In Part 8 of this series I described the development of a unique pneumatic / hydraulic suspension system, being used on a recumbent pedal trike.
By filling the existing Firestone airbags with water, their function altered from being air-springs to being hydraulic displacers – and by connecting these displacers to plastic bottles and then pressurising the system, damping and springing could be integrated into the one system. Regulation of the water flow through connecting hoses allowed damping force to be altered to suit the conditions.
That’s the system in a nutshell – but how well did it work, and what further development was needed?
Different Air Volumes
By using different sized carbonated soft drink bottles, the volume of air compressed during suspension movements could be easily altered. I experimented with bottles ranging in volume from 300ml to 2 litres. However, when using the big bottles there were some unexpected results.
My initial desire was to use a single large volume with the front suspension, with both front displacers plumbed to it. This would interconnect the front springs and also reduce their natural frequency over using the air-springs conventionally. (That’s because a much larger volume of air was being compressed, so resulting in a lower spring rate.) Taking this approach should reduce pitch accelerations (see the previous part in this series for more on reducing pitch.)
However, in practice the ‘springiness’ of 2 litres of trapped air proved to be too low.
With the 2 litre bottle, the correct ride height could be established by using a suitable air pressure in the system, but the suspension movements over bumps were governed far more by damping than by the springiness of the trapped air. This gave quite weird outcomes:
any weight change (eg even by the rider just leaning forward) caused an alteration in ride height
setting the pressure to give the correct ride height was a finicky procedure, with even small differences in pressure making large differences in ride height
front ride quality felt odd – perhaps because the recovery from a bump or rebound motion was so slow, or perhaps because suspension resistance was based largely on damping forces that depend on the speed of suspension movement, rather than the distance the wheels moved up and down
At the back, the suspension (that uses a much higher leverage ratio) also didn’t work particularly well with a large air volume. Weight change caused the same type of ride height variation experienced in the front suspension, but the recovery to the static ride height position was even slower. To stop bottoming-out over bumps, the damping had to be set stiffly – which in turn degraded ride quality.
So while on paper using large volumes looked like it could be really effective, the on-road results didn’t support this.
In fact, after trialling volumes of 2 litres, 1.5 litres, 1 litre and 600ml, what I found worked best was a single 300ml volume for the rear spring and individually plumbed 600ml volumes for the front displacers. (Incidentally, the small 300ml Coke bottles are available in bulk packs from supermarkets. These bottles, that appear to made from the same gauge plastic as larger bottles, are also very strong.)
Chalky is just a prototype – a work in progress. That explains the lack of proper powder coating, the bits that have been added and then removed – and so on. The final machine will be much better finished.
The initial plumbing set-up was made with garden hose attached to brass and plastic fittings with hose clamps. I used garden irrigation style plastic stop-cocks to regulate fluid flow and so alter damping. However, it proved very difficult to eradicate leaks, so when the overall pneumatic / hydraulic approach showed that it was likely to be effective, I swapped to better quality plumbing.
To regulate fluid flows I now use stainless steel ball valves. Hoses are braided stainless. These hoses, complete with fittings (including space-saving right-angled bends), are available at any hardware store – they’re used to connect domestic taps to plumbing.
I silver-soldered together some brass fittings that:
incorporate a female thread to take a fitting for a Shrader valve (ie a normal tyre valve), allowing the system to be pressurised
connect to the fittings used on the braided water hoses
adapt to a smaller brass fitting suitable for connecting to the pressure bottle cap
The caps of the bottles were replaced with heavier duty plastic caps with plenty of ‘meat’ to take threaded fittings. (I bought these caps on eBay a few years ago.) Inside the cap I use an O-ring to seal against the lip of the bottle; this is greased with petroleum jelly (Vaseline) – before they were greased, there were occasional leaks.
This plumbing is a lot heavier than the plastic hoses and fittings, but is vastly more durable and doesn’t leak.
Set-Up and Results
A lot of testing was undertaken, using different loads and on surfaces including smooth and rough bitumen, dirt, gravel, grass and concrete. Various suspension air pressures, tyre pressures and damping rates were trialled.
Back-to-back testing over the same 20 metre stretch of bumpy grass and dirt was carried out to test different suspension set-ups, and also to allow comparison to other pedal machines. An iPhone using ‘Vibration’ software was used to log and quantify peak and RMS accelerations in X, Y and Z axes, with these accelerations measured at the rider’s head.
The test results show that the trike rides better than a 28 inch wheel conventional commuter bike equipped with front suspension forks and seat post suspension (a Gazelle Medeo). It also rides vastly better than a Brompton folding bike equipped with 16 inch wheels (the same size as Chalky the trike), fitted with an aftermarket sprung seat and with standard small-travel rear suspension.
This pic shows measured accelerations in all planes for the three pedal-powered machines. (X = roll accelerations, Y = pitch and Z = vertical. Note especially the calculated RMS values. Click on the pic to enlarge it.)
With tyre pressures at about 40 psi, front suspension systems at 20 psi, rear system at 30 psi and the damper valves fully open, the trike rides very well on bitumen. (This is at a load comprising an 88kg rider plus a 3kg rear Tailrider bag.) The suspension is much more firmly damped that with previous airbag / hydraulic damper versions of the suspension, reducing movement with pedalling* and giving a firmer, more responsive on-road feel. For example, roll linearity is greater than previously – a very important characteristic in terms of giving the rider confidence in cornering.
*Movement of the machine with pedalling is not the same as occurs on a conventional suspended bicycle, where it is caused by the forces of the rider pushing down on the pedals. Instead, on the recumbent suspension trike, it is caused by the out-of-balance forces occurring as the mass of the legs and feet whizzes around.
Because the displacers are made of rubber, initial movement of the suspension to a bump is expansion or contraction of the rubber, rather than displacement of water. This reduces damping for small amplitude bumps (eg 5-10mm) and also provides an initially softer effective spring rate. Over larger bumps the displacers pump water into, and out of, the plastic bottles. This pumping of water can be clearly heard - a big bump is accompanied by an audible squirt. Set up this way, the trike can be ridden straight off 200mm high gutters without discomfort or damage.
On dirt the suspension set-up remains the same except lower tyres pressures (30 psi) are used. Despite using these low tyre pressures, ‘pinch flats’ (where the tube is holed by being pinched between the rim and the tyre) have not occurred.
For higher touring loads (eg additions of 6kg at the front and 30kg at the back) tyre pressures are raised by about 10 psi and the suspension system pressures are increased to retain the same ride height (a static deflection front and back of about 50mm). In these conditions about 25 psi is used in the front suspension units and 50 psi in the rear. Damping is then adjusted for more firmness by slightly closing the damper valves – as you’d expect from the much greater increase in rear load, it is the rear damper valve that needs the most adjustment. When carrying a touring load of this mass, the ride quality is exceptional, with rear bumps almost impossible to feel. In fact, you can ride the rear wheel through a 15cm hole and feel less discomfort than you’d experience riding an unsuspended trike over a matchbox toy car.
However, the suspension system does have a few failings.
On repetitive, small amplitude bumps (eg a dirt road with 10mm high, closely-spaced rocks poking through the surface) the rear tyre can skip around. Changing the rear damping valve makes no difference to this behaviour; it’s a combination of the unsprung rear wheel mass, the wheel’s small 16 inch diameter, the spring characteristics of the tyre and the initial lack of damping in the displacer. Reducing tyre pressures improves but does not get rid of this behaviour. This problem does not occur with a full touring load – it’s only when the trike is carrying a minimum load.
The other shortcoming is on large amplitude, long wavelength dirt-road corrugations, as pictured with a helmet included for scale. On small corrugations the suspension copes well, but on these high, long wavelength corrugations (eg 40mm in amplitude and 55cm in wavelength), speed must be substantially reduced or the ride quality is poor. Again, this characteristic is worse when the trike is carrying a minimum load.
Incidentally, for those who have been following my suspension trike development over a long time, the test regime I am putting Chalky through is far more severe than I did with my previous designs. It may well be that the previous designs would also have had problems on large corrugations and short-spaced repetitive rocks, but as I never tested them on difficult dirt roads and trails, I never found out!
So how much does each suspension unit weigh? The short answer is much more than a dedicated mountain bike spring/shock, and a little more than achieved with the air bag spring working with the separate hydraulic damper.
Specifically, measured per suspension unit:
...giving a total of 950g. This compares with 800g achieved with the air spring and separate hydraulic damper. (Note that the weight of the valve/hose/fittings/bottle could be substantially reduced if I went back to plastic hose and fittings.) As a comparison, a good quality integrated mountain bike spring/shock has a mass of about 500g.
One reason for constructing Chalky was to reduce weight over previous designs, especially when disassembled to allow long distance transport of the trike.
The new suspension system is easily drained of water, and the displacers can be removed in moments for independent handling, eg in carry-on luggage. The weight of the trike package (normally handled by freight carriers as a ‘bike’) is therefore still lower than previously achieved.
This may seem a semantic construct, but reducing weight of the ‘transport’ package is critical, and if 3kg can be easily removed and carried in your pockets (or backpack!), this is very useful.
The unique pneumatic / hydraulic system has major advantages over the previous design where airbag springs and hydraulic dampers were used. In my application it also has benefits over mountain bike spring/dampers. These advantages can be divided into three categories: design, set-up and operation.
No sliding seals
Everything (but for the replacement bottle caps) easily available off the shelf
Ends of the displacer do not need to be kept in alignment (eg by bearings as used in a conventional mountain bike spring/damper)
Long travel suspension units reduce required motion ratio of suspension, giving vastly lower frame forces
External valving can be used to regulate damping flows – if required, these could include one-way valves to give different bump/rebound damping rates
Working fluids are universally available (and benign) air and water – water can have anti-freeze added as required
By altering air bottle size, spring rates able to be easily trialled over a very wide range
Ride height variations able to be trialled
Different damping rates able to be trialled
By altering water level, different combinations of displacement versus damping able to be trialled
Ride height able to be easily adjusted for variations in load variations, both fore-aft and also side-to-side
Damping able to be easily adjusted on-the-go, individually for each wheel
Higher roll linearity than with air-only system
Reduction in unwanted vehicle movement with pedalling
Excellent ride quality unladen and exceptional ride quality when carrying touring load
Most parts able to be sourced widely, allowing easy replacement
Overall, the pneumatic / hydraulic suspension system gives the best compromise I have so far achieved when considering durability, ride comfort and handling.
Two further changes have occurred since we last visited Chalky.
The first is the use of a short cage rear derailleur, as the previous derailleur could easily contact the ground, especially in mud or deep gravel. In fact, in the derailleur cried enough and broke off. As can be seen in this pic, the shorter cage derailleur can still get near to being submerged (look at the position of the lower chain), but the gravel needs to be very deep indeed.
The use of a short cage derailleur (that takes up less chain slack) means that the chain needs to be moved from the largest front cog prior to the rear derailleur being changed to a low gear. When on either of the two other front cogs, no special care is needed.
The other change has been to the front suspension – and it was undertaken for just the same reason... to get greater ground clearance.
The original Watts Link, that prevents the front axle from moving sideways, could drag on the ground when the trike was traversing really deep single-wheel potholes, and the rose joints were also subjected to lots of dirt and mud splashes.
I replaced it with this design that locates all the links above the level of the axle. This increased ground clearance at the front from about 50mm to 120mm.
Note that as with the previous design linkage, I drew heavily on ‘Mechanisms in Modern Engineering Design’, the five-volume series authored by Prof Ivan Artobolevsky that contains literally thousands of descriptions of different mechanisms.
Touring with a Tag-Along
After the suspension had started working well, I brazed a fitting onto the carrier that allowed the attachment of a tag-along. This let my 5 year old son Alexander ride with me, something he loves doing. We attach a Burley Nomad trailer behind the tag-along, making up a vehicle we call the Roadtrain. At the time of writing, I take him to school each day on the Roadtrain, something that attracts a lot of attention from the other school students!
But as the pic shows, we’ve also been touring on the Roadtrain. Loaded up with two Arkel RT60 panniers, two RT40 panniers and a Tailrider, and accompanied by my wife Georgina riding another bike, we headed into the mountains of the Great Dividing Range near Hoskinstown in New South Wales. I’d like to say we spent a week away, but actually we went only for the weekend (all the time that was available). About half the roads were dirt and included some rough parts.
So how did the trike suspension cope?
Firstly, the rear suspension – with a vertical load of more than 60kg working through it – was magnificent. I inflated the rear pressure bottle to a static 60 psi and shut down the damper valve a little. This gave a ride that was quite amazingly good – I wouldn’t even feel the bump when the trike’s rear wheel met it; I only knew there'd been a bump when I felt the tug as the tag-along wheel passed it.
The front air reservoirs needed about 22 psi in them – slightly more in the right-hand one to counter an asymmetry in the pannier loads (water was in the right-hand pannier RT60 – 10kg). I varied the position of the damping valves for different road surfaces. On bumpy sections I opened the valves right up. However, on smooth sections taken faster this damper setting allowed the trike to lean – with the extra weight up high and the lowish roll centre, the trike could then feel tippy-toed – best characterised by flopping from one side to the other. So on smooth surfaces I positioned the damper adjustment so fluid flow was only just occurring, which stiffened roll damping.
Alexander stood on the pedals, knees flexed, for any large bumps I warned him about, and the tag-along’s large 20 inch tyre was set at a very low in pressure. The Nomad trailer’s tyres were also set low in pressure, but on bigger bumps it hopped around a bit. The trailer carried only sleeping gear so it was fairly light.
On this route I’d score the suspension at probably 7.5/10. Next time, with a similar high-mounted load, I’ll reduce the volume of the front spring pressure bottles from 600ml to 300ml (the same as I run on the back); this will give a faster rising front spring rate and so better control of body roll.