Roll rates in RCVD

[Z]

MORE ABOUT UWA'S INDEPENDENT ROLL-MODE CONTROL.
==================================================

Some more things that need explaining...

I assume everyone has a picture of UWA's system in their mind. Namely, there is a Lateral-U-Bar that has its "backbone" pivotted to the car-body, just behind the MRH. The ends of the LatUB have drop-links that go down to the two undertray tunnels, and these tunnels act as Longitudinal-Balance-Beams. The 2 x ends of each of the 2 x LongBBs can be considered to act directly down on the 4 x wheelprints.

Assume (to keep the numbers simple) that wheelbase is 1500 mm, and the drop-links attach to the tunnels/LongBBs at 900 mm rearward from the front-wheels, so 600 mm forward of rear-wheels. Thus any vertical force on a drop-link is transferred, via its LongBB, in the ratios 600/1500 = 40% to the front-wheel, and 900/1500 = 60% to the rear-wheel. That is, the forces are as per Archimedes' "Law of Levers" (so nothing new here! ).

Now consider the all-wheel suspension modes of:

HEAVE - With above picture in mind it should be obvious that the four wheels can all move together in Heave (ie. all equally up, or all equally down, wrt the car-body) with NO RESISTANCE from the above linkage (other than a bit of friction in joints, etc.). That is, the LatUB simply pivots about its "backbone", and NO PART OF THE LINKAGE IS STRESSED. Therefore, this linkage exerts NO FORCES on the wheelprints during Heave motions.

PITCH 1 - Assume that the "Pitching" occurs around the drop-link-to-LongBB ball-joints. Obviously these BJs don't offer any resistance (again, neglect any small friction), so NO FORCES on the wheelprints from this type of "Pitch".

PITCH 2 - BUT!!! What about "Pitching" about a lateral-axis at MID-WHEELBASE? Now there will be some rotation about the drop-link BJs, and also some rotation of the LatUB wrt car-body, as in the Heave case. But, again neglecting small friction, NO FORCES on the wheelprints from this type of "Pitch". (Please draw side-view sketch to convince yourselves of this).

TWIST - Assume that this "Twist" IS DEFINED similarly to Pitch 1 above, but with opposite rotations of the two LongBBs, as seen in side-view (eg. so left LongBB rotates nose-up about its Drop-Link-BJ, and right LongBB rotates nose-down about its DLBJ). So, again, the total linkage only has a small rotation at the drop-link BJs, and so again NO FORCES on the wheelprints from this type of Twist.

Thus, the above Roll-mode control linkage ONLY RESISTS Roll motions, and exerts NO RESISTANCE against any Heave, Pitch, or Twist motions, AS THEY ARE DEFINED ABOVE.
~~~o0o~~~

BUT!!!

WARP - Let us, somewhat perversely perhaps, DEFINE this mode to be the EQUAL and opposite vertical movements of the car's diagonal pairs of wheelprints. This means that if we consider the originally flat plane containing the four wheelprints to be Warped in this way, then the "Nodes" (= points of no displacement) of the resulting warped plane are along the longitudinal-centreline of the car, and along the lateral-centreline (= mid-wheelbase) of the car (see sketches somewhere...).

VERY IMPORTANTLY, this now also means that one drop-link must move upward, and the other drop-link must move downward, because they are rearward of the node points. The Lateral-U-Bar is thus twisted, and thus UWA's Roll control system RESISTS THIS TYPE OF WARP MOTION.

So, it is NOT WARP-SOFT!!!

BUT!!! It is still COMPLETELY TWIST-SOFT!!!!!
~~~o0o~~~

There is nothing paradoxical, at all, in the above explanation.

Here I should mention that I have very little idea of how the F1 "active-suspension era" engineers went about their work, or what DEFINITIONS they used. There is very little information about this in the public domain (and Dynatune ain't talking...).

But I do suspect that many of the workers at that time somehow "burned into their minds" the notion that the Warp-mode MUST BE DEFINED as above. And therefore any adjustment of LLTD away from 50:50 MUST NECESSARILY require an ACTIVE Warp-mode. I have certainly heard that view expressed several times, and used alongside phrases such as "Warp/Roll-mode coupling...".

This is why I have always used the name "Twist-mode" to distinguish it from the "Warp-mode" behaviour. I have also used "Bounce-mode" to distinguish it from the more symetrical "Heave-mode" (and this relates to Tim's earlier comments regarding "spring centres" and "aero Pitch sensitivity", which are (non-) problems that are easily solved with the appropriate definitions.) I would have chosen different terms for "Roll" and "Pitch" too, because these can also be defined differently, but I still can't think of good enough alternative words...
~~~o0o~~~

Bottom lines:
Sloppy, and unconsidered, use of DEFINITIONS = "we failed miserably".
Careful and considered choice of DEFINITIONS (right from the beginning!!!) = you can have a completely soft Twist-mode, AND any LLTD you want.

It really is much easier to solve problems, when you have a clear understanding of what you are talking about.

More to be said, but enough for now...

Z

[Pete Marsh]

"WARP - Let us, somewhat perversely perhaps, DEFINE this mode to be the EQUAL and opposite vertical movements of the car's diagonal pairs of wheelprints. This means that if we consider the originally flat plane containing the four wheelprints to be Warped in this way, then the "Nodes" (= points of no displacement) of the resulting warped plane are along the longitudinal-centreline of the car, and along the lateral-centreline (= mid-wheelbase) of the car (see sketches somewhere...).

VERY IMPORTANTLY, this now also means that one drop-link must move upward, and the other drop-link must move downward, because they are rearward of the node points. The Lateral-U-Bar is thus twisted, and thus UWA's Roll control system RESISTS THIS TYPE OF WARP MOTION.

So, it is NOT WARP-SOFT!!!

BUT!!! It is still COMPLETELY TWIST-SOFT!!!!! "


As the body is completely free around the roll axis this is not quite true. The body simply assumes a roll displacement proportional to the deviation of the link position from 50:50, and no force or spring stress is required. SO still soft warp. This is much same as a Trike (or bulldozer) if taken to the extreme, with the body roll angle determined by the one axle, but still with no weight transfer between the wheels due to the road surface change. There will obviously be inertial effects there if the motion happens quickly, but with a typical setting for a slightly rear heavy car very close to 50:50, and the relatively low frequency of the warp mode, we ignore them and treat it as zero warp. In our implementation there is some warp stiffness within the lower floor with it being a giant flexure so as to achieve the relative motion between the wheels. This is however very low, with measured warp stiffness less than 10% that of a 4 spring and 2 U bar car with equivalent roll stiffness.


Here is a good image of the system. Adjustment and function has been covered by others here already. Major difficulties include managing (mode separated)damping ratios with high unsprung without resorting to separate ones everywhere, achieving required displacement and relative stiffness profiles of "M" springs (or W if you prefer) and the lower floor, and steering system and front upright packaging.

Pete

[Z]

As the body is completely free around the roll axis this is not quite true. The body simply assumes a roll displacement proportional to the deviation of the link position from 50:50, and no force or spring stress is required. SO still soft warp.

Pete,

The above is actually what I was getting at when I mentioned the phrase "Warp/Roll coupling". With the body fixed in space, and NO Heave, Pitch, or ROLL-mode motions of the wheelprints, and the wheelprints then moved only in a symetrical (50:50) "Warp" motion, then the Lateral-U-Bar is definitely twisted, and this particular definition of Warp is resisted.

However, your car is still most definitely Twist-soft (neglecting any stiffness of the undertray), in accordance with the particular definition of "Twist" that suits the drop-link positions.

Anyway, the above might sound like I am splitting semantic hairs, but once the wrong definitions get burned into the mindset of people designing these things, I think it can lead to long term misunderstandings. As I mentioned before, one of these misunderstandings is that "If you want to change the LLTD away from 50:50, then you MUST have forces acting in the Warp-mode". This is true of the Warp-mode, but not true for an appropriately defined Twist-mode.

Z

[mech5496]

Z, in a symmetric (50:50) warp motion, the U-bar indeed twists, resisting the motion; the question is by how much? It is a question easily answered by doing some simple, mostly geometrical calculations. At your numerical example for instance (1500mm WB, 20mm movement in all wheels resulting in a 40mm 50:50 wrap motion) the pivot point moves upward ONLY by 4mm, and that's an extreme example (wrap movement too large, WB too small for FSAE). 4mm movement on pivot point, acting on a U-bar lever of let's say 200mm, gives an approximately 1.2 degrees twist in the U-bar. I think it is pretty straightforward to calculate the desired roll stiffness contributed by the U-bar in such a setup ant then calculate the U-bar contribution to the wrap stiffness, but still I think it would be insignificant.

EDIT: To compare with, a 2deg roll movement, assuming that the left-to-right pivot ponts are 1m apart, means a 17.5mm upwards/downwards movement on the U-bar links.

[Z]

Harry,

I will probably have to repeat this several more times, but the problem here is the high-level conceptual understanding of these things (or the lack thereof).

As I mentioned before, I have never been able to find any detailed papers, etc., about F1's adventures with this "modal" suspension thinking. But I have heard, quite often, mention of "THE Warp-Mode" as being ALWAYS, AND ONLY, of the "50:50" type (ie. as if the Longitudinal-Balance-Beams MUST pivot about their mid-points).

The problem with this thinking is;
If the GOAL is to have "The Warp-Mode" completely soft, then the Elastic-LLTD MUST BE 50:50!

As soon as you LOCK yourself into this type of thinking, then you can NEVER change handling balance by adjusting the Elastic-LLTD as UWA does, and as you and Loz correctly explained. You can NEVER have a "Twist-soft" suspension that gives 40:60 LLTD (as in my earlier post). And you certainly can NEVER have a "Tractor-soft" suspension (with 0:100 LLTD, as mentioned by Pete), even though these latter suspensions are just as "road-twist compliant" as the 50:50 Warp-soft ones.

Instead, when locked into the 50:50 Warp thinking, you might try adjusting Total-LLTD by changing the "Roll-Centre" heights (actually lateral n-line slopes!) at each end of the car, as Dynatune said they did in one of his few details. But, apparently, even the "brightest VD engineers" of the 1990s had great difficulty doing this. And here I must wonder why they even bothered with RC heights, when the "variable-LLTD-Twist-soft" approach is so much easier.

So, to repeat again, the problem is a conceptual one. Namely, locking yourself into a particular way of thinking because someone, somewhere, sometime, ARBITRARILY DEFINED something in a particular way (eg. "Warp MUST be 50:50"), even though that definition may not be useful to you.

Z