Traction Control Testing & Tuning

[KaPower]

I'm currently testing traction control (and launch ) functions utilizing the front vs rear Vss% method, and thought I should share my findings. (Currently running Alpha pre 1.5 - 4)

Object: Educate myself and others on tuning methods for traction control, and hopefully help the code developers by testing and confirming the control algorithms. (Also want to maximize my 0-60ft times!)

Test Platform: 1989 Nissan 240sx (rear wheel drive), 315mm Hoosier A7's all around (Autocross/Pro Solo purpose built race car), wheel speed sensor on the RF and LR wheels (Cherry hall sensors reading OEM nissan ABS 45 tooth wheels (PT4 and Nitrous-in for the inputs)). Taction control threshold % (0-5v) Pot mounted on a switch panel.

Engine: 4cyl (Ka24de) 400-500rwhp, Twin scroll Gtx3071 turbo system, Twin disc OSG clutch with plenty of inertia.

Background: The car is a purpose-built for autoX/pro solo. The main objective is cornering (and driving), however with Pro-Solo events, drag strip style starts/launches on *unprepared asphalt* and or concrete are part of the game. The straights are usually <200ft. For rear wheel drive platforms, sub 2.0 0-60ft times are rare (although a few have achieved 1.79's on concrete of late). My car averages without 2.0's and 2.1's, and on a few occasions have dipped into the 1.9's. All wheel drive chassis frquently dip into the 1.6's, (GTR's frequently see 1.5's).

If any one else has experience and or observations utilizing launch, and traction control, I think this would be a good thread to discuss them, and the features of Megasquirt.


My observations: I have tested Traction control (VSS slip % method) at 2 events within the last 2 weeks. I have used launch control on most all starts. The car tends to wash the tires right at the start, then bog when TC kicks in. Originally I wanted to have a soft traction control, where it would simply pull boost and retard timing, and then cut spark as a last resort. Unfortunately the code has a bug that that has been confirmed, so all my current testing has not utilized the boost reduction reaction curve. Not being able to log the threshold pot position, spark cut %, and boost reacting curve is also an issue.

Boost reduction reaction bug discovered on my first outing: Duty cycle is increased instead of being reduced, 250% + dc and 30 psi boost in 1st gear, all while the megasquirt is fighting itself cutting spark, and retarding timing.

Traction Control dc inv.png

[KaPower]

Second test session observations:

1. I tried to utilize the front wheel speed sensor for Vss1 and Vss2 as a rear wheel, so the launch control would not be disabled until the front wheel started to move. The idea was that this would keep the rear tires from washing on the hit - no such luck. TC does not work unless VSS1 is a rear wheel. It may be helpful to have launch control select which wheel to dectivate from.

2. Traction control reactions appear to be a little late to respond. The rear tires would spin up to 34mph (front tire 3mph) before the traction retard actually registerd in the datalog .389 to .437 seconds from the hit... The car would wash the tires, backfire and shoot flames everywhere (tc spark cut), bog, then take off. We eventully managed a couple 2.1's and a 2.0. Not good so far, but we have data to sift through and analyze! I utilized several different settings to see how the car would react.

This is very enjoyable to play with! That's all for now...

-Jason

[AbatelliCristian]

very interesting for me.
I hope to have development

[Speedy_G]

could you please also post your msq with datalog msl?

There are quite a lot of problems with the Traction Control. I already started a thread and showed those problems, but like always no help received.

1) The Slip Threshold is not transfered to TS (Most important when you use the knob slip threshold)
2) Not even the actual slip is transmitted to TS
3) TC Spark Cut is missing
4) TC Add Fuel is missing.

So how are you supposed to tune your TC properly, when you dont even know how it was reacting while driving...
I solved those problems with missing values by a custom firmware for me. But i had to misuse other variables, which i didnt need.
For example nitrous and boost_target. And there are always people who would need those.
So a official fix of these bugs would be the best.

And if i remember correct, the MS doesn't care about the decimal places at the slip calculation.
For example you got 12.99 % Slip but MS still thinks you have 12 %.


And by the way: The best launch control would be possible with DBW.
So for example that you can set a throttle target curve after launch. (With a few safety functions)
And would be even useful for holding the rpm while still standing.
But i doubt that there will be any development.
You could also use DBW for the Traction Control. So after all the "fast" spark stuff at a later stage that the throttle will close a little bit as a function of slipxtime.
So just like the other reactions.

[papiogxl]

Have you tried using timing retard after launch?

TC is fine once you're moving, but it doesn't replace proper power management off the line.

[KaPower]

could you please also post your msq with datalog msl?

There are quite a lot of problems with the Traction Control. I already started a thread and showed those problems, but like always no help received.

1) The Slip Threshold is not transfered to TS (Most important when you use the knob slip threshold)
2) Not even the actual slip is transmitted to TS
3) TC Spark Cut is missing
4) TC Add Fuel is missing.

So how are you supposed to tune your TC properly, when you dont even know how it was reacting while driving...
I solved those problems with missing values by a custom firmware for me. But i had to misuse other variables, which i didnt need.
For example nitrous and boost_target. And there are always people who would need those.
So a official fix of these bugs would be the best.

And if i remember correct, the MS doesn't care about the decimal places at the slip calculation.
For example you got 12.99 % Slip but MS still thinks you have 12 %.


And by the way: The best launch control would be possible with DBW.
So for example that you can set a throttle target curve after launch. (With a few safety functions)
And would be even useful for holding the rpm while still standing.
But i doubt that there will be any development.
You could also use DBW for the Traction Control. So after all the "fast" spark stuff at a later stage that the throttle will close a little bit as a function of slipxtime.
So just like the other reactions.

Thank you for taking the time to chime in! I apprecate your contribution.

Regarding the missing traction control tuning parameters in Tuner Studio. I agree, it is very difficult to see exactly what is going on without knowing where the slip threshold actully is. Same thing with spark cut, thats a very important parameter.

Also, I think that the developers probably have there hands full of requests, and have to prioritize things. We just need to continue to flood them with reports, to make them aware of the issues we are seeing in the field, and report the facts without bias (as much as possible). Then hope that we can work our way up the priority list.

How can we add these variables? Can anyone help us? Maybe we can assist by adding a few lines of code for review? Could we not just copy the TC spark reduction format, and impliment that with the other variables? I saw (somewhere - can't find the link right now) that there is an option for custom code developement in the manual, maybe we could start a crowd funding thing to encourage a developer to contribute?



Here is my current MSQ.

[KaPower]

Have you tried using timing retard after launch?

TC is fine once you're moving, but it doesn't replace proper power management off the line.

I have tested with after launch retard, but did not try tuning it (too many things going on with TC reactions). To simplify testing and setup of the traction control reactions, I turned off Retard after launch, and boost reduction after launch curves.

I'm hoping to have TC limit wheel spin in first gear, and then disable it with a logic switch. (I know perfect launch is supposed to do this, but I want to use slip % control).

[KaPower]

Datalogs from my last two launches.

[93supercoupe]

I had similar problems when using TC.

The only way i could get it to work was spark cut, progressively. Once dialed in, it worked well.

[KaPower]

I had similar problems when using TC.

The only way i could get it to work was spark cut, progressively. Once dialed in, it worked well.

What was your method for working around not having any traction control data on the datalogs?

[aidandj]

RE Reactions: Boost is slow to come back once you cut it. Spark retard should be the first thing you try and do. I also am interested in this discussion and hope it continues. We have a small thread over on miataturbo.net. http://www.miataturbo.net/megasquirt-18 ... uts-87770/

Scott from DIYAutoTune replied. Maybe he could chime in here too.

[shainiac]

I've heard of multiple people having poor luck with spark retard on turbo engines because it acts similar to launch control and increases boost pressure. ymmv

[papiogxl]

TC is a reactive system. You're starting way behind the curve by relying on TC to catch up to wheel speed. Talk to anyone who drag races (especailly no prep or street guys) and power management on the hit is the #1 place to improve your time.

Looking at your logs proves this, tire speed instantly jumps to 25+mph, and TC frantically tries to play catch up, often over shooting and causing bogging. Give it a running start by pulling timing on the hit, something like 20 degrees for the first .4 seconds should help dramatically.

[letitsnow]

Hi there fellow Street Mod Competitor and facebook friend!

I'm also very interested in getting traction control to work on the MS3. I intend to use it on a FWD car and for the full run. In order for this to be useful, I need to use 4 wheel speed sensors and some logic that says to use the highest speed on each axle. I need this because FWD cars that turn carry a rear wheel in the air for a good portion of the time and it's common to both be on the power and turning. If this isn't something that makes it into the official release, I was intending on writing it into the firmware myself or going the Hondata route. Right now I can't be much help because I don't own the MS3 yet and the car is half built.

As a workaround for slip transmitted, it would be easy to setup a math channel in TS to calculate that. It's just (VSS1-VSS2)*100/VSS2. You may also be able to do the same for the slip knob if the ADC it's hooked up to is reported in TS. Until the slip threshold via the knob is included in the log or you figure out a workaround, I would use a fixed value and adjust with the laptop between runs. It's worth noting that the code normalizes slip by subtracting the threshold from the calculated slip before calculating what to do wrt spark retard/cut/etc.

Based on looking at the images of the logs in your first post, it kinda looks like you have the threshold set too high and too much power being pulled out. Setting the threshold lower should get it to activate faster, then reducing how much power is cut should smooth it out some. I think the way this TC system is coded, to be most effective it's going to be active at some low level for the entire launch, I don't believe you want it to act in an on/off fashion. Meaning, the slip threshold is going to be lower than the actual slip % the tire wants to see.

If you want to look at the source code, it's in ms3_misc.c starting at line 4798.

-Bob

[KaPower]

Hi there fellow Street Mod Competitor and facebook friend!

I'm also very interested in getting traction control to work on the MS3. I intend to use it on a FWD car and for the full run. In order for this to be useful, I need to use 4 wheel speed sensors and some logic that says to use the highest speed on each axle. I need this because FWD cars that turn carry a rear wheel in the air for a good portion of the time and it's common to both be on the power and turning. If this isn't something that makes it into the official release, I was intending on writing it into the firmware myself or going the Hondata route. Right now I can't be much help because I don't own the MS3 yet and the car is half built.

As a workaround for slip transmitted, it would be easy to setup a math channel in TS to calculate that. It's just (VSS1-VSS2)*100/VSS2. You may also be able to do the same for the slip knob if the ADC it's hooked up to is reported in TS. Until the slip threshold via the knob is included in the log or you figure out a workaround, I would use a fixed value and adjust with the laptop between runs. It's worth noting that the code normalizes slip by subtracting the threshold from the calculated slip before calculating what to do wrt spark retard/cut/etc.

Based on looking at the images of the logs in your first post, it kinda looks like you have the threshold set too high and too much power being pulled out. Setting the threshold lower should get it to activate faster, then reducing how much power is cut should smooth it out some. I think the way this TC system is coded, to be most effective it's going to be active at some low level for the entire launch, I don't believe you want it to act in an on/off fashion. Meaning, the slip threshold is going to be lower than the actual slip % the tire wants to see.

If you want to look at the source code, it's in ms3_misc.c starting at line 4798.

-Bob

Thank you so much for the information Bob! Great addition to the thread!

I'll try my best to implement your advice along with papiogxl's suggestion (and others) for pwr management @ the hit.

I've been waiting to respond furhter to this thread after testing the non-bug version, but seems like thats not going to happen anytime soon?

I tried pulling 20 degrees timing at the hit, and it had a HUGE effect. The engine bogged , then boost spiked to 25+ psi instantly and hit boost cut. I'm going to try pulling 15 degrees and also going to try the boost timer function after the launch.

I have a huge National Pro-solo to attend this weekend in Packwood. James and/or Ken, any chance you guy's could let me try out the bug fix???

[KaPower]

RE Reactions: Boost is slow to come back once you cut it. Spark retard should be the first thing you try and do. I also am interested in this discussion and hope it continues. We have a small thread over on miataturbo.net. http://www.miataturbo.net/megasquirt-18 ... uts-87770/

Scott from DIYAutoTune replied. Maybe he could chime in here too.

Thank you very much for the contribution! I'll take a look at the linked thread.