Measuring the angle of front wheels.

nikpro
Posts: 56
Joined: Tue Jan 23, 2007 12:29 pm

Postby nikpro » Wed May 06, 2009 6:23 pm

chrisp993 wrote:OK - angle at wheels should equal:

angle at steering wheel / steering ratio

Is this what you found?


Problem being is the ackerman angle gives different angles at the wheels - this is what I'm trying to calculate at the moment - it's going well and I'll update when we've finished testing.
Fraser E White

chrisp993
Posts: 47
Joined: Fri Apr 20, 2007 2:12 pm
Location: Bloomfield Hills, MI

Postby chrisp993 » Wed May 06, 2009 8:21 pm

I wondered about that - be really interested in what you find - please do update - I'd suggest including steering wheel angle in your data for comparison :idea:

osborni
Posts: 497
Joined: Fri Dec 07, 2007 9:08 pm
Location: USA, Michigan

Postby osborni » Wed May 13, 2009 1:36 am

If it is over steer that you want to look at.....

Image
Red = actual steering
Orange = Required Steering

Assuming that you have a steering angle sensor, either string pot or a linear pot, you just need to have a math calc to overlay them.

Required Steering = {(Wheel_base/[(Speed^2)/lat_accel]}*57.3

Taken from: "Data Logging Manual"

It figures out how much steering is theoretically needed given no slip. Over lay the actual input, and you have a comparison of calculated vs. actual.

Any large deviations means over or under steer.
BMW 2000 M Coupe

chrisp993
Posts: 47
Joined: Fri Apr 20, 2007 2:12 pm
Location: Bloomfield Hills, MI

Postby chrisp993 » Wed May 13, 2009 2:14 am

Yes - the calculation in my post above is a more accurate version of this same equation

STEERING_NEEDED = 15.8*atan(sign(VAR_0004)*2.301/VAR_0052)*(360/(2*3.142))

but what I found was significant deviation between STEERING_NEEDED and the actual STEERING_ANGLE. Much more than could be accounted for by oversteer or understeer.

The Data Logging Manual (great book BTW) makes the point (on page 68 ) that the problem with this equation is that it doesn't take account of slip - and at track speeds, slip angle should be pretty significant - at least a couple/few degrees, which (with steering ratio of 15.8 in my case) equates to tens of degrees of steering wheel angle, which is similar to the deviation I am seeing.

I'm pretty certain my calculated STEERING_ANGLE is correct - verified by checking in-car video - and this doesn't agree with STEERING_NEEDED. Your data seemed to show close agreement ....... low speeds? I noticed your overlayed graphs looked to have different axes / scaling - are they the same or different?

osborni
Posts: 497
Joined: Fri Dec 07, 2007 9:08 pm
Location: USA, Michigan

Postby osborni » Wed May 13, 2009 2:55 am

It was a 30mph 0.8 g chicane into the parking lot at work. :twisted:

The point isn't to get a perfect match in degrees of actual wheel turn to theoretical angle needed, but to look for large deviations at certain points. Unless you are over steering around the whole corner (that would be drifting, not minimizing lap times).

It's the relative difference that matters.

Under/over steer shows up in G plots, it's just more subtle. With the overlay between the two steering plots, you get a better picture of where that is and why.

I didn't work very hard to get them to overlay. Again, it's big deviations that matter, not absolute accuracy.

When I get a change, I'll do a calibration run down to the culdesac at the end of my street and see what happens.
BMW 2000 M Coupe

chrisp993
Posts: 47
Joined: Fri Apr 20, 2007 2:12 pm
Location: Bloomfield Hills, MI

Postby chrisp993 » Wed May 13, 2009 3:14 am

Well :roll: I can fudge mine to match through scaling differently - isn't this the same as building in a LAT_G dependent slip angle? The Data Logging Manual also describes several ways of fudging the data to do this - which is kind of a clue that the equation doesn't really work as described under track conditions :wink: And I agree with you that this does highlight major oversteer / understeer as deviations from the norm, but that was pretty much telling me what I should have seen from the LAT_G plot anyway.

I tried the calibration run you mentioned when I installed the string pot and things agreed fairly well - it was only when I got some on-track data that the match went to hell! At which point I started wondering why - which led me to slip angles

My curiosity is to pull info out of why and how they don't match i.e. how the variation in slip angle behaves and what that is telling me.

The problem I think is that any attempt to look at handling (oversteer / understeer) makes assumptions about slip angle - and vice versa. Since the handling and slip angles are pretty deeply interrelated.

pault
Posts: 64
Joined: Fri Nov 02, 2007 4:53 pm
Location: St Charles, IL

To what end

Postby pault » Wed May 13, 2009 7:29 pm

It sounds like what osborni is getting at is that the actual number of degrees of oversteer/understeer isn't really important. What is important is do you need to make a change to the car's suspension to make you faster or do you need to make a driving change to make the car faster.

Paul

osborni
Posts: 497
Joined: Fri Dec 07, 2007 9:08 pm
Location: USA, Michigan

Re: To what end

Postby osborni » Thu May 14, 2009 12:05 am

pault wrote:It sounds like what osborni is getting at is that the actual number of degrees of oversteer/understeer isn't really important. What is important is do you need to make a change to the car's suspension to make you faster or do you need to make a driving change to make the car faster.

Paul
Yes, thank you. Oversteer should show up as less actual steering then what is calculated, Understeer as more actual steering then needed.

The tricky part is figuring out that from normal G variation that the car will experience. In the trace above, the right hand turn goes over a smallish curb, so there is some variation in the g plot as the front and the rear tires go over it.
BMW 2000 M Coupe

nikpro
Posts: 56
Joined: Tue Jan 23, 2007 12:29 pm

Postby nikpro » Mon Jun 22, 2009 5:03 am

Got it working reasonably well with the front wheels callibrated on turnplates:

Image
Fraser E White

osborni
Posts: 497
Joined: Fri Dec 07, 2007 9:08 pm
Location: USA, Michigan

Postby osborni » Mon Jun 22, 2009 10:59 pm

Still not sure why you went into the trouble.

Using "turns" works just fine.

Evening at Gingerman two weeks ago.

The circles on the strip chart correspond to the circles on the track map.

Red = actual wheel turns
Orange = calculated from data
Blue = lat acceleration
Green = TPS

Image
Last edited by osborni on Tue Jun 23, 2009 7:18 pm, edited 1 time in total.
BMW 2000 M Coupe

osborni
Posts: 497
Joined: Fri Dec 07, 2007 9:08 pm
Location: USA, Michigan

Postby osborni » Tue Jun 23, 2009 12:29 am

Oh, the "Required Steering" formula I use is:

((CAR_1/((VAR_0015*1.46667^2)/VAR_0004))*57.3)/20

CAR_1 = wheel base
Var_0015 = speed
Var_0004 = lat accel

The /20 scales the numbers to match the steering angle which is really steering wheel turns.
BMW 2000 M Coupe


Return to “Sensor-related questions”

Who is online

Users browsing this forum: No registered users and 15 guests