|Home Made Carbon TT Frame
by Damon Rinard
Many of you who have read about the first composite road frame I made have asked me what new projects I am working on. In addition to my regular busy schedule, I am trying to squeeze in a new carbon fiber time trial frame this year. Unlike the last carbon fiber frame I made, this time I plan to use a plug to make a mold, from which I will mold the frame in two halves.
The benefits of using an external mold compared to wrapping wet cloth over an internal foam core are:
- smoother exterior surface (less hand finishing)
- more even wall thickness (due to less sanding)
- ability to make more than one frame by using the mold over again
However, I still think wrapping wet cloth over a foam core is easier for a beginner. Here are the benefits:
- less time, money and work required (don't have to make a plug and mold before you can make the frame)
- less waste if you don't like the design (no bad mold hanging around)
See How I Built a Composite Bike in My Garage for the moldless method I used.
Obviously, making a bike frame in a mold requires that you have a mold. The way I've chosen to make my mold is to cast it off a plug. A plug is an exact model of the desired shape. So I have to make the plug first, then make the mold from the plug. Of course, I want the frame to be straight, which means the mold should be straight, which will happen only if the plug is straight. So I made a simple jig to make sure the plug will be straight.
Here is the plug shown on the jig I am using. The jig is made out of a a solid-core door from Home Depot. I laid it on a card table in my garage to form a rigid base. I drew the side view of the frame in AutoCAD, then paid $20 to have Tiger Reprographics plot it full scale on cheap bond paper. I laid the paper plot on the door, and made various fixtures to serve as standoffs to hold the plug's foam core in alignment. At the front of the frame, on the right side in the photo, you can see I've left the head tube quite long. I use the extra length to rest on the jig's lag screws. You can also see a bit of foam lying on the table. When I bond it in place, it will form the head tube's leading edge.
Here is a view of the frame jig, looking from the bottom bracket area up toward the head tube. The bottom bracket cavity in the frame wil be formed on the tube mandrel in the foreground. It is 1.5" diameter aluminum, with both ends cut off in a lathe. I rely on the square ends to align the rest of the frame. The top edge of this bb mandrel is the datum plane from which I reference all other points on the frame. This plane is 34 mm from the central plane of the frame. To check centerline on any point on the frame, I rest a straightedge across the face of the bb mandrel and extend the length of the straight edge toward the point I want to check. Since the bb shell is 68mm wide, the centerline anywhere around the frame should be half that distance, or 34mm, below the edge of the straightedge.
I keep the head tube in alignment by using adjustable metal parts mounted on the table. At either end of the head tube, I use one long lag bolt and an angle bracket screwed into the table top. I adjust the height of the head tube by screwing the lag bolt farther in or out of the table. I adjust the head angle by sliding the metal bracket to tilt the head tube. You can see the rubber band that holds the head tube in place. I check alignment on both ends of the head tube using the straightedge as described above.
I used Yoram Leshinski's Airfoil Plotter shareware program (see source list below) to choose and plot aero cross sections to scale. Airfoil Plotter lets me use my windows printer to plot them, so I didn't have to take them to a print shop. I used two airfoils: GOE460 and EP478, both symmetrical. I don't know enough about aerodynamics to know which one might be better for use in a bike frame, so I decided where to use each airfoil based solely on where their proportions fit most conveniently.
I carved the plug out of two-pound-per-cubic foot rigid polyurethane foam. (see the source list below). In the future I will use heavier foam, because I think it will be stronger and more rigid. That way I may be able to get away from the need for so much support while glassing the plug, since the foam will be rigid enough and strong enough to support more of its own weight. The light foam sags a bit.
At this point the plug has two layers of 8 oz. fiberglass on it. I used a vacuum pump (see the source list below) to draw a vacuum in a clear garbage bag to provide compaction, but it didn't work very well: I did not use the breather/bleeder plies as directed, and my set up provided virtually no pressure. The pump does the job as advertised when I include the recommended breather/bleeder plies! I have a little body work to do to fix the wrinkles it left behind.
The plug's head tube is 0.049" wall 1.25" diameter steel. It will remain several inches longer than the frame's final head tube will be. The extra length on the plug will form a similar cavity in the mold. That cavity will hold a head tube mandrel that will in turn form a cavity in the frame that accepts a secondarily bonded aluminum head tube. That means the leading edge of the frame must be at least wide enough to surround the 1 1/4" head tube plus the composite wall thickness. That makes it about the same width as most Kestrels and Trek OCLVs, that is, slightly wider than most steel frames. I'd like to mold the head tube in an hourglass shape, narrower in the middle and wider only at each end where the head cups press in. But using two separate inserts in the top and bottom is a lot more complicated than simply using a single full-length piece of tubing, so for now I'll suffer with the minimal increase in width!
Here you can see the elliptical aluminum dropouts lying on the table. I cut them by hand out of 1/4" plate. (See the source list below). The derailleur hanger has Shimano-approved dimensions relative to the rear hub axle, even though the wheel exits toward the rear of the bike. I decided to give up the quick wheel changes a forward facing dropout would give. After all, it is a time trial bike, and if I flat during a race I would probably abandon anyway. I don't have a support vehicle following! In exchange for potentially slower wheel changes, I get to shape the chainstay so it approaches the cogs more closely. There might be some aero benefit, even if only imagined, and I like the styling. You can see the pink foam of the left chainstay in the right side of the photo. The background is the part of the plot that shows the dropouts.
On the left is a cross sectional sketch of the mold I will be making. I plan to mold each side with a small (1/4 inch or so) lip around the perimeter. This lip will turn inward from the frame's skin to lie flat on the bonding plane that coincides with the central plane of the frame. On the right is a sketch showing the two halves that will come out of the mold. After cure, I'll bond the two halves together to form the completed frame.
YorJoy Enterprise, 50 Durie Lane, Thornhill, Ontario L3T 5H5, Canada.
- Airfoil plotter shareware program to plot aero cross sections on my windows printer. May be unavailable on the archived site. An airfoil plotter is online at airfoiltools.com.
- Catalog. Cost $5.00 (available online now), but worth it. Highly recommended for the information alone! Plus you get $5.00 credit on your first order.
- Part number 7781-50 fiberglass, 8.92 oz./sq.yd., crow foot, 0.009" thick. Used to skin the foam plug to give it strength.
- Part number 1080-50 fiberglass, 1.45 oz./sq.yd., plain weave, very thin. I use this as a surface layer in the laminate whenever a metal part may contact carbon fiber. The fiberglass acts as an electrical insulator, preventing galvanic corrosion.
- Part number 01-11900 Polyurethane foam, 2 lb/cu.ft., 2" thick.
- Part number 716-38 Unidirectional graphite, 4.7 oz./sq.yd., 0.006" thick.
- Part number 01-38200 carbon fabric, 9.1 oz./sq.yd., four harness satin (4HS) 0.011" thick. I use this for the visible layer since it has that characteristic carbon weave. I also believe the 90 degree fibers may act to constrain the zero degree unidirectional fibers underneath from buckling outward under bending stress.
Aerospace Composite Products
- V-03 AUTO-VAC Electric Vacuum Pump
Last Updated: by Harriet Fell