This article was first published in AutoSpeed.
Last week in Part 5 of this series I briefly discussed all the different types of suspension, listing
their pros and cons for application in a lightweight Human Powered Vehicle.
Rather surprisingly, the top three were: solid front axle, sliding pillar (or
sliding kingpin) and swing axle.
However, each of these has substantial negatives.
The first to be wiped from the list was the solid front axle: I wanted the suspension movements of the front wheels to be independent. Then
there were sliding pillars, but the thought of trying to get a shaft to slide up
and down at least 100mm without stiction, and with low wear, was rather off
putting. (Furthermore, in such a system, how do you configure the steering to
So what’s left? – swing axles! Surely not swing
No – Not Swing Axles!
There are some serious problems with swing axles.
As the vehicle rolls, the force acting along the
line of the suspension arm tends to lift the vehicle body. This is, firstly,
because the roll centre is high above the ground, and, secondly, the swing axle
forms a short virtual swing arm. As a result, the outside, more heavily loaded
wheel can ‘tuck in’, giving a sudden loss in cornering grip. (That's especially the case if there's a mid-corner hump.) The short virtual
swing-arm also causes a lot of camber change of the wheel as the suspension
moves through its deflection.
Swing axles have been used on a number of car
designs, although usually on the rear rather than the front. The pictured
Lightburn Zeta is one of the few cars that has run a front swing-axle design.
Early Volkswagen Beetles, the Corvair and older Mercedes used swing-axle
Leaving aside the lifting forces on the body
(sometimes also called ‘jacking’), how much camber variation would actually
occur with a swing-arm front HPV suspension? If we assume a track of about
800mm, a maximum swing-arm length (taking into account the space of the pivot
point) is about 350mm. If the wheel travel is 120mm, the wheel camber will vary
by 19 degrees. So if we have zero negative camber at full droop, we’ll have 19
degrees neg at full bounce.
But while that sounds like a crazy amount of
camber variation, it’s not way out of the ballpark – my first HPV ran 0 degrees
neg at full droop, 5 degrees neg at normal ride height and 10 degrees at full
So how do you reduce the amount of camber
variation? You can’t use longer arms because they’d then be longer than half the
track. (Well, you actually can – see the breakout box at the end of this
But actually there is a way of making the
Rather than having the swing-arms at right-angles
to the vehicle’s longitudinal axis, why not have them angled forwards? If
they’re angled forwards by (say) 45 degrees, the amount of camber variation is
reduced. Furthermore, the roll centre is also lower!
One suspension design that uses angled arms is the
semi-trailing arm rear suspension used in 1960s and 1970s BMWs and Datsuns (and
plenty of other cars as well). As this diagram shows via the red lines, the
pivot axis for the arms is neither parallel with the long axis of the vehicle
(which would make them swing-arms) and neither is it at right-angles to the long
axis (which would make them trailing links).
Rotate the assembly through 180 degrees (or drive
the car backwards if you like!) and instead of a semi-trailing arm rear
suspension design, you have semi-leading arms. I have never seen or heard of any
vehicle using this system but it seems to me that it answers all the HPV design
It has the potential to:
Give appropriate negative camber increase on bump
(and so of the outside wheel with body roll)
Use only a single suspension arm per side, so be
lighter than a double wishbone system
Work with a spring motion ratio that can,
depending where along the arm the spring is located, vary from a low motion
ratio to a high motion ratio
Provide a large wheel travel
Furthermore, by using semi-leading arms (rather
than semi-trailing arms), the room forwards of the front wheels is left free for
pedalling legs and steering castor will increase with suspension deflection (so
under brakes, for example).
I imagine one reason that semi-leading suspension
has not been used previously on vehicles is because with conventional steering,
massive amounts of bump steer would occur.
Bump steer occurs when the steering input is held
steady but the wheels still steer as they move through their suspension travel.
This movement tends to be opposite in direction for each wheel. In other words,
the wheels might be parallel at normal ride height (ie toe of zero) but at full
bump might point inwards (toe-in) or outwards (toe-out).
For a given suspension, the amount of bump-steer
that occurs is dependent primarily on three things: the position of the inner
steering ball-joint, the position of the outer steering ball-joint, and the
length of the tie-rod. This diagram
[taken from Fundamentals of Vehicle Dynamics
shows the ideal length of the tie rod (here called a ‘relay
linkage’) and the correct position of the inner and outer balljoints to avoid
bump-steer in a double wishbone suspension.
If the steering geometry is incorrect, the amount
of bump steer that can occur is massive – it’s not hard to get 10 degrees of
steering change over full bump!
In a semi-leading arm suspension that uses
conventional steering tie-rods (ie positioned at right-angles to the long axis
of the vehicle), the tie rods will be much shorter than the suspension’s
semi-leading arms. As a result, the outer tie-rod ball-joint and the wheel hub
assembly will move through different arcs, causing the steering tie rod to pull
on the steering arm and so create massive toe-out on bump.
However, and here’s where it gets even more
interesting, some Greenspeed recumbent trikes use a steering system that angles
the tie-rods forward in a way that’s very similar to the positions that would be
adopted by semi-leading arms. So if the steering tie-rods and the suspension
arms can be made of similar length and run parallel with one another, bump steer
could possibly be avoided.
But hold on, there’s another problem! As many of
you would know, semi-trailing arm suspension tends to toe-in on bump. (That
works well on the rear suspension of a car, because the rear toe-in stabilises
the cornering car by reducing the propensity for oversteer.) Turn this through
180 degrees and that becomes toe-out on bump. So if there are toe changes
built into the design of semi-leading arm suspension, won’t bump
steer inevitably occur?
As described above, bump steer depends on inner
and out ball joint positions and the length of the tie-rods. But to put it
another way, for a given steering system, the position of the
inner and outer suspension arm mounts and the length of the suspension arm will
determine bump steer outcomes.
So it should be possible to position semi-leading
arms to work with the Greenspeed steering system, compensating for the bump
steer that would otherwise occur with that type of suspension.
But at this stage that was all just theory – and,
untried theory at that...
Next week: mocking up a system and testing for
camber, castor and bump-steer changes
Long Swing Axles!
mentioned in the main text, swing axles normally can’t be longer than half the
track of the car.
definitely the case with powered axles. But what about with non-powered axles?
here is the rear suspension of a 1960s Honda 1300, one of the first cars made by
Honda. As can be seen, the suspension system uses leaf springs and swing
axles....and the swing axles pivoted from the opposite sides of the car! This is
perhaps amongst the simplest car suspension systems I have ever seen. Not only
does it have dynamic camber negative increase, you can see that the leaf springs
are progressive in rate. Furthermore, with the lightweight leaf springs doing
the wheel location, there’s no need for additional suspension links...