atleast-3

thats what i want to know, what purpose does it serve? and is it completly necessary to have on a track car versus a street car?

moridin2004

To zero out the bars... or to not. Most often, in my experience, is to make sure that there is no preload on the ARB (anti-roll bar) at static ride height. Of no use (most of the time) on a stock ride height car. Lower the car, then yes. Even more important when corner-weighting the vehicle.

atleast-3

so by preload you mean when you lower the car the tension thats put on the sway bar? so you have the adjustable endlink to like put it at a stock like location?

moridin2004

<TABLE WIDTH="90%" CELLSPACING=0 CELLPADDING=0 ALIGN=CENTER><TR><TD>Quote, originally posted by atleast-3 &raquo;</TD></TR><TR><TD CLASS="quote">so by preload you mean when you lower the car the tension thats put on the sway bar? so you have the adjustable endlink to like put it at a stock like location?</TD></TR></TABLE>

A simple explanation of how ARB's work is that when you compress the right front wheel it acts on the ARB through the end link. In turn, the ARB then acts on the left front wheel through the ARB. If the car is lower and the end links are the stock length, then the bar already has tension on it, which is generally not desirable. Adjustable end links allow you to take the tension out of your car for a specific static ride height. The other bonus is that you will go from a bushing material to a spherical bearing which has little to no deflection.

So, to answer your question simply, yes.

johnlear

<TABLE WIDTH="90%" CELLSPACING=0 CELLPADDING=0 ALIGN=CENTER><TR><TD>Quote, originally posted by moridin2004 &raquo;</TD></TR><TR><TD CLASS="quote"> If the car is lower and the end links are the stock length, then the bar already has tension on it, which is generally not desirable. Adjustable end links allow you to take the tension out of your car for a specific static ride height.
</TD></TR></TABLE>

Lowering the static ride height has no affect on ARB pre-load if the lowering is equal side to side, and previously no pre-load existed.

With the car parked on a level surface, if you fit one end link to the ARB, then found that in order to fit the other end link you needed to flex one end of the ARB up or down for the link to attach, then the ARB will be pre-loaded when everthing is in place. An adjustable link allows you to shorten or lengthen the link so that it can be fitted to the ARB without needing to 'force' anything in order to fit properly.

ARB pre-load can be a problem at any ride height, and most stock cars are likely to have at least some pre-load, which may get worse as the car gets older and things like rubber bushes compress / distort and move slightly away from their exact as new stock location (which will have some tolerance anyway).

I just fitted some new ARB bushes to my CB7 today, and while doing this found that with no links attached, when the ARB ends 'drooped' onto the ARB attachments on the control arm that the ARB contacted the control arm mount on one side before the other side, leaving a 5mm gap on that side.

If I had ignored this then when I installed the links and tightened it all up there would have been some pre-load in the ARB, so I fitted a 5mm spacer into the link to account for the difference.

A pre-loaded ARB will cause roll stiffness to be greater when turning in one direction vs the other, so the car will handle differently in left vs right hand corners. Just how differently of course depending on just how pre-loaded the ARB may be (and how stiff it may be, with the affect being stronger with a stiffer ARB).

moridin2004

I wasn't thinking clearly about my previous post. I guess I'm used to all cars after being lowered, having a preloaded ARB. The other concern is the end link angle, especially for through ARB style end links when lowering.

beanbag

<TABLE WIDTH="90%" CELLSPACING=0 CELLPADDING=0 ALIGN=CENTER><TR><TD>Quote, originally posted by johnlear &raquo;</TD></TR><TR><TD CLASS="quote">

A pre-loaded ARB will cause roll stiffness to be greater when turning in one direction vs the other, so the car will handle differently in left vs right hand corners. Just how differently of course depending on just how pre-loaded the ARB may be (and how stiff it may be, with the affect being stronger with a stiffer ARB).

</TD></TR></TABLE>

I don't think that it will, since all it does is preload the springs. It will make the car un-level, though, unless you compensate with adjustable springs.

Even if you have a twisted ARB:
Right wheel goes up -&gt; left wheel goes up
right wheel goes down -&gt; left wheel goes down



Modified by beanbag at 3:59 AM 2/24/2008

johnlear

<TABLE WIDTH="90%" CELLSPACING=0 CELLPADDING=0 ALIGN=CENTER><TR><TD>Quote, originally posted by beanbag &raquo;</TD></TR><TR><TD CLASS="quote"> I don't think that it will, since all it does is preload the springs. It will make the car un-level, though, unless you compensate with adjustable springs. </TD></TR></TABLE>

It pre-loads the ARB itself (how can it not, the ARB has a twist in it when connected), but is unlikely to be noticable as the ride height being uneven unless the pre-load is quite substantial. To some degree this will cause the springs to be unequally loaded, but again it won't be a significant affect unless the ARB is very pre-loaded.

Springs and ARBs do not do the same job, even if they both do contribute to roll stiffness in their own manner. If you start pre-loading springs in an attempt to compensate for a pre-loaded ARB then you're going to be introducing other problems on top of the existing one.

<TABLE WIDTH="90%" CELLSPACING=0 CELLPADDING=0 ALIGN=CENTER><TR><TD>Quote, originally posted by beanbag &raquo;</TD></TR><TR><TD CLASS="quote">Even if you have a twisted ARB:
Right wheel goes up -&gt; left wheel goes up
right wheel goes down -&gt; left wheel goes down </TD></TR></TABLE>

With a pre-loaded ARB, the chassis will already be 'seeing' some of the ARB stiffness before it even starts turning in a given direction, and the response will be sharper in that direction with less roll than in the other direction.

When turning in the other direction the chassis will initially be 'seeing' an effective 'negative' roll stiffness from the ARB, and the chassis will have to roll X amount before the ARB is even at zero roll resistance (i.e. reach a point where zero load exists on the ARB), and will have to roll more for the roll stiffness to be equal to that when turning in the first direction. Roll will be greater and response less sharp.

beanbag

<TABLE WIDTH="90%" CELLSPACING=0 CELLPADDING=0 ALIGN=CENTER><TR><TD>Quote, originally posted by johnlear &raquo;</TD></TR><TR><TD CLASS="quote">

With a pre-loaded ARB, the chassis will already be 'seeing' some of the ARB stiffness before it even starts turning in a given direction, and the response will be sharper in that direction with less roll than in the other direction.

When turning in the other direction the chassis will initially be 'seeing' an effective 'negative' roll stiffness from the ARB, and the chassis will have to roll X amount before the ARB is even at zero roll resistance (i.e. reach a point where zero load exists on the ARB), and will have to roll more for the roll stiffness to be equal to that when turning in the first direction. Roll will be greater and response less sharp.

</TD></TR></TABLE>

The only part of the chassis that "feels" the effect of a twisted ARB is the LCA's and the rear crossmember. It is an internal preload/force, and the rest of the chassis does not "know" about it. I still disagree that the roll stiffness is different in the two directions. Whenever you add springs together (at least linear ones) then you always ADD spring constants, i.e. you can never make a spring softer by adding an opposite one. All you will do is preload it.

For example, if you look at the force on one wheel, it is

F=k0(x0-x)+k1(x1-x) where
x is the displacement
k0 is the spring rate of the spring
x0 is the natural length of the spring
k1 is the effective spring rate of the ARB
x1 is the displacement of the other wheel.

The roll stiffness is dF/dx = -(k0+k1)

In other words, you can set x1 to whatever value you want (e.g twisted ARB, same as having the other wheel displaced higher or lower). It will provide positive or negative preload but does not affect roll stiffness. Another way to put it is that the roll stiffness does not change when x passes thru x1.

Your point is valid if you assume the ARB is a non-linear spring, e.g it is softer near zero twist, because the bushings are softer when unloaded, or if the ARB is loose when unloaded.

johnlear

Beanbag,
I'm not going to try to refute your maths, my maths isn't good enough to make proper sense of yours in any case!

I'll relate what I've been doing with my car today.

For a while now I've had a minor problem with the steering being significantly lighter in left hand corners and heavier in right hand corners, accompanied by the steering response and handling being significantly sharper in right turns and less so in left turns, and the car also being a bit more 'rolly' than it used to be. I was pretty sure it was to do with the cracked up OE ARB rubber bush I recently found (i.e. 1 dead bush, I've just replaced all 8 with new poly bushes).

While I was replacing the bushes the other day, I noticed that when the ARB was allowed to droop onto the control arm ARB mounting points that one end of the ARB touched it's control arm before the other leaving a 5mm gap on the right hand side (this was on level ground etc, but with no driver in situ, which may have some import).

I then checked this with various measurements taken with both links attached and then only one link attached to verify the 'correct' size of the discrepancy, and it was the same, i.e. 5mm gap no matter how measured. I then shimmed the gap with a 5mm spacer in the link so no pre-load existed when both links were attached.

Driving the car the steering was still noticably lighter when turning left and noticably heavier turning right, and the car cornered substantially 'flatter' when turning right vs left. Steering and handling was still more responsive in right turns, and less so in left turns. A bit confused at this point.

Thinking I might have messed something up in my measuremnts I added another 5mm spacer on the right link expecting this to most probably fix the problem (because my thinking suggested that increasing link length on the right side should pre-load the ARB in such a way as to increase roll stiffness in left corners and decrease it in right corners, which is what I was attempting to do) , but, the extra spacer made the problem worse! More confused.

On an impulse I then removed all added spacers on the right, and fitted one 5mm spacer to the left ARB link. Problems now gone; steering, handling and roll to left and right almost exactly the same (may need some very slight tweaking, but I haven't driven it enough as yet to decide). The car is generally much nicer to drive even if it now isn't so good as it was in right hand turns. This is all quite marked, definitely not a placebo affect...

I have to say I'm now still confused, because my thinking suggested that with the right link lengthened the pre-load should be increasing roll stiffness in left hand corners and decreasing it in right hand corners, but it seems to do the exact opposite!

At any rate, from today's experimentation I'm still pretty convinced that pre-loading the (relatively stiff) front ARB by even a relatively small amount has a surprisingly strong affect on left vs right roll stiffness, and also significantly affects other things such as steering weight (left vs right) and general reponsiveness left vs right (the rear ARB is a lot softer so changes may not be so obvious).

I am however still trying to figure out why the affect seems to act in the opposite direction to that which my theorising was suggesting it would???

beanbag

How does it drive with the ARB disconnected?

johnlear

<TABLE WIDTH="90%" CELLSPACING=0 CELLPADDING=0 ALIGN=CENTER><TR><TD>Quote, originally posted by beanbag &raquo;</TD></TR><TR><TD CLASS="quote">How does it drive with the ARB disconnected?
</TD></TR></TABLE>

Don't know as I haven't tried it, but I suppose I ought to for diagnostic purposes.

I do know that all other alignment parameters are good. I did a lot of work on checking alignment when trying to track down a mystery pull some while ago (which turned out to be tyre related), even to the point of checking all the chassis spring perch heights (correct to 1mm), spring lengths, measuring relative thickness of rubber spring seats, defomation of all the damper lower rubber bushes etc etc. The dynamic problems I've apparently cured with the 5mm spacer involved steering weight / response, chassis response, and roll, but not pull (i.e. despite the other problems, the car wasn't actually pulling to one side).

The car has almost 5° of caster angle (modified front radius rods), which does make the steering a bit heavier and any difference in left vs right 'weight' more noticable than perhaps it would be with the rather minimal stock caster angles and the overly light steering that the stock caster creates.

I have to say that I'm very happy with the result of what I've done with the ARB spacer, but frustrated that I'm having trouble explaining why it seems to work in the opposite direction to my expectation! I'm also at something of a loss as to why adding any ARB preload on any side is making the car so much more consistent left vs right, when I'd expect having zero pre-load would be the desirable condition.

I'm going to think some more about what you said in your ealier post to see I can deduce some "internal pre-load" involving the ARB and springs etc. I haven't really examined the contents of your earlier post properly yet, I was too tired when I read it earlier. Maybe I'll finally have to order that copy of 'Milliken and Milliken'?!

atleast-3

well ive got a question for you john, if to undid the endlinks and let the arb droop one side touched the control arm and the other was 5mm higher no matter what you did.... then wouldnt that mean that your arb is twisted? isnt the arb under tension, so if it were twisted then that might create a feeling of turning better one way then the other, because one side would be under more of a load? maybe by disconnecting the arb you would be able to see if the car handled the same left to right, even though it would really tail happy, assuming were all talking about the front arb which i think we are.

beanbag

CAn you give more detals about your suspension setup?
So if I understand correctly, if you don't add any endlink spacers at all, it tries to pull the right wheel upwards?

johnlear

If we’re taking about single wheel bump or in roll, then it’s (as far as I can see) impossible to add a ‘negative’ spring into the system by any means other than by a pre-loaded ARB.

It is an internal preload/force, but the pre-loaded ARB effectively 'adds' an initial negative rate on one side and an equally increased positive rate on the other. The sum of left + right total spring + ARB rate can't change, but the pre-load will alter the total rates at each side individually in an equal but opposite manner.

If the crossmember 'knows' about the pre-loaded ARB, then by definition the 'rest of the chassis' must also 'know' about it (the crossmember isn’t free floating, it’s an intergral part of the chassis). Increasing ARB spring load lifts the chassis (at the crossmember via the ‘D’ bush location) on the side with the lengthened link (I can easily see this affect on my car, especially when I experimentally lengthened the link even more), and by implication must lower it on the other side with the stock length link (I can't say that I actually noticed this before I took the spacer out). I think this must affect diagonal weight distribution, i.e. if we lengthen say the left front ARB link then the LF wheel gets heavier as does the RR, while the RF and LR get lighter.

So what does changes if roll stiffness doesn’t change? I suspect you’ll suggest that side to side ride height will change!

I would suggest that when pre-loaded the ARB is effectively a non linear spring, at least insofar as (in roll) on one side of the chassis it starts roll motion with an effective negative rate which then changes to a positive rate as roll progresses, and vice versa on the other side of the chassis.

If we have a pre-loaded ARB then the chassis will have to lean by X degrees before that pre-load becomes zero, i.e. before the ARB becomes unloaded and thus is transferring no loading (positive or negative) from side to side, and adding / deducting zero rate from that provided by either coil spring. The side with the longer link must see an increase in overall rate, and the side with the shorter link must see a reduction in overall rate.

I think this is a non linear affect, e.g. as the chassis initially rolls toward the side with the shorter link this softer side moves in bump with the ARB effectively creating a negative rate of it's own that deducts from the overall positive rate (i.e. effectively deducting from the coil rate), until the ARB reaches zero pre-load and then starts to add a positive rate to the overall rate. If this is happening at the soft side, it must be happening in reverse at the stiffer side with the longer link. By the time all negatives and positives have been ‘equalized’ the chassis already has a degree of roll on board, or a degree of less roll in the other direction.

This idea that increasing ARB pre-load creates an increase and decrease in overall rate on each side has been backed up by my recent experience, i.e. lengthening the left side link has made the left front suspension feel noticeably stiffer (in general), whereas before it was feeling softer (to the point I was wondering whether the left front damper was seriously damaged). I drive on bumpy roads a lot, so have had plenty of opportunity to feel this affect with the lengthened left side link, and bumps that previously felt 'soggy' when hit by the left front wheel now feel significantly 'sharper' (I can actually feel the smaller left side bumps quite clearly whereas I couldn't before). Previously the car has been permeated with this vague feeing of the LF suspension being too soft, and now with the 5mm spacer it feels much more equal side to side.

I still suspect I may have a LF damper that is too soft in bump relative to the RF. I hit a really big bump very hard a couple of months ago with the LF, and it subjectively hasn't felt quite the same after that, though the alignment checks out OK. However, I can feel the LF is just a little softer than the right front when I push down hard on the fenders. This may be muddying the waters with my perception of roll being oppositely affected to my expectation with ARB pre-load, but I still haven't figured out why it would.

Perhaps the roll affect I’m experiencing really is a placebo affect caused by the fact that pre-loading the ARB with a longer left link has increased steering weight in left corners (which is definitely a real affect). Still experimenting, or will when / if it ever stops bloody raining!

beanbag

I appreciate the time you took in thinking about this and doing careful testing. Anyway, I think you have spring rate and force/preload confused with each other. Spring rate is the derivative of force, i.e. how much the force changes when you change the displacement. So for example, consider the case of a wheel with a preloaded ARB. The spring wants to push the wheel down with a force of Fs, and the arb wants to pull it up with a force of Fa. The total force is F up - F down =Fs - Fa.

The spring has a stiffness of ks and the ARB has a stiffness of ka. What happens when you push up on the wheel?

The spring gets compressed by more, and it INCREASES it's downwards force by ks.
The ARB gets unloaded and REDUCES it's upwards force by ka

So the total downwards force is Fs + ks.
The total upwards force is Fa - ka.

So in the end, the total force is F up - F down = Fs - Fa + ks + ka

Now to get the effective rate, you subtract the final force from the initial force, and you end up with ks + ka. The point I am trying to make is that relaxing a stretched spring gives the same RATE as compressing an already compressed spring, even if they give opposite FORCES/preload.

Now going back to your case...
Any time theory and experience clash, theory has to go back to the drawing board.
So yes, a twisted ARB will make the car sit crooked. This might affect your corner weights, which affects left to right steering.
The other thing I can do is refine the math and say, fine, you have a non-linear swaybar which gets softer as you unload it because of some free play in the bushings. But when I say softer, I mean that the stiffness decreases . The rate/stiffness does not and never will go negative just because I twist the ARB the other direction. It is the force that goes negative, but we aren't talking about forces, we're talking about rates. So anyway, yes, there will be a certain softness in the roll if you lean the car one way or the other and unload the ARB. However, it is still true that an ARB, even if twisted, even if the bushings go soft, will never ever decrease roll stiffness.

After all this lecturing, I still have to agree with you that setting the endlinks to unload the swaybar when the car is level should create equal roll stiffness in either direction .In the end, I'm just as confused as you are!


Modified by beanbag at 9:41 PM 2/27/2008

johnlear

<TABLE WIDTH="90%" CELLSPACING=0 CELLPADDING=0 ALIGN=CENTER><TR><TD>Quote, originally posted by atleast-3 &raquo;</TD></TR><TR><TD CLASS="quote">well ive got a question for you john, if to undid the endlinks and let the arb droop one side touched the control arm and the other was 5mm higher no matter what you did.... then wouldnt that mean that your arb is twisted? isnt the arb under tension, so if it were twisted then that might create a feeling of turning better one way then the other, because one side would be under more of a load? maybe by disconnecting the arb you would be able to see if the car handled the same left to right, even though it would really tail happy, assuming were all talking about the front arb which i think we are.</TD></TR></TABLE>

The <U>front</U> ARB wasn't higher on one side no matter what I did, but no matter how I measured it (but I didn't measure in every possible way). What I did was to fit a 5mm spacer and eliminate the gap.

The ARB may be twisted (i.e. permanantly deformed or incorrectly made), or something else may be going on such as inadequate measurement procedure (that would be me), or worn / deformed bushes somewhere, or a sagged spring etc (last time I had the springs out each pair was exactly the same length, but that was a few months ago).

Ideally the measurement should be made with all four wheels on very level ground, but since I had no such very level ground I had the two front wheels on the ground (on ramps actually) but with the rear of the chassis jacked up from the centre (from the tow bar) so the chassis was more or less level with both rear wheels were hanging free. In theory this ought to place the front wheels under an equal loading as they woud be with all four on level ground, but to exactly duplicate all four wheels being on the gound it does assume no diagonal pre-loading to exist in any way when all wheels are on the ground (with one end of the ARB disconnected). This may not be the case, but I don't see why it wouldn't be. I did the same thing in reverse to check for rear ARB preload and found none.

Today I took all added spacers out to see what this does (back to the 5mm gap on the right), but it's pissing down rain today and I couldn't drive safely at any significant speed through enough corners to generate enough significant weight transfer and see what this actually did (can you spell 'aquaplane', I mean it's really raining today!).

All I can say is that with no added spacers on either side as it is now today, the steering weight still seems equal side to side. It may be that the original 'lightness' steering to the left was a product of the dead ARB bush on the right side, but it does still seem strongly associated with the front ARB. I've yet to try it with the ARB completely disconnected.

At this point further experimentation is driven solely by intelectual curiosity to solve a puzzle (and to further improve my present understanding of chassis dynamics), more than any real need to fix any significant handling / steering problems, but will have to wait for drier weather and free time co-inciding.


Modified by johnlear at 11:30 PM 2/27/2008


Modified by johnlear at 11:34 PM 2/27/2008

meb58

The force vector changes when one endlink is shorter than the other.

In any event, to the OP, set the endlinks as vertical in two axis as possible and try to make sure the bar ends are horizontal...horizontal may or may not be possible, but this orientation will give you a feel for how long the links should be given your ride hieght. You can then play with the endlink length - equally side to side - if you want to feel how a longer vs shorter measurement affect the 'laziness' or sharpness in response since the length of the endlinks affect the ARBs leverage - bar ends relative to the torque tube location - this relates to the opening sentence.


Modified by meb58 at 8:55 AM 2/28/2008

johnlear

Thank you for your input (you bastard, you’re making me think hard!).

I’m not sure my testing has been careful enough, and I suspect some of my assumptions (or at least one) may be incorrect. In particular I’m thinking some of my perceptions may have been clouded by the original dodgy ARB bush (one of the little ‘donut’ ones). Considering this, in the following I’m using an example of a preloaded ARB with a lengthened left side link, creating greater ARB preload on the left side. I’m not meaning my particular car, but a hypothetical example.


Hmm. I’m having a little trouble with your explanation, but if I’m interpreting you correctly;

I can easily get my head around this with an isolated suspension on one side of the car. E.g. if we were to have the coil spring, and a supplementary spring (that would look a lot like an ARB but with the ‘other’ end anchored to the chassis and not to the other suspension, i.e. a torsion bar acting only at one side), then if we preload this TB by lengthening the connecting link then the chassis will rise (in the same way as raising the coil’s seat would cause the chassis to rise).

But, the coil will also decompress and lose an equivalent ‘pre-load’ (i.e. +X at TB –X at coil = 0 pre-load gain, despite the chassis rise, i.e. the only position at which increased pre-load exists would be with the suspension at full droop and no weight on the wheel).

All we achieve is to raise the ride height and redistribute some of the coil’s static load onto the TB. The overall spring rate (coil + TB) doesn’t change. The same will happen in reverse if we shorten the connecting link.

After much thought experiment trying to disprove this with an ARB rather than a supplementary TB, I’ve concluded that this must also be the case in roll with a true ARB, by which I mean I’m agreeing with you. The ARB can’t affect (because of pre-load) overall spring rate on either side of the suspension to which it’s fitted, merely raise and lower the chassis on each side (with pre-load). However, I don’t think this is the entire story.

If we were looking at a hypothetical vehicle with only two wheels on a single lateral axle line, then with a preloaded ARB there would be no affect on weight distribution either (only the lateral ‘lean’ of the chassis), but, a real car has two axle lines and four wheels, which increases the complexity of the puzzle.

Because Newton was very smart (if a bit weird and rather batty), we know that; “for every action there is an equal and opposite reaction”. So, raising say the LF front corner of the chassis (by lengthening the ARB link on the LF side) must cause the diagonally opposed chassis corner to lower and this will load the spring at the RR corner and compress the spring. However, this must also mean an increased load on the spring at the original LF corner passing back from the now more heavily loaded RR (while simultaneously unloading the RF and LR to some degree).

In this case (with a real three dimensional car with four wheels) we must be looking at the pre-loading of the front ARB to be causing the LF suspension (as a whole, i.e. coil + ARB) to effectively be more loaded than the RF, and the RR to be more loaded than the LR.

This won’t change the roll couple as such (as we’ve decided that pre-loaded ARBs don’t change roll stiffness), but will mean that in left vs right hand corners at X lateral weight transfer (i.e. total weight transfer front + rear at Y lateral acceleration) weight carried on the rear wheels will be different (left vs right), as will the weight carried on the front wheels.

This is because the weight transfer itself doesn’t change, but the static weight distribution starting point is different in left vs right corners. With a front ARB preloaded to the left (i.e. longer left link) we could expect slightly more under-steer in right hand corners because at X lateral weight transfer the rear wheels would be more equally loaded (and the front wheels less equally loaded) than will be the case in left hand corners.

This may be accompanied by a perception of roll stiffness being different in left vs right corners, and may feel as if there is less roll in right hand corners and more in left hand corners simply (there’s a dangerous word!) because front roll stiffness is typically greater than rear roll stiffness, i.e. the car may feel to roll less in right hand corners than left hand corners….?

See above desperate theorizing. I think the premise that a pre-loaded ARB doesn’t affect total corner suspension stiffness or roll stiffness only holds up when considering a single axle line in isolation, when we add another axle line it’s not so simple (that dangerous word again!). I could be wrong!

The affect on steering weight in left vs right corners is I think due to the pre-loaded ARB causing greater weight to be carried on the LF than the RF (as in the e.g. I’ve been using above). I think this is also associated with caster angle (and possibly increased by interaction with Ackermann affect and KPI - king pin inclination, or aka steering axis inclination).

As the steering is turned, caster angle causes the part of the contact patch that is to the outside of the steering axis on the IF wheel (‘IF’ being to the inside of cornering direction) to lower relative to the chassis (i.e. since the road is an immovable object caster lifts the chassis on the IF, and has an opposite affect at the OF). Ackermann increases this affect at the IF by turning the IF further than the OF. KPI decreases this affect at the OF (maybe even negates it depending on the relative values of caster and KPI), but increases it at the IF.

This is also affected by the steered camber changes caused by caster angle (greater with greater caster), the IF gaining more steered pos camber and thus gaining increased positive scrub radius (effective if not nominal), and the OF to gain effective negative scrub radius due to neg camber gain These affects may also be increased by use of higher tyre pressures (causing the IF especially to carry more weight toward the outside edge of the contact patch).

The net affect (greater with greater caster) is for steering action to lift the chassis significantly more at the IF chassis corner than the OF corner, which also loads up the IF contact patch and spring etc more than at the OF (another good affect of high caster angle). Lifting the chassis requires force input from the driver, and the reactive force from the chassis is greater at the IF than OF.

With greater left side ARB preload the IF will be carrying more static weight (than the IF would be when turning to the right), and this greater weight will exist for any degree of steering input as the driver physically lifts the IF of the chassis.

This underlying static weight (when added to steering weight caused by ‘lifting’ the IF corner of the chassis) will be perceived by the driver as greater steering weight in left hand corners than right hand corners, because when steering right the IF will start off being lighter and remain lighter for X degree of steering input. This is why (I think) increasing ARB pre-load on the left increases steering weight to the left and decreases it to the right.

The only other source of ARB ‘progressive rate’ that I can think of would be dynamically changing angularity of the linkages, which I think will tend to effectively increase ARB rate as the angle between the ARB and the links moves farther away from being 90°. This would occur with any degree of roll, but unless roll was substantial (and / or the lever arms on the ARB particularly short) I can’t see a significant affect.

As I said, I think I have to revisit my experiment more rigorously. It was only an accidental experiment in the first place, and conducted in a rather ad-hoc manner. Having said that, it did seem a real affect (i.e. the apparent reduced roll to the left when I pre-loaded the ARB on the left side (or ‘un-pre-loaded’ it on the right side). I’m not utterly convinced just now that it wasn’t a placebo like affect, though you’d think a placebo affect more likely when the result coincides with expectations, not against them!


Modified by johnlear at 5:09 AM 2/29/2008


Modified by johnlear at 5:12 AM 2/29/2008

meb58

You guys are killing me...in a good way. In any event, if one endlink is shorter than the other, or one longer than the other, a preload will exist and the car will handle asymetrically thru turns.

johnlear

Just a stray thought on the driver's perception of roll. Considering that the driver is located substantially off centre, could it be that for a car that is known to roll equally in left and right corners, that the driver may still possibly percieve it to be different in left vs right corners?

This might be because when the car rolls in one direction the driver's body will move downward with the roll motion, but when steering the other way the driver's body will rise with roll motion. This alone might have a subjective affect on the driver's roll perception even if the lowering and rising motions the driver is subject to were equal in magnitude, but in reality they aren't.

Typically (for X lateral acceleration in either direction) the chassis will lower on the outside more than it rises on the inside (which is due to not all weight transfer occuring through the 'elastic' vector of the springs, but some also ocurring through the vector of the suspension geometry), so in fact when the driver is located to the outside of the corner he/she will move vertically more so than he/she will when located toward the inside of the corner. So, we have equal chassis roll motion in either direction, but even so the driver is subjected to unequal vertical motion depending on which direction the chassis is rolling, giving a sensation of greater chassis roll stiffness one way vs the other.

My car is right hand drive, so in my case there may be a tendency for roll to feel stronger in left hand corners than in right hand corners (as my body will lower more in left hand corners than it will rise in right hand corners)? This may give rise to a false sensation that the car has more roll stiffness when steering to the right than to the left? Of course if this is correct then reducing roll in actuality will reduce any false perception of roll stiffness being unequal in left vs right hand corners.

beanbag

<TABLE WIDTH="90%" CELLSPACING=0 CELLPADDING=0 ALIGN=CENTER><TR><TD>Quote, originally posted by meb58 &raquo;</TD></TR><TR><TD CLASS="quote">You guys are killing me...in a good way. In any event, if one endlink is shorter than the other, or one longer than the other, a preload will exist and the car will handle asymetrically thru turns. </TD></TR></TABLE>

If the ARB is twisted, in what way will the car handle asymetrically and why?

meb58

in a friendly way, you guys know how a force vector diagram works...if one endlink is shorter the bar is twisted - lets assume on level ground. And without answering the question directly, adjusting both endlnks to be as long as possible vs both being adjusted as short as possible will result in different behavior - one will be lazy and one will be more responsive...thus, one short and one long will result in lazy response in one direction and more response in the other - all else equal. The leverage changes.

beanbag

John,

From your posts, it is clear that you know waaaaay more about suspension than I do, since you are able to bring up all these different effects and how they all inter-relate to each other. I pretty much agree with everything you mentioned regarding cross weights, KPI, etc. You may be on the right track to figuring out your steering asymmetry. I also agree with your notion of driver perception. Sometimes when I am trying to test the damping of my shocks by swerving right or left, I find that I am always more aggressive about swerving left (which puts me in the opposite lane) vs swerving right, which puts me off the road.

Here's some other things to consider:

Does your car pull to one side?
If you make one side of a car heavier, will it pull towards that side, away, or do nothing?
One of my friends with a Corvette once had his steering rack taken off and put back on. Afterwards, he complained that steering was more difficult in one direction than the other, even though the car did not pull to either side.

johnlear

I suspect it’s related to pre-load of some sort, but it might not be the ARB, it might be pre-load in the struts, possibly caused by unequally deformed lower strut bush rubbers? This might go some way to explaining the 5mm gap to front ARB link on one side…

At any rate the problem is subtle, but definitely exists. It’s one of those problems that you think you might have fixed when you do X to the car, but after a while you realise that it’s still there, and that all you were feeling just after you did X was that something was slightly different.

Is that really a perception of roll, or just that steering toward oncoming traffic is more scary than steering toward a tree, or vice versa?!

Yes, but it’s not a straightforward story.

It now pulls ever so slightly to the right, but it used to pull more strongly to the left. I chased this for ages checking every bloody thing I could think of very carefully, as well as making experimental asymmetrical changes to various alignments (front and rear) and even deliberately preloading the upper spring seats to create pre-load. Nothing I did stopped it pulling to the left, though with some changes it did get slightly better in this respect, but always became worse in some other way.

So, I set everything back to normal and just put up with it pulling left (which was really irritating). One day I switched the front tyres side to side, not for any strong reason just as an experiment the result of which I had no idea. Problem instantly and unexpectedly cured, the car steered dead straight with no pull at all. However, it wasn’t perfect, the steering still felt lighter when turning to the left, and the car still felt to roll more turning to the left, and generally steered and handled less ‘sharply’ turning left than right.

The other week I found a split front ARB link bush. Replaced all link bushes with new poly items and the car now rolls less in either direction, but still feels a bit less ‘sharp’ in left vs right corners (though less difference left vs right than there was before). However, the car now pulls very slightly to the right.

This actually makes sense now. When I swapped the front tyres side to side and lost the pull to the left, it was obvious that the left pull must have been tyre related (i.e. not wear as both looked identical, it had to be tyre ‘conicity’ being unequal on each front tyre). It did puzzle me that with this obviously being the case, why then didn’t swapping the tyres side to side cause a pull to the right? Now I think it must have been that there was some pre-load affect caused by the split ARB bush that was causing a counteracting pull to the left, i.e. cancelling out what theory told me should have been a tyre related pull to the right.

This must also be why the original left pull was stronger then the right pull that I have now (which is very slight), i.e. there were two different things combined causing the original stronger left pull, i.e. the tyre conicity and the pre-loaded ARB added together (i.e. the effective pre-load caused by the split bush).

Why are things never simple?

I still have the steering that feels a bit lighter left than right, and the handling is still not quite as sharp left vs right, but I’m determined to figure this out!

In theory it depends on how much scrub radius and caster is present. An extreme example that makes the principle easier to understand is karts, which have quite large caster angles (roughly in the 10° to say 15° range) and massive scrub radius. The steering axis intersects the ground way to the inside of the contact patch, by which I mean not in the contact patch area at all but several inches to the inside of the inner edge of the contact patch (i.e. huge SR).

The rearward leaning caster angle (in conjunction with the SR) creates a situation whereby when (say) the left wheel is steered to the left it lowers relative to the chassis, and the right wheel rises relative to the chassis (this affect being stronger the greater the caster angle and SR are). Since the ground refuses to move, the affect of this is lift the chassis at the IF, and to ‘jack’ weight from the OF wheel onto the IF wheel, and also from the IR onto the OR wheel.

This affect is multiplied by the scrub radius, i.e. the greater the SR the greater the steered wheel height changes will be at both front wheels (they change height substantially, though less so at the outside wheel and more so at the inside wheel due to the affect of KPI adding and subtracting steered height change on each side).

These fairly extreme front end geometries are used on karts to create a weight jacking effect at turn in to unload the IR wheel. This is because karts have no differential and very grippy large rear contact patches. If the steered weight jacking effect didn’t exist then the IR wouldn’t be substantially unloaded (jacked) at turn in, and the combined grip of both equally loaded rear tyres (both unalterably rotating at the same rpm) would create drastic understeer.

With a twisted kart chassis (analogous to one side of a car having a pre-loaded spring or ARB), one front wheel will be higher than the other, and when you place the kart onto the ground one front wheel will contact the ground before the other, start taking load before the other and the steering will swing away from the side of the chassis that is lower when unladen (with front wheels pointing straight ahead) toward the side that is higher (i.e. the front wheels will try to find a common height / load which they can only do by turning away from the straight ahead).

With both front wheels turned back to the straight ahead position the ‘lower’ wheel will be heavier than the ‘higher’ wheel. When driving the kart it will want to steer toward the side that is higher when unladen (i.e. the toward the lighter front wheel), and the driver has to ‘steer’ it the other way to track straight, i.e. the kart will pull toward the side with the lighter front wheel.

However, if SR is zero, then the affect will disappear because the caster angle can’t then cause a steered height change at either wheel, so even if one wheel is heavier than the other there is no force causing a pull. The same would be true if SR were present but caster angle were zero (though in this case some slight pulling affect may exist toward the heavier wheel due to increased deformation and rolling resistance at the more highly loaded tyre, and be somewhat stronger the greater the SR was).

Most FWD cars have SR at or close to zero, so the above affect will typically be zero or small with most FWD cars, i.e. when travelling in a straight line any pull caused by pre-load in the springs or ARB (or chassis twist) will be slight at most. However, the steering may still be heavier in one direction than the other. This is because even with nominal zero SR caster angle can still cause a weight jacking affect associated with steered camber changes caused by caster angle, and these camber changes can affect the ‘effective’ SR.

As the IF wheel is turned it either loses neg camber or gains pos camber, and as this happens the area of greatest loading on the contact patch traverses across the face of the contact patch so that ‘effective SR’ changes (i.e. becomes other than the nominal SR because the actual centre of contact patch loading changes position as camber changes).

The opposite affect occurs at the OF wheel. The affects at each front wheel are opposite but not equal because Ackerman causes the IF to be steered more than the OF, and KPI creates more camber change at the IF and deducts camber change at the OF. The net effect is that even though nominal SR is at or near zero the chassis is still lifted at the IF and lifted less or not at all at the OF. This affect is stronger with wider treads and stiffer tyre casings, at least one reason why wide tyres tend to 'self center' a bit more strongly than narower tyres.

Because of this, if the chassis is pre-loaded, then it will be easier to lift when steering in one direction than in the other direction, thus causing steering to be slightly heavier one way vs the other.

Why are things never simple?

I can’t say whether this might be related to unequal side to side weight, or unequal geometry, or some non linear binding or whatever in the steering mechanism itself…



Modified by johnlear at 8:11 AM 3/9/2008