Wheel shimmy

[MCoach]

I'm moving the current conversation on wheel shimmy and all its fun to here as to not continue to detract from the UTAS build thread.



I believe it would get to zero or negative (loaded side) easier than you could get 100 (unloaded side).

You know, you're absolutely right. I was just trying to point out how large of a variation it can be, but you're right, it's much more likely to go the other direction. Silly me.


Interesting, MCoach. Would be great if you can find a video. I wonder if you have a picture of the chassis tube and your ackermann layout too?

I can't take pictures of the car currently as we packed it up this week and sent it off to the Detroit Auto Show. But let me see what else I can dig up. It definitely wasn't what we planned, but I misunderstood a frame rule and it was too late by the time it was clarified. The required added tube went right through the rack, so we had to move it about.


There are many possible oscillators in automotive suspension/steering/chassis. Olley wrote about some of these and we combined his several monographs into one chapter in "Chassis Design" (SAE R-206) -- Chapter 6 Oscillations of the Unsprung, about 70 pages. His examples seem like a good way to start thinking about these problems -- although you haven't said enough to say with certainty that he discusses your specific problem.

Olley's notes were actually my inspiration for how to tackle the problem so maybe... I need to go review that chapter and get more familiar with it.
Let me see if that chapter rings a bell....

[MCoach]

CWA, best pictures I could find of our steering geometry for now are these:

https://www.facebook.com/KetteringFSAE/photos/pb.209138079117577.-2207520000.1420859174./788212254543487/?type=3&theater

https://www.facebook.com/KetteringFSAE/photos/pb.209138079117577.-2207520000.1420859142./796138317084214/?type=3&theater

The rack sits in front of the upright and the tie rod picks up on the back of the upright. The tie rod is the member just above the front link for the front lower control arm.

[Z]

MCoach,

Interesting topic...

But I reckon the original quote would make a catchier thread title.

"Under threshold braking the car would enter a "death wobble"..." :)
~o0o~

More pics or videos would help, but I think I see the root cause of the problem.

1. Braking loads cause...
2. Rearward compliance/movement of both upright-lower-BJs.
3. Which cause LARGE TOE-OUT COMPLIANCE of both front-wheels, because of plan-view-angle of tie-rods.
(Edit: Above happens even with Offset (aka Scrub Radius) = 0, but big Offset makes the toe-out even worse.)
4. Next, side with bit more grip causes car to veer to that side (say R) (remember, large toe-out!).
5. Subsequent body-roll and SD forces cause other side (L) to now become more heavily loaded.
6. So car now pulls more to that side (L) (because large toe-out!).
7. Repeat from step 5, swapping Ls and Rs each cycle.

So, ... perhaps a "braking-load toe-compliance test" is in order.

1. Lock R&P and one front-brake (vice-grips).
2. Tie ratchet-strap to front of wheel, then under tyreprint and back to rear of car, but with fish-load-scale somewhere in the line.
3. Apply load via ratchet, measure load with fish-scale, and measure resulting front-wheel toe-compliance.

then ...

4. Print big poster for shop-wall with picture of above test, and,
"This sort of TOE-OUT-COMPLIANCE NEVER TO BE REPEATED!!!".

Z

(PS. Try this simple test. Set both front-wheels at 5 degrees static-TOE-OUT, each. Drive car slowly. Familiar?)

[DougMilliken]

So, ... perhaps a "braking-load toe-compliance test" would be in order.
An alternate setup is described here http://www.shopeshop.org/tim.3.htm Bill Shope gives directions for a simple "traction dyno" which will work equally well as a "brake dyno" if the chain runs the other way -- lock the front brakes and pull on the chassis at the CG height. There are some subtleties if you get really serious about this kind of testing, but a simple setup might be enough to get a rough idea and determine if compliance is the problem.

[BillCobb]

Since its at the Detroit Auto Show, have someone there snap some pics. When its done there, detour it to Milford and get some weekend MTS K&C tests done at the GM Proving Grounds. I Have a name and a number to see about the possibilities for making this happen. A FA braking compliance test (in phase and out of phase inputs) is usually a standard procedure. The only problem you would have is welding or fastening on the frame mounts for the magnetic clamps. Because of the short wheelbase, it might be necessary to run one axle at a time, but I happen to know they have done a few FSAE cars there.

Ya know, a mechanical drive on K&C test device on a trailer suited to FSAE cars could be a real money maker for some post graduates who can't give it up. You could probably get by with grease plates instead of air bearings and maybe some Iphone apps for wheel angles (steer and camber) etc. Apply loads forces and moments air cylinders on the load plates and presto. Elevators on the wheel pads can do a pretty good ride and roll input for small displacements.

Extra points for meeting your FEM predictions.

[CWA]

Good idea on the new thread MC, I have a keen interest in this topic. Is this an issue you/your team are currently working to fix, or is it on an old car that nobody is using any more? If you manage to try out the suggestions given above by other members on your car, it would be great to hear about the results.

Where Olley seems to differentiate between castor wobble (oscillation of steering system only) and tramp ("roll oscillation" of the front axle, also causing steering system oscillation), in my head your shimmy is closer to castor wobble, as may be seen on motorbikes and light aircraft U/C etc. Would you agree with this?

Some further questions I have: what is the frequency of the oscillation? Is it speed dependent; was the shimmy present at all speeds, did frequency change with speed ?

My own investigation into a particular case of castor wobble showed that a similarity between 'steering system natural frequency' and the 'natural frequency of the body-when-steered mode' was a causal factor. We could remove/add mass from the body to remove the shimmy behaviour, likewise adding mass to the steering wheel also removed the shimmy behaviour. So, whilst there must have been some forcing mechanism (an example of which Z described), which you could call the root cause of the shimmy, this forcing mechanism only caused the steering system to visibly oscillate (only caused us a problem) because of the matched frequencies. Removing the causal factor did enough to stop the oscillations, and removed the problem for us. To this day, we do not know what mechanism we can call the root cause, how exactly these reversing forces were generated (AFAIU, Olley's work could not lead to a definitive explanation for this either), but it turns out that we did not need to. Perhaps this could be the same in your case, too.

So, the next thing I would be interested to know is, does adding mass to your steering wheel remove this behaviour?

[Pete Marsh]

At UWA we have seen some of this with our new beam axles as well. We have put it down to two issues.

I agree with Z as to the driver. We see it only at wheel lock, and usually get out of phase alternate locking front wheels to go with it. The Goodyear tyres have quite a sharp drop in traction when they lock, giving a strong difference in longitudinal forces across the axle.

First, I think the ratio of scrub to trail is important, with higher scrub than trail to be avoided. I think we used to get some camber gain that moves the effective centre inboard and improves this when we had independent suspension, and as the beam does not do this, the static settings need more trail or less scrub compared to independent systems. Clearly, as scrub approaches 0, the steering becomes immune to any longitudinal loads, but it seems you don't need to go all the way to zero, although with tyre flex included I bet some cars go close.

Our new tractor style uprights have all roller element bearings, and hence have MUCH less friction than any of our previous designs. Could only be good right? Well the reduction in steering work/effort is substantial, (and valuable), and force based steering (or "feel" if you prefer) is improved, but I think the friction in older designs went a LONG way to damping this shimmy effect. SO - we tried a motorbike steering damper fitted to the rack........and instant fix.
Interestingly, the resistance of the damper is imperceptible to the driver, but the shimmy is totally gone and you can merrily go on your way flat spotting tyres now.

Pete

[Z]

To add a bit to my above post.

If MCoach's steering-tie-rods were more purely lateral, then rearward movement of the uprights would not cause as much change in toe-angles, and the problem might be cured.

This approach is very common in today's production cars, which allow quite large "wheel recession" (perhaps 2-4 cm, for less "NVH" with low-profile tyres), but with the steering-linkage arranged to give as little change in toe-angle as possible. This usually done by having the steering-tie-rod and front-wishbone-arm close to lateral and forming a parallelogram, wishbone-inner-front-BJ quite stiff, and wishbone-inner-rear-BJ quite rubbery. However, this "wishbone" nowadays looks more like an "L-arm" (or "boomerang-bone"?).

Z

[MCoach]

MCoach,

More pics or videos would help, but I think I see the root cause of the problem.

1. Braking loads cause...
2. Rearward compliance/movement of both upright-lower-BJs.
3. Which cause LARGE TOE-OUT COMPLIANCE of both front-wheels, because of plan-view-angle of tie-rods.
(Edit: Above happens even with Offset (aka Scrub Radius) = 0, but big Offset makes the toe-out even worse.)
4. Next, side with bit more grip causes car to veer to that side (say R) (remember, large toe-out!).
5. Subsequent body-roll and SD forces cause other side (L) to now become more heavily loaded.
6. So car now pulls more to that side (L) (because large toe-out!).
7. Repeat from step 5, swapping Ls and Rs each cycle.

So, ... perhaps a "braking-load toe-compliance test" is in order.



I'm not sure if this was the issue, but I'm willing to explore it on free time. I will recognize that the spacing on the pick-up points for the control arms are not the optimal for taking loads, but we were trying to give some room for lots of steer angle and plenty of room for opposite lock. Now that we have a better idea of how much steer is really needed we can spread them out and dial it back from the F1-esque control arms.


Since its at the Detroit Auto Show, have someone there snap some pics. When its done there, detour it to Milford and get some weekend MTS K&C tests done at the GM Proving Grounds. I Have a name and a number to see about the possibilities for making this happen. A FA braking compliance test (in phase and out of phase inputs) is usually a standard procedure. The only problem you would have is welding or fastening on the frame mounts for the magnetic clamps. Because of the short wheelbase, it might be necessary to run one axle at a time, but I happen to know they have done a few FSAE cars there.

Ya know, a mechanical drive on K&C test device on a trailer suited to FSAE cars could be a real money maker for some post graduates who can't give it up. You could probably get by with grease plates instead of air bearings and maybe some Iphone apps for wheel angles (steer and camber) etc. Apply loads forces and moments air cylinders on the load plates and presto. Elevators on the wheel pads can do a pretty good ride and roll input for small displacements.

Extra points for meeting your FEM predictions.

I'd be very willing to take you up on that. Give me a 2-3 more months and I can bring two cars down, the one with the problem and the one succeeding it.
I've been interested in building a tire testing rig for miniature race vehicles such FSAE cars, karts and the like. It might not mean much to the latter because they usually have a habit of doping the tires beyond all recognition anyways.


Good idea on the new thread MC, I have a keen interest in this topic. Is this an issue you/your team are currently working to fix, or is it on an old car that nobody is using any more? If you manage to try out the suggestions given above by other members on your car, it would be great to hear about the results.

Where Olley seems to differentiate between castor wobble (oscillation of steering system only) and tramp ("roll oscillation" of the front axle, also causing steering system oscillation), in my head your shimmy is closer to castor wobble, as may be seen on motorbikes and light aircraft U/C etc. Would you agree with this?

Some further questions I have: what is the frequency of the oscillation? Is it speed dependent; was the shimmy present at all speeds, did frequency change with speed ?

My own investigation into a particular case of castor wobble showed that a similarity between 'steering system natural frequency' and the 'natural frequency of the body-when-steered mode' was a causal factor. We could remove/add mass from the body to remove the shimmy behaviour, likewise adding mass to the steering wheel also removed the shimmy behaviour. So, whilst there must have been some forcing mechanism (an example of which Z described), which you could call the root cause of the shimmy, this forcing mechanism only caused the steering system to visibly oscillate (only caused us a problem) because of the matched frequencies. Removing the causal factor did enough to stop the oscillations, and removed the problem for us. To this day, we do not know what mechanism we can call the root cause, how exactly these reversing forces were generated (AFAIU, Olley's work could not lead to a definitive explanation for this either), but it turns out that we did not need to. Perhaps this could be the same in your case, too.

So, the next thing I would be interested to know is, does adding mass to your steering wheel remove this behaviour?

It's our car that competed in the 2014 competitions and still in active use to evaluate our designs for the 2015 car. The problem is currently 'cured' but I'm pretty sure we could get it excited again, I don't think I have the tires nor the time to do an in-depth exploration of it. I don't remember if the shimmy was dependent on speed or not. I wouldn't want to go down the path of a steering damper, as that seems like a big band-aid on a problem that needs surgery.

I'm trying to remember, but want to say we also saw The Wobble at it's worst case under acceleration at around 20 m/s as well.



At UWA we have seen some of this with our new beam axles as well. We have put it down to two issues.

First, I think the ratio of scrub to trail is important, with higher scrub than trail to be avoided. I think we used to get some camber gain that moves the effective centre inboard and improves this when we had independent suspension, and as the beam does not do this, the static settings need more trail or less scrub compared to independent systems. Clearly, as scrub approaches 0, the steering becomes immune to any longitudinal loads, but it seems you don't need to go all the way to zero, although with tyre flex included I bet some cars go close.

Our new tractor style uprights have all roller element bearings, and hence have MUCH less friction than any of our previous designs. Could only be good right? Well the reduction in steering work/effort is substantial, (and valuable), and force based steering (or "feel" if you prefer) is improved, but I think the friction in older designs went a LONG way to damping this shimmy effect. SO - we tried a motorbike steering damper fitted to the rack........and instant fix.

Pete, do you think the ratio of scrub radius to caster trail is still applicable at low values? We made a lot of changes for the 2014 car, so I wouldn't be able to pin it down to one major factor. Looking for commonalities here, we picked up some ball bearings with much, much less friction than the previous large combination roller element bearings. The old bearings had so much friction that we needed to push our cars down hill! We also switched steering racks to one that has half the steering ratio that we used previously, quite a difference there in steering inertia. But, with it being a rack and pinion I guess that goes both ways...

[Pete Marsh]

MCoach

The new uprights use roller bearings in the STEER axis, as well as the wheel axis of course. It is the huge reduction in steering system friction I think contributed to the shimmy in our case.

I personally believe a good dose of actual lat force feedback is important to driving the car. You need to let the steering angle float as normal force changes to keep the car in a set through dips and crests, and I don't think the tyre self aligning torque alone is enough to give you the feedback you need. So I would not like to see near 0 trail. Also, it is hard to package near 0 offset without adding KPI, which dilutes your camber gain from caster.

So I guess I have never considered the effects of near on centre steering with the UWA cars, as they all have some trail. With the roller element uprights removing masses of friction, the "feel" is much improved, and there is no doubt the feedback forces could also be reduced, but we are yet to try this, but I think the steering could easy end up too light, and too hard to find the bite on turn in. I have driven some of Monash's cars, and they have a very impressive solution that both provides good feedback feel and does not load up excessively with the very high lat forces they get. I suspect their solution involves both near on centre geometry, and also friction reduction, such that self aligning torque is dominant. In their case the floppy 10s and big down force would allow high KPI and low caster (or lots of offset) without too much ill effect IMO.

Pete

[MCoach]

Ah, the STEER axis, now you have my attention. Makes more sense, too.

I still need to get back to read up on shimmy from Olley... I've spent the last few days trying to get projects together for the snowmobile team, it's 110% season around here right now.

Maybe it's worth removing some "feel" to instill some stability in the vehicle -- and driver by allowing friction to exist as a damper? That seems counter intuitive and more like a band-aid solution. Hm...I'll read up more before trying to infer too much. In the case of Monash, I'll have to go scrape what information I can from pictures. Do you, or perhaps someone else, know some one at Monash that'd be willing to talk with me? I'm curious what else they may be taking into account that I could consider to further my concept, such as if aero plays a balance that may have more effects on the driver and feedback than numbers alone could tell me right now...

Thanks.