Somewhere between woodwork and framebuilding: bamboo bike #2

http://farm3.staticflickr.com/2858/12751550945_1b28a12861_h.jpg

INTRODUCTION

A couple of summers ago I built myself a bamboo bike frame in my kitchen, using about £100 of materials. I rode it hard, leaving it outside most of the time, and had great fun with it until around Easter, when I broke it supporting some friends doing the Devizes-Westminster Canoe Race (which is to say they were doing the race, I was cycling alongside. If I'd literally been supporting them, I wouldn't have been surprised that I broke it). I'm sure it was fine when I put it on the rack, but when I got home I found that TT/ST joint had gone rather wobbly. I could mend it, but I can't be bothered. It's time for iteration #2,* with all the lessons I've learned put into use...

This thread is intended to force me to properly document my build and decision process. Last time I hardly documented it at all, and I regretted not having photos of the process. Hopefully this will be interesting in its own right, plus helpful for the growing number of people considering making a bike out of bamboo. Bamboo is by far the easiest material for a novice or more experienced hobbyist framebuilder, requiring only the most basic tools. But it can also yield frames that are great to ride - fairly light, shock-absorbing, and great-looking.

Hope people enjoy reading!

• It's coincidentally also bamboo bike #2 currently on the forum - see also http://www.lfgss.com/thread105686.html
  

Definitely intrigued by this. £100 you say? How many hours did it take you recon?

**PLANNING

**I always end up spending way more time planning and researching my projects than actually making them. There's so much great information out there on the internet which has been utterly vital. All of it is easily found so I won't bother citing too many inspirations apart from where they are a bit more esoteric.

Bamboo choice
The first decision I had to make was the variety and size of bamboo to use. My previous bike, like almost all bamboo bikes you'll see on the internet, used quite oversized tubes for maximum strength and stiffness. Because it's a natural, highly variable material, bamboo is usually sold in width bands, and the 40-45mm band is generally used for the main triangle. However, my first bike was a bit over 2kg, and I fancied being lighter this time. Also, I am certain that the vast majority of a bamboo frame's stiffness comes from well laid-up joints made with a suitable material - 45mm tubes seem massive overkill. My last frame was bendier than I liked, but that was because my joints were shit (see below), not because there was any real flex in the tubes.

The next size down is 30-35mm, which is much closer to the tube size of steel frames. They are still extremely stiff and strong. I'm going to push the boundaries of hobbyist framebuilding a bit this time, and use the thinner tubes - if it comes off, I should get a frame that's significantly lighter than my previous effort. If it's too noodly, I'll rework it slightly to add a second top tube a la big Rivendell frames, but I'm hoping it won't be needed. For the chainstays and seatstays, I used 20-25mm poles. If I could handpick the poles, a smaller diameter might be possible, but the smaller you get the more weaknesses there tend to be, so I prefer to overbuild the rear triangle as I don't need cassette clearance.

Bamboo variety is something that's discussed a lot by framebuilders. The general consensus is that the strongest species are iron bamboo, tonkin, henon and guadua. However, in the good old UK it's very difficult to get any of these in the right sizes. Moso and black bamboo are widely available - moso is your classic green/yellow bamboo, while black is a very similar species with darker surface coloration. There's also tiger bamboo, but it's hard to tell whether it's *real *tigerbamboo, which is the same species as black, or something like moso that has been selectively scorches. It's also a lot more expensive. I'm sure that moso and black are easily strong enough, so I'll be using moso as I'd just sand off the black layer anyway.

I bought my bamboo from UK Bamboo Supplies - unfortunately they've just introduced a new shop system with a larger minimum order (10 poles of each size) - if you call them up they might be able to do smaller packs (it used to be 5). Their shipping is expensive and slow, but there's not much choice in the UK. The Bamboo Bicycle Club also sells bamboo kits as well as running courses. They are much more expensive, but they do hand-pick good bits and treat them well. I'm cheap so went for the UK Bamboo option, but you run the risk of getting wonky, cracked or funny-shaped bits so you need to buy with that in mind. Both times I've bought one 5-pack of thick poles, 1m long, and one pack of thin 1m poles, and that's worked out. But you may need to be crafty with the bits you choose.

***Lug choice
***Because I was being even stingier last time, I didn't use carbon fibre for the lugs. Instead I used sisal twine. In hindsight this was a poor idea, because while it's stronger than the hemp tow most non-CF builders use, the tight weave of the twine makes it hard to wet out. You also need to really, really overbuild the joints, which end up being very big. This time, I'm going to use 12k carbon fibre tow from East Coast, which is about 12p/m. I'll need several hundred metres, I think, but I need to research that part more.

If you read up on bamboo framebuilding, sooner or later you'll come across Craig Calfee. As well as his better-known carbon frames, he also makes some great bamboo frames, and his sideline Bamboosero sets up framebuilders in Ghana. Calfee's of the opinion that CF shouldn't be used for lugs because the coefficient of expansion of the two materials are too great and will cause the joints to fail over time. I can see the science behind this, but I have never seen a CF joint on a bamboo frame fail. I have, however, known several cases of hemp / other natural fibre joints failing. I'm sure that Calfee's hemp-lugged frames are as strong as CF and longer-lasting, but it's very easy to make a shitty natural-fibre joint - they're harder to wet out, prone to internal voids, and difficult to get the fibres running in the right directions. It's not just a matter of building up material around the joints. CF is much better suited to epoxy wetting-out, and has a greater margin of error in terms of actual strength. I don't think the thermal expansion issue is a problem considering the temperature changes the bike will experience. Watch this space!

For the glue I'll be using West Systems 105 epoxy. It's strong, odourless, not too unhealthy to have about, and easy to work with. I'll also be using bamboo sawdust to built up joint fillets before wrapping with CF.

***Metal bits
***Unless you're brave or very inventive, you'll need a few metal bits - dropouts, BB shell, and headtube. I'm getting these from Ceeway. Last time I used normal horizontal dropouts and found that they were fiddly to build up strongly, and I made the chainstays slightly too short and had to deflate the tyre to get it in. I'm not a track end fan, so will be going with the Ceeway sliding dropouts. This will allow me to adjust the effective chainstay length, easily run mudguards, remove the wheel and replace it without changing the chain tension, and change the inserts if I ever want to run derailer gears or a Rohloff hub. They're a bit heavier though, and a bit dear.

***Geometry
***Before I could do anything, I needed to decide on my geometry. Last time I think I just copied the Steamroller frame design; this time, I've done it from the ground up. Sheldon Brown links to numerous websites which have sensible modern bike sizing and geometry advice, which I used to get an idea of what I wanted. I then used BikeCAD to draw up my design:

Bike geometry is boring as fuck if you don't care about such things, but basically it was built to have short chainstays (for fat tyres and full 'guards, anyway - I wanted to stiffen up the rear - a wishbone seat stay will be used for the same reason), stack and reach which suit my size, and trail of about 55mm, maybe a bit less. This is a relatively low amount of trail which will give fast handling. For that you sacrifice high speed stability but I'm not going to be doing many 60mph alpine descents on a fixed gear...

Definitely intrigued by this. £100 you say? How many hours did it take you recon?

Oh, ages! Something like 100 hours I think altogether, because I was doing it in little chunks here and there, and with no power tools. That time does include building two wheels though, I think. Yeah, I reckon the frame cost about £100 - about £30 for the bamboo, £30 for the glue and twine, and £30 for the metal parts, plus sundries (gloves, brushes, etc.). CF is more expensive than twine though, and the metal parts can go up exponentially from there - for example, the natty sliding dropouts (stainless main bits, alu inserts) I plan to use are really dear, £40 or something mental,* compared to £12 for cast (mild steel) horizontal dropouts.

I reckon you could do it in three to four days if you were efficient and that was all you were doing. The main limiting factor is that you have to wait for glue to go off at various points - it'd probably take a day's work to cut, tack and prep the frame, and another two days to do the joints (with most the time spent waiting), then maybe another day's finishing, depending on what finish you choose to do.

***BAMBOO SELECTION AND PREPARATION

***
Poles by comatus1649, on Flickr

Preliminary drying
The bamboo arrives in a mostly-dried state, which is a good start, but needs some stuff doing to it before it's useable. It's important that you let it stabilise to the humidity of wherever you're working, or else you risk it changing size or cracking later on. I recommend buying the bamboo a reasonable time in advance so you can leave it to dry - you don't want to put it on radiators, in airing cupboards, or anywhere that's really warm, just leave it inside somewhere with an even, warmish temperature. You should also break as many of the internal nodes as you can to encourage even drying, and to help with later steps ...

From what I've read, I don't think you should heat treat it as many people recommend. Everyone talks about caramelising the sugars in the bamboo to make it stronger, but all the evidence I've found suggests that it reduces the strength by about 20% and makes it more brittle. I found a very useful paper called 'Bamboo in the Laboratory', which was the only properly scientific test of heat-treating bamboo I've seen. It needs to be taken with a pinch of salt because the needs of bamboo rodmakers (who split culms into smaller bits before reforming them into rods) are different to those of framebuilders, but the evidence is useful - heat-treating bamboo up to about 150 degrees C (I estimate - I don't have it to hand) permanently reduces cellular water levels in the bamboo, making it lighter and stronger. Beyond that point, after which the coloration changes, the bamboo is internally scorched and made weaker. (This is why I wouldn't recommend using 'tiger' bamboo unless you can check that it's the species with the natural markings, as some is just burnt moso or similar.)

I don't have a big enough oven to heat my bamboo properly, and all you can do with a propane torch is burn it. So I decided that just letting it dry thoroughly was the best option. Over time the bamboo turns from green to a nice tan colour - if you buy it from the internet, it will already be well into the tan stage, and will need little further treatment. Lots of hobbyist builders have used propane torches and had no long-term problems, but why reduce the strength of your key material?

***Checking the bamboo
***It's quite likely that your bamboo will have some problems. The initial checks are to find and discard poles that are too curved, cracked, or insect-nibbled. Curved poles can be useful if they're curved in one plane (chainstays in particular are better if they're curved) - if they're curved in two directions, like a very stretched out spiral, they'll be very hard to use. Curved culms often grow due to a restricted or light-starved position and may be weaker another reason to beware of extremely-bent poles (and those that are much greener on one side than the other).

Look out very carefully for cracks, both at the ends and on the nodes. Cracks are bamboo cancer. They will only get worse and spread. If, by the way, you find a crack in a pole when you've already built your bike, don't totally despair. Drill out the crack at either end to prevent it spreading, and wrap the pole tightly with CF tow. That should stop it spreading much further. Likewise, if you find any cracks on any of your poles (even ones you don't think you'll use), cut after the crack to keep the rest useable.

Also look for woodwork, rot and other grass maladies. There's often surface discoloration that's not a problem, but sometimes there are bits around the nodes on smaller poles that concern me slightly. Avoid using them if you can.

Marking and cutting by comatus1649, on Flickr

Marking, choosing and cutting
As I go I mark up all the problems with a soft pencil or Sharpie. This way, they're easier to spot when I'm trying to choose poles for particular parts of the bike. In the picture above, I've marked up the bit where the harvester has torn off the leaves, leaving open grain. This isn't a problem for me, but it increases sanding. If I was going for a natural finish on the bamboo, I might chuck this bit.

Choosing the right poles for the different parts of the bike is the real art of this stage. Looking at bike FEA models, the majority of the stress is around the bottom bracket, especially on the downtube. This joint wants to have the sturdiest, densest parts of your tubes.

Take into account your seatpost solution as well, by the way - on my first bike I used a normal seatpost, so I needed a seatpost that had an internal diameter of about 28mm (bamboo in that size has a wall thickness of maybe 4mm or so). It also had to be bolt straight, and I didn't want too many nodes or I'd just have to internally rasp them away (faff!). Some people cut a slot in the seattube for a seatpost clamp on a normal bike - I think this is a terrible idea. I used a double clamp made by eXotic which clamps around the uncut tube, and also onto a seatpost - I shimmed it down from 27.2 to whatever size I had that fitted into the bamboo tube.

However, this time I'm using 35mm tubes that have an internal diameter of about 22mm, so a normal seatpost is out. I intend to extend the seattube almost as far as the saddle, and bond in a 22mm BMX seatpost for the last little bit, with a plain-top saddle clamp. Unless I find an affordable integrated seatpost topper before then ...

As for the chainstays and seatstays, the clearance in this region is pretty tight, but you don't want to use very thin tubes. It's a hard balance to get right, but a very important one. You'll find that smaller tubes have flat sides, which end up being rather like the crimps on steel chainstays.

Once I think that I know which tubes I'll use, I then cut the pieces roughly to length, making sure that I have plenty of leeway. A pull-saw is the ideal tool for this job but anything will do really. The skin may splinter slightly when you cut it because of the longitudinal fibres (the reason it makes such great frames), but resist the urge to peel off the splinters - often if you go overboard they run the whole length of the node and look rubbish. You could wrap the pole in tape before you cut to prevent this but it's no big deal really.

In the next installment, I will cover my top-secret bamboo frame innovation, and my first great error of judgment ...

Dennis looks on approvingly by comatus1649, on Flickr

This is awesome, looking forward to the finished frame! Out of interest how would you describe the characteristics of bamboo more like steel or alu?

I've not had that many different bikes so I can't really say, but the general consensus is that it's much more compliant than alu, so more like steel in that respect, but with carbon's ability to absorb road buzz. I'm aware that this sounds like marketing woo, but this is what people say. Also, I think what people mean when they're talking about stiffness is generally stiffness at the bottom bracket joint, whereas compliance is a matter of absorbing vibration, so I think being stiff and compliant is possible.

Structurally bamboo is very similar to unidirectional carbon fibre (which is being increasingly used for bike parts), for what that's worth. I found my last one to not be stiff enough, verging on noodly before I reinforced it. But as I mentioned above, I'm sure that the noodliness is down to my poor joints.

***SURFACE PREPARATION (AND GREAT BOREDOM)

***On my first bike, I left the bamboo finish as it was. Bamboo has a great waterproof and durable finish, but unfortunately it does degrade when it's not longer growing and left outside for a few months. When I had to go without indoor storage for a while, my bike began to go mouldy on the surface. At the time I scoured it all off and oiled it, which more or less did the trick, but it wasn't a very good fix. Surface water would also eventually sink in, which is a Very Bad Thing for a material so prone to cracking.

So this time around I wanted to give it a good artificial coating that would last really well, and look better than my mouldy old one. I thought about painting it with epoxy, but a) it's pretty thin and would be prone to running, and b) it's very stiff. I don't know how much the bamboo poles actually bend - probably an imperceptible amount - but I was pretty sure that it would eventually crack and then blister, and then I'd have to strip it all off... Some people have used spar varnish, the finish used on yacht masts, but it's really expensive and not very nice to work with.

A bit of a brainwave happened when I was reading a rodmaking forum. A guy called Tim Anderson had been using Gorilla Glue to seal his rods, and this seemed like the perfect solution. I love Gorilla Glue and will be using it to tack the frame together, so I had just bought a big pack of it. Applied in very thin coats it makes a really hard, really shiny coating with very little effort, and it is also very slightly flexible. It's essentially a catalysed PU treatment, curing with the humidity in the air. It also gives the grain a really nice deep appearance. I used a cutoff to try it out:

Finished sample of wood by comatus1649, on Flickr

It's not the best picture, but I think it looks pretty damn good considering I quickly filed off the enamel, rubbed the glue in with my finger and let it dry. You can just about see the dust nibs, but they shouldn't be too hard to avoid (for a start, I won't be putting it on just after sanding).

So I was decided on the finish. Unfortunately, the very resistant natural coating on the bamboo has to go first, otherwise nothing will stick to it. And that means a lot of sanding.

A lot more sanding, in fact, than I had ever realised. And once I'd decided to sand it down to the grain (potentially a bad call), I couldn't really back out. The enamel skin on bamboo is quite thick, and really hard to sand. I was about a quarter of the way done on one of the chainstays when I realised that this was going to take a very, very long time. After hours and hours of hand-sanding, I decided it was powertool time:

Sanding gear by comatus1649, on Flickr[[/URL]

This greatly cut down the time, but working through different grades of paper by hand to 400grit still takes almost two hours per pole. The big poles take ages because they're big, the small ones because the grooves and dents where the leaves sprouted off are really tricky to sand. However, when you get down to the grain the poles start to look much nicer than the garden-centre chic look they had to begin with:

[URL="http://www.flickr.com/photos/78998990@N03/8890033461/"]](http://www.flickr.com/people/78998990@N03/)
Sanded seatstay by comatus1649, on Flickr

They'll look brilliant when they're coated with GG, can't wait! That's all for now. I've still got the TT to finish off, and all of the sanding on the DT and ST to do...

***OTHER UPDATES

***My lug decisions have changed a bit. As it's the same price, I'm going to use unidirectional tape rather than tow - it's the same kind of thing, but in 25mm-wide strips rather than 5mm-ish, which should be easier to handle and more economical on the epoxy. I also don't really like the way that tow-wrapped joints look, and don't see the point in using loads of bidirectional fabric like Brano Meres has - about 40% of the strength (and weight) of the fabric is wasted on most of the bike. So I'm going to use tape, then do an overwrap with fabric, like Cognitive Cycles have done on some of their bikes (like here, for example). I think I'll use a carbon-kevlar blend - it's lighter, more abrasion-resistant and less brittle and so good for a top layer. But the main reason is just cos I want a bulletproof bike.

This is an amazing thread. Thanks for so much detail. I've enjoyed seeing bamboo bikes. The idea of a homemade bike really inspires me to make one myself.

Hi, really liking your thread. I have just taken the plunge and bought some bamboo to make a bike from BambooUK, they are doing bamboo bike kits now. It costs a bit more but they choose the bamboo for you.

I really am a bit (lot) of a novice in these matters but it should be fun learning. Your blog is invaluable for UK bamboo bike builders as most are US centric. Thanks for this, please, please keep this blog up to date.

Any chance I could PM you if I get stuck?

Cheers,

Phil

Ha thanks for the mention! :) I'll be watching this as I am going to start mine properly in about 3 weeks when I get back from my holiday in spain. I was thinking about putting some CF tape as a top layer just to get a nicer looking finish as well so I'm glad you mentioned that. I haven't even had time to sand the paint off my donor frame as I'm in the middle of exams :(

BTW check out rattleCAD. Design your bike frame for free, unbelievable:

http://sourceforge.net/projects/rattlecad/

Any chance you could give me details on exactly which drop outs and head tube you are getting from ceeway ? I find their site a little hard to get my head around!

Also is this the carbon/kevlar tape you are thinking of using ?

http://www.ecfibreglasssupplies.co.uk/p-2826-carbonaramid-hybrid-225g-plain-weave-tape-25mm-wide.aspx

Do you think 99 meters would be enough considering it is 25mm wide?

Any help appreciated :-)

Metres not meters!

Subscribed!

This still happening???

Hi everyone. Sorry for the big lapse in communication - I've just moved house which has meant I've only had mobile internet for the last few weeks. I'll answer questions and get some lunch, then I'll type up the next big update...

Djangoberry/Phil: Sure, feel free to PM me if you have any questions. I can't claim to be an expert by any means but, as you'll find, there are a lot of things you pick up even building only one frame that you just wouldn't have thought about before.

Thanks for the link to RattleCAD - I've never used it before for some reason as I find even the free version of BikeCAD easy and powerful enough. It does lack some things though (and I'm not going to pay hundreds of pounds for the full version just yet...), so I might find some use in RattleCAD in future.

As you say, Ceeway's site is notoriously difficult to use, but their customer service is great. I decided to splash out on the TSS7 stainless dropouts with TSS2 and TSS8 inserts - they came to a pretty serious £56 before VAT and are so far the single most expensive part on the whole thing. I liked the functionality of the sliders though, and the increased space to strengthen the joint compared to horizontal dropouts. I went for the CYRK18 headtube from the Thron tubeset.

I had planned to use this UD tape, with a thin overlayer of this fabric. I wouldn't advise using a CF/kevlar blend for the whole thing because kevlar has a lower tensile strength. A top layer of it would protect against abrasion and impact damage (how much I really can't say; possibly barely at all, but more than CF alone). 99m of a 25m tape would be plenty, but I'd use unidirectional rather than woven tape for most of the joint, or half of your material (/weight/money...) is going in a direction that achieves next to no extra strength, - you're better off using UD tape or tow with the fibres aligned according to the load on the joints. In other words you'd need about twice the length of woven tape compared to UD tape, assuming they have the same weight per metre.

From my research (which you may wish to check) a bare minimum of 100g of unidirectional fibres should be enough to make all the joints (plus of course a roughly equal weight of resin). This corresponds to about 20m of UD tape or about 130m of 12k tow, 260m of 6k tow, etc... I will probably buy 40-50m of UD tape to be on the safe side as I'm using such skinny tubes and may wish to really build up the joints. If you're only using woven materials, you'll need something nearer to 200g as a minimum (Brano Meres used ~360g of woven cloth on his beautiful bamboo MTB).

Hope that helps - sorry it sounds so complicated!

Right, update on the way...

FIRST JOINTS - BUILDING THE JIG AND TACKING THE REAR TRIANGLE

In the long time since my last update, I've made a reasonable amount of progress. It's been a bit slower than I would have liked, and moving house has prevented me updating until now. Also, I start a residential job for 6 weeks in July, and probably won't be able to work at all on the bike until I'm done. So this project is going to take a lot longer than I had planned.

***Building a jig
***As more illustrious builders like Ron Cooper have shown, it's perfectly possible to build a great bike without the precision engineered and extremely expensive jigs that most builders use. My first bike was also built jigless and came out reasonably well - with lots of measuring and some creative thought, I believe that you can build a near-perfect bike this way. But this time, I fancied having a go at building a jig, albeit a very different one from those commercially available.

Commercial jigs are very, very expensive. For example, a lovely Henry James jig costs $4000, plus $750 for the stand to put it on. If you read the Paterek Manual you'll also find him recommending precision-lapped plates made from stone or metal to take exact table measurements, costing thousands more. Even homemade jigs using aluminium extrusions come to several hundred pounds. I believe that the fundamental assumption of jig designers and framebuilders, namely that all the variable parts need to be held precisely in place* at the same time*, is the reason why it's so difficult to build an affordable jig. I also believe that it's a wrong assumption - the builder can only ever work on one joint at a time, even if all the others are held tightly in place for him or her. So my jig design was built from the ground up not to hold the rear axle, BB shell, head tube and at least seven other tubes all in place at once, but to deal with one joint at a time.

My philosophy is that it is easier for a small-scale builder to take the time to precisely lay out each join on its own, rather than make a large jig which has the entire bike laid out. Working from simpler geometrical principles, much less precise machining and setup is required. Consequently, it is modular, compact, and very cheap. It cost me under £20, and while I've already found some flaws, I'm happy with the principle.

Once I've finished the bike and adequately tested my design, I'll write it up more fully, but you should be able to see how it works from my pictures throughout. The raw materials are easily obtained - 1m of 17mm inside diameter aluminium U-channel, 2 x 500mm M10 threaded rod, about 14 M10 nuts and a few washers. The nuts fit snugly into the extruded aluminium and can be glued into place, allowing rods to be attached to the extrusions to adjust lengths or join parts together. The completed jig can be disassembled into a few simple parts:

0: An assortments of nuts and washers used to attach parts and lock threaded rods into place
1: 1 section of aluminium 120mm long, into which two nuts are glued about 5mm from either end; a length of threaded rod (about 350mm) is attached at right angles to the centre of the aluminium, on one side (ie. so that it can be laid flat in a T-shape with the open part of the U-channel upwards). This is primarily to hold the dropouts.
2: A 350mm section of aluminium with two nuts near one end, and a 68mm section attached at a precise right angle at the other end, with the open part of both channels facing up. This is primarily to hold the BB and the poles attached to it.

1. Another long section of aluminium identical to 2, but without the short section at right angles.
   
2. A separate length of threaded rod 500mm long.
   
3. A short length of threaded rod about 150mm long.
   
   

As will be seen below, this can be set up in numerous different ways...

MAKING THE FIRST JOINTS

Bottom bracket and seat tube

The first joint in my process was the ST/SP to BB junction, a simple right-angle T. I have taken several photos which can be found in my Flickr photostream, but I'll only briefly describe the process here for clarity. Because bamboo poles are not entirely straight in all planes, you need to decide which way they will point. My ST pole was only very slightly bent, but I wanted it to bend away from the rear tyre and be straight when viewed from above - by comparing it with a straight edge and just rolling it in my hands I found and marked its forward face, and made sure to mitre it correctly.

Also be very sure to double check the BB orientation as well - most shells have a groove cut on the inside (past the threads) on the right hand (drive) side, but double check with a BB as well. I have read a build report where the guy had it all completed before he realised he'd put the BB in backwards, and had to run it as a self-undoing left hand drive.

Cut the mitre roughly with a saw first, as coarse rasps are prone to leaping off and taking bites out of the user if they don't have something to guide them. This whole process is of course much easier if you have the piece in a vice (either with rubber jaws or, better, a pipe clamp (which can be improvised from bits of wood)). Check regularly as you go - you want to end up with nice tight mitre that doesn't wobble, holds the other piece at the right angle, and doesn't have any visible spaces. It's not the end of the world if you're no good at it (I'm not, but I'm getting better...), as it will have very little effect on the joint's eventual strength, but it makes the initially glued joint stronger and easier to work with.

A pretty good mitre by comatus1649, on Flickr

I was pretty happy with that one.

Next stage is to have a dry run before gluing it. With my jig, it's easy - the short perpendicular section at the end of jig part 2 holds the BB exactly parallel to the longer aluminium section made up of part 2 and 3, joined with part 5 and some nuts - the BB is held in place with bands, as is the ST, in such a way that the joint is pulled together at precisely the right angle.

ST and BB shell in the jig by comatus1649, on Flickr

Having measured and remeasured and satisfied myself that it all worked fine, I applied a thin even layer of Gorilla Glue to the bamboo, having first misted it and the BB shell with water. Held in place, it had 80% of final strength in two hours, and was ready for the next stage of gluing.

Chainstays to BB and dropouts

For attaching the chainstays, the jig needs to be changed to configuration 2. Parts 3 and 5 (the long piece of aluminium and the short piece of rod) are removed from part 2 and replaced with part 1. Part 5 is used to hold the dropouts in place in part 1. The chainstay length is changed using the threaded rod part of part 1, screwed into part 2. The BB shell and attached ST can be rotated, and is held upright by the tension on the bands, as seen below:

Quick pre-glue layout by comatus1649, on Flickr

The BB/CS joint is very hard to get just right at this early stage because of the small mating area, the numerous angles to get right with the mitre, and the tight clearances. Bamboo's greatest weakness as a bike material is that it's quite straight and fairly uniform in cross-sectional shape, whereas for chainstays you want a bent, ovalised or crimped stay to create space for the tyre and chainring. Smaller bamboo poles have alternating flat sides which can be used to some extent. You also need to leave enough space outboard of the chainstay to build up the joint without interfering with the BB cup.

A great deal of measuring and head-scratching took place at this point, along with some test fitting of the chainring. I eventually satisfied myself that it would probably fit, though I would most likely have to file away some of the bamboo on one or both sides. I should be able to strengthen this just fine using CF, but if you're using other materials for the joints you may wish to consider using a wider BB shell to increase clearances (up to 100mm is available, but this limits BB choices and confuses chainline; I decided to make a standard 68mm shell work somehow).

The seat tube needs to be set to the right angle - this means that the rear axle needs to be elevated. I planned a BB drop of 55mm, so put DVDs and CDs underneath the right-angle section of part 1 until the centre of the axle was at 95mm from the table, compared to the BB's 40mm (in the centre, of course). I could then use a large protractor to set the angle of the ST. The inner tube bands hold the BB firmly enough when you pull the slack to the correct side - in other words, as it was leaning to the left, I needed to make the band tight on the right hand side. In jig configuration 2.5, parts 3 and 4 are used to provide a somewhat haphazard support as well, just in case:

Jig configuration 2.5 by comatus1649, on Flickr

Eagle eyes will note even more inner tube bands around either ends of the chainstays - on the BB side, I arranged them to overlap the edge of the bamboo by a fraction of a millimetre in the hope that it would stop them slipping down the BB shell. It didn't work very well, and I should have just scratched up the shell more. I didn't because it was a machined aluminium shell with a light textured finish rather than a smooth cast steel one, but I had several problems with poor adhesion. On the dropout end of the chainstays the bands were used to pull the chainstays tightly against the metal to improve the joint. Because the tubes were filled with foam the joint was much easier to get right this time around - the tabs on the ends of the dropouts are much smaller than the insides of the tubes. Some manufacturers make dropouts with plug ends rather than tabs, which I would imagine would work very well.

By the way, the angle of the dropouts was something I wasn't sure about. Eventually I went for an angle similar to horizontal dropouts which made the bottom of the aluminium inserts parallel to the floor. I think this looks fine, though you loose a little horizontal travel because it's diagonal. It does (in principle) mean that with a rear rim brake, the rim would stay aligned with the pad regardless of chain tension, which is the reason why horizontal dropouts are really a bit diagonal and track ends are actually horizontal.

Seat stays to dropouts and ST

Now for the final stage to complete the rear triangle - attaching the seat stays to the dropouts and the seat tube.

[By the way (as we shall see...), when I say 'complete', you shouldn't have any great expectations of strength for a frame only tacked together in this way with glue. It is reasonably rigid, but requires very gentle handling. The purpose of this stage is to get it all held together; epoxy filleting strengthens the joints enough for further modifications, a brief fiberglass wrap prepares the way for a durable CF wrap, and the CF wrap is what actually gives the joint all of its strength. Because the CF wraps entirely around the whole joint, what is inside it is almost of no importance.]

I wanted to do a wishbone seat stay design for this bike to make the rear triangle a little stiffer, to allow good tyre / mudguard clearance even with compact geometry, and to make it look cool! Making 6 joints all line up at once would have been a real headache, so I did it in two stages - first I connected the seat stays to the horizontal part of the wishbone, and secondly I connected the vertical part of the wishbone to the ST and to the horizontal part, and connected both dropouts. The first stage was done in situ to make sure it was all aligned correctly, with the horizontal part of the wishbone resting against the ST. After that had set, I trimmed the horizontal part to size and rounded the edges so that it would eventually have a more natural shape:

Roughed out wishbone by comatus1649, on Flickr

After that, it was trivially simple to rasp the tiny vertical part of the wishbone to shape and glue it in place. Glue can be a bit frantic to work with because it needs to be held in place just right and you're constantly fretting about it going off or dripping onto things, but by doing it in two stages this was made very straightforward. Again, inner tube bands came to the rescue holding it in place at the top, while I used hair bands for the bottom (see Flickr).

A couple of hours later I was feeling pretty happy with the whole thing, and wanted to take it out of the jig to set up the next configuration to attach the top tube. And then this happened:

Oh snap. by comatus1649, on Flickr

The right hand side had broken at the wishbone bridge and the bottom bracket - I forgot to remove the hairbands that were tensioning the joint before I took it out of the jig, and it snapped the two tiny joints just like that. With better preparation (roughing of the BB shell in particular) this might not have happened, but it was mainly a human error. However, it was very frustrating to have to awkwardly reglue two joints which had just been made so easily, and broken even more easily. I had been overkeen in trimming off the excess glue as well - this weakens the joint, but allows the next stage (filleting) to take place properly (it is stronger to have a joint with excess glue trimmed off and an epoxy fillet made around it than a fillet over the top of excess glue). Though I'd still advise trimming excess glue, leave it until you're just about to do the fillets, and be very gentle with the frame.

On the other hand, this was nowhere near as frustrating as what happened next. As you, oh patient reader, have read above, I precisely measured the ST angle with the correct rear axle elevation. However, because it didn't seem to be broken and I thought it had already set firm, I didn't remeasure it after a visiting friend knocked the damn jig off the table. Remeasuring didn't even occur to me until I had carried on, glued the wishbone seat stay and fixed the rear triangle semi-permanently in place. When I did remeasure it, I found that I had created a bike with a 69.5deg seat tube, an accidental bamboo version of Bauer's stealth bike in the 1993 Paris-Roubaix. It was such a massive error (3.5deg) that I can't believe it was anything other than it getting knocked over and me not thinking to check it.

This required a major redesign; I could rip it all to bits but I'd put in a lot of work and would have to remake some parts. I decided to work around it.

My original design had a very high BB, with a drop of 55mm (which would be high for a track bike, let alone a road bike). I decided to raise this by 10mm - this still makes it a little higher than a Surly Steamroller and will be just fine - it may even have some handling advantages but I'm not so sure about that. More importantly, it will make the ST angle about 71.5deg - still slack by modern standards, but I will be using a Brooks saddle. Because of their rails, they often need a layback saddle on modern frames, but should suit this frame fine. It seems perverse to design a frame to suit a saddle, but that's part of the fun of custom framebuilding!

I had also made a couple of other minor changes, such as making the rear triangle more compact. I couldn't get a headtube cut to a sensible length from Ceeway, so with the freedom / hassle of cutting it myself, I went for a 147.5mm HT, the strange result of aiming for 150mm and having more trouble with squaring it off than I had expected. Maintaining the same TT length for the sake of argument in BikeCAD gave me a slightly shorter reach or front centre which actually on reflection is no bad thing, so I kept it. It should tame the otherwise fairly long wheelbase a bit. There may be a touch of toe overlap but I can deal with that. Here is the revised design:

Revised geometry by comatus1649, on Flickr

That's all for now. In the next few days I will be working on the front triangle, using still more jig configurations. I'm going to order the composite bits today, so may wait to fillet the rear triangle before I start on the front, to avoid repetitions of previous 'Oh snap!' moments...

Thanks for waiting ;)

Ben

***STARTING THE FRONT TRIANGLE

***More tacking...
With the rear triangle all tacked in place, I decided to carefully start on the front while I waited for my epoxy to arrive. There are several ways the jig could be set up to do this including holding the rear triangle at such an angle that the TT needs to be vertical and can easily be checked with a spirit level, but I couldn't make this work. In the end I just went for a simple configuration to hold the two bits in place after I had set the angle (71.5deg ST and 9.2deg TT = 80.7deg between the two). A loop of wool was used to help me set an accurate angle, as the ST has a slight bend to it:

Jig configuration 3 by comatus1649, on Flickr

Somehow, this simple setup was incredibly difficult to get right. I couldn't really clamp this joint very well, so I would hold it in place for up to 20 minutes at a time, get what seemed to be a fairly firm self-supporting joint, go to make a cup of tea and it would have fallen off when I got back. Regardless of what I did, it just would not stay put. I began to blame supernatural causes for the difficulties I was having, and invented elaborate rituals to appease the spirits of the house. After a while I think I just held it in place for ages and ages and it did the trick. Luckily, thanks to the big mating areas, this is a fairly strong joint that's not prone to suddenly breaking like the CS/BB joints can be. If I make another bike using this jig I will make a little bracket to support the end of the TT.

When this was set, I turned to the DT. Both DT and TT were only roughly cut to length at this stage, as it was too difficult to work out how long they would have to be in isolation. The mitre on the BB end of the DT is particularly complex because it has to accommodate a small part of the ST as well as the BB shell. The same jig configuration was used, with the DT running in the channel. Unfortunately the nuts in the channel protruded just slightly too far and made the tube go off to one side, so I had to use my multitool as a shim to keep it centred. The distance between the TT and the DT at the end was set by eye, holding the HT up against them. This turned out to be surprisingly accurate.

Filleting
When these tubes were both set, I wanted to fillet several of the joints in the rear before going any further - filing the mitre for the HT in situ would put a lot of strain on the joints, and I needed them to be fully reinforced first. Filleting is the process of building up thickened epoxy around the joints to round off the edges and really increase the surface area of the joint. This not only makes the joints stronger themselves, but also allows a more curved shape to be achieved with the carbon fibre, making it much stronger.

Because I'm cheap, I used the sawdust I had been collecting all along to thicken the resin rather than any of the additives West System make. This worked fine, though the bits of sandpaper that had ended up in the dust even after I sieved it made the fillets speckled. Getting the epoxy the right thickness is the first and most important stage - too thin and it will just run everywhere and be a nightmare to handle; too thick and it won't adhere properly or add much strength. You're aiming for a peanut butter sort of consistency (smooth, not crunchy), though this depends a lot on what filler you're using. With my grainy sawdust it was not as smooth as peanut butter, but it had a similar thickness - it is more of a paste than a liquid in the mixing cup. After a few goes you get a hang of it; if you get it just right the fillets almost make themselves - they're thick enough to stay where you put them, but thin enough to flow out slightly and leave a fairly smooth finish. If you find the epoxy is dripping, it's too thin, and if it's grainy like wet sand, it's too thick. And don't mix up more than you need - the filler makes it go a lot further than you might have thought.

It's also really, really important to mask things off carefully before getting epoxy near the frame. If you get it in the BB threads, it'll be a nightmare to get out. You should also mask off the parts of the bamboo that won't end up part of the joints, but I forgot at first. Later on I used clingfilm between the bands of masking tape to create nice sharp edges to the joints.

Once your mixture is at the right consistency, you want to work it into the joints and round them off nicely. It's such a fiddly process that it's impossible to properly describe it, but using a rounded flat spreader you can slap on the epoxy, then carefully work it in to the right shape - a light touch and having made the mixture just the right consistency are the key requirements for success. Build up a little bit at a time, and don't obsess about getting it just right. Once I was done, I left it in the sun to go quite hard (an hour or so perhaps, but it varies so much), then moulded it into exactly the right shape with wet fingers. When it's nearly touch-dry, it can be worked without getting you all sticky as long as you keep your fingers damp. It's also worth saying that although I do this without gloves on and am fine, if you have sensitive skin or don't know you're not allergic to epoxy resin, you should wear gloves.

ST/TT/SS and ST/DT/CS joints filleted by comatus1649, on Flickr

**Setting the headtube
**This is the happy stage when it really begins to look like a bike frame. Using a protractor, I set the frame in the jig so that the headtube angle was vertical, and marked where I had to cut using a double-ended plumb bob:

Getting the HT angle right by comatus1649, on Flickr

Using this I knew where the deepest point of the mitre had to be on both tubes, and could carefully cut from the end of the tube to this point. Then it was a simple but laborious matter of cutting away the excess until the mitre was tight and even. For this I used a rotary tool with a sanding drum - I didn't use it much on the other joints because it's noisy and creates a lot of mess, but here it was worthwhile to avoid the risk of jamming the rasp and breaking one of the joints. It's also a lot easier to use a power tool than carefully rasp two mitres at once, especially when it's nearly impossible to clamp it in a vice. To finetune the mitres, I wrapped a bit of emery cloth around the HT, and used it to perfect the mitres.

When it was about right, I set up the jig in configuration 4, which uses the part previously used to hold the rear axle to hold a channel vertical, to which the HT is attached. With the jig set up so that the HT has to be vertical, its alignment on both planes can be checked with a spirit level rather than faffing around with a protractor. This join also required a fair bit of holding in place because I hadn't glued my jig together very well and it was all starting to come apart, so there was a lot of play.

Setting the HT from another angle by comatus1649, on Flickr

Final filleting
With all the poles in place, I could now finish the filleting and at last have a fairly robust frame shaped object which I didn't have to worry about breaking. The ends of the CS and SS were much bigger than the dropout tabs, so I wanted to cut them down a bit to make it more streamlined. This was done with some very careful saw-work.

Trimming the CS and SS by comatus1649, on Flickr
After this picture was taken, I also removed the visible expanding foam so that I could plug the gap entirely with epoxy. This should give me a really strong join between bamboo and metal, even before the reinforcement.

Filleting the HT and the dropout intersections was much easier than the previous few joints, partly because I was getting better at it and the joints were simpler, and partly because the mix was slightly thicker and much easier to work with. In general I think you should err towards a thicker filleting mixture - it's easier to work with and lighter (filleting can add a lot of weight to the frame).

At the moment, the frame is out in the garden curing. The thickened epoxy takes a really long time to cure, so don't rush on to the next stage if you've got all the parts to hand. The one possible exception to this is that there's a technique where you lay fibreglass on top of still-wet fillets, but I think that would be difficult to do on the complex joints on a bike. As it happens, I need to get some fibreglass from a friend so I can isolate the aluminium BB shell from the carbon fibre to avoid galvanic corrosion, so can't go much further at the moment. I start a residential job for six weeks next week, and I expect I probably won't be able to work any more on the bike until that's over. In the mean time I'll still be able to check the forums and answer any questions people might have...

Ben

If you read up on bamboo framebuilding, sooner or later you'll come across Craig Calfee. As well as his better-known carbon frames, he also makes some great bamboo frames, and his sideline Bamboosero sets up framebuilders in Ghana.

I saw some of these bikes in Ghana a few months back - there was a row of them outside a shop/workshop. Unfortunately, it was locked up (the bikes weren't tho!) so I couldn't have a chat to the staff about them. Took a few pictures tho. They looked pretty sturdy.

They're nice bits of work, overbuilt but very functional. Pretty ideal for their function, I guess. They're built using quite a clever PVC tubing jig - they're built off a table, on their side, rather than in the more normal orientation for building bikes, which makes using very cheap plastic tubing a possibility. A lot of thought has really gone into them.

The first bits of CF went on in the last couple of days - I've been working away so haven't had time to do much. Will post pictures when I have time to take good enough ones - it's hard to get them when you're covered in epoxy.

Reflections so far - unidirectional tape is much easier to handle than tow, and the thin strands of fibreglass used to hold the tape together provide a very useful visual guide of whether the carbon is properly wetted out (they go virtually clear). Peel ply tape is also very useful, though it is a bit difficult to get it tight on complex joins - it's 50mm thick and doesn't stretch. Perforated electrical tape over the top of peel ply would probably give you the best of both worlds, as the PP is very good at making sure you use just the right amount of epoxy and get super bonds between layers, and the electrical tape would compress it even better. I'm slightly concerned that using just vinyl tape would not encourage strong lamination as you end up with a thin layer of smooth epoxy on top of each layer of carbon. PP sucks out excess epoxy (though in practice you generally have to add more to wet out the PP fully anyway) and leaves a great texturised finish.

For reference, you want appropriately 11-12 grams of epoxy to 10 grams of carbon to wet it out fully. Considering the density of epoxy is about 1.1-1.4 g/cm^2, a reasonable rule of thumb is a millilitre of epoxy to a gram of carbon; 25mm 200gsm UD tape is 5g per metre. Add a little bit for pot/brush wastage.

This is one cool thread! Keep up the good work.

GETTING STICKY - CARBON FIBRE TIME

3 weeks later, I really should write a proper update. I haven't had as much time to work on it as I had hoped so progress is slow, but I have a few things to say further to my initial reflections.

Surface prep
It's obviously very important to get strong bonds between the composite joints and the bamboo or metal parts, and the first stage in this is effective preparation. The shiny outer skin of the bamboo has to be stripped off if it hasn't been already, and the surface abraded using coarse sandpaper or a file.

Likewise, the metal parts need to be roughed up with a file and cleaned thoroughly. Most things that degrease would be fine - water and washing up liquid (leave to dry well), alcohol, kitchen spray, chain degreaser, etc. The one thing I've come across that you really do have to avoid is white spirit - it leave residue which prevents a strong bond. I also sanded down the fillets to make them smooth and have a gradual transition into the bamboo, which will avoid having voids filled with lots of epoxy and little carbon fibre. Similarly if after putting on a layer of CF there are gaps left, it is sometimes worthwhile sanding them down flat, but there are no hard and fast rules.

First layup - headtube joint
Because it is the simplest joint on the bike, I decided to begin with the HT/TT/DT joint. It is also the most visible, so if you're not happy with your abilities then do some practice joints on scrap bamboo first (it would be very instructive to make some mock joints, then cut them up to see how good they are internally, like learning how to braze lugs). Also consider having a dry run of wrapping the tape without any resin.

This is the most technical part of building a bike, and as I'm only on my second (and indeed my first CF one) I don't really feel qualified to give detailed instructions. There is a lot of information out there - in particular I recommend this post on a very helpful blog, which links to a wrapping guide. You have to modify it slightly if using tape as I have done, or more significantly for joints which are totally different to the norm - I had to play it by ear entirely for the seat cluster / wishbone joint.

In brief, the principles of bike building wet layup are pretty straightforward. Mix up your epoxy as per the manufacturer's instructions, making sure you have enough to finish the job. Using a flat roller tray rather than a pot makes it go off slightly slower. Mix and mix and mix some more because it it's not properly mixed it won't work at all, and you won't realise until it's too late. Decide what wrap you're going to do, and cut a suitable length of tape or tow (you may want to get special composite shears) before getting all sticky. An arms' span was about as much as I could easily handle at once. Also cut about the same length of peel ply tape. Before you go on, think about what you're wearing - I know it sounds stupid but a) epoxy never comes out of anything and b) I get disproportionately hot and sweaty while laying up carbon, and it's a bit late to change when you're all covered in resin. You will probably want to wear gloves at this point as well.

Now wet out the area on the frame around which the tape will wrap. Don't worry about going overboard here as it will just drip off if you use too much, but you can't lay carbon fibre onto a dry surface and wet it out afterwards. Anchor the tape as mentioned in the wrapping guide by wrapping it over itself and pulling it taut. There's no need to pull the tape mega tight as you wrap it around, but don't let it go slack. Whenever tape crosses over itself the lower bit has to be fully wet out first - you can either wet out all the tape as you go along, or (my preferred, less messy option), only wet it out at the end or when you have to go over a part. A foam (AKA Jenny) brush helps to work the resin deep into the weave of the tape - when it's about right, the polyester threads holding the unidirectional tape go almost transparent.

When all the epoxy is applied and the carbon is fully wetted out, wrap peel ply tape tightly around the joint to press it flat and squeeze out any excess. If necessary, add more epoxy to wet out the peel ply. Make sure you leave the end of the peel ply un-stuck down so you can peel it off later. You'll be left with something that looks like this:

HT wrapped in peel ply by comatus1649, on Flickr

And after leaving it for 8 hours and removing the peel ply (which doesn't properly adhere to the epoxy, you can build up the next layer. Because the tape is thick, I only had to do probably 3 wraps (of about 2m each) to complete the HT joint. After removing the last lot of peel ply, I painted more epoxy on top and let it cure into shiny prettiness:

Pre-sanded HT joint by comatus1649, on Flickr

... then sanded it down to make it nice and smooth for the skinning process to come:

Sanded HT joint by comatus1649, on Flickr

Seat cluster joint
The principles of the other joints are just the same - at the time of writing I've 80% finished the seat cluster joint, and not started the BB joint. Both are rather more complicated than the HT joint, but being methodical, thinking about what direction the joint needs to be strong in, and following the wrapping guide make the process not too bad. I was soon left with something that looked a bit like this:

Seat cluster from another angle by comatus1649, on Flickr

Most of the whiteness in the picture is from the flash, but as a matter of fact this last wrap was slightly under-saturated with epoxy, and the white polyester binding can see be seen in real life. As long as each layer is wet out full before the next one is put on top, this isn't too much of a problem - I can put more epoxy on top and it will soak in, but it is easier to get it right first time. The last thing you want are voids of unwetted carbon (or for that matter bits of epoxy with no carbon fibre) within the joint to undermine the fantastic composite properties of cured carbon fibre that we're aiming for.

Dropouts
As mentioned above, I cut away the excess foam in the dropouts and plugged the ends with thickened epoxy so the dropouts were firmly wedged in place. This meant that I didn't have to get too technical about wrapping the dropouts with the fibres running in the right directions - just wrapping tape a couple of times around the joint will strengthen the bamboo enough that the epoxy/metal plug will never be able to break out. This was much, much easier than my technique on my first bike where I needed to reinforce an initially weak joint. It never worked and broke twice which, I don't mind telling you, is quite disconcerting when you go down a curb. Because each of the 4 dropout joints only use a little bit of epoxy and maybe 4-6 inches of tape per layer (I might do a double layer), this is also a good way to use up the little bit that's often left over after you've done a bigger joint.

Test fitting sneak peek
So despite the fact that I took obsessive measurements and checked them numerous times, I've been worried since the beginning that I'd fucked something up and the wheels wouldn't fit. So to set my mind at rest (even though it was much too late to actually do anything non-drastic about it), I took the wheels off my other bike and gave it a quick go. The headset's not in yet, of course, so the front end will rise by 15mm or so, but it gave me a quick idea of how it might look. It also proved that there is plenty of room for the 32c tyres I have at the moment so the 35mm Kojaks I plan to use should be great. I reckon I could probably fit up to 50mm tyres in there if I was careful, which might be fun if I can find a light slick set (fat treadless tyres are the mutt's nuts). I'm liking the look of the Kinesis DC37...

Mockup 1 by comatus1649, on Flickr

That's all for now. I'm getting a bit bored of wrapping lots of tape, so I might skin the headtube next... I need to get some fibreglass from a friend before I can give the BB joints a go.

Ben