For some time now I have been mulling over the idea of building a recumbent. Not having any metalworking experience I was quite unsure as to how much work would be involved, both in gaining enough metalworking skill to be able to complete the job, as well as the part of actually building the bike. I decided to start with an interim project, that of building a tandem bicycle out of two upright bicycles. I reasoned that this would require some metalwork, but not too much and at the same time it would get me thinking about some of the details one needs to take into account in order to build a decent working bike. If this project worked, then I figured it would prepare me for the somewhat larger task of building a recumbent.
I had read about DIY tandems before on Sheldon Brown's website, and it seemed that quite a number of people had successfully built tandems by combining two uprights. Indeed that website contains a great deal of information for the would-be builder.
For the joining of the frames I considered the cost of purchasing the equipment required for various techniques including welding and brazing. Research on the Net suggested numerous advantages to brazing, including less likelihood of warping the base metal (that is, the frame) as well as being easier to produce quality continuous joints. I investigated the torches used for brazing, and ended up with a Bernzomatic unit that can run on either propane (blue cylinder) or MAPP gas (LPG methylacetylene-propadiene mix, yellow cylinder). People on the Net had noted that MAPP gas is hotter than propane. Guidelines from users of such units suggested that for tubes the size of bicycle frame tubes propane would not be hot enough so I bought a kit including a MAPP gas cylinder.
I started looking around for suitable donor bikes to build the tandem. The criteria for the frames were that in order to be able to braze them they had to be made of steel and not aluminium. They also had to be the right size for the riders. I'm 1.8m tall and planned to be the captain, while Hanna, my partner and stoker is 1.6m. I ended up with two mountain bike frames. These are stiffer than road bike frames, and of course are made for 26 inch mountain bike wheels, which are nice and strong. Ideally I wanted 4130 steel frames, but the lack of identifying labels on the frames I ended up using suggested they might just be high tensile steel, not as stiff. With a limited selection of old bike frames at hand I decided to compromise and went with these.
I'm happy that the rear frame has dropout spacing of 135mm, the standard for modern mountain bikes. This made finding a suitable donor wheel a little simpler. The rear frame however was designed for use with an 'adapter claw' derailleur and therefore has no mounting point for a screw-on derailleur, as most higher quality derailleurs are. Using a budget adaptor claw derailleur is probably not such a big limitation, and indeed the Shimano Tourney unit I have ended up with works surprisingly well. I have also since then obtained from Harris Cyclery in the US a screw-on derailleur to adapter claw conversion mount in case I later wish to upgrade the derailleur. Brazing on a mount would also be an option.
I planned to cut off the front of the rear frame's head tube using a hack saw, and join that to the seat tube of the front frame. This did not work in practice because the seat tube on the front frame was double-butted, but on the outside - that is to say the inside diameter of the tube remains the same while the outside measurement has a step. There was no way the rear frame's head tube would fit nicely against it. Instead I ended up griding off (with a Bunnings Warehouse $20 angle grinder) much of the rear frame's head tube, in effect using the remaining metal of the head tube as a flange. The key here was to maximise the area of metal to metal contact in the join so I could get in a lot of braze material in order to make a strong joint.
Rear frame top and down tubes connect to front frame seat tube
640x853 image and
1728x2304 image also available
Joining two frames in this manner can limit the amount of room the stoker has to avoid knee handlebar interference, as well as for keeping their nose out of the captain's back. Being too cramped is bad. To compensate, the rear frame should have as long a top tube as possible. I ended up with a 46cm frame with a 56cm top tube. In my limited experience this is quite a long top tube for the frame size. At 1.6m tall Hanna can safely straddle the frame with minimal clearance, and yet get the benefit of the long top tube. I suspect that finding a suitable frame for a much taller stoker would be difficult.
During the design stage we took numerous photos of each other sitting on various bike frames. I then used a photo editor program to splice photos of Hanna sitting on her frame to photos of me sitting on my frame. This let us see quite clearly how much space Hanna would have on the completed bike, before we actually built it. Note that when using this technique, you must make sure both photos are at the same scale before splicing.
I looked at a number of options for ways to connect the two bottom brackets together to complete the triangulation of the frame. In the interest of aesthetics I decided not to re-use the chain-stays from the front frame as is commonly done, but instead to mitre and braze in a new tube that I would obtain from somewhere; unfortunately, needing to be quite long at around 58cm in length I did not have a donor bike with any tubes long enough. By far the easiest solution to sourcing the tube was to go to my muffler shop (Safe-R-Exhausts). The chap gave me a piece of tube for free. 1.6mm thick, 1.25 inch OD that would match the other frame tubes aesthetically. I decided to use regular exhaust pipe because at 1.6 mm it has a very thick wall, which means I would be able to get a reasonable amount of braze material in the joint. For the bottom bracket connecting tube I did not have any lugs or flanges to braze into as I did with the connection at the rear head tube/front seat tube join, so all the strength for the bottom bracket connecting tube joints would be from butt joints at each end. I mitred the tube at each end using a free program called Tube Miter that I found on the Net, first printing out the shape of the mitre, cutting it out, wrapping it around the tube, then using a felt tip to draw along the edge. This worked surprisingly well. I used my trusty angle grinder to shape it accordingly. I then used a half round file to tidy up the joint and to make sure the joints at both ends were nice and snug. Of course I didn't realise at the time, but I had messed up where the mitre was to go and cut too much off the tube in one area! Doh!
After cutting off the chain-stays from the front frame at the bottom bracket, I found two holes in the bottom bracket, presumably used as water drain holes for the chain stay tubes. I instantly suspected Murphy would be watching, and that any tube I used to connect the two bottom brackets would overlap these holes. Sure enough, when centred on the bottom bracket, the muffler pipe went over the holes. I figured that this might significantly weaken any joint, so I moved the whole connecting tube over, so that it's off centre. This means full contact is achieved around the entire tube edge.
In order to hold the frames together while I fiddled with the connecting tube, I bought some ten-inch vice grips from Bunnings. These let me clamp the flange of the rear frame (what used to be the head tube) to the seat tube on the front frame. With the frames upside down on the garage floor I could then play with positioning and filing of the connecting tube that would run between the two bottom brackets. The vice grips worked a treat and I don't know what I would have done without them.
Aligning the frame was the next thing I undertook. This involved running a piece of string from one dropout of the rear frame, around the head tube of the front frame, and back to the opposite rear dropout. I was careful to make sure the string passed on the inside of each dropout, and not on the inside of one and the outside of the other, which would mess up the following measurements. I measured the distance between the front seat tube and the string on each side and adjusted the frame alignment at the vice grips to make both measurements equal. Incidentally, I found when I had them bang on, the down tube of the rear frame looked slightly out of alignment when sighting along the rear frame's seat tube to the front frame's head tube. I can only think that the frame was built that way at the factory. At this point I decided that things were about as good as they were going to get so I commenced brazing.
A couple of weeks earlier I had tried out my new brazing kit (Bernzomatic, MAPP gas, around $100 from George Henry, plus some grade 4 brazing glasses) on a piece of bike frame and some gear/brake braze-ons scavenged from my junk pile. The metal heated to a nice red colour in not too long, and the flux-covered brazing rod (George Henry, stock sell-out) melted nicely. Things went pretty well and in short order I had a fairly solid, if not very pretty joint.
With so much practice under my belt I started on the tandem frame. The joining of the rear frame's head tube to the seat post of the front frame went reasonably well. It was quite slow going, trying to get enough heat into the joint to make the braze material flow nicely. Next I brazed the connecting tube that runs between the two bottom brackets. This did not go as well as hoped. I simply could not get the area hot enough. The result was a rather pockmarked mess. Superficially it seemed strong, but I had little confidence in it. Rather stumped, I gave up and over the next couple of weeks pondered what to try next. The joint needed more heat so I decided to try using my hot air gun to keep the surrounding metal warm, while at the same time brazing the joint. Not having enough arms (how I envy Zaphod Beeblebrox), I called in Hanna to apply the hot air; at up to 600 degrees C if you believe the box. This method did not work. After blasting the joint with heat for about five minutes on end I gave up.
Mentioning my problem on the Kiwi HPV website, Scott Campbell came to the rescue by offering the use of his oxy-acetylene kit. We made a time and I turned up with the frames, joined by my messy attempts. Scott showed me the basics of the oxy kit and I took over, finishing the rest. Compared to the MAPP gas, the flame was much smaller, and brighter - thank goodness for the grade four goggles. The braze material wetted so much more nicely, and in not too long the joint was finished. This time a much nicer joint.
Offset bottom-bracket connecting tube connects to front bottom bracket
800x600 image and
2304x1728 image also available
Back home again I worked out what I needed to finish the bike off. Bottom brackets, chains, half-links, chain-rings, tandem-length brake and gear cables. In fact the rear brake inner cable and rear derailleur inner cable are the only two tandem-specific parts on the bike, needing to be substantially longer than the equivalent cables for a single rider bike. Some of the parts I sourced locally, some from Harris Cyclery in USA. For those parts from overseas it was just a matter of waiting for everything to arrive.
Commercial tandems I had read about use a crossover chain design; the chain between captain and stoker is on the left hand side of the bike, while the chain between the stoker and the rear wheel is on the right. This allows the use of a front derailleur on the stoker's chain-rings thereby providing more gears, but has complexity trade-offs. Instead I had decided right at the beginning to go for a single-side drive design. That is, both of the chains are on the right hand side. This means there's less chance to get grease on your clothes. It also means you don't have to source special tandem cranks for the stoker, and in my mind the most important reason is the reduced torsional force on the stoker's bottom bracket. If you consider a crossover design, the stoker's bottom bracket has twice the torsional force applied to it as on a single rider bike. With the same-side design the torsional force is less than for a single rider bike, and hopefully will result in both the frame and bottom bracket lasting a long time.
The drawback of my intended single-side design meant not having multiple chain-rings for changing gears; this would limit the range of gears available, but I figured most of our riding would be on the flat anyway; tiny or huge gears would not be required. I used my own software First Gear ( www.gadgets.co.nz/firstgear ) to work out a suitable chain-ring/cassette combination to use to make the most of the eight gears that would be available.
For the single-side design a standard double chain-ring on one crank can be used effectively in the stoker's position to link both chains together. As it turns out, the spacing between the two chain-rings was marginal. The first two chain-rings I used, there was a small (less than 1mm) space between the two chains, which is fine. Later however I swapped the outer chain-ring for a Rocket brand ring and put on a new SRAM PC58 chain. All of a sudden I had both of the chains touching, but only slightly. Having the two chains rubbing is not nice, and it was possible to feel this through the pedals. My solution was to add spacers in the form of thin copper washers (bought from Blacks) to move the inner chain-ring inwards about 1mm.
Single-side drive and chain tensioner mounted on rear derailleur hanger
800x600 image and
2304x1728 image also available
I spent some time figuring out the width of the bottom brackets to use in order to produce a decent chain line. Starting at the rear hub I worked out the approximate chain line for the middle of the cassette. I aimed to select the stoker's bottom bracket such that it would place the outer chain-ring in line with the cassette mid-line. The resulting position of the inner chain-ring then dictated the width of the captain's bottom bracket; the captain's single chain-ring would connect with the stoker's inner chain-ring. In reality the width I wanted for the captain was less than what's commonly available, so I chose the narrowest, and scaled up the width of the stoker's bottom bracket as well. Note that while playing with different brackets, I found that overall width of the bracket is not a reliable way of predicting the resulting chain-line. There seems to be some variation of the 'offset'. I ended up going to Wheels-n-Deals, and measuring the distance from the face of the bottom bracket shell to the end of the spindle for each of their bottom brackets. Sometimes there's no substitute for a local bike shop!
Bottom brackets installed, it was a matter of waiting for the mail-order items to arrive. This didn't take more than a few days. At that point there was a mad dash to put everything on the bike to see if it actually was rideable. Some old wheels, a derailleur (set on a middle sprocket using the derailleur adjustment screws), one brake, some old chains. Some way to adjust the tension of the front chain was in order. Commercial tandems use an eccentric bottom bracket in order to allow the spacing between bottom brackets to be finely adjusted. I had no such bottom bracket, as it requires a special larger-than-normal bottom bracket shell to be installed. Instead I had figured on using phantom chain-rings as tensioners for the front chain. I hoped that these, in combination with a chain half-link, would allow me to get a reasonably decent amount of tension in the chain. This did not work. There was not enough room between the two bottom brackets to install two suitably large chain-rings in free flight. So not to be beaten, I left the chain sagging and went for a ride.
It worked! Steering like a barge, that feeling of moving sideways when turning sharply with such a long wheelbase. Hanna hopped on also and with no stoker bars yet fitted it was precarious, but we made it down the footpath in one piece.
The solution for the chain tensioner came to me in the form of a Surly Singleator I happened to have lying around. This mounts onto a threaded rear dropout in the same way as a modern derailleur. Lying in the corner of the garage was the rear derailleur of the front frame; derailleur mount included. This was quickly liberated from the chain stay using the grinder, and brazed to the bottom bracket connecting tube. I tried using my MAPP gas kit again, but there was not enough heat. I considered calling Scott for help, but didn't want to annoy him again. It turned out a friend from Autothority (thanks Wayne!) is a bit of a welding and brazing wiz, and he attached the dropout for me in no time at all.
The only other modifications I made to the frame were to move some of the gear cable braze-ons. The old path took the cables down under the bottom bracket, whereas the rear frame was designed for the cables to come down the rear seat stay. I moved the cable path to run along the top tubes and down the seat stay. Interestingly, the MAPP gas was perfectly suited to this task. Standard bike tubes must be thin enough that the heat cannot soak away, compared to the thick 1.6mm exhaust tube. Incidentally, after checking numerous times to make sure they were aligned nicely, I brazed both of the gear cable braze-ons round the wrong way. Doh!
The last part to arrive was the stoker stem. In actual fact, I bought this off TradeMe. It's an Ahead style stem designed for a less common 25.4mm Ahead steerer tube. I planned to flip it around and use it on the captain's 25.4mm seat post. Alas Murphy was watching, and the captain's seat post turned out to be 26.0mm in diameter. The solution here was to have an engineering shop bore out the stem to 26.0mm (thanks Gary!). This worked a treat, and soon I had some 26.0mm time-trial bars in place. Alas, Hanna did not find the positioning of these very comfortable, and opted for a straight mountain-bike bar. With a thinner diameter, a shim is required to make them fit properly in the 26.0mm stem.
Bike in need of paint, but otherwise finished!
800x600 image and
2304x1728 image also available
The bike rides surprisingly well. Prior to starting the project we had hired a tandem bike and gone for a couple of hour ride, just to see what it was like. Well I'm happy to say that for me, the captain, the homemade rides better than the commercial we had. The handling is pretty responsive by comparison, possibly due to the reasonably short wheelbase for a tandem. The steering is fine with no issues. For the stoker, the ride is also quite fine, the only negative being that Hanna cannot see directly ahead as she is still positioned quite closely behind me. A longer top tube for the stoker would be an improvement.
This tandem is intended for road use, and I have fitted it with 36 spoke (straight-gauge) wheels with double-eyelet rims and 1.5 inch slick tires. These are working well so far but because of the lower profile slicks, the ride height is reduced compared to big knobbly tires. Pedal strike when cornering was a concern during the design stage as the stoker's bottom bracket was a little lower than on the original mountain bike. I countered this by fitting some short 162.5mm cranks for the stoker. This has worked well in practice.
Two things I thought about during the design, but didn't actually do anything about were the steering trail, and the hump-back/bowed nature of the bike. The steering trail on a tandem is supposedly better if it is shorter than on a single bike. This is so that inadvertent leaning by the stoker does not cause the bike to 'steer itself' (like one does when riding hands free). One way to adjust the trail is to use different forks with less trail. This bike however has a very long head tube, and finding any other donor forks that fit would have been a mission, so I didn't do it. Besides, they would probably have similar geometry. Building the tandem as a 'hump back' by using a shorter connecting tube between the bottom brackets would also have achieved a somewhat similar effect. This would have resulted in more ground clearance; a good thing. The problem here was the connection between the rear frame's top and down tubes to the front frame's seat tube. There was no way to alter the angle of this connection without significantly compromising the strength of the joint. So the result is a bike with very similar steering geometry to the original single rider bike. In practice, and to me, it rides just fine. Perhaps it's not possible to ride along a 5cm wide path on a rough shingle road (we tried it!) due to some wavering, but it's still pretty good.
The last stage in the build was the paint. I had planned to have the frame garnet blasted, then to fill in my messy brazing with car panel filler, then to have it powder-coated. It turns out that powder-coating only adheres to metal and powder coating over filler would not be an option. Instead, thanks to fellow HPV member Martin White, I found out about a company called Industrial Painters. They do a baked enamel finish which looks to be pretty durable, and it will work over a metal surface that has been filled. Most surprisingly however is that it would not cost much at all; in fact less than powder-coating. I had been lead to believe by the powder-coating company that any other finish would cost at least $200; more than I wanted to spend on this project. So I first took the frame to Escort Metal Polishers to have it garnet blasted; this cost $40. Next, I took it home and spent some time filling in imperfections with Bondo. After that it was off to Industrial Painters for a generic white paint job. This cost all of $50 and took around a week. I think this is a real bargain. The other components on the bike I either polished up with Autosol on a rag, or, if too far gone, painted black using a spray can.
Bike finished!
2048x1536 image also available
