Ill try giving it a look tonight and at least figure out how I want to do it. Maybe I'll get ambitious even

 

Could one of you please explain why you think that bends in the B section of the bar do anything more than make the value of B slightly more than merely the distance from corner to corner? That's what the math says that it does.

 

Its not the bends in the bar were talking about, its the bends past the mounts were talking about. Part of the equation deals with the bending of the arms past the bar which effectively reduce spring rate. More length is just more bar itself to bend.

 

OK. I think I got it. And, yes, the math pretty much assumes that the brackets for the bar are all the way out at the corners, such that section B is only in torsion (twist) and does not contribute at all to the flexibility of the lever-arms. But we know that most bars (other than spine-drives) have brackets well in from the corners and that the part of B that is between the bracket and the corner does play a roll in the flex of the lever-arms.

But, man, this is not something that I'd do by math (and, if you know me, that says a lot). This is a good, old empirical question. So, I completely agree with the idea of making a bench version of the brackets, locking one end of the bar in place, and just measuring the darned thing. I'd be especially interested to know if having brackets not at the ends of section B causes the entire thing to cease to be linear.

 

I could be way off here and honestly I haven't put much thought into it, but the sections of B that are not inline with the centerline of the shaft will induce bending moments in section B and cause reaction loads in the brackets. B is not in pure torsion at that point.

Combine that with the unsupported sections of B and yes, I think there could be a signifigant amount of error induced by these designs features.

A quick "paper clip check" seems to agree with my thoughts and you see bending in the section that is not on centerline with the torsion component.

 

A paperclip test cant real be use due to strain hardening. Bent parts of that paper clip are going to be significantly stiffer than the unbent parts. My guess is a swaybar is stress relieved after bending.

I don't think the bends are going to have more than a plus/minus 5-10% error between the mounts. The legth of the bar past the end (denoted by C) will be my guess at the value that much change since that much length is bending. "A" will not change and stay measured from the mount bracketed to end link stud. That's my guess to get within 5-10%. But one of use needs to actually measure the bar.

I looked at how to do it last night and the factory mounts are awkward so bolting it down that way wont be to easy. Im now trying to come up with a way to add a few hundred pounds to one end of the bar while still in the car. I only have about 80lbs in dumbbell weights.

 

Strain hardening won't change a materials young's/shear modulus.

5-10%, 20%... doesn't really matter. We both have the same question, how close is that equation to reality for our particular application? If it's damn close, then we could make some pretty decent estimates on the stiffness of various bars out there and know we are in the ball park when trying to model up the suspension for roll stiffness. If its 25% off, then I wouldn't trust that equation to compare bars as the mounts, bends, etc. are going to vary and throw the equation way off to the point of it being basically useless.

I'll have to look at the bar again as I have the OEM one sitting around to come up with a way to hold it.

 

And, as you know, that isn't going to cut it. 80# is almost just noise.

Note that an Evo X gets almost exactly half of its roll resistance from the bars. The front sprung corner-weights are around 850#. Thus, the bars are transmitting in the neighborhood of 3-400# in a hard corner.

edit: the above was from memory and certain things that I did in the 70s ruined it. A bone-stock Evo X gets half of its single-wheel rate from the bars, so it's two thirds of its roll resistance

 

this is an incorrect assumption. the torque on the bar is dependant on the delta position of both arms. thus the entire length needs to be considered. each side gets the same force, one just being opposite of the other.


second, really you cant calculate one side at a time anyway because the two sides are coupled thus you have a coupled system and need to consider both at the same time to determine the roll resistance. the outside shock sees an additive spring rate from the bar, while the inside spring sees a negative spring rate from the bar. also, the main spring on one side effects the other side as well with a bar.

 

I totally agree with this, but every time I try to get a technical discussion going on the true single-wheel effects of bars (especiallu with regards to grip on gravel), it gets bogged down or dies out (or worse on Rally Anarchy). Most of all, the effect of a bar on single-wheel bump depends on the low-speed compression damping of the shocks. If you have an infinite amount of LSC damping (which is silly, but I'm trying to make a point here), then the other end of the bar is locked in place and it acts like the formula says. But if you have minimal LSC damping, which is typical of street shocks (i.e., too little LS damping of both sorts), then the other end of the bar does move the opposite wheel up and the bar is not having as much of an effect on the wheel that actually did hit the bump. And it's not just the LSC damping. The springs make a huge difference, as well.

You can also turn this around the other way. Since the damping of the shocks influences the effect of a swaybar on single-wheel bump, should we not be including the swaybars in our calculations of appropriate LS damping? I'd say Yes, but whenever I ask a shocks guru if he or she cares about the bars, they say No. They just want the sprung corner weight, the spring rate, and the desired percent of critical damping.

It's enough to make a math-geek cry.

 

I can model this in simulink pretty easily. I'll draw it up and see what I can come up with

 

If we can end up demonstrating why getting 40 to 50% of the roll-resistance from bars is usually best for tarmac, while getting zero to 25% is best for gravel, I'd be so darned happy. You can read these rules of thumb in various places, but no-one - not even folks like Puhn or Millikin - ever say why.

 

Thats the big thing Im trying to figure out. If my wheel rates are 500-600 where should my bar rates be. Well actually, its more like if I know I need 700-800lb/in of wheel rate to control roll, what is my balance of spring to bar. Running 900s and the stock bar worked great on smooth corners but didnt like bumps and that was significantly biased towards spring over bar. So now Im gonna try a more medium approach.

 

I think you misunderstood what i said. rather than try and explain it myself, i recommend reading this:
http://www.eviltwinmotorsports.com/w...ter-2011.2.pdf

 

the shocks make no difference in how the swaybar works. I believe the answer to your question involves looking at and comparing what your critical damping requirements are in heave(single wheel bump) and in roll. when you don't have a separate roll damper you're setting your roll damping using the ride dampers (which are themselves set by mass and spring rate) and an additional undamped roll spring. In general for sedans (read: it depends entirely on the specifics), with no swaybars and shocks set to 70% critical damping you will end up being over damped in roll so the additional undamped roll stiffness serves to speed up the roll response time.

i would speculate that a gravel car isn't targeting 70% critical damping and more likely is shooting for something in the 30% range to improve response over bumps. in that case having excessive sway bar stiffness is going to result in an under damped roll response, driving you towards less undamped roll springing. aka lower swaybar/spring rate ratio.

clearly if you run too much swaybar relative to your spring rate, you could end up having the car being under damped in roll which would have a tough time getting the car to take a set.