This article was first published in AutoSpeed.
Last week we looked at something which initially
seems straightforward but quickly becomes very difficult: sourcing the lightest
possible springs suitable for a Human Powered Vehicle with a wheel suspension
travel of about 100mm. As described in that story, I bench-tested torsion bars
made from plastics, rubber and steel; I tested rubber in compression and shear;
I tested a cantilever spring made from a carbon fibre snow ski and I even tested
things like carbon fibre golf clubs.
As a result of all this testing I decided I needed
a spring with at least 75mm deflection (ie the motion ratio would be kept fairly
low) and the lightest possible mass. But what sort of spring satisfied those
criteria? What about air bags?
For an AutoSpeed story I’d previously visited
the Airbag Man in local Brisbane (Australia). There we’d done a story on
aftermarket airbag suspension – see Airbag Suspension Systems - Part One. What I had seen at the time had
suggested to me that airbags might be suitable for HPV suspension.
So when I started to consider airbags for the new
suspension trike, I went firstly to www.airbagman.com.au. That quickly
led me to the Firestone site www.firestoneindustrial.com and
from there I went to the data sheet for the 4001 Firestone Airstroke / Airmount
– see www.firestoneindustrial.com/pdfs
The Firestone 4001 air spring has a maximum
diameter of 3.1 inches, a minimum length of 3.6 inches and a maximum length of
7.2 inches, giving a travel of 3.6 inches (91mm). The minimum air pressure that
can be used is 10 psi; the maximum is 100 psi. Mass is quoted as 1 pound (450
grams) but is actually 350g.
Unlike air shocks used in some mountain bikes -
and air cylinders used in pneumatic machinery – these airbags don’t use a piston
sliding in a precision cylinder, complete with precision seals everywhere.
Instead, the construction is almost ridiculously simple.
In the designs of airbag truck and bus suspension
– and of the unit described here – the airbag rubber rolls back on itself rather
like the way your mother bundled your socks together in your bedside drawer. A
non-precision (but specially shaped) piston moves within the airbag rubber and
as the piston extends, the rubber unrolls itself. Sufficient air pressure is
needed to separate the rolling rubber walls (so explaining the minimum 10 psi)
and the maximum travel is dictated at the different extremes of travel by (a)
when the rubber nearly fully unrolls, and (b) when the piston and the opposite
end mount nearly come in contact.
With increasing air pressure, the diameter of the
airbag expands slightly and, of course, applies more extension force.
So, compared with steel coil springs, there are
already three very important differences.
Air pressure is needed before the spring will
start to properly work
The external diameter of the spring increases a
little with increasing air pressure
The minimum and maximum lengths of the spring must
be set by external stops
But I lot of that I found out subsequently – at
this stage I was just wondering about the availability and cost of the Firestone
I rang The Airbag Man to find out both good and
bad news. The good news was that the airbag was in stock and available. The bad
news was that it had a retail price of AUD$300. That was about twice what I’d
hoped for but it’s still much cheaper than local prices for esoteric mountain
bike air shocks. (To give another comparison: the steel coil springs that I’d
had custom made for my first trike had cost me AUD$77 each.)
I ducked into Brisbane and picked up a Firestone
airbag. My first reaction was that no way this looked like it should be worth
AUD$300. But then when I got it home and started playing with it, I slowly
changed my mind.
All Firestone’s literature suggests you shouldn’t
inflate the airbag without it being constrained between stops. But that didn’t
stop me connecting a high pressure bike pump and pumping a few strokes, only to
find to my amazement and consternation that the airbag suddenly unrolled itself,
popping fully open. Fortunately, with the right internal pressure applied, it’s
easy enough to roll the airbag back over itself...
A normal, constant rate spring compresses by the
same increment when the same weight is placed on top. So a 1kg per millimetre
spring will compress by 1mm when 1kg is placed on it, 2mm when 2kg is placed on
top and so on. On the other hand, rubber (eg in compression) has a spring rate
that rapidly rises – in fact, as you reach maximum compression, the rate is many
tens of times higher than initially. Air springs are different again. As with
rubber, the rate rises with compression. But the increase is much more
progressive through the travel, which can also be quite long.
This slowly rising rate characteristic of air
springs makes aspects such as designing the ride height (and so setting the
amount of bump and rebound travel available) much more complex than for linear
rate springs. In fact, I found that despite having available pretty detailed
specs sheets on the Firestone spring, by far the best data were gained by
testing the spring’s characteristics in a mock-up suspension on the bench.
I initially tested the Firestone 4001 air spring
with a simple bench-mounted set-up. A piece of timber was drilled horizontally
at one end to take a long, snug-fitting (but not tight) bolt. The bolt was
inserted through the timber and then its exposed end clamped in a large vice. At
various points along the timber I drilled holes in which the top mount of the
air spring could sit. Further along the timber I marked the spot where in the
finished design the wheel will go. This set-up allowed the spring to be placed
at various positions along the arm, so changing the motion ratio. By placing
heavy steel weights on the top of the arm at the spot marked ‘wheel’ I could
realistically create different loads, including static deflection, 1g vertical
acceleration in bump (ie weight doubling) and other loads.
In addition, by bouncing the arm on the spring, I
could directly measure natural frequency and see how much damping was present.
The first step was support the timber so that the
air bag couldn’t be fully closed up (ie put in place a full bump stop) and then
the second step was to place 30kg at the ‘wheel’. I then inflated the airbag
until the airbag was about half way through its travel. (Incidentally, seeing
the airbag inflate and then the load gently rise off the stop is a very
interesting experience.) With the 30kg design load (ie the sprung weight of the
trike and rider) applied through the appropriate length and ratio lever arm, I
could then bounce the ‘wheel’ up and down and watch the system’s behaviour.
However, it soon became clear that it wasn’t the
normal ride weight which was the critical design factor, but instead the problem
was coping with the 1g vertical bump, one which doubles the weight acting
through the wheel. To simulate this, I loaded 60kg on the timber at the wheel
position and then inflated the spring until the spring just lifted the weight
off the bottom suspension stop. Then, without adding or subtracting air, I
removed 30kg to return the load back to normal. With the load halved, the airbag
extended to what would be the normal ride height position. Maximum bump and
droop travel could then be measured from that position.
Doing this showed that with the 4001 air spring,
the closer that the motion ratio could be to 1:1 (ie wheel and spring movement
the same), the better the system worked. However, the nearer that the motion
ratio is 1:1, the closer the wheel travel matches the spring travel – so
limiting wheel travel to being in this case the maximum travel of the air spring
- 3.6 inches. A motion ratio of 1.3:1 was then used, which gave a maximum wheel
travel of about 4.75 inches, distributed a little more in bump than droop.
what was the problem with using a high motion ratio? After all, since a stiffer
spring is required, can’t you just add more air to the airbag?
happened was that the higher the motion ratio, the greater the bump travel
needed to cope with a 1g load. So to cope with a 60kg load (ie 1g bump), figures
like a 3.5 inch bump travel and a 1 inch droop travel resulted - obviously
getting very asymmetric. You couldn’t lower the air pressure (and so increase
the droop capability) because then the system wouldn’t cop with an extra 30kg
imposed on it without bottoming-out. In other words, to use a motion ratio
higher than about 1.3:1, a larger diameter airbag was needed that would develop
more force for the same air pressure. This I wanted to avoid because larger
airbags are heavier.
is the same as when using any springs with a high motion ratio – they have to be
very stiff and so heavy.)
when experimenting with using the airbag at a high motion ratio, one very
interesting characteristic was present. Because of the higher air pressure
required and the extra weight the airbag saw acting on it, the natural frequency
of the system dropped substantially. In fact, natural frequencies could get as
low as 1.4Hz!
As indicated, the spring rate of an airbag changes
with both static internal pressure and spring deflection. And, making things
rather complex, the air pressures rises with deflection!
Let’s take a look at that. With the spring
supporting 30kg at a motion ratio of 1.3:1, about 30 psi gave a ride height
roughly halfway through full travel. Pump the spring up harder and the ride
height also increased – the spring was stiffer. But even with no air added or
subtracted from the spring, the measured air pressure within the spring could
How? When the spring was compressed!
Adding more weight (the equivalent of experiencing
a bump) compressed the air within the spring into a smaller space and so its
pressure rose, so increasing the spring rate. This is what makes it so hard to
work out what spring rate will be present at what deflection, and from what
starting pressure. Firestone’s tech sheets do not show these figures. This
varying rate is another reason why I suggest the spring pressure and motion
ratio firstly be optimised for maximum bump.
stated, Firestone’s tech literature doesn’t include data on the dynamic
characteristics of their air springs, including the 4001. However, competitor
Goodyear does have this information available.
Goodyear 1S3-011 spring is similar to the Firestone 4001 (although it is taller,
slightly large in diameter and heavier).
is the dynamic data with a starting ride height of 5.5 inches. As can be seen,
with a starting height of 5.5 inches, a load of about 80 pounds (which,
according to another graph, needs about 25 psi in the spring), compression to
3.8 inches results in an increase in load to 225 lbs. So, over this 1.7 inch
compression the average rate is 85 pounds/inch. But according to the
graph, over the last 0.4 inches the rate is 125 pounds per inch!
Because of their internal air pressure change
characteristics (and that’s purposely kept vague because I don’t fully
understand the mechanism by which it occurs!), air bags have a very low resonant
frequency. Depending on the motion ratio (and so the required air pressure),
bench testing showed that frequencies in the range of 1.4 – 1.6Hz were quite
possible. Translated, that means far better ride comfort without requiring a
tall spring package.
already described, the air spring needs to be constrained so that it cannot
extend or contract too far. If the spring exceeds its minimum or maximum length
(especially if it does so with some force) it will be damaged. This is
especially important in bump, where a very large bump (eg 2 or 3g vertical
acceleration) could smash together the internal aluminium piston and the
opposite end mount. Not only would spring stiffness then rise hugely, but the
impact would likely cause damage. The answer is to use a full-bump rubber
car suspension systems, the rubber bump stops (especially the full bump stops)
are not add-ons: they are part of the suspension system. In other words, the
wheel rate that occurs as full bump is reached is caused not only by the main
spring but also by the auxiliary spring that comprises the bump stop. To achieve
a smooth transition from the (spring rate alone) to the (spring + bump stop
rate), the bump stop must be appropriately designed. So the bump stop must
prevent compression of the spring once the compression reaches a certain point,
but must only progressively slow the spring compression up to this point. This
requirement for a steeply rising spring rate makes rubber a good choice of bump
stop, especially if it is shaped to firm up over half an inch or so of
found the cheapest and most progressive rubber stops could be found in
lightweight cars and bought several from a wrecked Daewoo Matiz. They have a
mass of only about 80 grams.
requirements for suspension droop bump rubbers are less complex. The maximum
acceleration that this bump stop will be impacted by is provided by the spring,
with this occurring each time the rider gets off the HPV and the suspension goes
to full droop. So while a stop must be present, it can be a simple block of
very hard to picture what the airbag looks like in action unless you’ve seen it
it is at near full compression (note how it copes with the angular disparity
between top and bottom)...
here it is at full extension.
Air bags have some major advantages and
disadvantages over alternative springs. (And I think that applies for cars as
well as HPVs!)
The advantages are:
Adjustable spring rate / ride height (but note:
not separately adjustable!)
Increasing spring rate with deflection
Tolerant of ends which are angularly displaced (so
they’re like coil springs rather than being like pneumatic cylinders)
Lower resonant frequency than comparable coil
The disadvantages are:
But there’s one other killer advantage that I
discovered on the test bench, an advantage that I think makes the disadvantages
pale into insignificance. That advantage is that the behaviour of airbags can be
dramatically changed with simple valving...
Next week: interconnected suspension