MoTeC traction control slip tables

[racedata]

Hi everyone !

My name is Marc, I come on this forum quite often but have never posted before so this is my first post. I'm graduating as an engineer this december after 5 years (damn you FSAE!!).

I've got a bit of motorsport experience (F1 as an intern for 6 months, World Endurance Championship as no.2 eng. on a GTE car, Touring Cars, GT3 Cup data eng., etc.) but no experience relating to engine calibration.

I currently work for a race team as a data/race engineer on race weekends. I'm basically in charge of damper and aero setup on a GT car.

We're running Motec M800 ECU. We recently got traction control and I've started dialing it in. We went out testing and from the data acquired I built a 2D table with speed and throttle position with target slip values...very basic I know, but the goal for now is to get it working

Here's my problem: Can I import a matlab or excel based file into the TC aimed slip table ? instead of changing the values one by one in the MoTeC ECU manager ?

I installed a 12 position driver adjustable switch and configured it to read the values in the aimed slip table and modify them by a +/- percentage. It's a quick way of doing changes at the track, but it modifies the whole table by a set value.

Any clues ?

[romoman]

No experience myself, but talk to Geoffrey Ring at http://www.racetek-engineering.com/

[racedata]

Thanks romoman,

I worked it out by myself, I imported the raw data from MoTeC as a .CSV file in MatLab and made a program that selects the slip values that yeild the highest positive acceleration (coming out of corners) at different TPS% & speeds. I pogrammed it to make a table with target slip values that can be imported in MoTeC as a .CSV file.

Works well so far as I've seen from testing today and yesterday, still some fine tuning to do but it's much quicker than before. I'm using ignition retart then cut. I adjusted the slip filter & PID for control to make the TC less intrusive while driving on kerbs coming off corners.

Works fantastic !

[racinray]

Thanks romoman,

I worked it out by myself, I imported the raw data from MoTeC as a .CSV file in MatLab and made a program that selects the slip values that yeild the highest positive acceleration (coming out of corners) at different TPS% & speeds. I pogrammed it to make a table with target slip values that can be imported in MoTeC as a .CSV file.

Works well so far as I've seen from testing today and yesterday, still some fine tuning to do but it's much quicker than before. I'm using ignition retart then cut. I adjusted the slip filter & PID for control to make the TC less intrusive while driving on kerbs coming off corners.

Works fantastic !

That is very cool! I bet that race team is glad that they hired you!

Ray

[racedata]

That is very cool! I bet that race team is glad that they hired you!

Ray

Thanks Ray!

The team didn't have anyone taking care of data before, a few weekends here and there they'd hire someone for help, but no more than that. It's only club racing but it got more and more competitive and they figured they needed someone every weekend.

I'm the only engineer on the car right now and it's definitely a lot of work because the rules are quite open and you can change everything on the car: dampers, springs, aero balance, ride height, engine mapping, traction control, gearing, final drives, brake bias etc. etc. etc. it's a good learning experience though, and definitely a good way to get some experience!

[Neil_Roberts]

Tire test data for longitudinal grip is usually presented as ( result ) vs slip ratio, but the physics make a lot more sense when the data is viewed as ( result ) vs slip velocity. That is, the speed difference between the tread belt and the road speed. If you ever get into drag racing, that will make the analysis and development job much simpler. In that case, you will have to deal with the full range of speed, not just a near-constant speed at 2nd gear tip-in.

[racedata]

Tire test data for longitudinal grip is usually presented as ( result ) vs slip ratio, but the physics make a lot more sense when the data is viewed as ( result ) vs slip velocity. That is, the speed difference between the tread belt and the road speed. If you ever get into drag racing, that will make the analysis and development job much simpler. In that case, you will have to deal with the full range of speed, not just a near-constant speed at 2nd gear tip-in.

Yes you're right, at lower speeds you obtain higher grip at higher slip ratios i.e. below 120kph. At high speeds this ratio somewhat reduces because a high slip ratio would result in a higher slipping velocity between the road and the tire.

Also if you look at the thrust vs speed graph and the aerodynamic drag, you realise that at higher speeds the slip ratios will be somewhat lower because it's not phisically possible to make the tires slip more because the engine is not powerful enough to overcome these forces.

So yes, working with slipping speed is also a good way of mapping traction control, but the aimed slipping speed will still have to be reduced at higher speeds because of the other forces acting on the car at those speeds, ie. you'll never be able to reach those.

[David Ferguson]

So, let me see if I get this right. You need less traction control at higher speeds because you have ... traction. Makes sense.

I figured the traction slip (% or velocity) was the maximum you would allow, not the amount you want to induce

[Phil Creighton]

Dave
I assume from my TC days that aero loads take care of it
Phil

[racedata]

So, let me see if I get this right. You need less traction control at higher speeds because you have ... traction. Makes sense.

I figured the traction slip (% or velocity) was the maximum you would allow, not the amount you want to induce

Yes it's the amount of slip I allow, if the engine can't reach that amount of slip even on full power, then you can't accelerate any faster...TC or no TC.

But you can still enter a maximum slip value at high speeds, i.e. if you hit standing water, one wheel in the grass, etc... the TC will come in.

[2BWise]

Is the Motec TC table 2D or 3D? More specifically does the controller allow you to take into account lateral acceleration and then modify the TC target accordingly? Are your drivers leaning on the system from the apex out or are they driving it as they normally would just relying on TC to keep from sliding the car too much on exit?

[racedata]

Is the Motec TC table 2D or 3D? More specifically does the controller allow you to take into account lateral acceleration and then modify the TC target accordingly? Are your drivers leaning on the system from the apex out or are they driving it as they normally would just relying on TC to keep from sliding the car too much on exit?

Well, you can build it the way you want,

The table I have right now is a 3D table. ie. speed along the X axis and throttle position along the Y axis. These two values combined give a target slip value. It's a basic way of calibrating traction control that works. Although the driver has to know how to drive without it and know that you can't go 100% throttle at the apex...That being said he can't just floor it on the apex and hope it works well because the car WILL go into oversteer...It's more of an intervention if he feeds too much throttle in.

You can also build a 3D table with lateral and longitudinal acceleration from a properly defined GG diagram.

Edit: seems like people are really interested in this topic but don't post much ? over 425 views ? And 200+ in the last day ?

Do people visit this forum in hopes of getting advice without giving anything back ?!??

[grumpyf14d]

"Do people visit this forum in hopes of getting advice without giving anything back ?!??"

I find it a fascinating read as I enjoy the technical aspect to racing and have utilized data analysis to the greatest extent possible however, my technical background (lack there of) would preclude me from contributing anything to the discussion.

[racerdad2]

As a new guy coming very late to racing, Apex has been invaluable. As this is a new hobby, not a career, all I can contribute is enthusiastic support for the wizzards of speed that I have found here :thumbup:

[racinray]

Well, you can build it the way you want,


Edit: seems like people are really interested in this topic but don't post much ? over 425 views ? And 200+ in the last day ?

Do people visit this forum in hopes of getting advice without giving anything back ?!??

I'm fascinated by what can be learned from data acquisition and I'm doing my darndest to learn more about it. I think that many have an interest in it but don't have the time or money to use it to its full potential. I suspect the interest is what attracts them to this post. And btw, most of our cars don't have traction control so there really isn't much that we can contribute.

Ray

[racerdad2]

My traction control is a combination of right foot & tired Pinto The opportunity to begin to understand the complexities of TC is greatly appreciated. As is the incredible engineering in engines, chassis & aero. The brilliance of race engineers never ceases to amaze me. All of you have our well deserved respect & gratitude.

[2BWise]

Well, you can build it the way you want,

The table I have right now is a 3D table. ie. speed along the X axis and throttle position along the Y axis. These two values combined give a target slip value. It's a basic way of calibrating traction control that works. Although the driver has to know how to drive without it and know that you can't go 100% throttle at the apex...That being said he can't just floor it on the apex and hope it works well because the car WILL go into oversteer...It's more of an intervention if he feeds too much throttle in.

You can also build a 3D table with lateral and longitudinal acceleration from a properly defined GG diagram.

Why did you choose TPS? It would seem to me that TPS would become more of a boundary condition. For speed below XX mph any TPS value above XX% would give you wheelspin. Therefore your 3D table would be more planar than anything else. Yes I agree that this would work. But, I think you should give some consideration to using lateral for the Y axis. I think you'll agree that as lateral increases your longitudinal accel decreases. Therefore, if you use lateral as your second input you can control acceleration essentially from the apex out and optimize corner exit long accel.

[racedata]

Why did you choose TPS? It would seem to me that TPS would become more of a boundary condition. For speed below XX mph any TPS value above XX% would give you wheelspin. Therefore your 3D table would be more planar than anything else. Yes I agree that this would work. But, I think you should give some consideration to using lateral for the Y axis. I think you'll agree that as lateral increases your longitudinal accel decreases. Therefore, if you use lateral as your second input you can control acceleration essentially from the apex out and optimize corner exit long accel.

First of all,

Thank you for the honest answers, my impression was that people came on here to get advice and n.ot give anything back...as in hoping to get some speed secrets and not sharing (or at least some of) theirs ! Good to see people are actually interested !

2BWise,

I built the TC map with no previous TC experience, I wanted something simple that could be setup quickly, and most of all, be easily changed with compensations on the fly. The tire data shows us that best acceleration occurs at an xx% slip value, but in real life with changing tarmac conditions, tire temperatures, camber change & varying vertical loads, that xx% is not the same.

You're right about the lateral acceleration VS longitudinal acceleration. This method would be best to get the most out of the tire coming off the corners. Here is an example of a GG diagram, along the x axis is the lateral acceleration and along the y axis is the longitudinal acceleration at a local track: Notice the data is biased to the right because this is a counter clockwise track.



Now here's a breakdown of a driver that could get the most out of the tire:

1. Starting at point 1 on the straight full throttle, no lateral acceleration, only positive longitudinal

2. In braking he would go from point 1 to point 2, no lateral acceleration, only negative longitudinal

3. The hardest part is going from point 2 to point 3 without longitudinal going to zero, i.e. braking while still turning in feeding in the steering and tapering off the brakes.

4. At point 3 you're in full cornering mode, lateral acceleration, no longitudinal.

5. As the driver feeds in the throttle he would go from point 3 to point 1, feeding in throttle and releasing the steering wheel.

TC would especially helpful going from point 3 to point 1. i.e. making the most out of the tire in combined lateral and longitudinal acceleration. One must be careful using this method however because the taction circle will change shape with speed, especially on cars with downforce that will have more grip at high speeds than low speeds. Also, how much of the traction circle is actually used greatly varies on suspension geometry, tire pressures, camber, temperature, track surface.

So, yeah, mapping traction control with lateral and longitudinal is the best way, but also the worst way because you have to take into account all these things !

[2BWise]

I built the TC map with no previous TC experience, I wanted something simple that could be setup quickly, and most of all, be easily changed with compensations on the fly. The tire data shows us that best acceleration occurs at an xx% slip value, but in real life with changing tarmac conditions, tire temperatures, camber change & varying vertical loads, that xx% is not the same.

Yes I agree that the ideal slip value is going to change as conditions change, but for a dry track I think you'll find that you can throw out a lot of those variables and assume them to be fixed. The other upside is that (it sounds like you've already used it) with a PID controller you will have some ability to work around the various parts of the track that may be the anomaly (off camber, crest, etc). It is all about compromise, but you can focus on the most important corners and skew the tuning to work best in those parts of the track. With some time I think you will find that the compromises might not hurt laptimes very much.

TC would especially helpful going from point 3 to point 1. i.e. making the most out of the tire in combined lateral and longitudinal acceleration. One must be careful using this method however because the taction circle will change shape with speed, especially on cars with downforce that will have more grip at high speeds than low speeds. Also, how much of the traction circle is actually used greatly varies on suspension geometry, tire pressures, camber, temperature, track surface.

So, yeah, mapping traction control with lateral and longitudinal is the best way, but also the worst way because you have to take into account all these things !

I have no idea of your driver's ability to multitask and provide feedback, but you have a very useful tuning tool in the cockpit. By using that 12 position switch he can do some of the leg work and make changes on the fly to provide you feedback as to which settings provide the best performance thru certain corners. This will help develop the TC table with speed. Does the system log switch position? So that you can reference the target slip value if the driver is actively making changes.

Secondly, just to clarify. My suggestion is not to tune using long accel and lat accel. It was to use speed and lat accel. That way you cover the varying speed range.

[racedata]

Yes I agree that the ideal slip value is going to change as conditions change, but for a dry track I think you'll find that you can throw out a lot of those variables and assume them to be fixed. The other upside is that (it sounds like you've already used it) with a PID controller you will have some ability to work around the various parts of the track that may be the anomaly (off camber, crest, etc). It is all about compromise, but you can focus on the most important corners and skew the tuning to work best in those parts of the track. With some time I think you will find that the compromises might not hurt laptimes very much.


I have no idea of your driver's ability to multitask and provide feedback, but you have a very useful tuning tool in the cockpit. By using that 12 position switch he can do some of the leg work and make changes on the fly to provide you feedback as to which settings provide the best performance thru certain corners. This will help develop the TC table with speed. Does the system log switch position? So that you can reference the target slip value if the driver is actively making changes.

Secondly, just to clarify. My suggestion is not to tune using long accel and lat accel. It was to use speed and lat accel. That way you cover the varying speed range.

Quick response !

True, some of the variables can be eliminated from track to track. After all this is not Formula 1, and i'm not after than .001 second improvement

The driver is pretty good, however since this isn't professionnal racing and he owns the team and he's paying to drive the car, I can't just say: you're not good enough we'll hire someone else !

The system logs the TC position setting, used it this weekend, started at a low aimed slip and asked him to turn it up as the race went on. Found out that after a few increments, cornering exit speeds were similar as the engine on full power can't slip the wheels more than a certain percentage with no TC intervention in certain corners.

Sorry for not understanding your post, lateral acceleration and speed would also be a good way, and i'ts the way i'm leaning towards for the next TC upgrade, as it would take into account the downforce produced from the wings at high speeds.

[2BWise]

Sounds like you've got some pretty good ground work started. Since this car appears to make some significant downforce it seems as though you'll be able to ignore speeds above a certain value once you've discovered that point. Sounds as though its a GT1 car? Got any pics?

[racedata]

Sounds like you've got some pretty good ground work started. Since this car appears to make some significant downforce it seems as though you'll be able to ignore speeds above a certain value once you've discovered that point. Sounds as though its a GT1 car? Got any pics?

It's a Porsche 993 GT2 with a twin turbo aircooled engine (hence the need for TC )

It's not a factory racecar from Porsche Motorsport, it's a car we built from a bare tub ourselves, basically a replica with better electronics and a few other goodies.

[924RACR]

What Bill said - speed and lat are the most important factors to use.

Of course, the devil is in how you shape the curves...

There are other useful refinements you won't likely have, but the lack of those can be mitigated with driver training.

Good luck, and have fun tuning...

[racedata]

What Bill said - speed and lat are the most important factors to use.

Of course, the devil is in how you shape the curves...

There are other useful refinements you won't likely have, but the lack of those can be mitigated with driver training.

Good luck, and have fun tuning...

Such as ?

Am I missing something ?

I'm still a beginner

[924RACR]

LOL... sorry, that's proprietary, you'll have to figure it out yourself.

[2BWise]

Such as ?

Am I missing something ?

I'm still a beginner

Nothing you're really missing. Its just that Vaughan and I both develop traction and stability systems professionally. We have access to some pretty elaborate control algorithms. Motec likely doesn't have the refinements you'd see in a production system.

[racedata]

Nothing you're really missing. Its just that Vaughan and I both develop traction and stability systems professionally. We have access to some pretty elaborate control algorithms. Motec likely doesn't have the refinements you'd see in a production system.

AH okay !

I see you're located in Michigan, makes sense you may work for one of the big 3 there! Bosch Motorsport is also in Michigan

Anyways thanks for the input ! Helping a young engineer

[924RACR]

Exactly.