wheelspin detecting boost controller?

[acannell]

lets collect info about a boost controller that detects wheel spin and reduces boost, like traction control.

how would we do this?

we already got wheel speed data for the passenger wheel via the speed sensor

now we just need it for the drivers wheel

and then for a back wheel to detect if the front wheels are spinning

what would be a cheap way to sense wheel speed?

i could design the electronics from that point.

to control boost we could use the factory waste gate solenoid.

so then it becomes a matter of figuring out what kind of algorithm works best.

you could set it up for minimum wheel spin, maximum allowed time for wheel spin...etc...

could make launching much easier...maybe

for wheel speed sensors, im thinking a black/white alternating strip, like a sticker, that goes around the drivers axle, and a simply infrared detector. for the rear wheel something similar but you'd have to put the reflecting strip somewhere else, like on the wheel itself (maybe inside rim?) or perhaps on the spindle somehow

[Gutfeeling]

Mount a VSS sensor to the backside of the hub knuckle facing the sensor part pointing at the axle. Then make a notch in the axle housing indicating break point for signal. Have the VSS pick up on how fast the signal should be picked up to determine wheel spin. Or have a VSS out back and one up front to compare signals like you originally said since it would take less programming for just using one sensor upfront.

[BAKES]

A friend and i are working on this i using the speed and distance senor and installing a abs sensor and tone ring on the rear hub but im not using the boost controler im going to use the soft rev or better yet the retard on the MSD box, if it sees a a 12 % slip on the front wheels it turn on the retard on to take to timing out but still keep boost up

[BAKES]

for the rear senor i am using the rear hub and sensor from a minivan that had a bendix 10

[90TurboVan]

Infrared sensor would be bad outside under a car, would get very dirty. ABS tone ring and VR sensor is your best option, one on a back wheel with one on each front wheel (unless you've got a LSD of some type). You could also use engine RPM and detect acceleration of the crank to determine wheel spin. The stock speed sensor is not a good choice since it's only picking up off of the passanger's side wheel. That'd be a manual only type thing though. Since boost is a very "fast" control, you'd have to have the controller learn the proper boost curve. Maybe have different boost maps for rain/dry/street/track since all are very different traction situations and having the controller re-learn for each would be slow. Probably just have it increase boost until it it gets to a certain wheel slip or decrease if it gets to wheel spin. Once it finds a value that works, that goes in the map and next time it just increases/decreases from that starting point.

From there, the electronics are the easy part. 3 VR conditioning circuits, a dsPIC, a low current transistor driver for the wastegate solenoid, a MAP signal input, and a good automotive grade power supply. (7805 with good filtering/zener clamps/current limit/reverse polarity protection)

As you can see, I've put some thought into this. Even tried making one a while ago, but didn't have quite the skills back then.

Edit:

You should probably watch EGT's if you're going to be retarding timing in boost at high power outputs.

[BAKES]

never had a problem with the egts when activating the traction control on the 911 twin turbo a 20 psi and we used to data log it it just seem to happen so fast it did not seem tweek the egt to much

[acannell]

yes the ABS tone ring/sensor is superior to an optical method...but where to get everything needed for an abs sensor setup for front AND rear (tone ring, hall effect...etc..)?

if we can get accurate wheel speed data for the front two and one rear wheel, we are in business i would say. controlling the wastegate solenoid should be simple, since the method used would be exactly like the stock ecu method (variable duty cycle with the stock solenoid)

lets do this!!!

im up for the electronics and firmware...i've got a full setup ready for programming and electronics development ( i do it for work )

we could always use the optical method to get started

[BAKES]

any caravan with bendix 10 uses the cv joints the have tone rings on the them , i cant remember if the sensor bolted to the knuckles or they have mounting hole cut into knuckle i would have to look . and for the rear just need the hub with tone ring and sensor

[Reaper1]

To get rear tone riungs just find a '90-'93 Y-body(Imperial/NewYorker) with ABS. The hubs interchange and they have tone rings built on them. The only thing is the sensors bolt on to the adapter plate...which is different from the smaller cars, so it takes some mods to get it to work(I've done it...fairly easy), but the brakes are also different. You *may* be able to mod the stock adapter plate to accept the sensor....

[Shel-Game]

It can be done with only the single SDS.

Generally speaking, the acceleration of the car is linear when plotted vs. the natuaral log of time. Any deviation from the line can be considered wheelspin. Because our SDS is on the RH output shaft, over the line is wheelspin by the RH tire, under the line would be the LH tire. Hard to explain without a chart, I guess (I can't post pics here). Anyway, take some speed data, calculate the avg accel for each period and plot that vs. log of time. You'll see what I mean.

I'm working on code to sense wheelspin in the Chrysler ECU.

The single sensor method will not be as accurate as a 4-wheel sensor system would be. You have to have 'error bands' above and below the accel line to judge wheelspin, and they need to be tuneable. There will be a certai amount of wheelspin that you will have to allow. Or, you will be cutting spark too often for fasle detections. It will be less accurate for auto cars launching becasue the torque converter efficiency is so low (hi slip). And, this method will only work when going straight ahead. If you are turning, it will sense the differential L/R wheels speeds as wheel spin. So, really, it would be a drag race application only. Not much use for road racing.

But, it can be done with a little math...

[Reaper1]

If you have a Pelquin, Quaife, or OBX, would the one sensor method not only be more accurate, but also work in corners(to an extent)?

I do like you idea! Is the acceleration plot figured out on the fly? If so, how?

[acannell]

wait a second..does the acceleration/single sensor method depend on the car being accelerated in a certain fashion by the driver? what if i decide to let off the gas for a second or im only using 1/2 throttle then I punch it, will these events bet seen as deviations from "the line"?

for that matter, wouldnt a four wheel setup see turns as wheelspin since different wheels would have different speeds?

[acannell]

okay so tone rings and sensors can come from:

caravan with bendix 10 (what the heck is that????)

for the front, apparently these are part of the halfshaft and may or may not be removeable froim the shaft, but i could just swap the shafts in. unclear if the sensor will mount or not to my daytona, for instance.

for the rear, the hubs apparently will plug right in, unclear if sensor will mount.


OR


'90-'93 Y-body(Imperial/NewYorker) with ABS

but again, the sensor may need a special mounting adapter


So it looks like the main issue is compatibility with sensor mounting. What is bendix 10, the abs code name?

[Shel-Game]

Quote:

Originally Posted by acannell wait a second..does the acceleration/single sensor method depend on the car being accelerated in a certain fashion by the driver? what if i decide to let off the gas for a second or im only using 1/2 throttle then I punch it, will these events bet seen as deviations from "the line"? for that matter, wouldnt a four wheel setup see turns as wheelspin since different wheels would have different speeds?

Well, you could look at throttle position, too and see that the throttle was lifted and assume that the change in wheel speed was due to the throttle change.

I'm not sure how a 4-sensor system would know if you were turning...

[Shel-Game]

Quote:

Originally Posted by Reaper1 If you have a Pelquin, Quaife, or OBX, would the one sensor method not only be more accurate, but also work in corners(to an extent)? I do like you idea! Is the acceleration plot figured out on the fly? If so, how?

Maybe, didn't think about that. You'd have to look at the acceleration and see if there is a regular trend.

You can calculate accel on the fly, but figuring out if it is wheel spin is a little more tricky. I'm still trying to work out the math. It's easy to detect looking at a complete run, but less so when looking at a small sample of data.