953 /Carbon Semi-Lugged Road Bike

953 /Carbon Semi-Lugged Road Bike

I have previously built a Road bike using Reynolds 953 and Llewellyn lugs with a Dedacciai carbon fibre rear triangle. This was something of an experiment as, whilst it was a standard thing to do, I was unable to find any real information about how to go about it, especially in relation to the glueing process. I have now been riding the bike I built for two and a half years and have had no issues with it other than a minor problem I will mention later. Using the experience I now have I am going to build a similar frame. This will again use 953 oversize  main tubes (35.9,31.7,31.7) with stainless Llewellyn lugs. The rear triangle  is supplied with it’s own bottom bracket so this has to be fillet brazed in. Although the rear triangle is supposed to be a  Dedacciai Firebox as before, from Ceeway, it is a slightly different shape from the previous one and is marked “Black Box and Black Tail”.

I have also obtained some Columbus Minimal carbon forks, which are supposed to complement steel frames and Ceeway supply these un-laquered with the transfers separate. These are significantly cheaper than the finished fork usually sold retail and suits me as I like to have them painted with the frame.
Overall the last bike weighed in at 8.2kg, so I am hoping this time to get well below the 8kg mark with the final build.
I managed to design the frame using BikeCad to a bike fit using the 73 deg lugs and amazingly managed the exact angles to within 0.2 deg.
First step was to sand down the lugs and braze-ons to 400grit to make the final polishing easier after assembly.
The rear triangle comes with custom bottom bracket and socket for seat stays unfortunately not available in stainless steel.

Because the bottom bracket is an odd shape and the seat tube to bottom bracket mitre will not be purely semi-circular I feel that the first step in assembling the frame is to fix the seat tube/bottom bracket joint to prevent movement of the bottom bracket during soldering of the rest of the frame. To this end the first step is to fit and mitre the seat stay unit into the bottom bracket.
Cutting and filing carbon fibre creates potentially hazardous dust so I wear a mask and glasses for this. It is a little arbitrary how you shorten the chain stay unit, and there will not be a need to shorten it by much. I removed about 5mm from the chain stay tips and found filing the bottom bracket projection to shape was all that was necessary to reduce the length to the design measurement. A standard 2nd cut metal file seems to be fine for the mitering.
fullsizeoutput_52b fullsizeoutput_52c img_2941
The next step is to set the position of the bottom bracket so the seat tube can be mitred in the correct position.
fullsizeoutput_52e fullsizeoutput_530 fullsizeoutput_52f fullsizeoutput_531
I used a 40mm hole saw for the mitre then hand filed the rear flat section to fit.

Before soldering the seat tube in I feel it is better to make the holes for the bottom bracket guide as it is easier to measure up at this stage.
fullsizeoutput_534 fullsizeoutput_535
The only supplier I have been able to find for these one piece guides is from CarbonCycles web shop (http://www.carboncycles.cc/?s=0&c=86&p=930&tb=001) which is run by disco brakes.com.
fullsizeoutput_537 fullsizeoutput_536 fullsizeoutput_538 fullsizeoutput_539
The carbon insert can be drilled through after glueing.
Also I drill drain/vent hole in the top of the BB shell connecting to the seat tube. Following this the BB shell has to be fixed exactly in the correct position using the chain stay unit. The chain stay unit is a loose fit to allow for epoxy filling so the pieces of cardboard are to wedge it to keep it central.
Once clamped up tight the chain stays are removed and the seat post tacked, I use a full circumferential tack for this.
fullsizeoutput_542 fullsizeoutput_543
Before progressing further I need to add a gear cable boss to the underside of the right hand chain stay. Ceeway supply an aluminium boss for this which needs to be riveted in place. Also the shape is incorrect and needs to be filed to a better fit.
I also discovered that the tip of my rivet gun was too wide meaning the rivet would go in at an angle potentially cracking the stay so I had to modify the tip with a short cylinder (another lathe job) to ensure it could be introduced squarely.
fullsizeoutput_545 img_2991
Galvanic corrosion
Apparently carbon fibre has a propensity for corrosion when in contact with other metals. It follows the rules of the galvanic series and the further apart the metals in the series the more likely corrosion, due to the creation of small currents, will occur. Of course this is most likely to occur in the presence of electrolyte solutions such as salt water. It is therefore important to use an epoxy resistant to this, and I assume using a marine grade is the correct thing to do. Carbon fibre is at the top of the galvanic series and aluminium is the most likely metal we are likely to use to corrode. In theory titanium would be the best to use or stainless steel. Mild steel is not particularly good either. I presume there are simply no commercially available titanium or stainless bosses suitable for riveting, neither do I know where you might obtain carbon glue-on ones. Therefore it is important to make sure there is a layer of epoxy between  the metal and carbon fibre components. With regard to the gear boss I paint the inside of the chainstay where  the boss is to be mounted with the epoxy mixture (hence the need to do this stage before final assembly) and coat the underside of the gear boss and the rivets themselves before assembly. Also stainless steel NOT aluminium rivets should be used. I intend to see if I can make some stainless rivet-on bosses in the future. Notably the supplied rear dropouts are also aluminium alloy.
Usually I have used the the lugs themselves to draw on the frame tubes then cut and file the mitre to match. this time I decided to mitre to the tubes themselves as when fillet brazing, feeling that maybe this will give a better fit.
As the lug angles were an almost perfect match the lugs fitted in nicely.
I like to line up the bottle cages by eye then mark the position rather then use set positions so I can ensure the aesthetics are correct, making sure the front derailleur boss will be clear, which is unlikely to be a problem with such a large frame.
I would suggest the bottle boss holes are drilled before the frame is assembled especially with thin stainless tubes such as 953. I think most framebuilders now do it this way. Because the tubes can be paper thin, the drill can suddenly dig in after penetrating the tube and rip it open. Much better to have it secured in a drill press or similar, also they act as additional tube vents. Main tubes now ready for soldering up the front triangle.
Frame now soldered up. I do the lugs in the jig and the bottom bracket fillet in a stand. The seat-stay socket is added after the main triangle. Mitred and lined up with the rear triangle in place then the rear triangle carefully removed and tacked in the jig before fully soldering in a stand. I have found securing the seat-stay socket with a three way clamp and a v-block underneath very effective.
fullsizeoutput_567 fullsizeoutput_566fullsizeoutput_56d
It seems sensible to mirror polish the lugs and braze ons at this stage, as they are more accessible, before glueing in the rear triangle.


Lessons from the previous build
I did say earlier, that I had experienced only one small issue with my previous build. Unfortunately after I commenced this build I had a further failure, though this has come at a good time as it has enabled me to analyse where I may have gone wrong.
The first issue was a loosening of the non-drive side seat-stay screw where it joins the rear dropout. This is of course a hinge joint so that the rear triangle can be adapted to different frame sizes.
img_2203 img_2204
In fact I did not glue this myself but in retrospect it looks like a plain epoxy mix was probably used, i.e. without filler. It was simple enough to remove and clean up the screw and thread and re-epoxy with the same mixture I used on the rest of the parts (with filler), applying the epoxy to the mating surfaces and screw thread. According to the manufacturer using a filler in the mix greatly increases the strength of screwed joints. Thus far I have had no further problems.
I now have had to deal with a more major issue, the loosening of the main drive side chain-stay socket.
fullsizeoutput_55c fullsizeoutput_55b
After removing the seat-stay screw it required some persuasion to remove the plug in dropout. This is a problem I have heard has occurred in bikes from major manufacturers.
fullsizeoutput_55d fullsizeoutput_55e fullsizeoutput_55f
Examination of the parts was very helpful in determining what went wrong. The epoxy seems to have adhered to the carbon fibre chain-stay very well, but hardly any epoxy remained on the aluminium alloy plug. There were also two defects in the epoxy coating inside the chain stay which must have been missed during application. My analysis of this is clear – I need to do more to prepare the surface of the plug. Aluminium forms an oxide layer on the surface very quickly and this is not good for adhering to epoxy resin. It needs to be thoroughly abraded to remove this surface coating, or etched if you are able to do this, and glued quite quickly afterwards. I suspect I did not abrade the surface thoroughly enough for fear of making the dropout too loose in the socket. The unfilled spaces were possibly due to poor initial wetting of the inside of the stay with the plain epoxy before adding the filler. This should be done with a brush as should the application of the filled mix. I used a rod for this – wrong. Careful inspection with a bright light before inserting the dropout is also clearly a good idea. I glued the whole rear triangle in one go, which should be OK using longer acting resin, but I also made the mistake of mixing far too much resin for the job. This greatly shortened the “pot time” which is the length of time the epoxy is useable from the mixing vessel. Large quantities generate a lot more heat and I remember the epoxy was going off towards the end of the process leading me to probably hurry the last phase which was the gluing of the rear dropouts. I should now be able to apply this knowledge to the new build.
I have now replaced this rear dropout after cleaning it up and trimming down the residual epoxy in the chain stay with a burr on a rotary tool which makes light work of it. I decided, rightly or wrongly, to cross pin it to see whether this is more effective or indeed detrimental in the long term as if it fails again I doubt it will be salvageable. At least a plain plug in can be re-done.
After glueing I drilled 2 x 2.5mm holes across the centres and glued two stainless steel dowels through.
fullsizeoutput_564 fullsizeoutput_565
After filing off the excess I wrapped some epoxy soaked carbon fibre tape around the joint, mainly for cosmetic reasons.
I do not intend doing this with the new build though I have wondered if there is any way of improving the anchoring of the plug in dropout. I thought maybe cutting some longitudinal grooves in the plug would be a sound idea but the plug is hollow and only 2mm thick so I have abandoned this idea. I may drill some holes in the metal to try and anchor it better.

The Epoxy Stage
For the Epoxy I have chosen to use West Systems products. The reason being only that they are widely available, often through marine chandlers and sometimes Axminster. I obtained mine from Seamark Nunn who had all the products in stock at the best price. West Systems had a lot of information on selection of the right product on their website with detailed instructions, although carbon fibre is not referred to by name, only composites are mentioned. I attach their user manual here, along with also some useful preparation information from the manufacturers of Araldite.
West User Manual 2015
Based on my research I chose to use the West 105 resin with the 206 slow acting hardener. The usual hardener is the 205, but I have used this since my first build and found it is fine for small jobs but to do multiple joints at one go it would go off too quickly. Various pack sizes are available, and for one frame a “junior pack” would probably be adequate. Unfortunately nobody seems to sell a smaller pack including the 206 hardener so I was forced to use the larger standard packs.
fullsizeoutput_544 fullsizeoutput_546
I also use the 850 cleaning solvent (which has very nasty vapour by the way) to clean off uncured epoxy at the end of the process, and as a pre-cleaner, though also used surgical spirits to thoroughly clean up the metal parts before glueing. West sell several fillers, which needs to be added to the epoxy to improve it’s strength and hardness. I have only used the 404 filler which is the strongest and recommended for steel, but there may be different opinions on this.
The process involves first mixing the resin and hardener in the correct quantities. For the larger packs you can purchase pump dispensers which are convenient but I found for smaller quantities syringes were the best option. The initial mix is then used to coat all the surfaces, which should be carefully brushed in. Then the filler is added to remaining epoxy mix and mixed well in. The filler can be added to suit but needs to be reasonably stiff to stay in the joints and I used a thickness somewhere between “mayonnaise” and “peanut butter”. This is then painted liberally in to all sockets and to the fitting parts, the excess can be wiped off afterwards. Obviously the rear triangle was glued in the jig until dry to maintain the correct position.
fullsizeoutput_572 fullsizeoutput_573
Metal surfaces have to be etched, grit blasted or abraded with emery, the latter being the most practical solution, then thoroughly cleaned. With aluminium parts the glueing should take place quickly, within 20-30minutes at most according to the guidance. In the end I decided to do the alloy rear dropouts initially, then the rest on a subsequent occasion so I could complete this process quickly. I also decided to drill 3 x 3mm holes in the dropouts to try and improve the key.
fullsizeoutput_574 fullsizeoutput_575
In fact only the chain stay sockets were glued, the seat stay bolts were fixed subsequently when the main sockets were glued. The drying and setting times of the epoxy are usually quoted at room temperature; I left my workshop heater on full blast for a few hours after completion to encourage the setting.
fullsizeoutput_576 fullsizeoutput_578 fullsizeoutput_579 fullsizeoutput_57a
The mix can be used to smooth over any joint irregularities if desired as it sands very easily and previously I experienced no issues with cracking of the surface
fullsizeoutput_57c fullsizeoutput_57d
Chain stay socket blended with chain stay unit, as were dropouts, but seat stay socket left as a lugged look.
Completed frame now goes to be painted. I will post the result in due course. Frame weight is 1.6 Kg. Bear in mind it is a large frame (approx 58cm) with a lot of head tube and lugs.
I  append two videos which  are a crude attempt to demonstrate how the filler greatly increases the strength of the epoxy resin, the first without filler and the second with. I accept that the unfilled disc is a but thinner:

Completed frame and bike. Bike weighed in at 7.3 to 7.6Kg, scale seemed to give several different readings!
fullsizeoutput_5f1 fullsizeoutput_5f2 fullsizeoutput_5f3


Ian Cornwell

about 2 months ago

Amazing, just found this. I'm currently building a near identical bike! I've had a 753 tube set and the same carbon back-end in my workshop/shed for nearly two years, while other things have kept me too busy. Finally had a day off work today and cut and mostly mitred all the tubes today. So glad I can learn from your experience with the bonding, it was partly dreading attempting that without any proper knowledge that kept me putting this off. This will be full frame number 4 for me, with a franken-tandem, and a old-frame-made-taller thrown in. My self-built bikes: https://goo.gl/photos/E6WZqr9HZhmRXUCD8 https://goo.gl/photos/gysCbiQhTSVZsSEMA https://goo.gl/photos/2okeNXDoaCsRZWfcA https://goo.gl/photos/AnyHmJxegWuuJPah8 Now I'm going to keep reading your site....


Stephen Hilton

about 2 months ago

Dear Ian, Nice to hear it may have been some help. You certainly have some unusual bikes.


Leave a Comment

Leave a Reply