Megasquirt Support Forum (MSEXTRA)

It has been a while since I have paid attention here, so forgive me if this has been discussed.

I think it would be useful to add a "rate of rise" traction control scheme for drag applications. The "perfect run" implementation certainly works ok until slip gets bad enough to truly be "behind" but after that point wheel speed is so low relatively to the timing curve it is effectively "all over." The true physical thing we are controlling is at any given wheel speed there is a maximum "rate of rise," and above that it is slip.

This would obviously require a new calculated parameter for wheel speed rate of change. Then also a calculated "% slip vs. rate of change." The user would need to input max allowable rate of change vs. wheel speed. For simplicity purposes %slip could simply be calculated as "rate of change/allowable rate of change." This could then easily be fed into existing perfect run traction control reactions. I am sure the curve would look slightly different, but the same reactions could be used.

In general, at slip onset their is a relatively large change in rate of change. Advantages to doing this in this manner in addition to what is discussed above is you completely remove the dependence on the launch timer.

Yes I agree. It would be a simple traction control but very effective. Make it self learning and make big $

I agree Wes

I like this idea as well. A maximum rate of acceleration vs speed could work in quite a few situations, including (at least theoretically) an all wheel drive rally car where it could possibly spin all four tires and a launch timer would not be applicable.

I think this could definitely be a better way around traction control for cars without speed sensors on undriven wheels.

There's a technical challenge to this. Very frequently you can blow the tyres away on an initial launch. Measuring the VSS and rate of change of VSS from rest is the most difficult time. If you aren't able to achieve this fast enough, you can be spinning away at a near constant VSS. i.e. getting the balance right between smoothing and response time is not easy.

James

We thought about this years ago. I wanted to use the gear ratio in the calc and watch the teeth of the crank rate of change (RPMdot) to get the resolution you need to to effectively do traction control. The problem get into this when you account for automatic transmissions.

Andy

That is certainly true to a point James, but functionally the same "issue" applies in the current strategy. I think applying instantly from zero speed likely is a loosing proposition anyway when dealing with a "driven" only scheme. I think the only way this works is to begin performing the calculation "speedot" from zero, but don't start applying corrections until you exceed the "minimum speed for VSS" number.

When I am looking at a wheelspeed graph, their is almost always a drastic increase in wheelspeed when spin is present, and it is very easy to spot. I am quite confident this would be a useful traction control addition, particularly for radial tire automatic drag cars. I can provide you with logs of spin vs. no spin for a typical car, it makes this quite obvious. There almost always is some necessary slip "on the hit" even on a good run, but the wheels don't "come off" until 10-15mph generally. At that point on a bad track wheelspeed may increase at 80 mph/s instead of say 35mph/s for a good run. If you would like to give it a shot, I also have a customer car currently in testing that would be available to provide feedback (at least until racing season is over in the southeast US).

I do concede I am fairly ignorant as to how quickly this calculation can be updated and applied to a correction within the ecu. I do know the existing systems that do it certainly talk up how quickly they perform the calculation/corrections.

I certainly understand every one thinks "their' proposed updates/strategies are "must haves" and you can't spend time on every one. I can think of no "con" compared to the current perfect run, and if it did nothing but remove the dependency on the launch timer this alone would be beneficial (then you don't have to account for transbrake release delay, which in and of itself varies slightly).

There has any mind why traction control spark retard is max 30 ? Can it make to like 40 ?
Motor still make too much power =) Not want use Spark Cut because make noise, some place too much..
Boost duty is too slow ..

wes kiser wrote: I can provide you with logs of spin vs. no spin for a typical car, it makes this quite obvious.

If it is 'obvious' in a datalog, then there's a good chance that the data can be used.

James

Rather then wheel speed, could accelerometer input be used?

Ive been thinking about something like that for a while. Something that adds or subtracts power based off accelerometer feed back. Havent been able to fully wrap my head around so i havent made a thread.

93supercoupe wrote:Rather then wheel speed, could accelerometer input be used?

Ive been thinking about something like that for a while. Something that adds or subtracts power based off accelerometer feed back. Havent been able to fully wrap my head around so i havent made a thread.

In a straight line, an accelerometer wouldn't be able to tell the difference between part throttle and wheelspin.

You could maybe use an accelerometer to maintain a specific drift angle.

Base it against time after launch. And fix the timers so they actually match real time.

jeffmarsh750 wrote:Base it against time after launch. And fix the timers so they actually match real time.

The existing method is already explicitly tied to time after launch, as is already an entire host of "from launch timers" (boost, timing, nitrous, fuel, etc....). The advantage of what I am suggesting, is linking it to allowable acceleration vs. wheel speed completely removes the time from launch dependency, and is the basis of how the current better standalone TC systems handle driven wheel speed only traction control. The problem with linking anything about traction control to launch, is by definition if the run gets slowed down at any point (excessive wheel spin, pedaling it, random issue) it can no longer work at all.