Custom 3D Printed Carbon Fiber Bike Frame

If you’re really into cycling and are on this site, chances are that over the years you’ve at least toyed with the idea of making your own frame. There’s something truly special and deeply rewarding about riding something that you’ve built yourself, and being able to choose whatever custom geometry and extras that you want only adds to the appeal.

At that point you probably looked at everything that you needed to do to make a bike. And that’s when things start to get daunting. Before you’ve even thought of the bike, you need to think about the jig. Fabricating a jig with basic versatility and functionality for framebuilding takes time and patience, and the raw materials costs are at least $200 [Instructable example!]. Rapid-prototyping has pushed the costs of pre-made bike jigs from $1000s down to $300 or so [such as the Jiggernaut], but if you don’t know how many frames you really want to make, it’s still money that you aren’t putting towards your first frame.

Next, you need to consider the tools that are specific to typical framebuilding. If you want to learn to braze steel frames, an oxy-acetylene starter kit with everything you need to get brazing is something over $200 (plus the price of an additional gas cylinders, brazing rods, and flux that you’ll use). A TIG welder suitable for working with thin-walled steel and aluminum bicycle tubing is more like $1000.

Next, the materials for the bicycle frame itself. While it would be possible to braze a frame with commodity straight-gauge 4130 (chromoly) steel tubing, it would be very heavy and wouldn’t ride very well - bike specific tubesets are double- or triple-butted to save weight and place extra material at the lug junctions. Bicycle specific tubesets will cost you upwards of $100, and a set of lugs and dropouts is about another $100.

If you’ve been adding up the tally, you’re already at more than $600 for your first frame – hopefully you won’t make any mistakes!

I wanted to design a process for building bicycles that allowed an enthusiast to spend less overhead, less time on finicky details, and put the emphasis on actually designing and making a bike that you want to ride.

Step 1: The Process

The process that I came up with takes advantage of the growing availability and affordability of CAD and 3D printing to allow people to build themselves a unique custom bicycle with unparalleled design flexibility. 3D-printed socket-style lugs are used to join commodity tubestock, and those joints are then reinforced with carbon fiber and epoxy. Since the rigid 3D-printed lugs define all the angles and the sockets provide a hard stop for the ends of the tubes, once the tubing is cut to the appropriate length all the pieces simply snap together, removing the need for a jig to accurately maintain the frame’s geometry during construction. 3D printed pieces can also be used to create 2-part molds for the carbon fiber lug reinforcements, improving both strength and appearance.

In addition to eliminating the jig, this method of construction has some major benefits. Since the design is generated in CAD, it lets you create whatever geometry you want, rather than using standard preset angles that brazed lug construction requires. Whether you want road, mountain, track, even cargo or recumbent geometry, the freedom is there. In addition, since the tubes are bonded together with epoxy and carbon fiber, you can use whatever material you’d like for the tubing – aluminum, steel, carbon fiber, titanium, or bamboo! My goal was to shift away from being restricted by the materials required by the tools, and instead enable you to realize your personal creative vision for what you want your bike to be.

Disclaimer: While I'm not a composites engineer or a framebuilder, I'm an industrial designer working in the aerospace industry who has prototyped carbon fiber aircraft parts and has read an awful lot of composite manufacturing theory. Still, when following these instructions and building your frame, think critically and safely and design conservatively.

Step 2: What You Need

Absolutely essential:
• Dimension-based CAD software
• 3D printed lug joints
• Tube stock (metal, carbon fiber, even bamboo)
• Bicycle-specific parts (bottom bracket shell, brake bosses, dropouts, headtube w/ correct ID)
• Cutting tools (pipecutter or hacksaw w/ miter box)
• Carbon fiber cloth
• Fiberglass cloth
• Fabric cutters (rotary cutter or high quality shears)
• Laminating epoxy (e.g. West System, MAS, Aeropoxy)
• 5-minute epoxy
• Mixing pots and stir sticks
• Epoxy squeegee (e.g. scrap piece of sheet metal)
• Sandpaper (variety of grits)
• Vinyl electrical tape
• Acetone / 99% isopropyl alcohol
• Nitrile or vinyl disposable gloves (lots!)
• Chemical cartridge respirator
• Safety glasses

Really handy:
• Digital scale & cheap calculator OR metering pumps for epoxy
• Rotary (Dremel) tool
• Utility knife
• Mini hacksaw
• Assorted files
• Sandable fairing compound
• Disposable razor blades
• Plastic release film
• Vernier calipers (especially digital)
• Disposable painter’s coveralls (for sanding)

For 2-part moulding (optional):
• Mould release wax (carnauba, car detailing supplies work)
• Extra hold hairspray (for mould release)
• Clamps
• Hammer and chisel (for persuasion)

Figure on spending between $250-500, depending on where you source your materials, what kind of tools and safety equipment you already have, where you’re getting your 3D printed parts, and if you want to mould your lugs or not. Having access to a friend or local makerspace with a desktop 3D printer will save a lot of money over using commercial 3D printing services, especially since neither the finish quality nor structural integrity of the printed parts are critically important to the build.

Step 3: The Bike

For this project I wanted to make something unique that would showcase the versatility of the manufacturing process – after all, if I’m going to the trouble of making my own custom frame, why make something I could buy off the shelf?

I’d always been curious about crank-forward geometry (essentially semi-recumbents, diamond frames with extremely slack seat tube angles) that allow you to efficiently pedal while also allowing you to put your feet down at a stop. No one really sells a light, stiff  semi-recumbent with components targeted at enthusiasts, so that goal was the seed for the frame’s design.

The first step in designing your bike is deciding what you want it to be. Fast and twitchy, stable and aerodynamic, sturdy and versatile, upright and relaxed? These traits and the riding position determine the tube angles and the kinds of components that you want your bike to accommodate.

BikeCAD is a great way to experiment with different designs, and uses a free browser based applet. If you aren’t interested in delving into the intricacies of geometry design, copying a bike you know that you like is the easiest and fastest way to succeed.

Designing a bicycle in all its detail is a fairly in-depth endeavour and beyond the scope of this Instructable, but for the CAD modelling process and manufacturing that follow, you’ll need to decide:
1. Tube angles and fork geometry
2. Wheel size
3. Dropout spacing (e.g. 120 mm for track bikes, 130 mm for road bikes, 135 mm for mountain/touring bikes, or 170 mm for fat bikes)

Once you have these basics, you can lay out your CAD model.

Step 4: The Materials

For my frame, I decided to use 6061-T6 aluminum tubing. It is lightweight, corrosion resistant, readily available and inexpensive, and well-proven in the cycling industry.

My choice of tubing diameters and wall thicknesses was based largely on precedent from other bicycles, and the commodity tubing that was available (I bought mine from Aircraft Spruce). My main tubes are 1.5” OD x 0.035” (0.9mm), the seat tube is 1.375” x 0.058” to provide an ID of 32mm that allowed the use of a standard oversized seatpost, and a 34.9mm seatpost clamp. The chainstays are 1” x 0.035”, and the seatstays are 0.75” x 0.035”. While these are straight-gauge tubes, the wall thickness is the same as the thin middle section on butted bicycle tubing – I don’t need the extra thickness at the joints because I’m not introducing heat to join them, and I’ll be reinforcing them with carbon fiber anyway.

Make sure that you do your research! My bicycle is going to be used for city riding by a 160 lb rider. These wall thicknesses are not appropriate for mountain biking or heavier riders. Obviously, using different materials requires totally different tubesets as well.

Tubeset manufacturers like Colombus, Dedacciai, Henry James and Nova Cycle Frames are good resources to find out what wall thicknesses are appropriate for different materials, applications, and locations on your bike. If you’re trying something new and aren’t an engineer, be conservative (i.e. bigger tubes and thicker walls) – your safety is worth more than a few grams of extra weight!

The carbon fiber I used was primarily 0.012" thick, 4.7 oz./sq. yd. unidirectional graphite tape (it comes on a 5” wide roll), 11 oz./sq. yd. 8HS satin weave fiber (for maximum conformability), and some 5.7 oz./sq. yd. twill weave, for easier handling with smaller pieces. I also used a small amount of 12K carbon fiber tow. The fiberglass I used is extremely light .004" thick crowfoot weave (4HS satin). I also purchased this through Aircraft Spruce, but these materials are available through many online suppliers.

I used West System 105 Resin and 206 Slow Hardener for my layups. It is designed for the home builder and thus is fairly forgiving as epoxies go, and is also less toxic / lower odour than some alternatives. That said, you should always take precautions to minimize your exposure to the uncured epoxy.

For ease of assembly, I used a stock head tube, bottom bracket shell, dropouts, and V-brake bosses rather than fabricating them myself. You can buy these parts individually, but I liberated mine (read: chopped them off with an angle grinder) from suitable broken donor bicycles at my local DIY bike repair space in Toronto, BikeSauce. If you’re okay with using old bits on your new bike, this is a great way to recycle!

Step 5: The CAD

Once you’ve decided on the design of your bike, you’ll need to create it in your favourite CAD package. Parametric, dimension-based solid modellers like Solidworks, Autodesk Inventor, or Pro/ENGINEER are the easiest to work with for this purpose, but there are lots of options that can accomplish what you need and export to 3D-printable formats.

If you don’t have a CAD modeller at your disposal already, I would highly recommend trying Autodesk’s surprisingly powerful (and free!) 123D. It has all the basic functionality you need for this process (sketching, extruding, sweeping, filleting, splitting, and shelling).

Modelling the parts to be 3D printed requires reasonable proficiency with CAD, but it’s certainly not beyond the ability for enthusiastic amateurs to learn for this project, so don’t be discouraged. I’ll describe the basic modelling process that I used but the details will be specific to your CAD application of choice. I used Solidworks to model this, so that’s the shorthand that I’ll be referring to for the basic operations.

1. Draw 2D sketches that lay out all the tubes and other key dimensions (such as dropout spacing)
2. For tubes that don’t fall on XYZ planes (seatstays and chainstays), project 3D sketches
3. Create transition sketches as necessary for curve sections e.g. the semi-circular paths for “wishbone” stays.*
4. ‘Extrude’ or ‘Sweep’ along your sketches to create the main tubes.
5. ‘Combine’ all your extrusions together, and create ‘Fillets’ at the tube junctions to create the transitions that will make carbon fiber moulding easier. Large, gentle radii are your friend.
6. ‘Split’ apart the lug junctions from your main tubes.
7. ‘Shell’ the tubes to the correct wall thickness.
8. ‘Extrude’ a cylindrical plug from your lug junctions that will fit the I.D. of your tubing. The tolerances on these plugs need to be tight so that it will fit snugly into your tubes for accurate self-jigging of the frame.
9. ‘Shell’ the lug sockets to about ~2mm so that they use less plastic. This saves weight and money!
10. You may need to ‘Split’ up your lug sockets into two pieces depending on the build size capabilities of the 3D printer you’re going to be using.
11. ‘Export’ your lugs as separate parts in a format that can be read by 3D printers (commonly .STL)
(12). In my case, because I was using dropouts from a donor bicycle I cut apart, I also needed to reverse engineer these in my CAD model. Integrating dropouts into your model is much easier and more accurate if you’re using stock flat plate dropouts or cutting your own dropouts.

*N.B. I separated my bottom bracket region into two separate parts - the junction of the seat tube, down tube, and bottom bracket shell became one, and a wishbone part that created the angles for the chainstays. This facilitated two-part moulding of the carbon fiber, since I only needed one parting line.

Step 6: Bonus CAD: the Moulds

If you’ve decided to use the 2-part moulding method to lay up your carbon fiber, there are a few extra modelling steps you need to complete in order to create your moulds.

1. ‘Thicken’ a body around your lug sockets and the first 1-2” of all the frame tubes that join to it. The thickness of this body defines the wall thickness of your carbon fiber. I used a uniform 2.5mm wall.
2. ‘Extrude’ a flange at the parting line. This will create a flat gusset in between the tube joints, providing a huge amount of bonding area for your carbon fiber parts, and adding stiffness and strength. ‘Combine’ this flange with your first body. This gusset is a total of 2.5mm thick, meaning you need to use 1.25mm thick carbon fiber per mould side.
3. ‘Extrude’ a box around your entire junction area. This is the base part of the mould.
4. ‘Subtract’ the first body from the mould box. You’ll end up with two blocky parts that surround your frame joint.
5. ‘Extrude’ guidance/alignment tabs to one side of the mould. This will ensure the two pieces clamp together straight.
6. ‘Split’ your parts as necessary to accommodate the 3D printer.
7. ‘Shell’ your moulds to reduce the amount of plastic that you’ll need to print (greatly reducing the print time and cost). You’ll be reinforcing this with filler later so that they are still strong enough to clamp.
8. ‘Export’ for printing.

If you designed your bike with moulding in mind, almost all of the lugs can be created with a simple 2-part female mould. However, due to the bottom bracket shell, your BB lug usually requires 3 moulds (2 to form around the frame tubes and the top of the BB shell, and 1 to form around the bottom of the BB shell).

Step 7: The 3D Printed Parts

A large part of the design intent with this frame was to use the 3D-printed lugs to remove the need for a jig. For this reason, mine are printed out of ABS (using the FDM process). Desktop 3D printers are the easiest to use for this, especially since the part itself isn’t cosmetic. Just remember if you’re outsourcing to a printing service like Shapeways, make sure that you are using a rigid material, not something like SLS polyamide – it won’t have the dimensional stability you need unless you use a very high wall thickness.

The prints for my parts and moulds came from an Up! 3D printer (pp3dp.com). The resolution, part quality, accuracy and ability to print support material very easily made it perfectly suitable for the needs of this project. I can’t comment on the capabilities of other consumer / hobbyist desktop 3D printers, but the latest crop of products on the market from a variety of manufactuers (MakerBot, RepRap variants, Ultimaker, etc…) should also be capable of outputting suitable parts.

Step 8: Introduction to Composites

For those of you who already know about working with fiber-reinforced plastics (composites), feel free to skip this section. But even those who are familiar with fiberglass and polyester resins should note that carbon fiber is a slightly different animal. The basic technique of using long-strand fabric to reinforce plastic resin is the same, but carbon’s properties are slightly more complicated.

1) Carbon fiber is only strong in the direction that the fibers are running. It is 10x stronger when oriented with the fibers than across them. This property is known as being “anisotropic” (unlike metals, with uniform strength in all directions, which are isotropic). Being anisotropic allows the designers to put the strength in a carbon fiber part only where they need it. This saves weight, but requires that you know exactly what stresses the part will be subjected to! If you don’t design for all the forces experienced, your part could fail. See the attached picture to see how sharply the modulus (stiffness) of carbon fiber drops off when the load is not oriented with the fiber.

2) Composite parts are made up of a number of layers called plies. The number of plies used and the direction determines the strength of the part.
a. 0 degree plies (with the fibers running along the length of a tube) are strongest at resisting bending, tensile and compressive loads. Bike tubes have most of their fibers running in this direction.
b. 90 degree plies (running circumferentially around a tube) are used to provide hoop strength, resisting crushing. This is the least important direction on a bicycle.
c. +-45 degree plies are the best at resisting torsional (twisting) and shear forces. This is important, especially at junctions like the bottom bracket.

3) Carbon fiber comes in woven cloth and in unidirectional tapes. The woven cloth has fibers running at 0/90 degrees perpendicular to each other. Most bicycles are made primarily of unidirectional fabric with a cosmetic “insurance” ply of woven fibers.

4) In practice, most parts will be required to resist some combination of all of the above forces. For example, a part with 8 unidirectional plies laid up @ 0, 90, +45, -45, -45, +45, 90, 0 degrees will be able to resist most forces equally. This is what is called a “quasi-isotropic” layup, since it approximates the way a metal part would work. For very complicated junctions like the bottom bracket lug where there are a lot of forces at work that are difficult to predict, this kind of layup will be heavier, but safer.
5) Strength comes from continuous fibers, which create a direct load path through your part. Be careful when sanding your part – if you sand down a lumpy layup for the sake of aesthetics, you may be severing the long-strand carbon fibers and compromising its integrity.

6) You can’t push on a rope. Carbon fiber is just a fancy fabric, so while it can take pulling (tensile) loads, it can’t take pushing (compressive) loads without some help. This is where the epoxy resin comes in. The epoxy is the matrix that holds all the carbon fibers in place and turns the individual layers into a strong, unified whole.

7) When laying up your carbon fiber, you want to make sure that each of the thousands of individual strands have been wetted out with epoxy. If your layup has too much epoxy, the only disadvantage is that it will be heavier (and a bit more brittle). If the fibers have not been properly saturated, it won’t be strong at all.

8) Only use epoxy resin. While more expensive than polyester, it is significantly stronger, less brittle, and creates much stronger bonds with the carbon fibers. It is also much less toxic and doesn’t smell like death.

9) Compression is important. Carbon fibers are stiff and will have a tendency to unfold from any curved surface you’ve stuck them on. The individual layers also won’t bond together into a strong uniform part. The key to preventing this is to compress the fibers together. This can be accomplished with a two-part mould, by vacuum-bagging, or simply by wrapping the part in vinyl electrical tape (sticky-side out) to compress it all together.

10) Carbon fiber fails catastrophically. Unlike steel - and to a much lesser extent, aluminum - which will bend when they fail, carbon fiber parts will snap. This can be sudden, and without warning, which can be dangerous if you’re riding! In practice this is not a huge problem, because a carbon fiber part that weighs as much as a steel or aluminum one will be much stronger, but just realize that unless you are an engineer, it’s better to resist the urge to make your carbon fiber joints superlight (at least for your first frame). A little bit of extra weight will go a long way for reliability and peace of mind.

Step 9: Make It Real

Once you have your 3D printed joints and tube-stock, carefully check the dimensions that you need to cut your tubing down to. Being really accurate at this stage will pay off later. The easiest way to cut metal tubing is with a pipecutter, as it will produce a clean, 90 degree cut. Measure twice, use plenty of cutting fluid and be patient. Also take note that the pipecutter can swage the cut end of your tube a tiny bit, so you may need to file out the inner diameter a bit to maintain a perfect friction fit with your 3D printed parts.

Do a test fit with all your pieces first. Everything should fit together tightly – play between the tubes and the sockets can lead to inaccuracy. If everything fits as it should, it’s time to start assembly.

Observe the surface prep rules – solvent wipe, rough sand, then solvent wipe again. Let your solvent flash off for a minute, and then mix up a batch of 5 minute epoxy. The 5-minute stuff isn’t nearly as strong as the longer cure-time laminating epoxies, but since your 3D printed plastic sockets aren’t actually structural, it’s more than good enough to tack them in place. The gluing process is neater if you smear the epoxy on the inside of the tube rather than the outside of the socket, since it won’t squeeze out.

It’s easiest to keep the whole bike assembled and glue one joint at a time, waiting for it to set in between. That way you can ensure that the whole bike is still perfectly aligned and that each tube is glued square in its socket, perfectly flush.

Note that I have designed my seat-tube lug socket so that the metal tube is actually continuous and slides through the joint, rather than butting against a socket edge the way the other joints are constructed. Make sure you glue it in place at the correct height.

N.B. Make sure your bottom bracket shell is oriented the right way when you glue it in! Mark the drive side so you don’t turn it around by mistake.

The dropouts are the trickiest area  - if there’s a chance for an alignment mistake to happen, it’s probably going to happen here. Make sure that you have a spacer clamped into your dropouts, since a bike’s rear triangle is pretty flexible when a wheel isn’t in place. This part is much easier if you’ve designed your CAD sockets around flat plate dropouts as it makes angular misalignment from faulty measurements much less likely. If you did it my way using donor parts that you’ve reverse engineered in 3D and have a little bit of play, clamp your wheel into the dropouts when you’re gluing them up. Having the wheel makes any slight angular misalignment very visible and easy to correct.

Once you’ve finished gluing in the dropouts, heave a sigh of relief, and admire your handiwork. It already looks like a bike, now you just need to reinforce it.

Step 10: Surface Preparation

When working with carbon fiber composites (or any adhesive), surface preparation is key. Most important is keeping everything clean and free from contamination. You also want to make sure that all the surfaces you’ll be bonding to are mechanically ‘keyed’ to promote a stronger joint – this means coarse sanding (under 100 grit), filing, or knurling to rough up the surface.

Here is a general method to follow when bonding to any surface:
1) Remove any dirt or other big contaminants if necessary. Use soap, water and elbow grease.
2) Degrease with acetone. Acetone contains no water (unlike alcohol) and the solvents flash off very quickly.
3) Rough up the surface with 100 grit sandpaper.
4) Final clean with acetone to remove any sanding dust and leave a clean, bondable surface.

If you are using epoxy and bonding with metal, add a final step.
5) After you’ve mixed up your epoxy, rub some onto your metal part and ‘wet sand’ the epoxy into the surface of the metal with 80 grit sandpaper. This removes the surface oxide layer that rapidly forms in just a few seconds and allows the epoxy to bond to the base metal, making for a stronger and longer-lasting connection.

Step 11: General Composites Safety

Epoxy is much less toxic than the polyester resin normally used for autobody repair or fiberglass canoes, but it’s still a good idea to take safety precautions as excessive exposure to epoxy can lead to nasty sensitization reactions.

Work in a ventilated area, wear a cartridge respirator, safety glasses, and nitrile (or vinyl) gloves. Latex isn’t sufficient. And if you do get any epoxy on your skin, do NOT clean it off with acetone. This gives the epoxy a highway into your circulatory system! Use 99% isopropyl (from the drug store) or denatured alcohol instead. Even plain white vinegar can effectively be used, and is much safer.

I like to wear two pairs of gloves when I'm doing layups, so that I can throw out the sticky outer pair when I need to handle other tools and not risk touching uncured epoxy somewhere on my workspace. It works best if  the inner ones are some high quality snug fitting mechanic-style nitrile gloves, with cheap looser disposable vinyl gloves over top, so that you can change them frequently.

Step 12: The Carbon Fiber

The dropouts are the easiest joints by far, so they make a good place to start and work on your technique.

The actual layup is always a little bit of a rush to make sure that you’ve wet up all your plies within the pot life of the mixed epoxy, so make everything as easy for yourself as possible ahead of time. I used West System 206 Slow Hardener to give myself a longer working time (it also has a higher tensile elongation than the 205 Fast Hardener, which makes the bonds to the metal slightly less brittle).

Lay out a sheet of plastic film and tape it down on your work table to do your layup so the epoxy won’t stick. Think about the joint that you’re working on, and the stresses it will see. Measure and cut out all the plies that you are going to need and spread them out on your work table, in the order that you’re going to use them. Make sure that you use a sheet of fiberglass to isolate the metal from the carbon as your first layer. For the dropouts, the order of plies in my laminate (called the layup schedule) was:
1. 0/90 (degree) Crowfoot weave fiberglass
2. 0 (along the axis of the tube) unidirectional carbon fiber (UD CF)
3. +45 UD CF
4. 0 UD CF
5. -45 UD CF
6. 0 UD CF

Plan on doing one joint at a time, and mix up only enough epoxy to do this to avoid waste. Make sure you’ve prepped the surface you’re gluing to. A clean, rough bond is a strong bond. If you’re wondering how much epoxy you should mix and have a precise scale (I use a $20 digital kitchen scale), you can measure the weight of the carbon plies you’re going to lay up and use it as a guideline. The correct ratio for a strong carbon fiber part is about 60:40 carbon to resin by weight, but a lot of your epoxy is going to end up on the table, on your gloves, or squeezed out onto the table after you compress your part, so I usually mix about 3:1 resin to carbon. For instance, for a dropout, I had about 9g of carbon, and needed about 30g of resin. I then add my hardener (6g) and mixed it up. If you don’t have the handy pump dispensers to properly meter your resin, having the cheap digital scale and a cheap, large-buttoned calculator make life much easier. The cheapest way is just to measure water out in the correct resin-to-hardener proportions and pour it into a paper cup, measuring on the side the level. Use that Master Cup to copy the markings onto your other mixing cups.

Smear a little bit of the epoxy onto the dropouts and take some 100 grit sandpaper and wet sand the epoxy into it.

Now lay out your first piece of fiberglass, and pour a small amount of epoxy into the center. Using a squeegee (a scrap of plastic, or in my case, aluminium sheet), spread it evenly. Flip it over and repeat the process. With fiberglass, you’ll know it’s wetted out properly when it turns transparent. Now gently scrape the cloth with your squeegee to remove excess resin.

Wrap it around your dropout and make sure that it is lying flat and neatly. If it’s bunched somewhere, give it a bit of a slice with a disposable razor blade.

Repeat the process with the carbon fiber. Carbon is less pliable (especially unidirectional), and is harder to tell when you’ve wetted it out properly. Work the epoxy through the cloth on both sides, make sure the whole ply is glistening, then scrape off excess globbing. Wrap your plies around the joint, overlapping them so that it is uniform.

For your biased plies (e.g. +-45 degrees), cut the ends at an angle so that they will finish flush and wrap smoothly.

Once you’ve done your layup, allow some time (depending on your hardener, 10-20 minutes is usually sufficient) for the resin to gel, and then compress it by wrapping it with vinyl electrical tape, with the sticky side facing out. Wrap evenly and with uniform pressure – if you pull too hard you’ll get deep ridges in your part. I also drilled holes in my roll of tape so that excess epoxy could squeeze out better.

Allow the part to cure for the required time (usually 12 hours or so) before you unwrap the compression tape.

Step 13: The Rest of the Joints

It is completely possible to do the rest of the joints the same way as you did the dropouts, laying up your carbon fiber and then using a compression-wrap to compact the joint. However, many of the joints on a bike are much more complicated, with multiple tubes meeting at the junction, complicating the wrap. Generous fillets will help, but if you use this method, you are probably best off using unidirectional carbon fiber only on flatter, smoother transitions to prevent bunching. Woven cloth (especially ‘twill’ and ‘satin weaves’) conform to compound curves much better and will help you get a superior part both cosmetically and structurally.

One disadvantage of the hand layup / compression wrap method is that it tends to be fairly labour intensive. It is more difficult to achieve a uniform wall thickness of carbon on your part, and there will almost always be a fair bit of sanding required to get a good finish. Sanding carbon fiber is a pain, and also breaks the long strand fibers that make the part strong.

I wanted to take full advantage of 3D printing technology for this project, and decided that I would create 2-part female moulds to complement the male “core” sockets that were used to jig the frame together in the first step. I could then lay my carbon fiber into the moulds and clamp them over the joint in order to get a compacted, uniform joint which would require less difficult post-mould finishing.

Step 14: Making Your Moulds

Moulding your parts requires a little bit more time in advance to do the extra CAD modelling, 3D printing and prepping of the moulds, but it pays off in easier layups that are less stressful, improved aesthetics, and hopefully improved strength and durability as well due to the uniform compaction.

I’ve already covered the modelling technique in the previous section, but once you’ve printed your moulds, they still need some preparation.

1. If you needed to split your into smaller pieces to fit the 3D printer's build area, glue them together. I used ABS cement (5 minute epoxy also works just fine, and is necessary if your printed parts are a different kind of plastic like PLA).
2. Use autobody glazing / spot putty and sandpaper to smooth over the “stair-step” effect caused by the 3D printer build layers. Since this mould is the final outer surface, make it nice and smooth.
3. Flip your moulds over (to the open shell side, not the mould side). Stiffen up your moulds and give a flat surface for clamping by pouring in some filler material. I used plain old concrete because I had it lying around and it works.
4. Wipe the inside of your mould with mould release. I used Meguiar’s carnauba-based mould release wax. Wax on, wait until it cures, and wax off. Apply several wax layers, and apply it to every surface – epoxy will drip all over when you clamp your moulds in place, and you really don’t want these to stick.

Step 15: Laying Up Moulded Lugs

The process of laying up the moulded lugs requires the same preparations as wrapped lugs. Make sure your space is clean, free of clutter and contaminants, that you’ve taped down your plastic worksheet, and that all the tools you need are readily at hand. Make sure you’ve got your respirator and your gloves.

Again, plan your layup schedule, all the pieces you need, where they go, and how they’re oriented. Draw it out on a sheet of paper and tape it up on your wall for reference so you don’t get lost midway through.

The inside of your lug moulds will have some pretty odd contours, so without a little bit of trial and error it’s difficult to tell what size and shape you should be cutting your plies out in order to fit. There is software that can help you “unwrap” surfaces with compound curves and project them onto a flat template for fabric patterning, but the easiest way is probably just to use a (clean) rag and incrementally trim it until it fits neatly into your mould.

In order to make up the 2.5mm wall thickness in my moulds, my layup schedule consisted of:
1. 0/90 8HS satin carbon fiber
2. +45 UD CF
3. -45 UD CF
4. 0/90 8HS satin carbon fiber
5. 0 UD CF (not on gusset)*
6. 0 UD CF (not on gusset)*
7. 0 UD CF (not on gusset)*
8. 0/90 fiberglass

You want to transmit the bending loads through the tubing of the entire frame, so the majority of the reinforcement is in the 0 degree axis. The +-45 degree reinforcement is important to resist torsional loads, but you’ll find that I haven’t added any additional 90 degree plies, except for the fibers in my woven plies, as hoop strength is not critical in these joints (certainly not when your laminate is 2.5mm thick, anyway!).

For the +-45 degree layers, I used unidirectional fibers for my first several layups. Because the UD tape is not woven, there is no stretch in it and it does not conform to compound curves well. To prevent the tape from bunching, I cut each piece into a series of thin longitudinal strips so that I could lay them manually. For the bottom bracket, however, I decided to use the 8HS satin weave CF cloth cut on a +-45 degree bias. This made layup and handling much easier, so I would recommend it.

*The final axial plies are cut to fit around the lug itself only, rather than across the entire flat gusset area of the mould. This is necessary to build up the extra 1.25mm wall thickness required to fill the mould in these areas.

It's important to ensure that you have the right number of plies to produce the correct wall thickness of carbon fiber in your moulds, so to verify that you've calculated it right, you can pinch your plies with a set of digital calipers and see if you've got the correct thickness (or slightly thicker, since it will be compressed in the mould).

Note that because I’m laying the cloth into the mould rather than onto the frame, the order of the layers is reversed, and I put the fiberglass into the mould last. Alternatively, you can also wrap the fiberglass directly around the metal tubing on frame in advance of clamping your moulds on.

Once you have your pile of plies cut into the right shapes laid out on the table, you can begin the layup process, as described earlier. Prep your surfaces, wet sand epoxy into the metal tubing, and isolate all metal parts with a layer of fiberglass. Mix up multiple small batches of epoxy throughout the layup so that they flow and wet out the cloth well, and squeegee off excess resin to reduce weight and allow better laminate compaction. Before you put your first ply into the mould, spray mould release agent into your waxed moulds – polyvinyl alcohol is commonly used industrially, but cheap extra hold hairspray works well, too! When you are laying the pieces into the mould, make sure they conform to the mould smoothly and don’t have folds or bunches. This is tricky on the first ply, since it will stick to your gloves better than the mould, so getting it to lie properly is more difficult than with later layers.

Once you’ve laid all your plies into both of your moulds, carefully align one half of the mould with your joint and press it in place. Then place the other mould.  Use multiple clamps to evenly distribute the clamping pressure across your moulds, and tighten them all incrementally so that the moulds are parallel with a uniform gap in the middle.

Once you’ve clamped the moulds, take a razor blade (or pair of metal scissors that you can clean off with acetone afterwards) and cut off as much excess flashing around the joints as you can. Sometimes this can be more easily done after a few hours when the epoxy has gelled but not yet cured to a rigid solid.

Start with the least complicated joints to build up your confidence and familiarity with the process. I started with the chainstay wishbone, then did the head tube lug, then the seat tube cluster, and then finally ended with the bottom bracket lug, which is the most involved of the layups.

N.B. When doing the bottom bracket and headtube lugs, make sure that you don’t get epoxy inside. In both cases I filled the tubes with a thick coating of grease, placed electrical tape over any drainage holes, and inserted printed caps into the openings.

Step 16: Demoulding

After the lugs have set up (at least 12-24 hours), you can remove the moulds. I found that snapping off the alignment tabs was necessary, and that in most cases some mild persuasion with a hammer and chisel was required to free the parts. Working the chisel in between the gap on alternating corners works well.

While I had realistically only designed the moulds for 1-time use, almost all of them survived the demoulding process reasonably intact (except for the tabs) with the exception of the bottom bracket moulds.

Step 17: Cleaning Up the Lugs

Cleanup is the most anticlimactic part of the build, because at this point your frame is basically complete from a structural point of view, but there’s a fair amount of mindless time consuming labour still to go.

If you used a manual wet layup method to sculpt your lugs, large amounts of tedious sanding awaits you if you want a smooth, uniform surface. Graduated sanding by hand is usually the way to do it, even though it takes hours. Large power sanding tools generate huge amounts of heat when sanding carbon fiber which can push the epoxy resin past its glass transition temperature, so you want to be very careful if you go this route. It also generates huge amounts of sanding dust, which you do not want to breathe in. Clean your shop after every sanding session (electrostatic cloths and vacuum cleaners are useful for this).

If you moulded the lugs, even with the greatest of care there is still going to be some cleanup required. You will definitely need to cut down the gusset plate to shape unless you were extremely careful trimming your plies to shape and laying them up. I find the easiest way to do this is with the cutoff wheel of a Dremel for the initial shaping, followed with the barrel sander, and then hand filing or sanding to finish.

Removing the mould release wax deposited on your finished part can be a chore as well – automotive or household products designed for this (like Goo Gone, for instance) are helpful, as solvents like acetone flash off much too quickly to remove a layer of hardened wax easily.

For the ragged edges that are around your tubes, the Dremel also comes in handy. If you want to get a nice straight edge where your carbon meets your tubing, you can wrap a piece of tape around the tube as a guide, and then very carefully use a mini hacksaw to score a cut in the carbon fiber all the way around. Stop when you hit your fiberglass layer (you don’t want to cut into the metal tube and damage it). Once you’ve scored the flashing all the way around, a utility knife or chisel (angled so that blade will be forced up away from the tube, not dig into it!) will snap it off relatively painlessly.

Sanding the edge of your carbon fiber lugs so that there is a smooth transition between it and and the metal tubing not only looks nicer, but it also reduces stress risers at the junction between the two materials, so is a good idea.

Step 18: Extra Layups (and Avoiding the Temptation to Do Them)

If after demoulding and sanding you find that there is a delamination, ridge, pocket, or void, have no fear. Simply grind or sand away the damaged area and add new layers as mentioned in the previous steps.

In my case, I decided that I wanted to do something a bit nicer with the edges of my lugs, get a sharp clean edge,  and feather them into the tube a bit more smoothly. I could have done this with some sanding, but I was lazy and wanted to do it by adding material instead. I cut some thin ribbons of fiberglass and carbon fiber tape of varying widths to create the transition, and then laid up a larger piece of the twill weave carbon for the outer layer. After that, I wrapped in compression tape as I'd done for the dropouts.

This turned out to be a big mistake. Getting nice uniform layups without the moulds was trickier than I'd thought, and I ended up with a bit of a lumpy mess. The carbon fiber had also slithered around a bit as I added the compression wrap (the key is to let the epoxy set up a little bit before doing this, rather than immediately after the layup). The end result was a bit of an aesthetic nightmare, exactly the opposite of what I'd hoped for, and in trying to fix some minor imperfections meant that I needed to do a huge amount of sanding to get back to the point I'd started at.

Learn from my mistakes, and avoid the quick fixes. Getting a nice finish on your parts will usually require some measure of sanding no matter what, so do it nicely the first time instead of doing it twice.

Step 19: Bonding on the Brake Bosses

For my frame I wanted to use V-brakes – I prefer their feel, strength, and ease of adjustability to caliper brakes (not to mention the lower price). This required that I bond a set of threaded V-brake bosses onto my seatstays.

You can buy these parts from a frame supplier, but in my case I liberated them from the same donor frame that provided my headtube and dropouts. I retained some of the tube that the brake bosses had been welded to, and then ground off the back half of the tube. I then used a half-round file that approximately matched the diameter of my seatstays and filed the interior of the tube.

I mocked up the pair of bosses on my frame and measured the spacing between them. I then drilled two holes with the same spacing in piece of scrap wood. Plugging the bosses into these holes ensured that they would be properly aligned.

I then used some high quality kneadable metal epoxy putty and put a layer on the back of both of the bosses. I then pressed these against my seatstays, pressing hard to squeeze out any excess putty but allowing it to fill any gaps between the boss and the seatstay, creating a solid bond. Make sure that proper alignment is maintained, because the metal epoxy sets up quickly (usually 10 minutes).

After the epoxy had set, I sanded away the excess and create a lashed joint with some carbon fiber tow (thin unidirectional ribbon that comes on a spool) for extra strength and insurance. I first wrapped the joint in fiberglass, then wrapped several Figure-8s with the carbon fiber until it was uniform and covered all the edges of the boss. I then wrapped it in electrical tape to compress it, as before.

Step 20: Finishing Touches

There isn’t much left to do! After all the clean-up, you need to cut the slot for your seatpost binder clamp. I chose to cut mine in the front, as with my dramatically slack seat tube angle I felt it would be stronger to have the seatpost bearing against uninterrupted tubing.

Drill a small pilot hole about an inch below the top of your seat tube, and then drill it out to about 7/32”. Then mark a line from the top of your seatpost to the center of the hole and cut the slot. I used a Dremel cutoff wheel, but it can be easily done with a hacksaw as well. Having the hole reduces the stress risers that can lead to a crack propagated from your slot (a bad thing). After you’ve cut the slot, remove any burrs and file the cut edges smooth with a flat needle file.

For aesthetics sake, paint on a final thin layer of epoxy resin with a foam brush onto your lugs. It’ll bring out the 3D lustre and depth of the fibers and generally make the whole thing 'pop'. Since the epoxy is in a thin film, it will take longer to cure than your other layups (at least 24 hours) so don’t get impatient and touch the thing while it’s still tacky.

After everything has cured, spray on a coat of 303 Aerospace Protectant (or some similar product). UV rays degrade epoxy, so if you want to leave your carbon lugs nude (and who wouldn’t want to show off their handiwork?), you need to protect it. You can clearcoat it with a 2-part urethane, but it’s extremely toxic stuff that I wasn’t prepared to mess with, so more regular maintenance with 303 suits me just fine.

And you’re done! Bask in admiration of your creation, throw some parts on it, and go for a test ride. It’s one of the more satisfying first rides you’ll ever have.