An easier to set up traction control

[Matt Cramer]

This one was inspired by some discussion of traction control on Yellow Bullet; some of the drag racers there raised a couple legitimate issues with current traction control in the MS3 compared to competitors such as Davis.

1. Wheel slip doesn't work if the non driven wheels are not solidly in contact with the ground.
2. The "Perfect Run VSS" mode takes considerably more setting up than the Davis algorithm.

Here is a proposed method that would be easier to set up and is similar to what Davis is doing.

1. Use a 1st derivative prediction algorithm to predict when the next VSS pulse will come in, with a user adjustable tolerance.
2. If there is a pulse received before the predicted time, check to see if the ECU still gets a pulse at the expected time, and make sure there is not a flag set for previous noise pulses. If it does do, throw the in-between pulse out as noise, but set a flag that a noise pulse was received.
3. If no noise pulse is seen before the expected time, clear the noise pulse flag.
4. If the pulse is received before the expected time and no pulse arrives at the predicted time, or if the flag indicates that there have been two rounds with "in-between" noise pulses in a row, apply traction control. The % slip is calculated by 1-(time from last good pulse to bad pulse) / (predicted time between pulses).

Does this sound workable? It should be pretty straightforward to tune; you'd just set a tolerance for the next VSS pulse, or a curve of tolerance vs speed, and the traction control reactions. The curve would be a bit more generic than the existing perfect run curve. It would also be able to work at any arbitrary time from launch control, so it could be used on a road race car, off road vehicle that is not using launch control, or other application where the traction control may need to kick in at some point other than immediately after using launch control.

[jsmcortina]

At a first glance I see probelms with that. Sampling tooth to tooth is severly impacted by torque ripple from the engine and any sensing inaccuracies. The current VSS algorithm average over multiple teeth to get a reliable wheel speed.

That being said, I have some ideas for changes in the way traction works. I plan to try these in the 1.5 firmware.

James

[dontz125]

1. Wheel slip doesn't work if the non driven wheels are not solidly in contact with the ground.

Why not? If the front wheels are turning more slowly than the rear because they're 3' in the air, you have significant slip, and the ECU should do its thing.

[jsmcortina]

1. Wheel slip doesn't work if the non driven wheels are not solidly in contact with the ground.

Why not? If the front wheels are turning more slowly than the rear because they're 3' in the air, you have significant slip, and the ECU should do its thing.

I disagree!
Haven't you seen ProMods run the quarter with front tyres barely touching the track? So long as the car is going in a straight line and isn't in danger of a wild wheel-stand, that's kind of normal.

James

[dontz125]

Not a pertinent example - Matt's comment was that slip % cannot be used to trigger TC if the front is up; the fact that some cars run the strip on two wheels is irrelevant to that point. They simply aren't using slip % TC triggering based on front wheel speed.

[jsmcortina]

The algorithm that Matt was referring to in point 1 does use the front wheels.

James

[dontz125]

Extreme case - If the front wheels are not turning but the rear wheels are, you have infinite slip and TC should engage. If the front wheels are turning a little, then you have a large degree of slip. and TC should engage.

[muythaibxr]

On a drag race car, if the front wheels aren't turning and the rear wheels are, the rear wheels could still have plenty of grip. If the front of the car isn't going go up that far, you don't want to pull power in that situation b/c you're still accelerating faster than you would if you did cut power.

The only time you want to cut power is if the wheels are actually really slipping.

This is different from how circuit racers would want things to work, but their cars are exceedingly unlikely to wheelie.

Ken

[dontz125]

I'm not debating or discussing whether wheel slip is a useful or undesirable method of triggering TC on a drag car or bike. I'm discussing this statement:

• 1. Wheel slip doesn't work if the non driven wheels are not solidly in contact with the ground.

Nothing anyone has said yet has changed my mind about whether wheel slip CAN trigger TC while the wheels are in the air. Whether you WANT airborne front wheels to trigger TC is a completely different statement and discussion.

[muythaibxr]

OK that's fair. A wheel slip method can work when the front end isn't in contact with the ground, as if the front wheels aren't spinning, yes you can assume that is slip and engage TC.

For a drag racer you don't necessarily want to do that.

I've been thinking of methods of TC that also involve accelerometers in addition to wheel slip that *could* work for both drag racers and circuit racers... I know that a lot of the new TC methods being used on bikes use a pretty sophisticated combination of wheel slip and accelerometers to control wheelies and traction separately from one another... meaning they'll allow a certain level of wheelie, and they'll allow a certain level of slip, and they can detect the difference between them.

Ken

[thokes82]

May I ask to let the current wheel slip% option at least in selectable as an option? It is exactly what is needed on a circuit and is the method which is widely used and known.
I understand the needs on drag strips but I am not not wheelie-ing out of the corners when I flat out at the apex... Thanks and Merry Christmas...

[MS2tester]

This one was inspired by some discussion of traction control on Yellow Bullet; some of the drag racers there raised a couple legitimate issues with current traction control in the MS3 compared to competitors such as Davis.

1. Wheel slip doesn't work if the non driven wheels are not solidly in contact with the ground.
2. The "Perfect Run VSS" mode takes considerably more setting up than the Davis algorithm.

Here is a proposed method that would be easier to set up and is similar to what Davis is doing.

1. Use a 1st derivative prediction algorithm to predict when the next VSS pulse will come in, with a user adjustable tolerance.
2. If there is a pulse received before the predicted time, check to see if the ECU still gets a pulse at the expected time, and make sure there is not a flag set for previous noise pulses. If it does do, throw the in-between pulse out as noise, but set a flag that a noise pulse was received.
3. If no noise pulse is seen before the expected time, clear the noise pulse flag.
4. If the pulse is received before the expected time and no pulse arrives at the predicted time, or if the flag indicates that there have been two rounds with "in-between" noise pulses in a row, apply traction control. The % slip is calculated by 1-(time from last good pulse to bad pulse) / (predicted time between pulses).

Does this sound workable? It should be pretty straightforward to tune; you'd just set a tolerance for the next VSS pulse, or a curve of tolerance vs speed, and the traction control reactions. The curve would be a bit more generic than the existing perfect run curve. It would also be able to work at any arbitrary time from launch control, so it could be used on a road race car, off road vehicle that is not using launch control, or other application where the traction control may need to kick in at some point other than immediately after using launch control.

did they try to use a GPS with pulses per meter travveled ?

[Nightstalker1993]

I'm in with Matt Cramer's idea. My car is pretty hard to install a wheel speed sensor to get the wheel slip traction control working, plus I ran out of digital inputs on the ms3x db37 connector. Would be really beneficial if there is a simpler method of setting up traction control

[Matt Cramer]

May I ask to let the current wheel slip% option at least in selectable as an option? It is exactly what is needed on a circuit and is the method which is widely used and known.

It's more likely that this would be added as one more method rather than being made the only method.

[Matt Cramer]

Posting this to make sure we don't step on Davis's current patent:

http://moretraction.com/PDF%20Files/Dav ... 7944__.pdf

[n20junkie]

I know this is an older thread, but I am a MS3 user that would love this additional traction control option.



I am looking to do some small tire racing this season and traction control has become part of the game these days. I will be using perfect run until something better comes along, but a faster "internal learning" (trying to stay away from patent or copyrighted language) or smart TC would be awesome.