Why Aren't Any Torsion Bars Progressive?

[jyl]

I read that one advantage of coils over torsion bars is that coils can be progressive and bars cannot.

Why are there no progressive torsion bars? I can imagine a solid bar where half the length is 26mm in diameter and the other half is machined to neck smoothly down to 24mm, for example. Wouldn't this give a progressive torsional resistance? Alternatively, two half-length bars joined in the middle by a splined collar, so you could mix half-lengths of different diameter.

Any thoughts about this? Would it work?

[Tyson Schmidt]

You are exactly right. A tapered torsion bar would be progressive. And you could have two different diameter torsion bars splined end to end for the same exxect as dual stacked coil springs. My guess is a cost-to-benefit ratio that just doesn't make sense.

Progressive springs are really only for making a car that is prepped for the track, a little nicer riding on the street. Progressive springs aren't really for an all-out race car, where they are so stiff, it's barely accurate calling it "suspension".

The dual coil set-up like on the front of Jack Olsen's car, has a tender spring that is basically fully compressed at normal ride height. It's basically only there to keep the main spring in place at full shock extension, and to keep the wheel in contact with the ground as much as possible in this same situation.

[350HP930]

Torsion springs are very linear by design.

Coil springs can be designed with varying geometry so that after the low rate section is compressed the spring starts expressing the tendancies of the less compressed high rate section of the spring.

[Chuck Moreland]

Originally posted by Tyson Schmidt
A tapered torsion bar would be progressive. And you could have two different diameter torsion bars splined end to end for the same exxect as dual stacked coil springs.

It doesn't work that way. I know it seems like it would work, but it doesn't.

Basically what you are creating is two springs in series. The formula for the resulting spring rate is:

1/(effective spring rate) = 1/(spring 1) + 1/(spring 2)

This is a linear function.

Progressive coils do not work by having two different spring rates. They work by spacing some coils closer together (the close wound coils may or may not have the same spring rate) The close together coils bottom out first, thus effectively shortening the spring and increasing its spring rate.

[Tyson Schmidt]

Originally posted by Chuck Moreland
It doesn't work that way. I know it seems like it would work, but it doesn't.

Basically what you are creating is two springs in series. The formula for the resulting spring rate is:

1/(effective spring rate) = 1/(spring 1) + 1/(spring 2)

This is a linear function.

Progressive coils do not work by having two different spring rates. They work by spacing some coils closer together (the close wound coils may or may not have the same spring rate) The close together coils bottom out first, thus effectively shortening the spring and increasing its spring rate.

Hey, that's not fair!

That's weird Chuck. You'd think that the small diameter portion of the torsion bar would twist until it reached a certain rate, then the larger diameter would start to twist as the rates began to overlap.

[island911]

Well here's a shocker (or two) for ya all.

First, the torsion bar suspension system of our 911s is non-linear. (progressive)
(think about the spring rate at the wheel. . .if the arm travel were allowed to go from 0-90°, no more force would be winding the spring . . ..it would all be lifting the car.)

Second, progressive torsion springs do exist.
One type has a tappered fixed "mandrel" down the center of a (very hollow) torsion spring. As the spring winds up, it starts clamping onthe center mandrel . . .shortening the effective length.

[Quicksilver]

Originally posted by Tyson Schmidt
Hey, that's not fair!

That's weird Chuck. You'd think that the small diameter portion of the torsion bar would twist until it reached a certain rate, then the larger diameter would start to twist as the rates began to overlap.

Metal will twist at with a linearly increasing rate until it reaches it's yield limit. Metal doesn't deform until it reaches a limit and then "lockup". It deforms with a steadily rising resistance until it reaches a point where it starts to permanently deform. At this point the force required to bend it levels off and then drops.

Say you had a bar that was tapered so the small end had an effective rate of 100 inch/pounds and the large end was 200 inch/pounds. The bars effective rate would be in-between as the small end would distort more and the large end would flex less. It would also fail easily as all of the stress would be concentrated in the small end.

It would be theoretically possible to create a torsion bar that would progressively bind on a rigid tube that surrounds it but it would need to be larger then we have room for. It would be hard to tune, be prone to failure, and probably of no real value.

The only real way to get a rising rate out of a torsion bar is to use a linkage geometry that provides the rising rate. If you want to go to that type of trouble then you should scrap the banana arm rear suspension and go with a more effective multilink system. (Hey, Porsche already did!)


Wayne

[island911]

Originally posted by Quicksilver
. . ..The only real way to get a rising rate out of a torsion bar is to use a linkage geometry that provides the rising rate. If you want to go to that type of trouble then you should scrap the banana arm rear suspension and go with a more effective multilink system. (Hey, Porsche already did!)


Wayne

Are you saying that Porsche went to a multilink systemto get a rising rate system?
. . .they already had that with the t-bar set-up. eg, If they wanted a faster rising rate, they would simply shorten the t-bar arm.

I believe the multilink system was intended more to handle the ABS and address the subtle geometry changes needed to optimize certain loading conditions.

[Tyson Schmidt]

That makes perfect sense.

Hey, what about a gundrilled torsion bar with a smaller bar inside it. Then you could twist the inner bar to a certain point, then the outer bar would begin to twist after the anchor made contact with the anchor for the inner bar.

Not that you'd go to that trouble, but I just like solving problems.

[Quicksilver]

Originally posted by island911
Are you saying that Porsche went to a multilink systemto get a rising rate system?
...

No, but if you wanted rising rate then that would be the way to get it. The banana arm arrangement was kind of a kludge. It was dated in the 70s.
The thing I love about it (and the 911) is it is amazing what can be done when you stick with something and refine it instead of scrapping everything and starting over every few years.

Originally posted by Tyson Schmidt
...
Not that you'd go to that trouble, but I just like solving problems.

That is the "other" reason to have a brain.
(The first reason is to keep your head from imploding.)

Wayne

[masraum]

Definitely no expert, but Wayne(quicksilver) sounds most correct about the progressive torsion bar theory to me.

I think Chuck has the workings of a progressive coil spring backwards, but I could be wrong. I don't believe they design progressive rate springs to coil bind. I believe the more widely spaced coils compress first and then the more tightly spaced coils compress, but again, I could be wrong. I found this page that seems to show that in an illustration, but who's to say the page isn't wrong.

http://www.jontreby.fsnet.co.uk/CV/usa/Page001.htm

Progressive rate torsion bars, this is an interesting thread.

EDIT:

After a little more research I found this info which also indicates that PR coil springs don't bind to increase the rate.

http://www.unofficialbmw.com/all/suspension/all_springs_101.html

[Zeke]

I'm no expert either. What I know is that the rate of a coil spring is determined by the dia. of the wire, the dia. of the coil wind and the no. of coils (length). Therefore, it reasons that the relationship of the coils to one another is not a factor until they bind. I am voting for Chuck's theroy.

[jyl]

Chuck, now I do vaguely remember the formula you used, from college physics. It's the same as electrical resistance in parallel, right?

Anyway, while looking into this, I ran across a very interesting information source:

http://www.mech.uwa.edu.au/DANotes/

Some of the topics are:

SPRINGS Types of springs; close coiled round wire helical compression springs; the spring characteristic; stresses & stiffness; buckling; wire materials; presetting; fatigue loading; spring design.
Appendix - presetting a torsion bar.

THREADED FASTENERS Thread geometry; screw thread mechanics; static failure; loads in an elastic bolted assembly; preload and its control; fluid pressurised joints; bolt fatigue; non-uniformly loaded bolt groups.

WELDED JOINTS Fillet welded joints; geometric properties of lines; traditional analysis; throat stresses and joint safety; unified analysis; conclusion.
Appendix - the compliant lap joint.

The part on torsion bars is at http://www.mech.uwa.edu.au/DANotes/springs/appendix.pdf

[ChrisBennet]

It's not a theory, progressive springs use coil bind to effectively make the spring shorter and increase it's spring rate.
-Chris

[island911]

Originally posted by Quicksilver
No, but if you wanted rising rate then that (multi-link) would be the way to get it. The banana arm arrangement was kind of a kludge. It was dated in the 70s.
. . .

Why is that?
Do you have any idea what you're talking about? It seems your just throw'n stuff out there, predicated on the notion that, if it's older, it must be kludgey by "todays standards."

. .. that would be a mistake.

[Chuck Moreland]

Originally posted by masraum
I think Chuck has the workings of a progressive coil spring backwards, but I could be wrong. I don't believe they design progressive rate springs to coil bind. ....... I found this page that seems to show that in an illustration, but who's to say the page isn't wrong.

http://www.jontreby.fsnet.co.uk/CV/usa/Page001.htm

Progressive rate torsion bars, this is an interesting thread.

EDIT:

After a little more research I found this info which also indicates that PR coil springs don't bind to increase the rate.

http://www.unofficialbmw.com/all/suspension/all_springs_101.html

Actually your references both support my statements about progressive coils. Here is a quote from the BMW reference:

"Normal constant pitch springs don't touch coils at all (except a little at the ends because of the slight change in pitch as mentioned above). Progressive springs do intentionally touch one or more coils to reduce the number of active coils and increase the rate."

And a picture from the other reference:



My formula for series springs is accurate for both coil springs and torsion bars. The resulting spring rate is always linear.

It is possible to create progressive torsion bars by creating mechanical stops that eliminate a portion of the torsion bar, or add additional diameter. This is the same way it's done on progressive coils.

It is also possible to create a progressive rate at the wheel by changing the mechanical advantage over the spring (either torsion bar or coil) as the wheel moves through its arc. I think that was Island's point. However the spring itself is not progressive, only the system.

[Randy Webb]

"banana arm arrangement was kind of a kludge. It was dated in the 70s."

- It's not clear how he means this. It was a big improvement over the 356's rear geometry (but that was in the early 1960's). Maybe he means that unequal length A arms were already well-known by the 1970s. Of course, the reason for the semi-trailing arms and the McPherson F stuts was to save space. It's hard to optimize everything at once. Given the requirement of a passenger car with a R engine, I don't see what else they wold do at that point in time. The Weissach axle design dates from the 928 period.

[Randy Webb]

"I believe the multilink system was intended more to handle the ABS and address the subtle geometry changes needed to optimize certain loading conditions."

I think it was to optimize tire contact patch size by better controling suspension geometry at various cornering attitudes. i.e. same as Isl. said depending on the def'n of "loading."

[Quicksilver]

Originally posted by island911
Why is that?
Do you have any idea what you're talking about? It seems your just throw'n stuff out there, predicated on the notion that, if it's older, it must be kludgey by "todays standards."

. .. that would be a mistake.

I don't get emotionally involved with how something is done. Sometimes I do get "emotionally involved in how it works! I have no love of "new". I have a love of "it works".

BTW, What's wrong with a kludge anyway? The most important thing is if something works. People often forget that it is most important to "just get the job done".

I said it was a kludge because it is pretty far from a great solution. It was a great way of being so far ahead in the 30s through the mid 60s. It gave you a reasonable camber curve, good antilift/antidive, and pretty good toe control for the tires that were available. This was done at a time where most sports car makers were still messing with beam axles. It was inexpensive and its pickup points are out of the way of the transaxle. It was an amazing solution but lets not get emotionally tied to it.

In the later sixties vehicle dynamics really started to be a science and tire technology started to jump forward. The whole purpose of a performance/sports car is to "Feed the tires". If you can do anything to create more grip then that is what people do. More power? It is only to feed the tires. If the suspension isn't doing its best to allow the best use of the contact patch then people will change it. The banana arm system started to show its limitations when tire technology ran right past it and when compared to multilink systems its short comings became obvious. Then we get to the part of my post that you didn't quote.

. . .it is amazing what can be done when you stick with something and refine it instead of scrapping everything and starting over every few years.

Objectively the banana arm arrangement has serious problems. It has a serious problem with adjustability. The unsprung weight is too high. Toe control is a problem. The chamber curve is linear whereas an unequal length multilink system can give you a number of adjustable chamber curves. The list goes on.

Porsche never was attached to the banana arm system. They left it behind for a system that works better. Name a purpose built on-road racecar that doesn't use unequal length multilink suspension. We will ignore examples that are based on restrictive rules or based on legacy automobiles. When you have any series where people can build and run what they want there is only one solution. Why? Not because they dislike "Old". It is because losing badly sucks.

BTW, Chuck is completely correct on how coil springs are made progressive.
How about this as a brain twister to throw in; "Coil springs are really torsional springs that are bent into a spiral". It is true but you can get a headache trying to figure out the how and why.


Wayne

[Tyson Schmidt]

Everything that Quicksilver says about the semi-trailing ara rea suspension is true.

But, it's shortcomings are greatly minimized when the bushings are replaced with spherical bearings and heim joints, and the suspension travel is reduced by extremely stiff suspension, so that camber curves and higher unsprung weight are almost non-issues.

After that, their superior durability becomes a huge factor. Look at how the early 911 absolutely DOMINATES vintage rally. They don't break.

That, and the fact that some of it's shortcomings become assets in the hands of skilled drivers. I like being able to steer the car with the throttle. I don't want a toe-correcting rear suspension trying to fix a problem that I don't want fixed. I want to drive the car, not be a passenger.