Straits Dispatch

Last week I got heaps of brazing advice from builders around the world. This afternoon I tried to synthesize it into some better practice brazes. I laid an honest 50cm of fillets. The quality was better than last week.

practice joint

I bought a couple ButBros #1 tips to use on my thinwall cromoly. I really liked using it because it gave me such fine heat control, and was slow enough that I could tell when the weld was ready more from the flux behavior than from the color of the metal. In fact the metal didn’t get anywhere near as red as my old school joints. Sometimes I could even see the semi-cool blueing on the edges of the HAZ. I figure that’s a good sign.

cool flux

I think another decent sign was that the flux melted clear and didn’t vanish. Only once, when I applied only a thin slurry, did it seem that the weld stopped flowing well halfway-through. I looked out from under my dymidium glasses and noticed that there was no yellow flare left from the flux — I must have exhausted and dissipated the flux, hence the terrible joint flow.

What else did I learn by doing lots of boring T-joint fillets? That, like TIG, pulling the welding bead works terribly — beads should be pushed. I am sure when I have done mitered joints in the past, as I’ve been screwing around, some parts of the joint were pulled, and consequently badly-done.

It isn’t like I found a single magic bullet to fix my brazes, it was just a bunch of things simultaneously: heat control, torch control, fluxing, and purposeful practice.

In my notebook, I only have one “to do next” item, that’s to miter up a bunch of practice tubes and start brazing miters again. The bad news is, for the next next two weekends I’ll be in Japan not in my workshop. The good news is, I’ll be in Japan.

Son Luke woke up at 6am to read me Clifford books. After that, and a family bike ride to 7-11 for strawberry-flavored milk (yuck), I hit my emails. I was incredulous when Hubert suggested I use an oxidizing flame instead of a neutral flame. I’d heard of using a carbonizing flame instead of neutral, but never oxidizing. But then Dave Bohm seconded Hubert’s advice and I got excited.

So I made an unplanned morning visit to my workshop and tried it out.

First I dialed in an oxidizing flame. It seems pretty easy to find — the blue flame tip makes a very discrete switch from a rounded, somewhat fuzzy tip to a sharp triangle with a hollow inside. And the torch howls a bit like a jet.

My first pass was sort of a disappointment, I actually had pretty bad porosity, even though I saw the heat zone was much smaller and I moved briskly. On second joint I noticed something important: whenever I pulled the torch away from the workpiece because things were getting too hot, I heard a crackling/sputtering sound. Hmmm

Then a few other things clicked…. my Lincoln Electric GTAW/TIG lessons kicked in: concentrate on minimal, tightly-focussed current into the weld puddle, moving fast and keep the joint shielded till it cooled off. As well as a bit of offhand advice from Steve Garn at NAHBS to use a slightly carbonizing flame to shield the braze.

Maybe when I pull the torch away, the hot zone is exposed to air and generates pores?

So on the next run, I assiduously kept the torch aimed at the hot zone. If I needed to reduce heat, I pulled away, but I kept the torch aimed at the hot zone until it cooled off from red.

Pre sandblast. I was concentrating on torch control, not fillet size… I got bigger as the weld heated.

That seems to be the answer. Whenever I keep the welding area nicely shielded with torch flame/gas, I am not generating porosity, or if I do, the pores are very small. Measured by calipers I see them about 0.1-0.2mm diameter only, and far less frequent.

Post-sandblast — few and tiny pores. Way better than ever other attempt.

And yes, I think the oxidizing flame makes a big difference. Why? Because it focuses the heat zone into a far smaller area.

At this point, it’s starting to feel a bit like TIG/GTAW welding with more hysteresis. In fact, yesterday’s practice with a neutral flame required a #5 sapphire tip to work. Today I switched to a tiny #2 sapphire tip and found the heat more than sufficient. In fact, now that I see how things are working, I’d like to try a #1 tip — I think that could could work.

My fillet brazes have been suffering porosity. Reading the archives, I hypothesized that I was cooking the brass, fuming off the zinc. (Clue: sometimes I’d have big difficulty remelting a fillet [raised melting point due to lack of zinc]).

Why was I overcooking the tubes? Because I wasn’t seeing “dull cherry red” — I was seeing bright orange metal because the burning flare of the sifbronze flux I used was blinding me from the underlying metal until I got it really hot.

Test solution? Wear lenses to block out the sodium-like flare wavelengths. I randomly chose one of the two vendors discussed in the archive (see footnote) and ordered AGW-300’s, as described:

“The AGW-300 is a revolutionary new filter designed for hot glass workers that are working at the furnace doing gathers and gazing into firepots as well as scientific and artistic glassworkers who are doing work with clear borosilicate only. This is also an excellend filter for the metal worker who needs high visible light transmission coupled with excellent infra-red (heat energy) filtration as well as sodium flare filtration. Soft glass workers will also find this filter useful to remove the heat energy that you are exposed to while working with large torches and/or large pieces.”

A month later, here I am in what were called “aviator” style glasses. I didn’t realize that when I was a nerdy fourth-grader back in 1982 that I was also wearing “aviator” glasses. But the glass is clearly something special — the lenses weigh an absolute ton. I have to wear a eyeglass strap to keep them from falling off my face.

THE TESTS

I did five fillet brazes, welding stubs of 1mm cromoly as 90d fillets. I started out using 1.6mm sifbronze-coated rod, sifbronze flux mixed into a paste, and a ButBro/Sapphire-style #2 torch.

The first fillet was absolutely awful. I couldn’t get the rod to melt, nothing flowed, and the result was terrible — massive pores and clearly a garbage joint.

As I was struggling to figure out what was happening, I thought to myself, “are these glasses even doing anything!?”

Then I looked over the top of them and was shocked by the massive orange rocket flare around my torch. I used to braze with that going on? It’s true, these glasses filter out the flux glare. The only color I occasionally see how from the torch is a light green. The violent orange glare is gone.

And I think that’s a clue to one thing that was happening — since I was seeing the underlying metal, instead of the flare, I spotted the red/orange glow way sooner than normal, when the joint wasn’t nearly as hot.

So I switched over to a #5 sapphire torch tip and tried another fillet. With much stronger heat output, the brass rod flowed a lot more quickly. When I finished, I was delighted at the initial look of the fillet — quite smooth considering I did no rework, just laid the joint and stopped. However, when I sand-blasted the joint, I could find patches of porosity underneath. This was a way better joint than the first fiasco, but it still had the pores I am trying to exterminate.

For my third test I plugged in a #3 sapphire. While I was welding it, I was excited, feeling like the torch was not running too hot, but still pretty sufficient to melt the filler rod when I brought it in. I was hoping it would be a nice balance of heat and control, but the result was quite crummy, lots of fine porosity around the joint.

SO WHAT’S HAPPENING?

Is the porosity coming from:

1) zinc fuming (still too much slow heating of the joint)

2) dirty joint from insufficient flux action or crappy flux

3) overheating the joint (but I am getting it everywhere from a #2 to a #5)

4) something else

TWO MORE TESTS

So I removed variables for the second suite of tests. I threw out the disgusting old jar of dehydrated paste that I had been re-watering and using before. I made a tiny, fresh slurry to pre-apply the joint. I used 1.6mm plain brass rods, no more coated sifbronze rods, and just did old-school, heat the rod, dip it in the jar of flux powder style brazing.

The first result of the second batch of tests. These three patches occurred on the side. The biggest pore is maybe 0.3mm?

The first result seemed pretty good, but again, after I sandblasted the joint, a few small patches of porosity appeared. The pores were way smaller than the previous tests. I noticed they occurred mostly on the side of the joint, not at the acute angles. That area gets hotter for sure, but I also wondered if I was using insufficient flux.

I still haven’t seen a “dull cherry red” glow to the metal, I think I’d describe everything as closer to dull orange than ever having a cherry red hue. And the glow extends to something between the diameter of a US quarter or fifty-cent piece. Is that still too harsh?

Also sometimes I hear a bubbling, spattering noise. I think this must be the flux. Is that the clue? I don’t believe I’m heating the joint anywhere close enough to start burning off the trace metals of the cromoly and shooting white sparks. (I don’t know if these lenses would make trace metal sparks invisibile — they might)

(footnote)I online-ordered many thousands of dollars worth of metal-working equipment during my one month vacation in Pittsburgh. The only vendor I had any problem with was Aura Lens. I would never, ever recommend using them. They took well over a month to get my lenses (clearly they’re just a middleman). Trying to contact them via email would elicit little more than late, three word responses. I only finally got confirmation that they had shipped the package ten hours before my self-imposed deadline for contacting PayPal Dispute Resolution.

Early in my oil career, at a long, drunken lunch, my colorful and successful manager declared, “the secret to success in this industry is having staying power.” It didn’t resonate me at the time, but I’ve always remembered it, and over the succeeding fifteen (OMG!) years I’ve seen its truth time and time again, including in my own results.

Now I’ve read a research report that objectively discovered the wisdom my old boss shared: experts are made from 10,000 hours of deliberate practice.

I had the same first reaction most people probably do: how many years is ten thousand hours? (Answer: five years of forty-hour weeks). Ouch.

But on second thought, this study is really exciting. It’s saying:

1) I have self-determination. If I spend 10,000 hours of deliberate practice on something, I will become good at it.

2) It helps me measure whether a venture is worth working on. The price of expertise is ten-thousand hours, no shortcuts.

3) It explains what ‘deliberate practice’ requires to succeed.

I can write a list three pages long of things I’d like to be good at: gardening, jiu-jitsu, drawing, bread-making, driving, japanese language, key telegraphy, french language, bicycle framebuilding, piano, programming, sewing, essentially ad-nauseum. But in moments of honesty with myself, I try to categorize them into “things I like enough to spend 10,000 hours mastering” and “the things I do not.”

For some items that don’t make the “worth 10k commitment” cut, it’s a death sentence, for others it just means they’re third-class hobbies that have a low-priority call on my time and which I’ll never be great at. Importantly, it depends on how the avocation responds to a less-than-10,000-hours commitment.

For vegetable gardening, it’s ok. It just means I have spotty vegetable production, the results aren’t spectacular, the garden sometime languishes and sometimes overwhelms me with too much bittergourd. It’s ok, it still makes me happy. But for jiu-jitsu, it’s a death sentence. I’m terrible at it. Being a bad jiu-jitsu player sucks. And just practicing a little bit doesn’t improve my skills almost at all. So there is no point in working at it all all.

My 10k list is not finalized. I am only certain that bicycle framebuilding is on the list. I’d like to add Japanese, but I’m not sure if language skills follow the same principles as other skills. The rest of the list comes in as an undifferentiated peloton. How many 10k slots can I possibly allocate? Three?

So I’ve gotten as far as deciding I will spend ten thousand hours becoming an expert bicycle constructeur. What are the study’s important implications to my work?

Not all practice makes perfect. You need a particular kind of practice—deliberate practice—to develop expertise. When most people practice, they focus on the things they already know how to do. Deliberate practice is different. It entails considerable, specific, and sustained efforts to do something you can’t do well—or even at all. Research across domains shows that it is only by working at what you can’t do that you turn into the expert you want to become.

Deliberate practice involves two kinds of learning: improving the skills you already have and extending the reach and range of your skills. The enormous concentration required to undertake these twin tasks limits the amount of time you can spend doing them. The famous violinist Nathan Milstein wrote: “Practice as much as you feel you can accomplish with concentration. Once when I became concerned because others around me practiced all day long I asked [my mentor] Professor Auer how many hours I should practice, and he said, ‘It really doesn’t matter how long. If you practice with your fingers, no amount is enough. If you practice with your head, two hours is plenty.’”

It is interesting to note that across a wide range of experts, including athletes, novelists, and musicians, very few appear to be able to engage in more than four or five hours of high concentration and deliberate practice at a time.

[T]rue expertise can be replicated and measured in the lab. As the British scientist Lord Kelvin stated, “If you can not measure it, you can not improve it.”

These excerpts illustrate “deliberate practice.” I have to conscientiously build frames, thinking before, during, and after “how do I do this better,” “how do I assess how well I did,” and “what do I need to focus on next round?”

A master framebuilder fabricates a frame quickly and accurately. I watched Carl Strong account for a 0.015″ bias when welding a frame together. My bikes #2 and #3 were more accurate than the first, but they still required cold setting.
At NAHBS 2010, Strong said that a master builder can fabricate a fillet-brazed bicycle with fork in 8-12 hours. In comparison, my first bike took six weeks. The next two, I built simultaneously during five weeks. Steve Garro brazes fillets that don’t even need cleanup, they’re created perfect. My fillets had pits from zinc fuming, a side-effect of overheating the joint.

Comparing myself to masters, I see three analytical criteria that follow:

1. Time-keeping:

   I need to be able to fabricate a fillet brazed frame in 8-12 hours, a TIG frame in 3-4. I can measure my speed emprically.

2. Tolerances: Masters build their frames extremely straight and with minimal cold-setting. I can measure the alignment of my frames as I build them. This is also empirical.

3. Joint quality: Master builders make perfect joints that need minimal clean-up. I can forensically examine practice joints and get master advice on the joints I’ve put on bicycles. It’s not empirical, but there are clear standards.

These criteria guide my next project. I need to take detailed notes and data. Conscientiously, I need to think through each step as I go, decide how I will do it better than last time, execute, measure my success, and then figure out steps to do this step better the next time. And logging this lot as I go, so I can study how I’m going.
I’ve already been trying this on brazing. When I brazed frame #2. I was disatisfied with the size and smoothness of the fillets. After reflection I realized that my technique of laying a big lump, then trying to flow it out was flawed. Instead I should be laying the fillet more like I TIG weld. So on frame #3 I tried it as if I was TIG welding. The results were notably better, bigger, smoother fillets. But still they’re not perfect — I had pits and I had issues of bronze that would not remelt.
Studying these results I realized I was overheating the joints. My goal is to get the minimal heat necessary, indicated by a dull cherry red, not orange. But I can never see that dull red. More research indicated I should use dimydium glasses which are an optical notch filter for the flare of sodium (quite similar to the flare of flux brazing). This will let me see the joint, rather than the flare, and spot the dulll cherry red color earlier. So before I braze frame #4, I’ll be practicing until I can identify the minimal cherry red melt point.

Arguably the most famous violin teacher of all time, Ivan Galamian, made the point that budding maestros do not engage in deliberate practice spontaneously: “If we analyze the development of the well-known artists, we see that in almost every case the success of their entire career was dependent on the quality of their practicing. In practically every case, the practicing was constantly supervised either by the teacher or an assistant to the teacher.” Research on world-class performers has confirmed Galamian’s observation. It also has shown that future experts need different kinds of teachers at different stages of their development. In the beginning, most are coached by local teachers, people who can give generously of their time and praise. Later on, however, it is essential that performers seek out more-advanced teachers to keep improving their skills. Eventually, all top performers work closely with teachers who have themselves reached international levels of achievement.

Money can buy me lots of equipment and material, but it cannot buy me good teachers, especially the “local” teachers, people who can give generously of their time. I turn out to be lucky in this regard, because my friend Sulaiman runs a respected professional bike workshop on the same floor as my shop (in fact we also share machines and a workshop). He’s always around and has lots of bicycle and machinist experience. He’s particularly good at two things with me: 1) cross-checking my (often foolish) ideas 2) helping me recover from problems and mistakes. Come on, I’ve made exceptionally gross errors on all the frames I’ve built so far, and each time, he lowers a rope ladder into the pit of despair I’ve dug myself, and each time I end up with a rideable, safe bicycle, somewhat better than its predecessor.

My goal is to build between six and ten bicycles in the next ten months, and get my process and techniques to the point where there are no more gross mistakes. By then it will be summer 2011, where I’ve reserved a weeklong 1:1 seminar with master bike builder Dave Bohm in Arizona. I expect he’ll help me with finer, more advanced aspects of efficiency, quality, and design. There’d be no point to go to him for a seminar when I’m still doing nit-wit things like chopping a tube too short or welding the rear dropout on their wrong side, etc.

I have a full time career outside bicycles. Of course I like to daydream that I retire and can spend all my time at the workshop (I did for three months, in fact, and it was great). But I must say one nice thing about having a ‘day job’ is that I have no concerns over the commercial side — all I want to do is build bikes and give them away for the forseeable future. It means I’m happy to go through lots of practice materials, to chop bikes I’ve made into bits and see where they weren’t good enough, and to take a break when it’s no fun. If I had commercial pressures on me, I don’t know how it would distort what I do. (And frankly, if I had commercial pressures, building bikes seems to be about the most poorly-paid skilled trade in the world, anyway.)

So these ideas are what guide my approach to framebuilding today. I’m still working through the ways I can master Japanese, which seems quite orthogonal to framebuilding, but I’ll address that in a different essay sometime.

leave a pair of spare sandals at the workshop if you don’t think wearing road cleats all afternoon sounds like fun.

Three hours and no splinters — amazing!

I hoped to finish brazing the Melakaveli fork today, but in the end I failed. The sanding blast cabinet workshop was locked and no one was around. So I was unable to clean the inside of the fork crown to my taste for silver brazing. I tried sand paper, but it didn’t work well enoguh.

Maybe this week I’ll steal away for a quick shot-blasting and then do the brazing that evening, after work. I’d like to get on to the main triangles of the frame this weekend.

Related, one bit of good news is that my jig hardware from Sputnik is less than two weeks away! If I’m lucky I can use it put together the Melakaveli.

I collected the freshly-painted 9Velos from the powder coater today.

Two frames, two forks hanging in the rafters

I’ve never powder-coated a bike before. Since these bikes are experimental bikes, I tried powder coating. It’s way cheaper ($40.00 instead of $400.00 for a good spray job).

Downsides? The apparent inability to putty a frame before powder coating and the paint’s minimal thickness means every single blemish on the bike is visible, everything from a light carbide center-line scrape on the top-tube to zinc-fuming pinholes on the BB fillet.

This would never happen in Japan

Unnecessary downsides? The shop didn’t demonstrate good paint hygiene. The frames are painted a semi-gloss brown that looks like a Hershey’s Kiss (this is what the guys asked for). But they allowed spray from other jobs (some sort of generic white) to mist over and hit parts of my fork and frame batch. And since powdercoats are baked-on, there’s no getting rid of it.

I need to do some machine work on the steerer tubes tomorrow and then the bikes will be ready for assembly and abusive testing.

My workshop is hot, hot, hot. On a normal day, I might change shirts four times, as they get drenched with sweat. The worst is when I’m brazing and have to turn off the ceiling fans to prevent drafts on my flame.

Mostly I toss the shirts into one of the many cotton laundry bags I loot from my hotel stays (nice cotton laundry bags are a a a key distinguisher between really good hotels and pretender hotels).

I thought using cotton laundry bag would let the laundry dry out as it sat there.

5kg Lumps of sweat-soaked clothes don’t dry out as fast as I thought. I picked up the laundry bag yesterday and saw that some fungus spores have been playing Farmville on the bottom. Yuck.

How many different species can you spot?!

Ling’s friend is a family physician. On my behalf, she ordered ten boxes of nitrile gloves. I wanted something to protect myself from chemicals and also to protect my raw steel bicycle frames from my sweaty hands. Nitrile gloves are more modern technology than latex.

So far I’m relatively happy with them. The best application I’ve found so far is when taking apart my engine lathe. The gloves do an admirable job of protecting me from “paper cuts” from all the sharp bits of metal and scrap all over the lathe. It’s also nice to have a barrier to all the oils and scum on the machine.

I’ve been going through about five pairs a day lately.

Didn’t get the bikes to the powdercoat yet. I reamed the seattubes, chased and faced the bottom brackets, and reamed and faced the head tubes. This led me to a problem on 9Velo1 — during brazing the thin headtube distorted quite a bit. When I reamed it, it leaves part of the wall way too thin for comfort on the bottom.

So tomorrow I’ll have to machine a sleeve to build up the exterior wall of the lower side of the head tube. Then it will be durable.

Shouldn’t prevent me from finishing everything up tomorrow and getting it to the powdercoat.

Ryan had some thoughts on my Visualizing Wheel Flop post.

The way I visualize the fall in height:

To keep things simple, imagine the wheel has zero width. With the wheel dead-ahead in normal riding position, I see three points:

C: the contact patch between wheel and ground
S: the point along the wheel where the steering axis intersects
G: the point on the ground where the steering axis intersects

Now imagine the wheel turns left 90 degrees. I think max wheel flop is experienced at a ninety degree turn. It also looks like at this position C=S=G

If that’s correct, then I think I can calculate the actual drop in height as a function of: trail, head tube angle, and wheel radius. This is a going to be a different number than the standard Wheel Flop Calculation.

Jan Heine and the Bicycle Quarterly back-catalog have been expanding my understanding of bicycle design at a furious pace. I’m particularly interested in front-end geometry. It’s not just about trail and fork-rake. Another important parameter is “wheel flop.”

Bicycle Quarterly describes it:

Wheel flop: The front of a bicycle is lowered as the handlebars are turned. This means that gravity reinforces handlebar deviations from center. The amount of wheel flop is determined by the factor “f”, the weight distribution of the bike, and an eventual front load attached to the bicycle’s forks. A bike with too little wheel flop will be sluggish in its reactions to handlebar inputs. A bike with too much wheel flop will tend to veer off its line at low and moderate speeds. See Vol. 3, No. 3; Vol. 4, No. 3; Vol. 5, No. 3; Vol. 8, No. 2.

It’s a clear definition, but I still couldn’t see it very well. So I put together a test rig to inspect the phenomenon close-up.

Test Rig

I suspended the bicycle from a Park stand horizontally and put a piece of plywood underneath the front wheel as an artificial horizon. I dropped a blue dot underneath the front axle and projected the steering axis with a chalk string. These marks define the trail.

Wheel in straight-ahead position

Now start turning the wheel to the right. You see two things: 1) The gap between the artificial horizon and the front axle point increases. If the bike wasn’t suspended, the front of the bike would have lowered down. 2) The contact patch moved forward

Wheel turning to the right

It makes more sense now. The wheel rotates and the bike falls until the contact patch is intercepted by the steering axis. Or as Jan Heine describes, “wheel flop and trail go to zero.”

When Bicycle Quarterly reviews a bike’s geometry, they include “f“, the “wheel flop factor.” It’ll be a number like 12.7 for example. I interpreted that as the total distance the bike drops when turning the wheel. Bicycle Quarterly says the formula for wheel flop is sin∂⋅cos∂⋅trail. My math is lame, so I needed to draw a picture to visualize how this formula relates to the drop of the wheel.

My first diagram of wheel flop dimension (wrong!)

But my “wheel flop” dimension bears no relation to sin∂⋅cos∂⋅trail. What the hell is that formula reflecting?

Re-draw the wheel with Geometric Trail (T) and Mechanical Trail (MT)

Where in that triangle is this trail formula???

The right triangle

Here is a right triangle. θ is the steering tube angle (normally around 73°) The hypotenuse is Geometric Trail. One leg is Mechanical Trail (MT) equivalent to Sinθ⋄Trail. The other leg, unnamed, I label “G” defined as Cosθ⋄Trail.

I still don’t see the formula in there. Drop a line, X, from the vertex.

Now with a new right triangle and some algebra, the formula becomes clear:

• Sinθ = x/g
  
• g⋄Sinθ = x
  
• Substitute the other definition of ‘g’
  
• (Cosθ⋄Trail)Sinθ = x

That’s it… The ‘x’ dimension is wheel flop! That’s definitely not the original dimension I expected. The ‘x’ dimension is clearly not the literal amount that the wheel drops in a turn. I double-checked this with Jan Heine. He confirmed my conclusions and said this number is just a factor and the actual lowering is proportional to this factor.

For completeness sakes, you can see from the formula:

1. as θ decreases, wheel flop increases

2. as trail increases, wheel flop increases

Friday afternoon I pushed on and finished both 9Velo frames. Now they need sandblasted and sent to the powdercoater. I’ll do that Monday. Friday evening I tore apart the frontbox of my HG-28 Shanghai Ning Jiang engine lathe because the power feeds were not working. Pleasingly, in an hour I had it diagnosed, repaired, and nicely reassembled. Now I can get on with the business of re-calibrating the tailstock setback and topfeed.

Here’s a photoset showing the mechanical guts of the lathe’s Norton Mechanism.

Is when the afternoon thunderstorms roll in. A strong cross-breeze sweeps through the floor and the rains cool off the stinking-hot air.

Cool water

Soaking the flux off the second 9Velo

Now I’ve got no excuses not to put the first two 9Velo’s to sleep. Their triangles are all finished. Now it’s just time for brake boss and brazeons. And frame cleaning. Ugh.

It’s way easier to build big brass fillets if you have a well-fluxed rod. And it’s easier to keep my rods well-dipped in flux if each dip gives me 4″ of flux rather than 1″. So I hacked together little device out of two scrap ends of bicycle tubing. It’s not a thing of beauty, but it allowed me to build up some quite solid brass welds last night.

The welds held together a nicely-aligned frame. After a false-start with an non-flat surface table (who ever heard of such a thing!?) I measured on the table at a nearby machinist’ shop and found that my frame is aligned to with 0.1-0.2mm. I must confess extreme pleasure with that. Now tomorrow will be a challenge to get the rear-triangle finished with similar accuracy. The chainstay jig is particularly shabby so I’ll have to do a lot of intermediate checking as I go.

I spent all morning making brass dust — polishing the brazes of Ryan’s 9Velo. After three hours of that, I was sick of it. Thankfully, a junior staff member arrived to take over the job. Remaining on Ryan’s 9Velo? two water bottle mounts, a brake bridge, cable guides, and the derailleur mount.

RYAN: (1) I’m setting up the derailleur hangar for a 52T Road ring ok? (2) two sets of water bottle mounts is ok? (3) Drilling the rear bridge for normal road brakes, nothign weird, right?

MICHEL: Your 9Velo’s front triangle is totally mitered up and ready for brazing. Depending on how tired I get of sanding Ryan’s fillets, I’ll be jigging your frame into the Henry James soon.

In June I’ll take a week’s course on TIG/GTAW welding at Lincoln Electric in Cleveland. I’m taking a TIG class currently at ITE Singapore, but it’s too basic. The Lincoln course should be better and I’ll concentrate on welding cromoly steel tubes.

Ryan’s 20″ (406mm) 9Velo disguised as a 700mm wheel

I sorted out Ryan’s 9Velo today — the main triangles are complete (and, thank goodness, satisfyingly square). Left to do are the brazeons and a chainstay bridge. I thought these guys wanted their 9Velo’s to be true fixies, so all I planned to put on was a front brake, but actually they want the cable routing and derailleur hangers mounted so that they can make a geared and fully-braked bike if they get sick of singlespeed life.

To save money we’re going to powdercoat this frame as an experiment. I think it will require me to do gross frame-preperation beforehand.

It cannot be a good randonneur bike if it doesn’t have a good frontbag. And I don’t like kludging on fittings as an after-thought. So before I even started the geometry of the Singaporteur, I visited the Guu-Watanabe shop in Kichijoji, Tokyo to order a customized randonneur bag.

Their workshop is fantastic — all sort of specialized tools for working with leather and heavy canvas. They showed me lots of cool tools: punches for cutting parts of leather and their stable of powerul walking-foot sewing machines.

They’ve already made bags for VeloOrange Decaleurs, in fact they had a sample, so there should be no problem getting a good fit. This will be my experiment with VeloOrange. I’m going to try to use their fenders and front rack along with the decaleur. I saw them at NAHBS 2010 but didn’t spend a whole lot of time talking to them.

I wondered if all S&S Stainless Steel BTC couplings are created equal in terms of size:weight ratio.

My graph indicates, practically yes, they’re all similar enough. The middle sizes are most “efficient”, around 25g lighter than the estimated 200g weight (equating to 12% improvement on theory). But 25 grams, even on my six couplings, adds up to only 150g weight saving, which is miniscule.

Rims (click for fullsize)

Sheldon Brown has a complete listing of all wheel size standards, as well as a guide to which tires fit which inner rim diameter.