Last week in Part 1 I covered the most important design criteria for a human powered vehicle (HPV). To recapitulate, these were that the machine needs to:
Be supremely comfortable in all road conditions
Have sufficient gears that a person can pedal at their chosen speed and effort all the time, irrespective of changes in gradient, wind and load
Not require balancing at low speeds
Have a very stiff fame that under normal pedalling loads, has only minute deflections
Be consistent in handling
I also added another criterion for my particular application: it needs to be able to be packaged into a compact assembly for freighting to a distant place.
So how best to address these factors in a new design?
On a road-going HPV, think that a minimum of 100mm of suspension travel is required, and for best results, 130 – 150mm is better. Note that I am not talking about riding over mountain bike terrain, jumping off dirt mounds or performing other stunts. I am talking about the required suspension travel to give a really good ride quality on a wide variety of real-world roads, including unexpected potholes and sudden drops. Use a shorter travel and you’ll need stiffer springs to avoid bottoming-out – and there goes your ride quality!
The static deflection (how much bump travel occurs with the machine loaded but not moving) needs to be 30-50 per cent of total travel. That is, with 100mm of travel, about 30 – 50mm is ‘used up’ by the normal loaded weight of the machine.
So why so much static deflection? Primarily, because you want the wheels to follow the road, both up over bumps and down into depressions. Let the tyres skip and you no longer have cornering control! On my Air 150 recumbent trike, I can watch the front wheels ‘walking’ up and down over bumps and hollows, and feel effectively no vertical accelerations transmitted to the rider. That’s very different to a suspension design where you fall into every hole and the suspension moves only on sharp, upwards bumps!
To perhaps state the obvious, suspension should be on all wheels. (Unless your weight distribution is all on one wheel
[and if it is, I suggest you design a unicycle and learn to ride it]
, you can feel the road input from every wheel. Also, as implied above, suspension is for handling as well as ride comfort.)
The suspension needs to have effectively no stiction – that is, when the machine is pressed down, the suspension shouldn’t stick before then suddenly ‘giving’. (High stiction suspensions will allow the occupant to feel every little bump. Go into any bike shop and feel the stiction in nearly all bike suspension systems!)
The damping needs to be strongly biased towards being in rebound, not bump. (High bump damping is just the same as having stiff springs. High speed, strong bump damping is just the same as having stiff springs plus stiction!)
The suspension should be easily adjustable so that the static deflection remains constant, irrespective of load. (Otherwise, bump travel will go down with increased load – or you need to use overly stiff springs that won’t compress as far with load, so giving a bad ride in unloaded conditions.)
These requirements rule out the vast majority of typical bicycle suspensions. If you also add the criterion that the suspension should use conservative motion ratios (eg the spring travel should be a least 40 per cent of the wheel travel), there are almost no commercial suspension systems that match these criteria. (These motion ratios reduce frame loads and also results in better linearity in suspension behaviour.)
Finally, in machines that do not lean into corners (in the way a bike or leaning trike can), a very strong anti-roll bar will be needed. This is because with soft, long travel suspension; a relatively high centre of gravity (there needs to be clearance for bump travel, so the ride height will be higher than in a non-suspension machine); and (on trikes) with only two wheels providing roll stiffness, body roll without an anti-roll bar will be excessive. (That’s even the case when a high roll centre suspension design is used.)
After a lot of testing and investigation, I think that the compact industrial range of Firestone rolling-lip airbag springs give by far the best combination of:
Spring characteristics - gently rising rate that steepens near full compression
Utility - the spring ends don’t have to be always parallel, and the springs are durable in rain, hail or shine
Mass - the 4001 series I use are 350 grams each
Stiction - there is none at all in the spring
Size - about as compact as the equivalent steel coil spring, but without the fatigue problems that can occur in metal springs not very carefully designed
Adjustment – the air pressure within the springs is easily altered by the use of a bike pump, allowing static deflection to be kept constant irrespective of load
However, the springs are expensive – about three times the price of having steel coil springs custom made. Damping also needs to be performed separately, which adds weight and packaging complexity to many suspension designs.
The best dampers I have found are custom modified motorcycle steering dampers - I use ones from the Yamaha R1.
This unit has an external cast passage into which valving can be inserted, allowing rebound damping to be made stiffer than bump damping.
Because the piston is a loose fit in the bore and the seals are very good, these dampers have almost no stiction. When mounted, the stiction they do have is completely undetectable. (The units have a max travel of 70mm, so in most applications, they need to be mounted at a motion ratio of about 0.6 or 0.7:1 – that is, they travel 60 – 70 per cent as far as the wheel.)
They’re also pretty cheap when purchased secondhand. However, at about 500g, they are not light.
To keep motion ratios low, suspension arms need to be long and the springs need to be mounted relatively close to the wheels.
The Firestone 4001 airbags have a 90mm travel, so if the wheel travel is to be 150mm, the motion ratio needs to be about 0.6 – that is, the spring moves 60 per cent as far as the wheel. If the suspension arm is 500mm long, the spring would then need to be placed 300mm from the pivot, or 200mm from the wheel.
If 30kg load is placed on the wheel, in this example the spring ‘sees’ 50kg. However, in a 1g bump this load on the spring doubles to 100kg – and that’s still well inside the 4001’s 180kg max rating.
A 1g Bump?
One ‘g’ means one unit of gravity – the force that tries to accelerate us towards the centre of the earth at 9.8 metres per second per second. If you are subjected to 1g vertical acceleration, your body weight doubles. If you are subject to 0.5g vertically, it’s like your body weight has risen by 50 per cent.
Using ‘g’ units when considering forces on a moving vehicle often make things quite easy to quantify.
A 1g bump doubles the effective weight that the springs hold up. Therefore, the sprung load of (say) 110kg might double to 220kg – and the suspension needs to do this without hammering into solid bump-stops.
The maximum braking you’re ever likely to get is 1g, so if the loaded machine weighs 110kg and has one front wheel on which all the weight is being placed under extreme braking, the force acting backwards at the road contact with the tyre is 110kg.
If it’s a trike, you can corner at 0.5g and the machine does not lean, a sideways force of half the total static weight is being developed by the three tyres, eg 18kg sideways at the bottom of each tyre. (So you can hold a wheel with its axle vertical and then hang an 18kg weight from the rim and watch the deflection that must actually occur at max cornering!).
A rule of thumb is to assume maxima of 1g bump, 1g braking and 0.6g cornering (0.8g in a leaning machine). All these are about 20-30 per cent higher than would occur in normal use.
For price, weight, lack of stiction, durability, range of sizes, and availability, I don’t think anything can beat sealed ball bearings being used for the suspension pivot points.
As important as the suspension is the seat.
More than any other part of the design of a traditional, diamond-framed bike, it’s the seat that is the greatest shortcoming. I mean, would you ever voluntarily sit on a seat the size of your hand…. for hours? No? Well, why do so on a bike? It’s downright silly to place such high pressures on your perineum, and not surprisingly, in many people it leads to numb penis and testicles, or labia.
It’s not only uncomfortable but, I suspect, dangerous to long-term health as well. (An awful lot of male cyclists seem to get testicular cancer…)
We’ve already decided the most uncomfortable seat in your home is on a bike – so what’s the most comfortable? A wide, supportive, reclining armchair? That certainly seems to suit a lot of people. So why not make the bike seat just like that?
The seat on my Air 150 recumbent seat is based closely on the seat design used on Greenspeed recumbent trikes. These seats consist of a shaped pair of steel tubes with a strong plastic mesh cloth stretched between them, tensioned by elastic ‘bungy’ cord laced through eyelets and tightened across the back and underside of the seat. The mesh allows air to pass through it, while the seat is shaped to provide lumbar, shoulder, back and bum support. It has an area dozens of times greater than a bike seat – and comfort about as proportionally better.
It’s also strong enough to be a structural component of the frame.
This type of seat design can be set at different angles – I prefer 35-40 degrees.
Of course, such a seat requires a recumbent riding position – that is, the pedals are in front of you, not below you. Rather luckily, such a position also makes it a lot easier to avoid pedal bounce on a machine with soft suspension (more on this in a moment).
The gears and power transmission parts can be taken directly from high quality bicycles.
As described last week, I think that 81 gears works very well – a triple front chain-ring, nine gears on the rear cluster, and a 3-speed internal hub. Specialist suppliers like Greenspeed provide all these parts, even for small (16 inch and 20 inch) wheels.
However, I think that most people way over-gear their HPVs – and for the vast majority of time, end up using gears in the bottom third of the range.
When thinking about gearing, go lower rather than higher. You can always free-wheel down hills, but when you’re loaded (or pulling a trailer) and you suddenly meet a big hill while facing into a head-wind and having tired legs, lower ratios will be very welcome.
While I have in the past used oval-shaped front cogs, I now don’t recommend them for suspension machines. They certainly make very steep hill-climbing easier, but they also produce a cyclically varying chain pull, which tends to cause suspension machines to bounce.
Cleated pedals improve power transmission and reduce effort considerably. Get mountain bike shoes and associated cleats – you can walk around in these shoes quite easily.
Balance at Low Speeds
Many HPVs use three wheels. This gives stability that a two-wheel machine (ie a bicycle) cannot match, especially at very low speed. Compared to a bicycle, a trike is also much more stable in slippery conditions. Put one wheel of a turning trike on ball-bearing gravel and you almost don’t notice it; do the same with the front wheel of a bike and you hit the road.
Trikes can be ‘tadpole’ (two wheels at the front, one at the back) or ‘delta’ designs (two wheels at the back, one at the front).
But the higher the speeds (especially when heavily laden), the less advantageous a trike configuration is. Or, perhaps more accurately, the more the disadvantages of a trike start to outweigh its advantages.
But what do I mean?
As indicated above, a trike with long-travel, soft suspension requires an anti-roll bar. Without it, body roll will be excessive and the trike will tend to lurch in quick S-bend manoeuvres. This anti-roll bar feeds large loads into the frame, which in turn means strong (and so heavier) frame attachment points, suspension pick-up points and a strong anti-roll bar itself.
In a trike the torsional (twist) loads on the frame are high, especially when the weight distribution becomes biased over the single wheel (as is easily the case when carrying lots of gear). Lateral (sideways) forces on the spoked wheels are also high.
As speeds rise, cornering forces increase – resulting in still higher anti-roll bar, frame torsion, and lateral wheel forces. The outcome is frame flex, which can result in fatigue and cracking, and may also cause odd handling and steering. Clearly this can be prevented by making the frame sufficiently stiff, but at the cost of weight.
A bicycle has none of these problems because as it leans, forces are fed straight down through the frame and wheels – lateral and torsional forces are much smaller than in a cornering trike.
So at higher speeds, having three wheels is a hindrance, while when going slowly, having three wheels is highly advantageous.
Solutions include trikes that lean, or bicycles that become trikes (or quads) at low speed, eg by the use of drop-down ‘trainer wheels’.
As indicated above, frame stiffness is important to vehicle handling and behaviour. But its greatest importance is in resisting pedal forces. Any frame deflection detracts from the power going to the drive wheel – yes, the frame may ‘spring back’ as the pedals continue to rotate, but I don’t believe this energy is recovered in anything like an efficient way.
As indicated last week, I was able to make a direct comparison of the hill-climbing behaviour of two of my HPV designs that were very similar, except in frame stiffness. The stiffer frame trike was a much better hill-climber, even though the frame stiffening added weight.
In a recumbent bike or trike, a good indication of frame stiffness can be gained by sitting on the machine, selecting the lowest gear, locking-on the brakes, and then pushing as hard as possible on the pedals. Some frames will deflect at the pedal axis by 15mm!
Frame stiffness is best addressed by:
The most important design aspect of gaining consistent handling is the steering. The steering should:
Have an appropriate ratio
Not have any stiction (caused by pivot point stickiness)
Not have any hysteresis (caused by flex in steering arms and pivot points)
Have good feel and weight (determined by amount of castor, trail and scrub radius)
Self-centre (determined primarily by amount of castor)
Not change in toe with suspension movement (ie not have any bump steer – this is the single hardest part to get right on long travel suspension machines, especially tadpole trikes)
If you are new to vehicle steering, you might look at the above list and decide it’s all mumbo-jumbo.
Unfortunately, there’s no real way of simplifying this – I can explain in more detail what each term means (eg see
Building a Human-Powered Vehicle, Part 3 and
Beginners' Guide to Front-End Angles, Part 1), but none can be omitted.
Realistically, within certain ranges, the best settings can be found only by experimentation.
When I describe the detailed build of Chalky, each variable (like castor, trail, etc) will be shown – for another example, see the specs list at the end of
Air 150 Recumbent Trike, Part 2.
Another important ingredient in handling consistency is braking. Wheel rim brakes, that alter dramatically in effectiveness depending on whether the rims are dry or wet, should be avoided. Instead, internal drum brakes or discs brakes should be used. Braking should not cause the machine to change direction.
On a non-tilting design, roll linearity is also very important in cornering feel. To put this another way, the degree of roll should be related to how hard the machine is cornering. A vehicle that initially rolls a lot, and then quickly stiffens in roll once a certain roll angle has been reached, feels terrible on the road. Roll linearity is optimised by careful attention to anti-roll bar links and mountings.
Suspension members should be rigid and should be rigidly located. I mentioned above the use of sealed ball bearings for suspension pivots; these have effectively no ‘play’ in them at all. An approach that allows some movement is to use rubber or polyurethane bushes (pictured). These absorb some road vibration but give the machine’s handling a less precise feeling. Don’t forget when considering the hardness of polyurethane or rubber bushes that loads being fed into suspension pivots can be in the order of hundreds of kilograms, especially under full brakes.
Long-travel, soft suspension is easily excited into bouncing, especially at its natural frequency. With the type of suspension systems I’ve been describing, this is likely to be at around 1.5 – 2Hz, an input often obtained by pedalling.
A recumbent, where the pedals are ahead and not below the rider, is much less excited by pedal bounce than a traditional diamond-framed bike. This is because each pedal stroke is not accompanied by vertical accelerations. However, without careful design, what will happen on a suspended recumbent is that the back of the machine will squat with every forceful pedal stroke.
To counter this, the chain pull path needs to be designed to extend the rear suspension each time a strong tension is experienced on the chain. Fine-tuning of the right chain pull path to give the best compromise for different gears and loads can be done on the road – a moveable idler wheel can be used to make the adjustment.
One advantage of building a suspension machine is that, with careful design, the machine can fold up around the pivot points integrated into the suspension. Taking this approach means that the strength is not reduced at all. If airbags are used, they can easily be either deflated (allowing the suspension to be placed in its maximum bump position) or unscrewed and stored separately (allowing for example the suspension arm to be swung around the other way).
Other design aspects that can allow the folding (or breaking-down into smaller pieces) include frame folding hinges, screw couplings (eg S&S couplings) and parts of the machine that unbolt. An example of the latter is where the seat unbolts from the main frame.
If the machine is to be foldable (or able to be broken down), this design criterion must be uppermost in the mind of the designer every step of the way!
The fact that it’s taken me well over 3000 words to outline in just broad brushstrokes the way some of design criteria can be addressed serves to illustrate how the more you get into these machines, the more you realise the complexity and difficulty in balancing cost, weight, stiffness, consistency in handling feel, and comfort.
Next week we’ll look at the front suspension design of Chalky
A Full-Size Car?
So how much of this is relevant to someone who doesn’t want to build a machine powered by pedals, but instead wants to build a ‘real’ car?
In short, most of it.
Aspects such as roll linearity, steering feel, steering self-centre’ing, scrub radius, motion ratio, spring rates, static deflection, frame stiffness – and so on – are all just as important on a larger, powered road machine.