Sway Bar Adjustment

[TimH]

It's like a balance or a see saw. 10 fat kids on the short side, 1 fat kid on the side that's 1/10th the distance from the fulcrum. Same balance as if you put 5 1/2 fat kids on each side with the fulcrum in the center.

I can't believe I'm arguing with Daryl...

Dave's point is that the 10 fat kids only move 1/10th the distance. He concedes that when we're talking 10% rather than 1000% the difference might not matter.

[BorkRacing]

First time since the FFU that I've ever disagreed with DaveW on some tuning/set up advice. Einstein and Edison, I'm certain were wrong at least once . Then there's me; a broken clock--I'm right at least once per day even if wrong all the other times.

Taking an extreme example: One lever arm 1" long and the other 10" long. The resulting degree of twist with any given force left or right hand turn is the same as when the two arms are 5.5" long.

If there is an upward force on the 10" long arm, the arm has lots of leverage on a very stiff "mount" (the 1" arm).

If you turn the opposite way and put the same amount of upward force into that 1" arm, you don't have near the leverage but your "mount" is very soft (10" arm).

The net result in terms of roll will be the "same". Where it gets wonky is not the difference in lever arm length but the angle that your drop links are acting upon those lever arms. When those drop links are vertical vs. more towards the horizontal.

It's like a balance or a see saw. 10 fat kids on the short side, 1 fat kid on the side that's 1/10th the distance from the fulcrum. Same balance as if you put 5 1/2 fat kids on each side with the fulcrum in the center.

No need to make it more complicated than it is.

And for the second time in 2 days I agree with Daryl...world must be ending...lol

[BorkRacing]

I can't believe I'm arguing with Daryl...

Dave's point is that the 10 fat kids only move 1/10th the distance. He concedes that when we're talking 10% rather than 1000% the difference might not matter.

Not talking how far the lever moves, talking what the total torque on the bar is. Teter-toter doesn't apply.

[TimH]

Not talking how far the lever moves, talking what the total torque on the bar is. Teter-toter doesn't apply.

We all agree the torque is equal unless there's binding in the mount. However, different amounts of motion left vs. right in the force transfer could lead to jacking, correct?

[Daryl DeArman]

I can't believe I'm arguing with Daryl..

Ha ha. Believe it or not, there have been instances where I've agreed to just disagree.

[problemchild]

Bringing a person's credentials into a technical discussion generally indicates that you have run out of relative points to contribute to the discussion. I have no engineering degrees but I like to learn from technical discussion.

I have found that an anti-roll bar is a single spring and will remain a single spring regardless of the linkage applied to it. If the linkage is not similar on each end it can still have the same net rate as if it has mirrored linkages. The spring rate curves with movement will likely differ with different linkages (which may or may not be significant in any application).

The main reason to have a 2 dimensional ARB with symmetrical linkage is that you can adjust it on the fly without scaling the car and resetting the linkage preloads. If your U-shaped ARB has symmetrical linkage, then you can set your cornerweights initially, then move your linkages uniformly, which will change the ARB rate, but not the cornerweights.

Another example of this is the more modern ARB systems with rotating blades. Some have just one blade, or blades of different rate. The complete ARB assembly works as a unit ...... not with one corner having a different rate than the other.

Another example is a bent (twisted) conventional U-shaped bar. As long as you reset the preload of the links after each ARB adjustment (on the scales), the car will never know the ARB is bent. But if you just move the links without resetting preload, then the car will know something is not right. It won't be the ARB itself, but the operator's poor attention to detail, and the resultant cornerweight problem.

Like on any race car, we all make our own choices about how relative any theory may be. I have had no problems any time that I had mismatched ARB linkage or bent ARBs. I just needed to be aware that preload was changing with adjustment, and adjust accordingly.

The 83-84 Reynard has nonsymetrical rear ARB linkage. It was one of the best cars of its time. I believe Adrian Reynard and his staff may even have some of those engineering degrees and credentials that people were referring to in previous posts.

[BorkRacing]

Bringing a person's credentials into a technical discussion generally indicates that you have run out of relative points to contribute to the discussion. I have no engineering degrees but I like to learn from technical discussion.

I have found that an anti-roll bar is a single spring and will remain a single spring regardless of the linkage applied to it. If the linkage is not similar on each end it can still have the same net rate as if it has mirrored linkages. The spring rate curves with movement will likely differ with different linkages (which may or may not be significant in any application).

The main reason to have a 2 dimensional ARB with symmetrical linkage is that you can adjust it on the fly without scaling the car and resetting the linkage preloads. If your U-shaped ARB has symmetrical linkage, then you can set your cornerweights initially, then move your linkages uniformly, which will change the ARB rate, but not the cornerweights.

Another example of this is the more modern ARB systems with rotating blades. Some have just one blade, or blades of different rate. The complete ARB assembly works as a unit ...... not with one corner having a different rate than the other.

Another example is a bent (twisted) conventional U-shaped bar. As long as you reset the preload of the links after each ARB adjustment (on the scales), the car will never know the ARB is bent. But if you just move the links without resetting preload, then the car will know something is not right. It won't be the ARB itself, but the operator's poor attention to detail, and the resultant cornerweight problem.

Like on any race car, we all make our own choices about how relative any theory may be. I have had no problems any time that I had mismatched ARB linkage or bent ARBs. I just needed to be aware that preload was changing with adjustment, and adjust accordingly.

The 83-84 Reynard has nonsymetrical rear ARB linkage. It was one of the best cars of its time. I believe Adrian Reynard and his staff may even have some of those engineering degrees and credentials that people were referring to in previous posts.

Greg,
I only mentioned the Milliken work because I received a snide PM of who do I think I am disagreeing with 2 ME.
Thanks for agreeing with me, another first...lol

As a side note I have heard you use non-symetrical linkages, was wondering if it was the preload or the dfferent arm lengths that was actually the effect. My suspicion is it was the preload.

Robby

[Daryl DeArman]

And for the second time in 2 days I agree with Daryl...world must be ending...lol

Ha, Ha. You can't be wrong all the time. Even I was wrong once. I thought I was wrong, but I wasn't.

I kid. I am very aware that people believe I'll argue about how many angels can fit on the head of a pin. My .84 posts per day average since day one of ApexSpeed suggests there's probably more validity to that belief than I might like to admit. Especially given that for 13 of my 18 years here on Apex Speed I haven't even owned a formula car. I enjoy formula cars, I enjoy a good debate. I may be guilty of stirring the crap from time to time, but if there was no crap to stir. . .

[BorkRacing]

We all agree the torque is equal unless there's binding in the mount. However, different amounts of motion left vs. right in the force transfer could lead to jacking, correct?

That depends on your roll center height and it's behaviour when the car rolls. Let's not even go there right now!

[Jerry Kehoe]

I'm sure you guys can go on and on regarding the adjustments but I thought Dave W's explanation of why ( at least on a FV) a stiffer bar reduces understeer rather than increases it. This reminds me of the RT4 where Wilbur Bunce had me increase the roll bar rate substantially and along with the stiffer springs it seemed the bigger the bar the less push which would have been opposite of the old norm of softening the bar to decrease understeer. Assuming Dave is still reading all of this on the subject the question is if the same basic explanation of the FV holds for other cars?

[DaveW]

I'm sure you guys can go on and on regarding the adjustments but I thought Dave W's explanation of why ( at least on a FV) a stiffer bar reduces understeer rather than increases it. This reminds me of the RT4 where Wilbur Bunce had me increase the roll bar rate substantially and along with the stiffer springs it seemed the bigger the bar the less push which would have been opposite of the old norm of softening the bar to decrease understeer. Assuming Dave is still reading all of this on the subject the question is if the same basic explanation of the FV holds for other cars?

I have no personal experience with a stiffer front bar reducing U-S. As you mentioned, in most circumstances, a stiffer front bar should increase U-S by increasing front weight transfer while decreasing it at the rear.

However, in the case of heavily rising rate front suspension, as I mentioned in the FV post, one can help minimize the wheel rate increase during cornering (which creates U-S) by minimizing roll. Positive camber during cornering could also be reduced by reducing the roll angle as in the FV if the camber gain is low.

So, IMO, while the effect of a front bar stiffness increase is usually more U-S or less O-S, in special circumstances the opposite can occur.

An effect I have experienced is that the front roll stiffness has more effect on the car's response rate than its U-S/O-S balance. The rear bar always has seemed to have a much greater effect on balance.

[2BWise]

Make sure the motion ratio of the adjustable link is consistent thru its travel. Or at least changes at a rate that matches the opposite side. You would definitely create a bar with inconsistent rate characteristics if the motions ratios of the two sides vary independently.

[alangbaker]

Different torque on each arm but same torque thru the entire system. You are treating each arm as a separate system, they are not. Your example would be true if the bar was held solidly by the mounting points, but they are not.

We are talking 2 different concepts here. I am referring to the rate of the bar and you and Dave are talking the distance the arms move. The reality is none of us will be using ARB setups with massively different arm lengths. So let's just agree we can't agree and move onto other more important tasks.

But it isn't the rate of the BAR that matters in the end:

It is the spring rate that the bar transfers to the tire.

Short answer:

Dave is correct. If you have severely different length arms at the two ends of the ARB, then you WILL experience different handling depending on which way you turn.

Do the math:

Write the equation for the force transferred from bar to pushrod to upright to tire where x is the distance from the axis of the ARB to the attachment point on the arm.

Now see how the force changes as x approaches zero.

[Daryl DeArman]

But it isn't the rate of the BAR that matters in the end:

It is the spring rate that the bar transfers to the tire.

Short answer:

Dave is correct. If you have severely different length arms at the two ends of the ARB, then you WILL experience different handling depending on which way you turn.

Do the math:

Write the equation for the force transferred from bar to pushrod to upright to tire where x is the distance from the axis of the ARB to the attachment point on the arm.

Now see how the force changes as x approaches zero.

You are assuming the other end of the bar is fixed. It is not.

[BorkRacing]

You are assuming the other end of the bar is fixed. It is not.

Exactly! A car rolls by raising one side and lowering the other, not only raising one side.
For the mechanical engineers in my world thats degrees/G.
But if you want to believe it's wrong fine with me.
Greg Rice does it "wrong" and his cars win consistantly...hmmm
Reynard did it wrong and their car was fast and stable...double hmmm

Bottom line is suspension tuning isn't just science (or ME) but also an art.
And as in art, what you find pleasing others find ugly and wrong.

I'm done, I wish the best for all in their setups.
What I found sad was many jumped on here to attack but no one helped the guy with the gearbox problem. (Except for Greg, Demeter and I)
Bad form guys, bad form.

[problemchild]

Interesting mind puzzle, I think the anti roll bar does not recognize which side of the car is stiffer or softer. It only knows how stiff it is .

But it isn't the rate of the BAR that matters in the end:

It is the spring rate that the bar transfers to the tire.

Somewhere, someone added the "If you have severely different length arms at the two ends of the ARB" qualification. What the hell does that mean to a guy trying to put sliding clevises on his soft external FF ARB? Nothing! I think we all agree that moving both clevises would be preferable to moving just one clevis on one side. Having once played around with an ARB with sliding clevises it was "hokie" from the beginning and required a lot of attention to keep functional. The floating clevises would float and introduce preload to the assembly. Eliminating movement on one side may have been a more positive solution as you could change the balance of the car while driving it, which was the whole purpose. Some of that change would be because you also changed the ARB preload, but what does that matter to the end result. Some potential theoretical rate change side to side is completely inconsequential to the question asked to start this thread.

[problemchild]

General note to racers using adjustable ARBs.

IMO. Adjustable might as well be a synonym for the phrase "having play". A perfectly tuned cockpit non-adjustable ARB will perform better than a cockpit adjustable ARB. The adjustable feature is a compromise. We have to decide that we would rather have extra play and adjustment ...... or less/minimal play but be unable to adjust while driving. Modern ARB designs tend to have reduced stiction, and better adjustment capabilities, but IMO, an external traditional ARB with very good bearing mounts, as used on most VFFs and CFFs may be the best performing ARB option. A bigger ID hollow ARB is best, and reducing the stiction in the bearing mounts is most important.

[Steve Demeter]

A thought to introduce:

If you have a bent into a u shape tube type of bar. Set it at 0 preload . Only 1 side is adjustable. Then when you adjust it on the track, does that not preload it? When I had those kind of bars, the way to take pre load out as to adjust the length of the links. By moving only 1 link along the side of the U are you not in effect doing the same thing?

[problemchild]

A thought to introduce:

Post #46.
That has not been contested.

[Daryl DeArman]

A thought to introduce:

If you have a bent into a u shape tube type of bar. Set it at 0 preload . Only 1 side is adjustable. Then when you adjust it on the track, does that not preload it?

Yes.

When I had those kind of bars, the way to take pre load out as to adjust the length of the links. By moving only 1 link along the side of the U are you not in effect doing the same thing?

Doing the same things as what?

Here's what worked for me, obviously YMMV:

Set the clevises at the same distance along each leg, both right and left. Allow for some amount of tuning both softer/harder. If that isn't possible utilize a different rate bar or adjust bar at opposite end of car proportionally to achieve the same roll stiffness ratio front:rear.

Get both drop links the same length, at the same angle both fore/aft and left/right.

Minimize friction at the mounting blocks/bushings/bearings.

Ensure that the drop links were exerting force directly into clevis at static height, not imposing a twisting motion on the clevis around the bar.

If any adjustments to the bar were made on the track, when I come off the track I measure where they ended up and split the difference. In other words, if I moved the adjustable end on the rear bar 1" stiffer as the front tires went away, then I would start with both legs 1/2" stiffer than session before, again going back to step one.

The only time I didn't have the adjustable bar at the mid point of its adjustment in the beginning of a session was if I thought there was a decent chance of rain, but not so much that I wanted to go out with bars disconnected. I would go out with the adjustments made so the adjustable end was at the end of its adjustment hard, giving me the ability to make it make a larger change towards soft than I otherwise could.

Hope that made some sense.

[alangbaker]

You are assuming the other end of the bar is fixed. It is not.

Nope. I'm not assuming that at all.

[Daryl DeArman]

Nope. I'm not assuming that at all.

Okay. Then I'll just have to disagree with your conclusion.

Load a progressive spring from the soft end or the stiff end, makes not a difference. A given load results in a given deflection. Put the input in the stiff end of the bar and hold it by the soft end, or input the load in soft end and hold it by the stiff end, the total relative degree of twist is the same for a given input.

[john f]

Okay. Then I'll just have to disagree with your conclusion.

Load a progressive spring from the soft end or the stiff end, makes not a difference. A given load results in a given deflection. Put the input in the stiff end of the bar and hold it by the soft end, or input the load in soft end and hold it by the stiff end, the total relative degree of twist is the same for a given input.

It's the length of the arm that is different, side to side. The longer arm generates the same torque with less force that the short arm does! That force, through the link, is what is applied to that corner of the car. If you apply 10# through a 10" arm you are generating
100 #" of torque. If the other arm is 5" long you need 20# to resist that torque. If long arm is on the right side of the car, in a left hand turn you are adding 10# of force to the right wheel and removing 5# of force from the left. A right turn would add 5# to the left wheel and remove 10# from the right wheel.

john f

[problemchild]

It's the length of the arm that is different, side to side. The longer arm generates the same torque with less force that the short arm does! That force, through the link, is what is applied to that corner of the car. If you apply 10# through a 10" arm you are generating
100 #" of torque. If the other arm is 5" long you need 20# to resist that torque. If long arm is on the right side of the car, in a left hand turn you are adding 10# of force to the right wheel and removing 5# of force from the left. A right turn would add 5# to the left wheel and remove 10# from the right wheel.

john f

Wow. Clearly that math is beyond question. Good luck with your racing.

[Offcamber1]

It's like a balance or a see saw. 10 fat kids on the short side, 1 fat kid on the side that's 10x the distance from the fulcrum. Same balance as if you put 5 1/2 fat kids on each side with the fulcrum in the center.
.

As long as you are not the 11th fat kid.

[BorkRacing]

It's the length of the arm that is different, side to side. The longer arm generates the same torque with less force that the short arm does! That force, through the link, is what is applied to that corner of the car. If you apply 10# through a 10" arm you are generating
100 #" of torque. If the other arm is 5" long you need 20# to resist that torque. If long arm is on the right side of the car, in a left hand turn you are adding 10# of force to the right wheel and removing 5# of force from the left. A right turn would add 5# to the left wheel and remove 10# from the right wheel.

john f

Why assume the force is unequally dispensed? Why wouldn't the forces normalize and there would be 7.5# of force delivered to each side with one side moving more than the other depending on where the longer and shorter arms are?

[Daryl DeArman]

As long as you are not the 11th fat kid.

I was going to saw the 6th one in half

[john f]

Why assume the force is unequally dispensed? Why wouldn't the forces normalize and there would be 7.5# of force delivered to each side with one side moving more than the other depending on where the longer and shorter arms are?

If you apply the same force to each arm (7.5# in your question) you would be applying 75 lb-in of torque to one end and 37.5 lb-in to the other end. seeing as the moment in the bar must be equal through the whole bar, this loading could not exist. You must remember that torque is is generated by 2 items, length and force. For a given torque, if one increases (force or length), the other will decrease. This was covered in statics 101. It is called summation of moments. Take a look.

https://en.wikipedia.org/wiki/Statics

[stonebridge20]

Anybody want to talk about pistons ?

[Garey Guzman]

Anybody want to talk about pistons ?

Mike,
How do you fix a piston that doesn't seem to go all the way to the top of normal travel? Meds?

[stonebridge20]

Mike,
How do you fix a piston that doesn't seem to go all the way to the top of normal travel? Meds?

Make a proposal to SCCA so J&E can build longer pistons ?

[stonebridge20]

Got pictures...lol

Yep.

[stonebridge20]

Ok, I'm out a here. I'm in enough trouble !

[BorkRacing]

Yep.

Not of Gary's short "piston" I hope!

[tlracer]

Getting back to the original question:

There have been points made on here about the bar not being 'fixed' but surely this isn't correct. Think of a very thin ARB. This will twist given a very small moment at one end, such that the other end will deflect little or nothing. What it won't do is transfer the whole angular movement and applied moment to the opposite side. Thus it is - essentially - 'fixed' to a given extent based on its stiffness (or lack of).

Now since all ARB's deflect (unless so massively thick they are simply a solid link) this will always be the case. In the following I have assumed the ARB is not that stiff so there is some rotational deflection within it when a moment is applied.

So (working with the two linkages of the same length), given a fixed upward force 'f' through the linkage at one end of the bar, applied at a distance 'l' from the rotation axis, the resultant moment will be 'f x l'. A quantity of that moment will cause twisting in the bar and the remainder will be applied as a moment to the opposite-end linkage. By dividing the latter moment by the length of that side's arm 'l' the resultant upward force transmitted to the wheel will be calculated.

If we shorten the linkage to, say, 'l/2' on one side only so the two are unequal lengths and apply the same force 'f', then the moment becomes ' f x l/2', so half the original value. When applied, twist in the bar will require the same moment as in the first example, so the remainder that is transferred to the opposite wheel will be smaller. The resultant force applied to the opposite wheel can be calculated in the same way as above by dividing by 'l' and will - as the moment is lower - be smaller also.

Now, if we keep the same setup as above, ie. with unequal-length linkages, and apply the force 'f' at the opposite end we will get a moment 'f x l'. Again, a fixed amount of that moment will cause twist in the bar and the rest will be transferred to the other end. The resultant force can be calculated by dividing the moment by the linkage length (ie. 'l/2') so the force on the wheel will be higher.

Edit: Please challenge this as, having written it with such initial certainty, I am reading it through and wondering...

Edit 2: ...and, having thought, all the above is possibly more relevant to when one wheel is in 'bump', ie. an instantaneous force is applied to one end of the ARB setup, so the bar is, effectively, working as an additional spring rather than to control roll.

In the latter case the applied moment is around the roll axis of the car and the 'lever' length is the distance between the CG and the roll centre for whichever end of the car (front or rear) that's under scrutiny. So there is a simultaneous upward moment applied to one end of the ARB and a downward moment to the other. Can we assume these are equal?

Some of this moment will be taken up by twist in the ARB and the remainder will be applied via the end linkages as a force onto to the wheels. If the linkages are the same length then the force on the wheels will be equal and in opposite directions.

If the linkages are not equal then the applied forces won't be either. My gut feeling is that the 'softer' end (ie. the one with the longer linkage) will either droop or jack further depending on whether the roll is towards or away from it.

[DaveW]

Hey, Robby, we actually agree on something (tlracer's post above) !

[racerdad2]

Engineers begin their careers knowing nothing about everything. As time goes on, they learn more and more about less and less. At the end of their careers they know everything about nothing.

I have the utmost respect for the engineers in our sport. Racers like Dave W. engineer, test, re-engineer, test, re-engineer the re-engineering, test. Tough to beat on track. Never cheat. Just compete.
Thanks to all for making our sport faster & safer !!!

[BorkRacing]

Hey, Robby, we actually agree on something (tlracer's post above) !

Dave, I respect you, your education and your race accomplishments. No question there. But if we don't question interpretations of suspension behavior we won't learn anything new, discover new set ups or discard old ways of thinking.

I suspect you understand that.

For those that don't understand that read Adrian Newey's autobiography.

[DaveW]

Dave, I respect you, your education and your race accomplishments. No question there. But if we don't question interpretations of suspension behavior we won't learn anything new, discover new set ups or discard old ways of thinking.

I suspect you understand that.

For those that don't understand that read Adrian Newey's autobiography.

Yeah - I've had more than a few of my ideas on how things work disabused over the years...

I read the autobiography. Absolutely fascinating !!!

[john f]

Getting back to the original question:

There have been points made on here about the bar not being 'fixed' but surely this isn't correct. Think of a very thin ARB. This will twist given a very small moment at one end, such that the other end will deflect little or nothing. What it won't do is transfer the whole angular movement and applied moment to the opposite side. Thus it is - essentially - 'fixed' to a given extent based on its stiffness (or lack of).

Now since all ARB's deflect (unless so massively thick they are simply a solid link) this will always be the case. In the following I have assumed the ARB is not that stiff so there is some rotational deflection within it when a moment is applied.

So (working with the two linkages of the same length), given a fixed upward force 'f' through the linkage at one end of the bar, applied at a distance 'l' from the rotation axis, the resultant moment will be 'f x l'. A quantity of that moment will cause twisting in the bar and the remainder will be applied as a moment to the opposite-end linkage. By dividing the latter moment by the length of that side's arm 'l' the resultant upward force transmitted to the wheel will be calculated.

.

I'm back
I do not want to cause trouble, but I have a problem with the above description. If the torque input does not transfer to the other end of the bar, where did it go? The bar itself does not "use up" some of the torque. That would be saying that when you use a torque wrench with an extension, you are not getting the full torque to the fastener you are tightening. The torque that you input in the driveshaft is not all delivered to the pinion? The torque you input to the steering wheel doesn't all make it's way to the steering box?

If you do what is called a "free body diagram" of the sway bar, there are a total of 4 forces being input into the bar. The 2 links at the ends and your 2 mounting points. Nothing else. Neglecting the weight of the bar, when the car is level, with no preload, these four points are carrying the weight of the bar. At this point there is no bending moment induced in the bar. The moment we move one link up or down, the loading on the other points change. The free body analysis shows the direction and magnitude of each of the mounting points. It will also show that we are now generating a bending moment in the bar.

If one link is moved up, by a given force, either the other link has to resist the force an equal amount (equal length arms) or that end will just move up. no resistance, no moment. Put your torque wrench on a loose nut and turn the wrench. no resistance from the nut, no torque.

Now, there is one caveat in the previous description. The mounting points of the bar on the chassis. As the mounts are a rotating bearing as such, this point can generate friction, which will require a torque to overcome. It is at these 2 points where a percentage of torque can be lost. If this amount is more than a few percent, you have got a bound up bar.As problemchild stated in post #45:

" but IMO, an external traditional ARB with very good bearing mounts, as used on most VFFs and CFFs may be the best performing ARB option. A bigger ID hollow ARB is best, and reducing the stiction in the bearing mounts is most important."

My description assumes good bearing mounts.

Please note that I am not trying to "one up", "bash", taut my education, or years of experience. The biggest thing I learned in school eons ago was "I learned how to learn". I am always open to learning, but I prefer it be correct info . That's my $0.02.

john f (john f boxhorn)