Controlling wheelspin electronically

[JohnK]

Any of you guys with software and hardware experience been thinking about building systems that can manage power delivery for the driver? It dawned on me that a limited-slip diff does this for the suspension, and a system that listens for wheelspin can be seen as doing the same for the drivetrain. Have played with the PIC microprocessor enough to make me think there could be something done in this regard, definitely inexpensively, and possibly without too much difficulty (he was last heard to say). I imagine doing the code to tune the behaviour of such a thing so it really increased drivability could be really interesting. Consequences: throttle technique becoming a lost art ('cept for people who like to ride trials motorcycles in the woods during the winter). Lotus did the entire bleeping suspension in the mid '80s for their F1 cars so how hard could it be?

Cheers,

[slomove]

Obviously this can be developed from scratch but there are already many solutions on the market, be it installed in luxury cars or after-market controllers. So why would I want to re-invent the wheel? Also while the actual control unit could probably be made cheaply the required wheel sensors and actuators will cost some money. I would also guess if the big guys spent a few man-decades in getting this to work nicely there may be some smart tricks to it.

 

That said, I would sound like an awesome project to me :cheers:

 

On the other hand, if I wanted a car to legislate my right foot I would probably buy a BMW M5 or so......it is so unsevenish.

 

Gert

 

 

[slngsht]

somebody makes an aftermarket unit - I don't remember who.

 

In my case, if I really wanted it, all the sensors are already in place. i'll have to replace the camaro throttle body with the one from the vette (Drive by wire), and put in the stability control module, accelerometer, etc...

 

Calibration of a system like this is no trivial task. You have to consider slip angles to get optimum traction. Also, need to compare wheel sensors (front and rear), input from accelerometer, as well as fault logic to disable the system when things don't make sense.

 

I suppose a simplified version could simply control the throttle, as opposed to spark control and ABS activation, but the result will be less than optimal.

 

I know at least for vettes, most people disable the system at the track, as it tends to cook the rear brakes. Also the system freaks out if the car goes airborne and lands.

[Davemk1]

Can't you just unplug spark plug wires one by one until it won't spin the tires anymore?

 

dave

[solder_guy]

Can't you just unplug spark plug wires one by one until it won't spin the tires anymore?

 

:rofl: :rofl: :rofl:

 

[JohnK]

Can't you just unplug spark plug wires one by one until it won't spin the tires anymore?

 

dave

 

 

EXACTLY! - It's just that you have to figure out how to get a microprocessor to do that for you, just enough plugs and just when you need it.

Cheers,

[JohnK]

somebody makes an aftermarket unit - I don't remember who.

 

In my case, if I really wanted it, all the sensors are already in place. i'll have to replace the camaro throttle body with the one from the vette (Drive by wire), and put in the stability control module, accelerometer, etc...

 

Calibration of a system like this is no trivial task. You have to consider slip angles to get optimum traction. Also, need to compare wheel sensors (front and rear), input from accelerometer, as well as fault logic to disable the system when things don't make sense.

 

I suppose a simplified version could simply control the throttle, as opposed to spark control and ABS activation, but the result will be less than optimal.

 

I know at least for vettes, most people disable the system at the track, as it tends to cook the rear brakes. Also the system freaks out if the car goes airborne and lands.

 

 

I should put a disclaimer in when I suggest these things - I'm a scientist by training and all of these issues are really interesting (at least to me) because they give insight into how the car behaves when you're driving it somewhere around a limit. What the limit may be like, what it is sensitive to, . . . It makes driving fast even more interesting that it already is, if that's possible. And the net is that you may be able to build something that allows you to explore what's going on.

 

I (and Carroll Smith) have found that the dirt-track guys are often a good place to start here because their stuff isn't subtle (as subtle as, say, F1) and is accessible to the beginner, not to mention that their stuff works. Talk to a serious dirt track racer about the benefit of traction control and see what they have to say about it's effectiveness. AND they know how to CHEAT (not that that's what I'm after but if you can hide what you're doing it adds sophistication to the system). In that regard, I recall reading about an add for a system that works without wheel sensors, figuring out when you've got wheel spin by looking at changes in in the rate of angular acceleration - all in a very little box that listens to the rpm of the drive shaft and grounds spark plugs in a random sequence to limit output / wheelspin. Pretty impressive. I spent years archetecting and designing information systems for a "large mid-west consumer products company" that were really complicated and learned about system decomposition and prioritizing. This is the same sort of thing, but FUN (even if it pays not quite so well). This sounds like an ideal project for the several months when it's too cold in Ohio to drive or work on the car. Again, Lotus did this in several years for a complete F1 suspension system, producing the Lotus-Honda 99T. Allan Staniforth does a good job of describing the nature and value of this accomplishment in his "Competition Car Suspension".

One of the the intriguing things about attempting something like this is the speed of programmable processors that you can buy for literally a few of bucks and which have interfaces built into them that can save months and months of work. When Ford delivered its ECU in 1986, the processor did all of the work necessary to figure out how much fuel to deliver into each cylinder and when to fire the spark plug for that cylinder in 1/4 of a crankshaft revolution at 6000 RPM. I should mention that any Ford that you see driving down the street is weighing the amount of air its engine is inhaling. calculating the A/F for the engine conditions for that mass of air, and calculating when to fire the spark plug based on the amount of oxygen in that mass and the delivered fuel - for each cylinder, and in real time. It's also listening to the angular acceleration of the flywheel and the amount of oxygen in the exhaust for that cylinder to trim subsequent fuel delivery and tell if there's been a misfire. So, how hard can wheelspin be to control?

Anyway, controlling wheelspin sounds like a really interesting area for those who like to experiment with technical tweaks to ho

[slngsht]

 

Yes, you can build a system completely based on angular acceleration of the driveshaft. Simply knowing that based on your tire type, track surface, weight distribution, and center of gravity height (affects weight transfer), your car will be able to accelerate at a known maximum rate. If your driveshaft is accelerating faster (faster than the max acceleration plus X% to allow for optimum slip), then you have wheel spin.

 

Problem with this approach is tire temp, and surface condition (such as wetness) make a difference. So if you calibrate the system for your typical street, then go to a place that has a grippy surface, the system will kick in too early.

 

An improvement would be to employ an accelerometer to determine rate of acceleration to rate of change of driveshaft speed. I think for a cheap system this might be the way to go. You'll have to deal with how to handle wheel spin when lateral acceleration is present, etc... An alternative would be to use an average signal from the front wheel rates.

 

Speaking of simple... I partially built a REALLY simple traction control device back in the 80's. I was a mechanical engineering student, but thought I'd try it anyway. The idea was to build a super cheap device. It never made it on a car, but I did build a device that could detect wheel spin on the bench. Looking back now, the approach was slightly comical, and pretty crude. I used all analog circuits, basically converting the wheel pulse to analog voltage, getting avg for the rear, and the avg for the front, then comparing the two voltages to produce one final signal that represented wheel slip. The idea was to then use a simple controller to act on throttle / spark to do something about it, but that's where I stopped. I have no idea if it would have been any good in real life.

 

[BobDrye]

"I was a mechanical engineering student"

 

 

I used all analog circuits, basically converting the wheel pulse to analog voltage, getting avg for the rear, and the avg for the front, then comparing the two voltages to produce one final signal that represented wheel slip.

 

This is not the way that a gear head talks! I would say that you have latent wire head desires.

 

 

 

:banghead: :banghead: :banghead: :banghead: :)

[JohnK]

I certainly agree that comparing the rates of angular acceleration for driven and non-driven wheels would be a solid indicator of wheelspin, but thinking of part count and implementation effort involved in this encourages me to look for a simpler solution, if a good one is possible My questions is, in order to identify the wheelspin condition, isn't it sufficient to calculate the time required for the drive shaft to complete small angular displacements, say every 5 degrees, on a continuous basis and compare the current time-to-complete-5-degrees interval with the previous? An difference in interval time greater than some empirically-determined value would identify wheelspin. I THINK this would be independent of road or tire conditions or weight transfer, and wouldn't need to refer to the rate or to changes in rate of acceleration of the car. Even if the car is very powerful, the rate that it can accelerate itself should be much less than it can accelerate the driveshaft (and diff and moments of the half-shafts and wheels) when there is suddenly reduced traction. And since the controlling value is the difference between the current and previous intervals, the value is relative and should be independent of which gear the car is in, or how much power it has, or how light it is. I'll guess that some programming would be required to handle some things like windup of a limited-slip differential. It also would be interesting to figure out how to synchronize the driveshaft speedup detection event with the engine firing cycle in order to reduce engine output.

One inspiration for this is the production Ford ECU program. It looks at crankshaft position change when there is a firing event due, and expects to see a shortened interval between timing mark detections due to combustion-caused crankshaft speedup. If it does not record this reduced interval, it signals a missfire condition to the ecu for a fuel management adjustment (a false too-lean condition). Aren't computers fun?

 

 

As an aside: Your student-days project reminds me of something done my an ME friend who had to design a timing circuit for a rocket, well before the digital revolution. He timed the run-time of the motor using a silver solution contained in a glass cylinder which was wired so that it completed an electrical circuit. The current flowing through the silver solution caused the electro-chemical deposition of the silver on the walls of the cylinder. The amount of current and the concentration of silver in the solution determined the run time.

 

[flexer]

I am running a racelogic traction control set-up in my locost. Hope to have it running in the next 2 months or so. I have installed a racelogic set-up in a BMW M-coupe and a S2000. They are AMAZING

 

J. R.

[locostv8]

It would seem to me the simplist way to accomplish this and get other benifits, doubling the tire contact patch and kinda ABS, would be to use an AWD system. The cost wouild be building a Locost and taking this into account durring the construction. The passenger would be a bit intruded upon. I believe I have figured out the components to keep this locost and would only require 4 machined parts to accomplish. Instead of cutting power you could increase power.

[JohnK]

I am running a racelogic traction control set-up in my locost. Hope to have it running in the next 2 months or so.

 

Been reading about the system at http://www.racelogic.co.uk/ . Look forward to hearing about it, and any details you may feel like sharing about how the implementation goes with your Lowcost.

 

I mentioned earlier that I drove a Porsche Boxter and found that it's version of traction control interfered with my driving. You said you liked how the Racelogic system behaved in your M3. Did it take you long to learn how to leverage it, and do you change settings when driving conditions change?