This article was first published in 2008.
major problem in the tiny minority field of recumbent trike design is that any
assessment of the worth of a new machine seems to be always couched in
respectful, group-hug terms. So someone produces a suspension trike design that
has clear and obvious design flaws – and then everyone falls over themselves to
tell the constructor how wonderful it is!
course, this approach results in the perpetuation of mediocrity...
a major reason why in this article, I present as much hard data as I can
assemble and, where that data isn’t obtainable, try to be as honest and
objective as I can be in my assessment.
way, those building new (and potentially better) designs can see if in fact they
1. Ride Quality
The main objective in developing a suspension
trike was to give a much improved ride quality. So, what was the outcome?
To data-log ride quality I use a battery-powered
accelerometer (a tiny chip on a very small PCB) and place it on the seat. I then
sit on it (it's only a few mm thick). I then log to a Fluke Scopemeter.
(Incidentally, the accelerometer and board are very cheap and could be logged to
an old laptop.)
On the Scopemeter screen, the bigger the peaks and
troughs of the graphed line, the greater the vertical accelerations that are
being undergone by the rider. That means you can immediately get a feel for the
data by thinking that if the ride experienced by the rider was absolutely
smooth, the trace would be a flat line. The greater the bumps being felt, the
bumpier the trace.
Here I have shown the logged
vertical accelerations for three different trikes over the same stretch of road.
The road is from my house, up a hill, down a slight descent to a roundabout,
then back home again.
The road is absolutely typical of the roads around
where I live - bumpyish bitumen. On the ride I am slow up the hills and about 30
km/h down them. It takes about 5 minutes.
The trikes are:
standard commercially produced trike, a Greenspeed GTR (20 inch wheels, 60 psi
- my first suspension trike JET (20 inch wheels, 60 psi tyre
- Air 150 (20 inch wheels, 60 psi tyre pressures)
- Air 150
with 30 psi tyre pressures (that I normally run)
The comparison is
fascinating - and revealing.
The unsuspended GTR is as it feels - lots
of bumps, some at about 1/3rd of the way into the ride very harsh. (For
comparison’s sake, let's call the max vertical measured acceleration - ie the
biggest bump - '10'.)
JET did very well indeed. It was a very heavy trike
(so had a high suspended mass) and used polyurethane bushes in every suspension
pivot. Steel springs were fitted (see series starting at
Building a Human-Powered Vehicle, Part 1). This design both absorbed
vibration and also bigger bumps. (On the same scale, the biggest bump is now a
The current Air 150
(airbag springs, ball bearing pivot points) does even better than JET on bigger
bumps. (Note that while the absolute voltages are different, the scale remains
the same so direct comparisons are still possible.) However, you can see that
it's not as good as JET on vibration (there are lots more jiggles in the line)
and medium size bumps can still get through. However, on the same comparative
scale, the biggest bump has now dropped to 4. (The huge bumps at each end of the
trace are when I get on/off the trike and also my driveway drop-off, over which
I varied in the speed on the different tests.)
The final trace with the
Air 150 at the 30 psi tyre pressures I use shows that the 'vibration' type of
bumps nearly all gone. The biggest bump achieved in the actual ride is only 2.2
(so over the GTR, the maximum bump acceleration has been reduced by a measured
78 per cent!) and you can see it better approaches JET in small bumps and
vibration, while being much better in big bumps.
The reason for the improvement with the lower tyre
pressures might be two-fold. Obviously the lower air pressure in the tyres means
the tyre vertically deflects more over bumps – it’s a softer spring. But because
damping of the front suspension is by track change, a lower air pressure in the
tyres also allows them to have more sideways wall deflection, softening the
damping on small bumps.
I think that in ride comfort the design has been
The measured reduction in the affects of large
bumps by about 80 per cent over a non-suspension trike is very much as it feels
on the road – the Air 150 has a luxurious, cushy but well damped ride that is
far less wearing on the rider that the relatively harsh, juddering ride achieved
on non-suspension machines over the same terrain. Note also that the above
data-logging is on just a bumpy bitumen road at normal speeds – the ride
improvement over a non-suspended trike on high speed bumps, and over really bad
surfaces at all speeds, is massively improved.
In fact, I would suggest that it has the best ride
of any small-wheeled human powered vehicle in the world.
following photo sequence is over a drop-off that, at its greatest height, is
about 200mm. If you think this is an inappropriate test, consider that just the
day before this sequence was photographed, I was riding on an urban cycle path
that ended abruptly and without warning in a 45 degree gutter of about the same
height. I rode straight off it.
The steering uses modified Greenspeed
“non-crossover” steering components.
By far the hardest thing to get right on a long
travel suspension design is the steering. This is primarily because it’s very
easy for bump steer to occur – this is when toe changes are associated with
suspension movement. For example, a vehicle might have toe-out on bump and
toe-in on droop.
One result of bump steer is a steering
non-linearity when cornering – for example, toe-in on bump will result in more
steering occurring on initial turn-in, making the machine darty.
Quite late in the development, I raised the outer
steering tie-rod ball-joint height. This was primarily to achieve better
clearance between the tie-rod and the frame but at the same time I also
dialled-out some bump steer that had previously been occurring. However, this
change also apparently altered the Akermann compensation and appeared to
increase the turning circle a little (I don’t understand why).
Whether it’s because the dynamic toe change is now
better controlled, or because Ackermann appears to be lessened, or some other
reason (or combination of reasons!), happily the steering is now much improved.
Specifically, it has superior road feel and better self-centre’ing. (For those
experienced on recumbent trikes, I think the negative scrub radius steering of
the Greenspeed GT3 Series II is still superior to my design, but by only a little.)
The feel of the steering is also quite dependent
on the tyre pressures being used – as you’d expect, it gets heavier and less
responsive as tyre pressures decrease.
The steering achieves the very difficult
compromise of being sufficiently light and responsive at slow speeds on cycle
paths and the like, but not too sensitive at high speed downhills. That said,
you still need a light touch at high speed – using more a pressure on the
steering handlebars rather than a definite push/pull.
as it is with cars, the suspension angles that are measurable with the vehicle
stationary and not subject to dynamic loads may well not be the suspension
angles being achieved in hard use.
the bench, the steering and suspension of the Air 150 have been very carefully
set up to give zero bump steer. But on the road, some pics appear to show toe-in
on bump – but then when you have major changes in camber angles with suspension
deflection, assessing toe by eye becomes rather difficult.
short, I don’t know what toe changes occur in action – and there’s no easy way
to find out.
The modified Greenspeed steering works very well.
However, if I was starting the project again, I think I’d look closely at using
negative scrub radius for the potentially better road feel and self-correction
that could result.
Many believe that, since a suspension trike has to
have greater ground clearance than a non-suspension design, the higher centre of
gravity will inevitably result in poorer cornering. However, this doesn’t take
into account the fact that the front track can also be increased.
Skidpan testing is an excellent way of determining
real-world maximum lateral acceleration - that is, how hard the machine can
constantly corner. In addition to assessing “over-turnability” (most trikes will
pick up the inside wheel before sliding), it also assesses:
- How well cornering can be sustained (it’s
usually easy to get a higher peak figure)
- Steering accuracy and response (without good
steering, it’s hard to stay tracking accurately around the circle)
- How well power can be applied (apparently, some
machines cannot be pedalled at high steering locks...!)
For these reasons, skidpan testing is much better
than simply placing the machine on a large board and then tilting it until the
trike and rider start to overturn (and then noting that angle of tilt). A tilt
test is vastly optimistic in the calculated maximum lateral acceleration. (So be
sure to find out how lateral acceleration ["g"] figures were obtained before
Of all the performance assessments that can be
made, skidpan testing is the easiest to conduct, is zero cost, can be done
nearly anywhere – and yet shows most accurately something normally hardest to
I live at the end of a quiet street at which is
located a smooth, bituminised and flat cul de sac. I use this area to lay out a
temporary skidpan testing circle.
A 6 metre diameter circle was marked out by the
simple expedient of temporarily driving a nail into place in the middle and then
stretching out a 3 metre rope, looped at one end around the nail. At the other
end I placed a piece of chalk and then stretching the rope tight, marked out the
circle. (And then removed the nail!)
With the diameter of the circle known (or actually
half its diameter – the radius) the equation to work out the centripetal
(lateral) acceleration is this:
39.48 x radius
---------------------- = Centripetal acceleration
...where radius is in metres, time is in seconds
and the answer is in metres per second per second.
So with the circle 6 metres and a time of (say)
5.8 seconds, the lateral acceleration is 3.52 metres/second/second. Divide this
by 9.81 to get the results in g’s – 0.36 g.
The tyre pressures were set at 60 psi and then
testing of the new trike was undertaken. (As noted above, 60 psi is much higher
than I normally run but it’s in the middle of the ballpark most people seem to
use on recumbent trikes.)
The results of testing are shown in the table
below. Greenspeeds GT3, GTR, GTC and X5 are all non-suspension trikes. JET
(Julian Edgar Trike) was my first suspension trike, Air 130 was the first using
airbag suspension and Air 150 is the current machine.
Greenspeed GTR (old model)
*ridden by Georgina Edgar – I’m too large to fit
safely on the machine
As can be seen from this listing, the Air 150 has
gone backwards over my previous two designs. This is because the higher seating
position gives a higher centre of gravity and so, with similar track and about
the same weight distribution as JET and the Air 130, the trike tips more easily.
However, it’s still as good as the Greenspeed GT3 and only a bee’s dick behind the
GTR and GTC.
But what if the suspension is lowered? (On a
smooth surface like a skidpan, you don’t need bump absorption.) I lowered the
airbag pressures (a 10 second job), so dropping the height of the machine. Note
that even at this ride height there was still about 1.5 inches (38mm) of bump
travel – in most people’s views, more than enough! Maximum measured skidpan
figure then rose to 0.39 – the same as I have measured with the Greenspeed X5.
There’s no reason that on smooth surfaces the Air
150 couldn’t be run at this ride height all the time, but I am certainly not
going to do that – so the measured number stays at 0.36.
While smooth skidpan testing is very effective, it
clearly does not tell the whole story. For example, yaw response (how quickly a
machine can change direction) will vary between designs, as of course will bumpy
I don’t have any way to measure yaw response but
subjectively I’d suggest that the lighter, smaller designs (like say a
Greenspeed X5) can change direction more rapidly than the Air 150. In comparison
the Air 150 feels ponderous when flicking in and out of say a line of witches
On bumpy corners the Air 150 suspension works very
well. On bumpy corners taken very quickly, the inside wheel will lighten
(sometimes to the point of being off the ground!) and then the outer suspension
airbag can be seen working all around the corner. There’s an important point to
note here. Because of the semi-leading front suspension design’s high roll
centre, even when the trike is on two wheels, the front outer suspension does not unduly
compress. That is, suspension jacking helps offset the outer wheel’s suspension
compression. In effect, this stiffens the suspension when it’s at maximum roll.
However, corner fast enough on really bad surfaces
and the rear suspension can develop a hop. Perhaps better damping could dial
this out but since it occurs only in extreme conditions, I figure it’s a good
warning that things are about to let go.
Pressures and Skidpan Testing
skidpan test of the Air 150 was run at the tyre pressures I use – 30 psi. In the
recumbent triking world this is seen as a very low tyre pressure but I like it
for two reasons – firstly, it improves the ride over high-frequency,
low-amplitude bumps (like coarse surface bitumen), and secondly, I think the
steering is better. The trade-off is increased rolling resistance.
after the skidpan test, I have another reason for the low pressures – the
trike’s grip level is improved.
noted above, a consistent 0.36g was obtained with 30 psi pressures. At 60 psi
pressure, this dropped to 0.32g – again, with consistency.
think that there are two reasons for the improvement in lateral acceleration
with lower tyre pressures. I would guess that the round cross-section tyres put
more rubber on the road (ie the tyre better ‘keys’ into the surface), and from
riding the machine with the two different pressures, I can say that it’s easier
to go faster with the lower pressures. This is because when the trike is on the
edge of lifting the inside wheel, the tyre deflection makes the steering less
nervous when you’re trying to balance the trike at that cornering force.
doubt anyone would have guessed that result – another reason to do careful,
I am quite sure that a low-slung dedicated racing
recumbent trike could exceed all the lateral acceleration figures listed above,
but my aim was to match on smooth surfaces the cornering prowess of general
purpose commercial trikes – and that’s mostly been achieved. And, as you’d
expect, when cornering on bumpy surfaces, the Air 150 is far better than the
non-suspension trikes – the tyres stay in touch with the ground rather than
skipping across it. (I'd also suggest that on wet bumpy surfaces my design would be better than all available general purpose trikes.)
Angle of Roll
skidpan is also a good place to measure the maximum roll that occurs. The Air
150 uses interlinked front airbags (softening bumps but also reducing roll
resistance) with the roll stiffness provided by the large sway bar. However, its
affect is diluted by it being connected to the suspension arms well inboard of
what is the maximum angle of roll (before a wheel lifts off, anyway!)? As this
pic shows, with a line drawn through the axles and another across the top of the
rear vision mirrors, the max roll measures about 4 degrees (not including tyre
sequence of photos used earlier in this article of the trike being ridden over a
sharp drop is just one of a number that were taken. The pic shown here, taken
from another sequence, very clearly shows the front suspension travel that is
being used in this situation.
red arrow points to one of the front airbags. This shows that near to full
travel is being used.
purple arrow points to the chain on the slack side – the downwards acceleration
that is being experienced can be seen in the chain being left behind (ie
appearing to fly upwards).
So what are the good and bad points of the design?
My summary (with ratings out of 10) is this:
Ride quality - outstanding over large bumps,
good over small bumps, not floaty, little pedal squat, improves with large loads, some rear suspension extension under brakes, a small front/rear
symmetrical bob at high pedal cadence – 9/10
Handling - predictable, precise, not upset
by bumps, adequate although not outstanding maximum lateral acceleration, very
predictable when on two wheels, not upset by large loads – 7.5/10
Steering - precise but not nervous, no bump
steer, good at both high and low speeds, can pedal-steer a bit at high speed in
top gears when pedalling hard – 8/10
Brakes - excellent fade resistance,
progressive and easily modulated, twin independent discs require even pressure,
can lift rear wheel under hard braking – 7/10
Frame - very little boom bending, small
amount boom twist, able to carry large loads, low weight for size of
machine but heavy in absolute terms - 8/10
Clearly, from this scorecard I am pretty happy
with the end result. For those who think the Air 150 too large (and so
overweight), something like perhaps 15 per cent of the mass could be taken out
by the use of smaller 16 inch wheels and a proportionally smaller width and
length. If the rider was prepared to put up with boom flex typical of some
commercial trikes, a little more weight again could be shaved off.
However (I guess as is self-evident since I made
it to suit me and no one else!), I love its size and the resulting roominess. I
think that ride/handling compromises can always be improved: there’s not a ride
I go for on the machine where I don’t think: “Hmm, steering feel was a bit
bereft then” or “Impact harshness over that bump too high” – but there’s also
barely a ride where I don’t think: “Hell, the trike handled that bump well” or
“Gee, the frame is stiff – I am climbing my 40 per cent gradient
driveway and I’m not even in bottom gear!”
But as the negatives above show, I certainly
don’t think the machine is perfect...
specifications are approximate and where appropriate, with trike loaded with
rider and with the carrier fitted
circle: 5.5 metres
base trim): 23kg
diameter: 20 inch
wheel Magura Big hydraulic discs, separate hand levers
Suspension: 30 degree semi-leading arms, 25cm high roll centre, anti-dive
geometry, interconnected Firestone airbag springs, 32 x 0.9mm anti-roll bar,
damping via track change
Suspension: longitudinal trailing arm, Firestone airbag spring, custom valved
Yamaha R1 motorcycle steering damper, active anti-squat chain pull line
and anti-roll bar pivot points: 30 x 9 x 10mm sealed ball bearings, 10mm high
tensile Allen-key through bolts
Greenspeed non-crossover, underseat steering
Greenspeed zero scrub radius, bronze bushes
loaded camber: negative 5 degrees
dynamic camber: plus 8 degrees to minus 12 degrees (plus 3 to minus 12
loaded castor: 7 degrees
castor: 6 – 8 degrees
maximum suspension travel front: 170mm (130mm usually used)
suspension travel rear: 130mm
change with full suspension travel: 100mm
deflection front: 50mm
deflection rear: 50mm
front and rear natural frequencies: 2.2Hz
steer (over full suspension travel): too low to easily measure!