Shimmy - Around Steer Axis - What do you know?

[CWAnthony]

Hi all,

I'm working on a project that on initial shake-down has shown symptoms of shimmy through the steering system. No data was gathered, and observation/driver accounts are not that great at the moment. We can't correlate it with any situational constants just yet (vehicle speed, a certain "bump" input etc). We hope to explore it in more detail within the next few weeks or so.

The vehicle has no compliant suspension, has a minimal, reasonable amounts of castor trail and castor angle, same for kingpin angle (zero scrub radius/kingpin offset - "centre-point steering").

I suppose I just wanted to gather people's general thoughts/knowledge/experience on the subject. I've got the Millikens' "Chassis Design: Principles and Analysis" book arriving tomorrow, which as I remember covers the subject in a large amount of detail.

I've been aware of this phenomenon in the past, mostly through university. I still remember one of my lecturers, very experienced with MSC.ADAMS work during his studies and work in industry saying that shimmy is one of the hardest phenomenons to recreate in simulations.

I've experienced it a few times in my VW beetle, and it occurred pretty much always on the same stretch of road, at the same speed, intiated by the same man-hole cover! This occurred in spite of my Beetle being fitted with a steering damper (no it isn't leaking!). I'm also aware that Land Rover Defenders, even the new ones, also have steering dampers, presumably to reduce shimmy which is a problem due to the presence of a solid front beam axle.

One of my initial thoughts then, is that shimmy may be reduced/removed if the wheel is able to laterally translate in relation to the body. Even though "independent suspension" is present on a beetle, the wheels cannot laterally translate, due to pure trailing arms and solid joints. Similarly, a Defender, or any solid front beamed-car can also not allow the wheels to translate inwards/outwards. Conversely, a car with a "wishbone" can allow lateral translation during vertical bump (wheel moves closer to body along "y-axis"/lateral axis of the car). Also, more strongly, wheels on more modern cars can laterally translate thanks to bush deformation, without any vertical bump movement needed.

My interpretation, based upon initial (internet) research, is that shimmy seems to be an un-pure form of gyroscopic precession (as opposed to gyroscopic nutation). The front wheel's rotation axis (hub spindle) cannot follow it's own complete circle when "shimmy" occurs, as classical "precession" theory seems to describe. Instead, the spindle axis almost seems to just oscillate back and forth about the kingpin's axis, hence the steering system movement. Perhaps this does make it nutation, not precession? Either way, this doesn't really get me anywhere, as I don't yet know what causes the oscillation, nor how it can be mitigated. If new Land Rover's still have dampers on, perhaps the problem can't actually be removed without adding in compliance in the way of a suspension system.

So, what do you know?

Chris

 

[BUGGAR]

Same problem with a beetle. Tighten the torsion arm link pins and that will cure the shimmy.

Zero scrub radius causing the vibrations because the wheels are "hunting" laterally?? Usually there has to be some play or deflection in something to allow vibration but zero scrub can exacerbate the issue.

Let us know what you find.

Bob

 

[CWAnthony]

Interesting thoughts Bob. So when you say "hunting laterally", do you mean there is potentially force/torque direction reversals occurring around the steer axis, caused by pretty standard changing contact patch forces/torques and the minimal scrub radius? Seems like an interesting theory, if I could adjust scrub radius on our car easily I'd give it a shot straight away, even with minimal confidence!

But this shimmy is not constant, it doesn't occur all the time. I am presuming that it's initiation is from some kind of road input (although to repeat, we're not sure on this yet), but I don't see how torque reversals due to small scrub radius could cause the constant oscillations that we are seeing. The mechanism you describe, I imagine, would have a much less "ordered" behaviour.. if that makes any sense.

Have you ever actually seen shimmy induced by low scrub radius before Bob, and seen it cured by scrub radius adjustment/increase? Or is it just an idea you've had after reading my post? No offence meant by this question, I'm not trying to test your credibility or anything, I'm just wondering if it's a "common problem".

Thanks,

Chris

 

[BUGGAR]

By “hunting laterally”, I meant just as you described. And as you said vibrations usually have to have a force input to at least get them started. The forces have to, of course, start at the tire contact patch and can be caused by scrub radius eccentricity – e.g. forces not acting in line with the king pin axis and trying to turn the wheels.

I like to use the “principal of exaggeration” to help understand a problem. If you have a real long scrub radius, there will always be forces pushing the wheel to rotate about the kingpin axis in the direction of that scrub radius. A big scrub radius will make these forces too great to be overcome by counter-vibrating modes which are trying to move to wheel both ways.

A short or no scrub radius will have scrub forces about the king pin too weak to overcome reverse forces from inertia vibrations (initiated by uneven road surfaces, suspension geometry, or?) and the wheel will wobble about the kingpin stimulated by the opposing inertia forces and you get shimmy.

To test the idea, try installing a wheel with a different offset/backspacing to change the scrub.

My scrub radius “theory” is just an idea with no basis other than my understanding of vibration analysis. I haven’t found this described anywhere so I’m trying to learn as I go. I played around with different wheel offsets on some off-road Land Cruisers and found that with zero scrub the vehicles would climb rocks without steering kickback but would wander all over the road at highway speeds and easily go into shimmy on rough terrain at high speed. These had solid axles and stock steering linkage with no dampers.

Finally, it takes either play or deflection in the components for vibrations to occur. Although it is best to get rid of any forces causing vibration, it is also necessary that everything be tight and stiff.

I hope this makes some sense. Let me know.

Bob

 

[CWAnthony]

Hi Bob/all.

Sorry for the delayed reply, but I have quite a substantial update on this subject which I hope many will find interesting. I hope the length of the post does not put people off.

First, I should point out that since reading Maurice Olley's notes on the subject, as published by the Milliken's in the aforementioned book "Chassis Design: Principles and Analysis", I no longer think of the subject in the way I described in my first post!

The main thing I have learnt is to distinguish between two phenomenons; shimmy, and castor wobble. According to Mr. Olley, shimmy is a coupled oscillation between steering system (around steer axis) and front axle in the "tramp" mode (rotation about longitudinal (x) axis). This is what is seen on solid-front-axle'd vehicles (Defender, Ford F-150, etc). Castor wobble is just rotational oscillation about the steer axis. Two key differences are that shimmy requires input from both front wheels to exist, whereas castor wobble can exist due to only one wheel. Therefore it is castor wobble that is seen on shopping trolleys, motorbikes and aircraft nose wheels, as well as some cars.

It is a case of castor wobble that our car is suffering with, as we have no (considerable) vertical compliance for any kind of "tramp" mode to be present.

So on the weekend, a few colleagues and myself investigated this phenomenon pretty heavily. It is actually very easily excitable, some road inputs may cause it, but most notably, the driver can cause it by simply 'knocking' the wheel with his hand and letting go. The oscillation does not exist when hands are on the wheel, and replacing hands on the oscillating wheel can easily damp out the oscillations.

Changing between wheels of varying trueness did nothing, the wobble is independent of speed, brake torque, brake torque imbalance, tyre pressure changes, tyre pressure imbalance, added kingpin friction, and imbalanced kingpin friction.

What did remove the oscillations completely were mass changes - removing mass from the front axle, or adding mass to the steering wheel!

This aligns with how Olley describes castor wobble, and also with some things I have read on this site. Castor wobble can be present with matching natural frequencies; oscillations of the vehicle body (in the "steering mode") and oscillations of the steering system itself about the tyres.

The former "steering mode" oscillation is what I find most interesting, this oscillatory motion is due to an obscure "rocking axis" that Olley noted the vehicle body seems to oscillate around during an instance of castor wobble. He states that even for a vehicle with 'suspension', the springs did not seem to be actuated, this was movement of the vehicle on it's tyres alone. He describes it as a mix of roll and yaw motion, around an equivalent, diagonally positioned axis. This I believe is motion excited by the steering geometry; the varying amounts of KPI, castor, offsets, will determine how much roll or yaw motion may be seen in these oscillations. These variations, together with the inertia properties of the body will give an effective natural frequency of body motion whenever the steering system moves it.

And so it seems that, perhaps coincidentally, the body's natural frequency in this mode currently must be the same as the steering system's. My next step then, is to try and confirm this, by making rough calculations/measurements of these frequencies.

I do also plan to investigate this issue properly and fully using multi-body analysis software. We still need to work out what a feasible solution would be in our case, even if this does turn out to be the cause. We currently seem to need a very large mass change in order to remove this wobble, well outside the current scope of our requirements. So currently I hold the view that some kind of steer axis change might be enough to alter the effective body natural frequency enough to avoid matching of the steer system's, without any mass/inertia changes to the vehicle being required.

And in our case there is also an added complication, as this problem has been intermittent, it has been present at one 'event', then not present at another, and now it is back again. So we need to try and explain this too. If interest remains, I'll keep you all posted.

 

[BUGGAR]

Fascinating. Please keep me/us informed. Unfortunately I don’t have Olley’s book but I am interested in his work.

I’m still trying to understand Olley’s vibrations related to the “rocking axis”. I’ll sleep on that one until I understand it better.

What suspension geometry are you using? You noted that you “seem to need a very large mass to remove this wobble”. Thinking out loud here: Change in mass in a vibrating system changes the frequency of vibration in that system and apparently in your case more mass will inhibit the vibration. I further understand that adding mass is not a viable solution for your vehicle so that is not an option.

As I further understand, you are investigating modifying the steering geometry to de-tune it from the effective body natural frequency. As noted above, I don’t yet understand the body frequency theory so please let me know what you find here.

I still find myself going back to vibration basics – you need an oscillating force input and play/deflection in the system to get vibration. Your vibration “system” occurs in only one wheel and is not transferred to or from the other wheel thus isolating your problem to that one wheel. If everything seems tight and stiff, could you be getting the cat on the Golden Gate Bridge vibration syndrome – very small inputs are amplified enough (through coinciding frequencies) to set the whole system into vibration? You can get rid of any play in the system but maybe achieving sufficient rigidity is impractical. Who wants six inch diameter steering links!

Absent any magic bullets to share with you, I agree with you that maybe it’s time to play with the steering geometry. I’m an old man and very fixed in my ways so I still think zero scrub radius is a suspect. Let me know.

A rhetorical question I have is why is this vibration restricted to one wheel and not translated (coupled) to the other by the steering mechanism? There’s gotta be some difference between the two wheels for this to happen?

I think this is a great topic. Please share some more.


Bob

 

[CWAnthony]

Hi Bob, I regret checking the thread so late (0130 over here), as now I can't resist replying! I do appreciate your response though.

There is a diagram in the book that describes what Olley means by this rocking axis probably better than I can in words. Perhaps I can scan it in at some point. Essentially imagine the vehicle roll axis (x), and the vehicle yaw axis (z), then imagine an axis at 45 degrees to both in the same plane (it would be sticking out the front/top of the car, like a unicorn's horn, for want of a better description!). This is the "rocking axis" that Olley felt that the car oscillated around when demonstrating castor wobble.

Please keep in mind that this is only a subjective description of his perception of the motion, it's existence is not set in stone. Nevertheless, it is not a complex dynamic motion to imagine, as I have interpreted it at least. It is actually very easy to picture. To expand, imagine a car with a huge amount of castor trail, let's say 1 foot! When you turn the steering from lock to lock, the vehicle body will yaw, won't it? And similarly for cars with large KPI angles/offsets, you get large amounts of "castor jacking", which causes a perceived "rolling" of the chassis (although perhaps this is twisting, to be more accurate). When the two are mixed, with whatever strengths, due to whatever steer axis design you may have, the steering system is moving the vehicle body in free space in some way, correct? It doesn't matter whether you want to analyse the body's movement as being around a "rocking axis" or not. In fact, this "rocking axis analogy" is probably about as useful as the "roll axis analogy"; it is useful as just that, as an analogy.

So, point being, steering system displacement also causes body displacement. And depending on your yaw inertia, and the portion of body displacement that is yaw, your roll inertia, and how much of the body displacement is roll, and also with translational movements/inertia of the body considered, the body will have some kind of natural frequency in this "steering mode". This natural frequency doesn't need a spring to be excited, it is like a pendulum natural frequency. The steering system, in this case, supplies the constraints of motion, and also the input energy (through the tyres).

This is as best as I can describe that idea, I hope it is a bit more clear to you. So my thoughts were, that this body natural frequency must be similar to, or some multiple of the steering system natural frequency. The steering system natural frequency of course being a product of all rotating/moving steering mass, from both front wheels all the way up to the hand wheel.

And so my further thoughts are, if I were to adjust the steering axis in some way, let's say, to double my trail, the vehicle body will yaw a lot more for constant steering displacement, and so if the same yaw inertia is kept, a lower natural frequency for the body in this steering mode may result. In the same way, I could adjust any other parameter of the steer axis and get a similar effect. Any adjustment that results in the body's natural frequency in this mode moving away from the steering's natural frequency should be enough to stop the wobble. So in a way, scrub radius adjustment could be a viable solution, but I don't think in the way that you initially described Bob. And neither is it the one-and-only solution.

Some additional notes in response:

To expand further on what I said, Olley says castor wobble may be CAUSED by only one wheel, but both wheels in our case, will still oscillate, because they are of course linked. We are not seeing just one wheel oscillating as I may have unintentionally implied, both our front wheels oscillate. But Olley states that in the case of castor wobble, only one wheel may be the culprit. This could be true in our case, but it's impractical for us to find out ourselves! I merely added this info in to explain how "shimmy" seen on aircraft nose wheels, trollies etc should more accurately be referred to as castor wobble, because they exist in single wheel systems.

Our vehicle does not have any suspension/suspension geometry, just steer axes. Hence why there can be no "tramp". Sorry, I don't actually seem to have made this clear!

Personally, I struggle to see why lash/play in the system, or any system in fact, is necessary for these types of oscillations to be seen. But I have not been taught any mechanical vibration theory past a very basic level. I can say though, that whether our system is free enough of play or not, this system property has not changed, and yet altering the mass properties of the system has both allowed the shimmy to exist, and mitigated it. This to me, is enough to imply that play in the steering system, at least in our case, is not the cause of the castor wobble.

Olley (as originally observed by Rolls Royce, in fact) states that the energy from shimmy/wobble in order for it to be sustained comes from the road. If you raise the oscillating wheels from the road, they cease to oscillate. So as far as I can work out, the oscillation may begin from, let's say a driver input. The steering system responds at it's natural frequency, causing steered wheel displacement and force build-up at the tyres. The tyre force acts to restore steer position to neutral, but at the same time as the steering system displacement of course, the body has been disturbed. The constructive interference between these two oscillating masses (body and steering system) is so that as the steering system passes it's trim position (due to tyre restoring force), inertia from the body acts at just the right moment to push against the steer system and continue to displace it in the other direction (de-stabilise it). This will cause another tyre restoring force build up, in the opposite direction, and the cycle repeats itself.

This I'm afraid is all I can add at the moment. It might be a bit scruffy or hard to read, but I'll make any necessary amendments in the morning! I hope at least some of it makes sense.

 

[BUGGAR]

Excellent write up CW! I'm going to enjoy taking some time to think it out.

Meanwhile, I can't add anything right now but I am looking forward to any test results.

Bob