use.

As for the people doubting the usefulness of 3D mapping, read the description of TC in MoTeC and ask yourself why the manufacturer of one of the most popular ECU in higher profile racing does it that way...

[letitsnow]

On the non driven axle it should be the faster wheel to compare too. This is to prevent issues when you have a locked up inner wheel while breaking into corners.

But are you braking into a corner while also trying to accelerate and spin the wheels?

The traction code is only designed to work when you are on the throttle accelerating.

James

It's very common for FWD and some RWD cars to carry the inside non-drive wheel through the entirety of the corner, even on power. If you trail brake at all that wheel slows significantly or stops until it touches back down. This would make an average of that axle not represent the vehicle speed accurately.

[jsmcortina]

Please review the options in pre-1.5 alpha 6.

James

[gjw]

Hi James,

The changes made in the 1.5 alpha code are great and solve a lot of the issues. A really great innovation I saw in a recent implementation of the TC was a an optional differential threshold across the un-driven wheels which would determine if the vehicle was moving straight or in fact negotiating a turn and then use this to have two separate tables one for straight line reactions and the other for when the car is in a turn. This allows more aggressive slip in a straight and a slightly more restrained setting in a turn. This could be coupled to what someone stated above an over all modifier of reactions for speed and another for turn.

Thanks
Glen