Caster vs KPI

[Tom Wettenhall]

Woodsy: As I understand it, KPI has two effects. It reduces scrub radius, which reduces steering effort, and introduces a component of vertical movement of the tyre with steer angle, which increases effort. (delta y dependent on scrub radius)

So, which one wins? That is, when you dropped the KPI on your car, did the steering effort go up, or down? (I know it's dependent on the specifics of that car's steering, but I'm just interested)


As an aside, running large amounts of KPI is going to do interesting things with diagonal load transfer.

[woodsy96]

Tom, the KPI was 0° before I joined the team and remained right through till after I left the team. So I can't comment directly on the difference inducing KPI has.

However, from conceptual point of view with regards to you vertical movement, if you imagine an extreme situation with

-Caster = 0°
-KPI = 6°
-Parallel steering

If you turn the steering wheel the delta z (I think that's what you mean by delta y) both wheels go up and down the same amount (which, as you rightly say, increases steering effort to do the work to raise the car dz mm). So, on that front, I am not 100% convinced on first thought (maybe some more thought is required on my part) that it does interesting or useful things with diagonal load transfer, in isolation.

If you look at the other extreme of
-KPI = 0°
-Caster = 6°
-Parallel steering

Then when you turn the steering wheel one wheel goes up (the outside wheel), the other goes down (inside wheel). Seems to me that this reduces the LLT on the front "axle" and increases the LLT on the rear "axle," and this load transfer is very close to instantaneous. Seems like this could only be a good thing, especially for getting the car pointing into the corner when the driver first turns in.

Also because one wheel goes down and one wheel goes up the chassis does not increase ride height from steering (neglecting camber change across the tyre width raising the car). This still seen as work through the steering wheel by warping the vehicle. If I could trade all of my "non-advantageous" steering effort from KPI to "advantageous" steering effort of caster I would (and did )

As an aside note reducing scrub radius only reduces steering effort because of the dz it induced in the wheel, so it only reduces the steering effort caused by (dz due to steering) from the caster. It does reduce the amount of steering "kickback" from braking (and accelerating on FWD). I think you might have mixed that up with mechanical trail? (In plan view the longitudinal forces are reacted through the steering wheel about the scrub radius, and the lateral forces are reacted through the steering wheel by the mechanical trail. Simplified of course).

Or am I missing something?

[Tom Wettenhall]

I stand corrected on the KPI load transfer thing. Sorry about that.

By 'interesting' I meant 'almost certainly bad'.

We use OptimumK software for suspension design. Either we couldn't find the change coordinate system option or it doesn't exist, but the end result is we have z as longitudinal, x as lateral and y as vertical. Apparently I've gotten used to it to the point of using the wrong coordinate system outside CAD.

The caster jacking effect you refer to is what go karts use to unweight the inside rear tyre. It's less important on an SAE car because the torsional rigidity is generally much less, simply because of wheelbase, but you could make use of it on a smaller car or a CF monocoque car.
Another reason why we (MUR) don't use heaps of caster for jacking beyond the obvious steering effort is because in order to de-jack the wheel with caster, you have to centre the steering. If you use high rear roll stiffness to unweight the inside rear, loss of traction will remove the jacking force immediately, giving you one more tyre to assist in slide recovery. Since we have a four cylinder engine, we get a lot of power oversteer, so being able to manage it is good for us.

Scrub radius will increase steering effort, indirectly, because in a situation like corner entry where you have different braking (or driving) forces applied to each contact patch, a net torque will be applied to the steering which is counter to the direction of steer. Pretty much what you said about kickback.

[Eric Hutchinson]

Tom,

You should be able to change the coordinate system in OptimumK.
See here.

That's for version 1.1, but I remember it being similar in 2.0

Eric - University of Wisconsin-Madison

[BillCobb]

Come on boys and girls, you are still missing the important reason(s) for a specific caster setting.

Hint 1: It depends on the tire construction, rim width and pressure.

Hint 2: Its especially needed by a novice driver.

Hint 3: The subject is 'racecar', either spelled forwards or backwards.

You should also be told that on some 'racecars' the caster angle is different on one side of the car. Now why would that be?

[M. Nader]

Originally posted by BillCobb:
Come on boys and girls, you are still missing the important reason(s) for a specific caster setting.

Hint 1: It depends on the tire construction, rim width and pressure.

Hint 2: Its especially needed by a novice driver.

Hint 3: The subject is 'racecar', either spelled forwards or backwards.

You should also be told that on some 'racecars' the caster angle is different on one side of the car. Now why would that be?

I would guess in Nascar the caster left and right would be different to compensate for each other (camber change)?

[MCoach]

Originally posted by BillCobb:
Come on boys and girls, you are still missing the important reason(s) for a specific caster setting.

Hint 1: It depends on the tire construction, rim width and pressure.

Hint 2: Its especially needed by a novice driver.

Hint 3: The subject is 'racecar', either spelled forwards or backwards.

You should also be told that on some 'racecars' the caster angle is different on one side of the car. Now why would that be?

Bill,

I would venture to say that it is all about catering to the steering feel. The tire will reach peak Fy at a certain slip angle and what the driver feels will be dependent on the pneumatic trail (nonadjustable other than tire selection, rim selection, and pressure) and the mechanical trail. Mechanical trail will be most dependent on caster angle and then pin lead (which I've only used for smaller adjustments). The novice driver, will need a warning before the tires start to give way. The drivers know this as "the steering going all funny and light". By selecting camber at a certain angle, an engineer can give that warning, without overpowering the feedback from the tire completely.

I would also venture to say that some cars run different caster angles because that side is on a banking (such as a 'Merican NASCAR oval), thus needing to adjust the caster for proper steering feedback.

[Tom Wettenhall]

Eric: Cheers, thanks for the info. I'm no longer a component-designing member of the team (silly one year program) but I'll pass it on.

I can imagine for a car which is significantly heavier on one side, such as a car with a large single radiator on one side, mismatched caster may help even up the left/right turning response. Difficult to justify the extra design time for it though.

Would I be correct in saying the caster angle for best lateral grip would be dependent on tyre pressure as increased pressure reduces sidewall deformation slightly, and as a result the maximum contact patch area, hence grip, for a given lateral and vertical load occurs at a lower camber?

Where kingpin geometry is non-adjustable, is there another way to improve steering feedback for novice drivers? Would help with training people who've never driven fast before.

[Z]

Boring Olympic stuff on the box, so may as well have a rant...
~~~o0o~~~

All this talk about "more castor improves steering feel..." is A LOAD OF CODSWALLOP!!!

This, no doubt, is the result of the failed education system, neglect of Euclid, consequent lack of clear thinking, rigour, etc., etc. Mumble, grumble, grrrrr....
~~~o0o~~~

Here are some of the relevant parameters.

STEER-AXIS GEOMETRY. This axis requires 4 numbers to specify its position wrt wheel or car body, such as;
1. Trail (also known as Mechanical Trail, Castor Offset, and many other names).
2. Offset (aka Scrub Radius, etc.).
3. Castor (aka Caster, Rake, etc.).
4. Steer-Axis Inclination (aka SAI, KPI, etc.).
All of these are best defined with the aid of drawings. Unfortunately, they are most often undefined!.

WHEEL GEOMETRY.
1. Diameter.
2. Cross Sectional Profile (eg. width, etc.)

And a whole lot of other stuff, such as tyre construction, 4 wheel spring rates, etc.
~~~o0o~~~

IMPORTANT POINT. When the textbooks make the above claim about castor/steering feel, what they actually mean is,
"Making suspension adjustments that increase castor AND TRAIL gives improved steering feel because of the greater self-centring effect of the greater TRAIL."

Castor, IN ITSELF, gives very little self-centring (or "steering feel"). With narrow tyres, like on bicycles/ motorbikes, castor gives almost no self-centring at all.

Like I said, SLOPPY THINKING, not befitting of engineers!

All this sloppy "castor" talk is probably also influenced by the naming of the little wheels under office chairs, or shopping trolleys, as "castors". Bizarrely, these have zero Castor angle, but lots of Trail!
~~~o0o~~~

Anyway, if you want to understand "steering feel", then start by clearly DEFINING the important parameters. Also be very clear about any ASSUMPTIONS you make (such as "Adjusting this screw, changes ...." what???).

Next, start with all six of the above parameters (ie. 4 x steer-axis + 2 x wheel) set to zero. Then vary only one at a time, and figure out what happens. A toothpick (= steer-axis), ball of blue-tack (= upright), and a coin (= wheel), can help with visualisation.

So with the wheel reduced to a point (zero diameter and width), and with the steer-axis passing through this point (zero Trail and Offset), it should be apparent that any Castor angle, or SAI (or both), will have no self-centring, either static or dynamic. Add wheel diameter, and SAI gives static stability, but Castor does not.

On the other hand, a vertical steer-axis (zero Castor and SAI) positioned in front of the wheelprint centre (positive Trail) has no static stability. But it does have dynamic self-centring, which comes from the lateral Fy forces acting on the wheelprint.

Now consider positive Castor with positive Trail (still with zero sized, or small, wheel). This is statically UNSTABLE, but develops dynamic stability. Which is bigger? Simple quantitative calcs suggest the dynamic stability, from Fy acting on the Trail lever arm, is of order 100x the static instability, from Castor + Trail.
~~~o0o~~~

In other words, Trail gives lots of self-centring "steering feel", but Castor, in itself, does not. Call a spade a spade.

Z

[Mike Cook]

Z,

Nice overview. Have you ever driven a shifter kart? At least on the ones I have driven, statically the caster causes quite a bit of aligning torque. Granted, they have lots of caster and very rigid suspensions (frame flex), but extrapolating a bit, I would think that caster has more than an insignificant role in self aligning torque on a fsae car. I have a torque gauge on our steering setup, and if I remember I will put the car on slip plates and measure the torque vs SWA statically.

Mike