This article was first published in AutoSpeed.
In the last article in this series I’d only just ridden my new HPV, a pedal
powered trike, for its first tentative test ride. Before riding it I’d been
concerned (and that’s an understatement!) that the radically changing camber and
castor - and the high roll centre - of the semi-leading arms front suspension
would cause massive problems in steering and road-holding. But as I’d hoped,
that proved not to be the case.
Instead, I’d found these aspects to be problematic:
1 high speed steering twitchiness, which was so bad the HPV threatened to leave
2 brake attitude change, where under braking the rear suspension quickly went
to full extension
3 rear damping, where the approach of linking the airbag to a separate
reservoir through a restrictor wasn’t all that effective
But maybe I wrote that article to soon, in a first flush of enthusiasm. And
why do I say that? Well, a further week of hard testing, including high speed
descents around long bumpy corners, found some more issues....
But first, solving last week’s problems.
High Speed Steering Twitchiness
I’d designed the front semi-leading arm suspension to give zero toe-in under
normal bumps, going at the extreme to a few millimetres of toe-in near full
bump. That seemed to me to be a reasonable design goal – neutral moving through
to understeer-reducing toe-in of the outer wheel under maximum roll.
However, when I told a few fellow HPV building enthusiasts about the problem
I was having with the twitchy high speed steering, one of them, Ian Sims (the
founder of Greenspeed recumbent trikes), suggested that in fact the cause of
this twitchiness might be the bump toe-in. His idea was this: toe-in with roll
will immediately increase the steering angle, in turn increasing roll, which in
turn increases steering angle... And this felt pretty well on the money: on
turn-in, the trike just lunged in the chosen direction!
Ian pointed out that changing the steering so that toe-out occurred
under bump may give better results.
I shifted the inner steering arm rose joints downwards by about 8mm
(something easy to do – just bolt them on under the steering arm bracket rather
than having them mounted with spacers on top of the bracket) and then set the
static toe to zero. The latter is very important – if you have toe that’s
varying with bump and rebound, by definition toe will vary with ride height. And
when you have airbag springs, ride height depends on load and air inflation
pressure... So setting toe to zero requires not only a designated ride height, but
also the correct airbag pressures and HPV loads to attain this height.
And with the inner steering rose joints relocated to give toe-out on bump,
the high speed steering was vastly more settled. Admittedly, slow-speed turn-in
was no longer as crisp, but since at slower speeds there is less body roll (and
so the outer wheel compresses less and therefore has less dynamic toe-out), the
change in low speed turn-in wasn’t enormous.
The toe-out that occurs in straightline bumps has no discernible affect on
Brake Attitude Change
How much the rear suspended extended on braking depended a lot on how stiff
the airbag spring was. If the rear airbag spring was connected to a separate
pressure canister, the effective spring rate was much reduced – there was more
air available to be compressed and extended, and so the spring was softer. As a
result, the spring extended more easily under braking.
I’d intended damping the rear air bag spring by using a restriction in the
tube that connected the airbag to the canister, but found that the damping
provided in this way wasn’t very strong. Therefore, the original reason for
having the separate canister was no longer valid, so I removed the canister,
which in turn made the rear spring stiffer.
By removing the canister and using just the air volume inside the airbag, the
extension of the rear spring under brakes was much decreased. Damping was
slightly reduced but taking into account the reduced braking extension, the
overall result was much better.
In Part 4
of this series I showed how if the front airbags were linked and a variable
opening valve was placed in the connecting tube, the airbags could easily be
rider-adjusted for spring rate. Furthermore, by using a valve that disconnected
the airbags when the steering was turned, the airbags could be automatically
stiffened in roll when cornering.
However, a bit like the rear airbag damping, the results on the road were not
nearly as impressive as on the bench.
In short, when the airbags were connected, their spring rate was simply too
soft. This was noticeable in any throwing around of the trike: acceleration,
braking, turning. Pneumatically separating the airbags reduced lurch, roll and
pitch – without much worsening the ride. Yes, the ride was firmer with the
airbags not interconnected but the overall dynamic result was better.
This was all made startlingly clear by testing with a multi-turn needle valve
in the front airbag interconnection. I rode around, adjusting the valve until
best results were obtained. Because of its many turns, I had no idea where I was
in the adjustment range (whether the valve was near fully closed or near fully
open, for example), but by then removing it and checking its flow, I found that
the best dynamic results were obtained with the valve effectively shut.
That’s not to throw the baby out with the bath water: I think potentially
very good results could still be gained by using external reservoirs and
variable airbag interconnections. (For example, lightweight electronic control
of solenoids could make things very interesting.) However, since at this stage
of development, an exceptional ride quality is achievable without the extra
weight and complexity of interconnection and separate reservoirs, it makes sense
to pursue that course first.
So with the steering twitchiness quelled, and the extension under brakes much
reduced, I went for longer and harder rides. And, as mentioned at the end of
last week’s article, the main frame tube then started to fail. Specifically, the
front spring support arms bent upwards. However, heating and straightening the
main tube, and then adding a small cross-brace, solved this problem with only a
little gain in weight and a lot of increase in strength.
And with the frame strengthened, and the steering now working well, I decided
to try the trike down my local test hill. I don’t have a gradient figure, but
this hill is clearly steeper than other local hills marked at 15 and 16 per cent
– it’s sufficiently steep that you need to be pretty damn’ fit to walk up it
without stopping a few times. In addition to its steepness (with some pedalling,
you can do 75 km/h down it!), it has some absolutely wicked bumps (both large
and small) and includes a long gentle bend.
The non-suspended Greenspeed trikes are extremely tricky down it: my GTR
model launches into the air over the big bump and comes back down to earth with
a shattering bang. JET, my first suspension trike, was extremely effective down
the hill, especially in its steering but also in its ability to absorb the bumps
without throwing the trike off line or discomforting the rider. (But it was just
so bloody heavy to ride back up the hill!)
So what would the new machine be like down this most challenging of
blacktops? There was no way I was going to try it at max speed, but (and without
a speedo on the trike I am guessing) I’d probably have been doing 60 km/h.
At the top of the hill I gave the machine a few encouraging pedal strokes and
then stopped pedalling and concentrated on steering and suspension feel. The
steering, while extremely sensitive (as you’d expect from a system with only a
small movement from full lock to lock) was settled, while the suspension was
almost contemptuous in its dismissal of bumps – the big test bump, especially,
was just a very rapidly passing dull thump.
But oh shit, what’s happening at the back?!
As the trike entered the sharper part of the long bend, the back tyre started
to patter. And it didn’t just patter in straight line: it started
to patter sideways.
Now when you have a wheelbase of only 108cm, the rear doesn’t need to get
sideways very much before you start to get very excited indeed. The pattering
didn’t go on long enough for any steering correction to be needed, but it was
more than enough to concern me. And concern me a lot. At higher speed, I had the
feeling that the trike could get very rapidly out of shape – perhaps enough to
So what could cause the rear wheel to adopt this behaviour?
At first I thought again of damping. Apart from internal damping of the
airbag, the rear suspension in this iteration had no damping at all. But by
definition, the frequency at which the rear suspension would get most excited is
its natural (or resonant) frequency. In the case of the rear suspension, that’s
less than 2Hz. But this pattering was at a much high frequency – perhaps 10Hz.
The excitation frequency (10Hz) is sufficiently far from the natural frequency
(under 2Hz) that the lack of damping would have little or nothing to do with the
patter. (To put this another way, if the rear suspension was wallowing up and
down a few times a second, and the trike was going sideways on each wallow, lack
of damping would be the obvious culprit.)
So what about the natural frequency of the tyre then? - it’s much higher than
the natural frequency of the spring. The tyre’s natural frequency is dependent
on the pressure (and so static deflection), but even a quick calculation shows
it more likely to be hundreds of Hertz – so again, a long way from the
OK then - what about the torsional stiffness of the rear suspension? I’d
tried very hard to make the rear suspension stiff in torsion (see
Another Human Powered Vehicle! Part 10 - Rear Suspension) so that the side loadings at the tread wouldn’t twist the
suspension, but if this member wasn’t up to scratch, it would explain the
behaviour. How? Well, the wheel would load up sideways, the suspension arm
twist, the wheel unload over a depression, and the wound-up suspension arm
rapidly spring back to its original orientation. As a result, the rear wheel
would skip. Yes, that could explain it...
Next week: revisiting the rear suspension design
Did you enjoy this article?
Please consider supporting AutoSpeed with a small contribution. More Info...