Bubble and Squeak
by Miles Kingsbury
This is now a finished account of my Trike build that I raced throughout the 2009 season.
I decided that a fully faired trike was the way I wanted to go as I want to go fast but have not got the balance or bravery needed for a bike.
Another consideration in my design choice was that our company helps out a local secondary school with building and racing 4 wheel electric cars in the Greenpower series of races around the country. We had some success in 2003 and became national champions in 2004. We achieved this largely by employing the aerodynamic principles we are all familiar with in HPV racing.
Unfortunately, some other teams then went the same way and the result was that quite a few inexperienced 11 to 16 year old pupils were racing in machines capable of 50mph. Not surprisingly, the organisers had to change the rules to try and reduce speeds. This was achieved partly by the elimination of battery charging during the event and insisting on a head out cockpit rather than bubble canopies.
I have based my new design on the bubble type machine which was easy to construct and quite fast.
The new school design will look something like this CAD image. Although the two machines look quite different, they will both use 406 size wheels and can share a lot of common parts, in particular wheels, brakes, steering components and structural materials.
Solidworks is my weapon of choice for the design work. It is a powerful 3D CAD package of which I probably only use about 10% of its features. You can build 3D assemblies and animate them to allow you to turn pedals and check knee clearance or move the handlebars and see where the wheels end up. In my real job, it has eliminated the need to build prototypes as well as producing pretty pictures for quotes, handbooks etc.
Weight 18kg TBC
Frontal Area 0.575sqm
Ground Clearance 25mm
Seat Height 40mm
Cassette 9 Speed 11-34T Shimano
Cranks 150mm Custom
Tyres Schwalbe Kojak 406 x 35
The bubble is the main element of the bodywork, it looks a bit scary to produce but in fact is relatively simple to make. We have made a few now with no failures and at low cost. Our technique has worked well for us but is quite slow and I am sure there is room for improvement in the process that someone more experienced would be able to employ.
A rather large fan oven had to be constructed. We made ours by using 75 and 50mm thick Cellotex (which is aluminium backed foam insulation board), 4kW of cooker elements and a cooker thermostat.
I cannot remember where the motor and fan came from but it is very important that it is big and powerful as an even temperature in the oven is a must. At least one small window is required to see what is going on.
We used 2mm clear Spectar Copolyester for the screen which can be blown at 100”C. It only needs a very small pressure that can be achieved by any type of pump, but a large volume pump may be needed for larger bubbles. We used a compressor but a car foot pump or track pump will work.
To make the former to blow the bubble I used 12mm MDF with two stiffeners on the back. There was a 60mm wide flange around the edge to clamp down the plastic. This was made from two layers of the 12mm MDF glued together.
The former and flange were drilled through with a 6.5mm clearance drill for the M6 clamping screws. The bottom edge of the flange needs a small radius to stop it cutting into the plastic. The whole former and flange was then assembled and heated in the oven to dry it out. It was then dismantled and painted with two coats of epoxy resin to make it air tight. MDF is surprisingly porous as we found out on our first attempt.
In the centre of the former, a pipe fitting was glued in for the inflation pipe. We used 6mm copper tube with a coil to warm the incoming air and to allow movement of the tube for connection outside the oven. On the top side there was a thin aluminium sheet taped on two sides over the air inlet point to stop cold air chilling the surface of the plastic as it entered.
The plastic was cut to the outside of the flange and drilled through for the M6 screws. The sealing tape was then stuck to the former. We used vacuum bagging tape that is like Blutack on a roll, we also used it when moulding the floor of the trike. The protective backing was taken off the plastic and the whole thing assembled with the screws as shown. The spring washers taking up the slack as the plastic thins.
The whole assembly was then be put in the oven and warmed to 100”C. A little bit of inflation at this point stops the plastic touching the former and being marked. We left it at this temperature for 2 hours to soak before beginning inflation. Our technique for inflation was to blow the bubble up until it stopped wobbling and then stop until it started wobbling again, continuing this process until the required shape was reached. There is a window in the oven and a couple of marks to sight across to establish the finishing height. The oven was then be cooled, keeping an eye on the shape and inflating as necessary as the bubble shrinks back a lot as the air inside it cools.
We have found that the nose and tail of the bubbles is always a bit flat compared with the middle; I imagine this is because there is more constraint in these areas. As long as this distortion is expected when the original layout is done it can be factored in. I am sure someone could put me right on this, answers on a postcard please.
Leaving the bubble stuck down to the former made it easier to cut the plastic flange off with a wood saw or similar. We were careful handling the bubble once the flange was removed as it becomes very floppy.
The wheels are made from Alesa alloy rims, Cellite 620 Board and a custom hub with Avid mechanical disk brake. This is a new departure for me as I have mostly used drum brakes and spokes in the past. These new fangled disk brakes seem to be catching on so why not give them a try.
The Cellite Board is a glass fibre and aluminium honeycomb which is a lot cheaper than Fiberlam at about £150 for an 8ft x 4ft board but it is not as light. The board we used comes out at a thickness of 13.6mm. The wheels and brakes will be used on the school electric car as well, so that is two 4 wheelers and a trike plus say three spares, making a total of 14 wheels!
The Cellite Board was cut into a circle on the bandsaw at an angle to give two gluing surfaces, one on the braking surface and one near the spoke holes. The faces to be glued were de-greased and roughed up and then stuck using Araldite 420. The assembly was put on a flat surface with a weight and left for 24hours.
The trike front hubs have two 12mm bearings and are designed to be used on the front and back on the electric car. The trike rear hub is a modified 9 speed Shimano Deore.
The centre hole in the Cellite disk was cut using a 32mm hole saw and the hub glued in.
I am going to use 406 x 35 Schwalbe Kojak tyres on my trike for starters as they feel very nice and flexible. Depending on how I get on we may well use them on the electric cars as well.
I have a cunning plan for the rear 9 speed block, I am going to take off the largest sprocket and fix a 34 tooth Stronglight chainring directly to the wheel. This will give me a nice low gear for Hog Hill but I am also hoping it will give me a reverse gear if I can get the chain length right.
I did consider using the same Cellite Board for the floor of the trike but it would have come out rather heavy, so instead I used up some very old carbon fibre pre-preg that I had lying in the freezer. One layer of 380g/m² for the underside and one layer of 240g/m² for the top plus extra layers where the wheel cut outs will be. The sandwich material was 6mm Nomex Honeycomb. The floor is basically flat but with an angled upstanding edge to locate the sides of the trike.
The Mould was a similar construction to the one used on the bubble but when I painted the MDF with Epoxy it reacted for some reason and bubbled up and blistered. To get round this I covered the mould with a thin sheet of aluminium.
I was going to mould in the wheel cut outs with the aluminium blocks shown in this picture but it was far to fiddly when I started laying up the pre-preg , so I took them out and cut the holes later.
The upstand was moulded with the removable flange shown. It is about 25mm tall and angled at 10°
It turned out that the aluminium sheet was too shiny or something as the release agent did not work very well and the moulding stuck to it.
The back of the mould was stiffened in the same way as the bubble mould and then covered with some felt cloth in preparation for vacuum bagging.
The floor was bagged up and cooked in the oven but the combination of aluminium and MDF distorted horribly as well as sticking (as I mentioned earlier).
It came out looking like a spoon with about a 100mm bow along the length and about 30mm across the width. This was cured by putting the floor without the mould back in the oven and weighting it down. It was then taken a few degrees above its original curing temperature and soaked for about 30mins. After that it was more or less perfect.
The vertical sides provide stiffness along the length as well as mounting points for the cranks, front wheels and seat. They were glued in on a flat surface with the same Araldite adhesive.
The seat was folded and glued using the Cellite Board. With a quick calculation or using the CAD a strip was cut and removed from one side of the board.
An MDF former held the position while it was gluing. The pulley brackets were also glued to the underside of the seat.
I decided to use Rose Joints for the steering pivots for the first time. Peter Ross was fitting them years ago on his early Trice designs. I am using them mainly to enable me to adjust the wheel camber angle to give maximum steering within the tight fitting bodywork. I have some concerns about the amount of friction in them but if this proves to be a problem, there is a ball bearing version available at about four times the cost. The geometry of the wheel mounts gives centre point steering and about 60mm of trail. The wheel mount assembly is made from four mild steel components brazed together.
I have used M6 Rose Joints for the track rod and steering linkages and this time I am using left and right hand threaded joints to make adjustments quick and easy. The whole assembly hangs off two 3mm aluminium load plates which are glued and screwed to the Cellite boards. There is a short turned and stepped tube between the two plates which should stop the board being crushed.
I decided to make what I call Z cranks for the trike because of the structural layout. We have used this style in the past on some pedal cars we used to race. I am not convinced this is the best way to go as they are more difficult to make and heavier than a more conventional design. The chainring assembly consists of a laser cut 80t sprocket with laser cut spacers and chain guides. The 150mm cranks are mild steel oval tube brazed to 12mm pedal axles and 10mm mounting spigots.The crank mounts look very light but in theory there is very little load going through them. This is because the pedals are on the same radius as the chain and therefore the larger loads should cancel out.
I was originally going to mount the rear wheel on a steel swinging arm with a rubber mount to smooth out the bumps but when I sawed up the tubes, the weight was rather alarming. As I only have 25mm of ground clearance, I cannot have any significant suspension movement, so I decided to rigidly mount the rear wheel.
I cut out the two Cellite wheel supports and glued them to the back of the seat and the floor. I machined two aluminium vertical dropouts and glued them to the supports.
There is an M4 tapped hole in each of the dropouts which will give a little ride height adjustment if needed.
When the Cellite board is cut, it leaves two razor sharp edges that would be dangerous if left untouched. I did consider gluing on some aluminium strips but found that car body filler did the job quickly and easily. I applied it using a flat spatula and then smoothed it down with a bit of steel tube sawn in half.
The bodywork was made from 1.5mm aircraft plywood with two 6mm plywood formers. The bubble flange was used to position and mark out the plywood sides and also as a template to cut out the top former.
A thin plywood strip was then glued to the inside face of the bodywork, providing a step to locate the bubble.
The mudguards were formed from 1.5mm ply and glued with corner beads and then glued into the main bodywork.
The joints and seams were then sealed with polyester resin. An aluminium tube was fitted to brace the two sides and the mirrors fitted.
I then had an amazing bit of good fortune finding a 2 for 1 offer on a suitable air intake diffuser.
A glass filled nylon inlet was turned to match the angle of the diffuser and the assembly glued into the nose. The larger end of the diffuser was fitted with a sleeve that would allow an extension to be used to direct the incoming air directly onto the screen.
The bodywork was painted with a small roller and shocking pink vinyl silk emulsion.
This was the colour we used on one of the original school electric cars which I thought made it stand out well.
PAINTING THE SCREEN
The screen was then held in place on the bodywork and an area marked out for sun shading. This was done on the outside with some thin black tape.
This shape was then transferred to the inside using masking tape and that area was painted with silver model paint applied with a small brush.
A pair of bottle cages was fitted and the seat covered with blue camping roll foam.
The numbers were then fitted on the front mudguards and the inside of the screen.
The screen was then taped into place using 25mm black tape.
The first test was done at Reading Velodrome but did not last very long. I gradually built the speed up over a few laps and at about 30mph there was a distinct burning rubber smell.
I stopped and had a good look over the machine with Slash, expecting to see one of the front wheels rubbing as the clearances are rather tight. We couldn’t see anything obvious so I went back on the track and built up the speed again but at about 32mph the rear tyre blew and I skidded sideways across the track and came to a very quick halt.
The blowout was caused by a lack of torsional stiffness in the rear wheel support which was easily cured by the addition of triangular brace on each side.
I was very pleased with the initial performance of the trike especially its stability during its sideways skid.
The first 2 hour race at Castle Combe was a bit tough as I was not very fit but I managed 4th place and averaged 28mph which was actually 30mph for the first hour and 26mph for the second.
As the season progressed, my performances improved as I got fitter and more used to driving the trike. It is a joy to drive on fast twisty circuits and I managed mainly 2nd and 3rd positions during the rest of the 2009 season and ended up 3rd overall.