The photos for the Excelsior story are arranged in the order the bike took shape; it seemed the logical way. This table will direct you to each of those stages.
|1. Engine Casing||14. Cylinder Heads|
|2. Fender Bosses||15. Valves & Ports|
|3. Head Tube||16. Engine Assembly|
|4. Bevel Drive||17. Three Great Photos|
|5. Rear Hub||18. Rockers & Struts|
|6. Rear Wheel||19. Front Forks|
|7. The Crank||20. Gas & Oil Tanks|
|8. Handlebar Casting||21. Final Assembly|
|9. Patterns and Molds||22. Ready to Go!|
|10. Saddle & Heads||23. On the Road|
|11. Engine Parts||24. At the Show|
|12. Camshafts/Towers||25. Race Night at the Track|
|13. Rockers||26. Driving Home|
The OHC Excelsior racer is based on a set of 1919 cases, modified to have a bevel drive for the cams and magneto. Below are stock 1919 cases. Note how the bottoms are quite corroded, probably from sitting in the ground for many years. Perfect cases for the prototype.
Here the right case has been set up in the mill, and the stock timing chest removed. This was a fun job, not too fussy!
Fixing the corrosion. Aluminum tubes were machined up, cut away to match the cases, then glued in place. Bondo is used to fill the gaps and corrosion. I have to get the outside shape of the engine first, then I can figure out how everything fits inside. I think it’s called “reverse engineering”.
This is the basic shape of the bevel drive. The case is sitting on a full size drawing of the engine. This worked OK for the first mockup of the bevel drive, but when I drew the engine in Autocad, the shape changed a little. I was learning positions of shafts and gears as the engine started to reveal itself.
The cylinders on this racer are larger than the stock cylinders. The fin spacing is different too. I decided to machine a couple from solid 6061 aluminum. Here is the start of the process. I don’t need a hole right through, they just have to look correct on the outside.
Lathe work is complete, so the bottom gets squared off to match the cases. The cylinder is held on the rotary table while the endmill does its job.
It was time to start on the frame. With a bicycle frame I always start with the front triangle, but a motorcycle frame starts at the back. That’s just how it is. In this picture the seat stays are being mitered while being held in a fixture on the mill. A hole saw is used at about 250 rpm.
I cut the dropouts from 1/4″ plate in my bandsaw, and milled the slots to 7/16″. The rear axle is 1/2″, with flats milled to fit the slots. I could have got them CNC’d, but making them by hand gives them a different look, an older look.
The fender bosses have been brazed on, and the axle adjusters are Tig tacked on. The final brazing will happen when the stays are attached.
The tubes are .125″ wall DOM (drawn over mandrel) tubing. The ends of the tubes were plugged, welded, machined smooth and radiused on the lathe, before slotting on the mill. I always say that anyone can make a frame; it’s just hard to make a good frame.
Progress is slowly happening. A rear triangle is mated to a set of cases with barrels. The OHC frame is based on a 1914 rear end coupled with a 1919 “front clip”. Bicycle frame terminology has no knowledge of “front clip”, so I had to learn it was everything forward of the seat tube.
Here is the basis of the project. An 85 year old photo showing a 1000cc OHC V-twin with bevel drive in a heavy duty bicycle frame with no brakes… exciting! Remember, when this bike was made, welding wasn’t even invented. The rider I have ungraciously cropped is Joe Wolters.
The mock-up made it this far before I realized the bevel drive was split down the middle, plus a few other small details, so it had to be made all over again. Making the upper bevel drive covers has turned out to be difficult. There are only five photos of the right side, all from different angles, and the lighting is usually harsh. A shape that works in one photo might not work in others.
Back to the frame. Here the headtube is being machined in the lathe. I can’t finish-machine the inside for the bearings until all welding is complete. It’s important to “think through” all the steps involved. I’ll need a fixture to hold the completed headtube assembly when I bore the ID (inside dimension).
The headtube has been Tig-tacked to the rest of the dummy headtube castings. If this shape works well on the prototype, patterns will be made for future castings. Everything is held in the frame jig; accuracy is important.
The welded headtube structure is mounted on the fixture (remember the fixture?) and the ID bored for the bearings. I figure a .001″ press fit is good.
You’re looking at the frame clamped in the frame jig. It’s not fancy, but it is effective, and didn’t take forever to build. When I’m finished with this project, I’ll take it all apart and use the metal for something else . . .
Here I’m building the bevel drive that splits. Basic construction technique is bits of Tig-welded steel mostly covered with sculpted bondo. I’m working off an Autocad drawing, so the important steel parts are carefully machined and accurately positioned.
Progress as the shape starts to emerge.
The two sides. Perforated steel is used for the shape of the outside cover. If you look closely at the inner cover, you can see beauty and simplicity of the bevel drive. The front cylinder drives from the crank, and that bevel drive spins the rear cylinder and mag drive, which are at a 90 degree angle. The scribed lines below illustrate the intersection point, which is about 7/8″ above the centre-line of the crank.
The frame is basically finished. Lower bevel drive needs threaded mounts for the tubes, but the magneto is mounted. Check the frame casting just in front of the magneto. The downtube is brazed into it.
Late night photo at the shop. Time to start thinking about a rolling chassis. The fork tubes have been ordered for literally months now. Without patience, a project like this will never happen. In the meantime, I can get to work on the wheels. . .
To build the rear hub, I started with a chunk of 303 stainless, a couple of sealed bearings, and an original sprocket to get the fit of the spline.
The middle of the stainless was bored out and mounted on a fixture. This would allow me to turn it around and make it easier to work on each end, while keeping the outer dimensions concentric with the bearing bores.
The basic shape of the rear hub is starting to appear. 303 stainless is not the strongest, but it does machine very nicely. If you look at the chips sitting in the lathe bed, you’ll see they are all very small and tightly curled. The long, stringy ones are the most dangerous.
The lathe work is complete, so the hub is mounted on the rotary table on the milling machine to drill the 40 spoke holes. The flange is in the way for this side, so the milling head was angled 6.5 degrees to compensate. I bought an extra long centre drill, and extended the drill used for the holes. Tricky little job.
The finished hub. At this point I’ve worked on the Excelsior project slightly more than a year. How those guys at West Cost Chopper can build an entire bike in 4 to 5 days is absolutely amazing. I take my hat off to them.
The finished rear wheel. They left the small sprocket on the right side, even though it wasn’t used with the bicycle cranks. It’s used as a dust cover for the bearing. Rubber is a Firestone Ribbed Racing Tire. Size is 28″ x 2.25″ and they have no steel bead. There’s a rubber flange that hooks under the rim; the rim is rolled over to catch this rubber flange. That’s all there is to hold on the tire. You can put the tire on by hand . . . the rubber just keeps on stretching and stretching! If you get a flat at speed, there’s nothing to keep it on the rim. That’s how dangerous they are. The concept of a steel wire in the tire’s bead wasn’t invented until the early 1920s.
Here is the eccentric. It clamps into the bottom bracket of the frame and allows adjustment of the chain from the engine sprocket to the jackshaft. The keyway keeps both sides of the eccentric in line with each other.
These are the jackshaft sprockets that mount on the eccentric bearings. The stock setup had a double-row ball bearing with the OD threaded (26 TPI reverse thread) so the two sprockets threaded on. Very complicated and almost impossible to duplicate, so regular shielded bearings were used.
Start of the pattern-making process. This is a rocker from the bottom of the front fork, to be cast in 4140, machined, then heat treated. I made the pattern a little “chunkier” for added strength. The distance between the centre lines of the holes was lengthened .044″ to allow for shrinkage: 1/4″ per foot. All surfaces to be machined were also built up slightly. It now goes to the pattern maker to be mounted on a board to suit the foundries’ specifications.
Yes, you can still buy Excelsior parts. This remanufactured crank arm didn’t have the shoulder in the right place, so a new tube was turned on the lathe, split in half, and Tig-tacked in place. That “thing” on the top left is a spigot, used to hold the crank in the four-jaw chuck for machining.
The crank after machining. Notice the collar was brazed after tacking. Also notice the flaw in the casting. Not good. This is 4140, a tough and hard steel. The chips are small and curled; you can’t get that with a cheap steel.
Even after moving the shoulder all the way to the left, the crank still wasn’t long enough to clear the bevel drive, so the middle section was turned down to fit a seamless DOM (drawn over mandrel) tube. It needed to be a 1/2″ longer. This is OK for the prototype, not production.
Bronze bushings (not shown) are machined up as spacers to position the side-to-side positioning of the cranks. Must remember to look for a pair of 1919 pedals.
This is the pattern being made for the handlebar casting. Pieces are individually machined from steel and Tig-tacked together. Notice they’re held to a 1/2″ x 2″ cold-rolled flat bar with two 6mm allen screws. The edge of the flat bar lines up perfectly with the edge of the pattern; this makes alignment very simple. The hole saw is set to rough cut the centre section, then a boring bar will take the final cut to fit the last piece.
The pattern now has all its pieces, and is ready for bondo and primer. Below it is the Excelsior handlebar casting I can buy. It’s poured from ductile iron, which certainly machines beautifully. It’s better than cast iron, which is known to be brittle. I’d rather have mine in 4140 steel. It’s tougher to machine, but I know it won’t break. Do you see how the hole on the right is a little thin on the inside?
The pattern is coated with high-fill primer, and now needs sanding with 320 paper, followed a shot of glossy Imron paint. Can you see the pattern is 1.021 percent larger than the standard part? Shrinkage (remember the shrinkage episode on Seinfeld?) is 1/4″ per foot for steel in a casting.
The bike is slowly starting to take form. The gas and oil tank shapes were made from cardboard and held with masking tape. A little bit of experimentation was needed to get the shapes right.
I couldn’t find a sheet metal brake to buy, at least not one of good quality. I wanted a straight leaf brake with an adjustable apron. A friend very kindly lent me this one. The “leaf” has been removed, and a 1.25″ shaft welded to 1/2″ flat bar, as shown below. This is what the 18-gauge steel is bent around. The “adjustable apron” makes this possible—see the adjustment on the left, underneath. The piece of wood is to sight down; to line up the end of the wood, the round stock and the mark on the sheet metal to make every bend as consistent and accurate as possible.
Mockup of the gas tank. Oil tank is the left side. The number 6 denotes the number of tries to get the shape right. It’s pretty close here, and number 7 became the final version.
The patterns I gave the patternmaker have come back; molds taken from the patterns along the “split” lines are then mounted on boards to be sent to the foundry. These will be used to cast the struts and rockers for the front forks. The foundry will use these boards to make sand molds and then pour in 4140 steel. The middle lump is the “gate” where the molten steel enters, and the two uprights at the ends are the “risers” where the steel exits, but then flows back into the mold as the parts cool. It is definitely an art.
I can buy reproduction saddles like the one below, but I wasn’t happy with the shape. Have you ever tried to ride a bicycle with the sides sticking out like that? Your inner thighs would get so chafed it would be unbearable. I decided there must be a better shape while still being true to the original. This is the start of the “plug” which the female mold will be made from. Pretty simple: a couple of 1/8″ wires welded and formed into a shape.
To turn the wire form into a 3D shape, aluminum welding rod was bent and glued into the correct profile, masking tape applied over that, and the inside then fiber-glassed.
These are the two “plugs” before final painting. After paint, many coats of mold-release wax are applied before the mold is started.
Not everything goes as planned. The mold wouldn’t release from the plug, and here the plug is destroyed in trying to save the mold. This whole “seat mold thing” has become very time consuming. In all, five molds are needed to make these seats.
Both molds ended up being undamaged, so that was a small cause for celebration. The yellow below is pattern-makers wax, because I need to allow for the thickness of the leather before the inner molds are started. The white (under the wax) is the gel-coat sprayed over the plug before the many layers of fiberglass are added. The pattern-makers wax has adhesive on one side, and a hairdryer is used to warm the wax slightly to help it conform to the shape of the mold. The heat also helps the adhesive to stick better.
The inner molds are done, and need to harden for several days before being removed from the outer molds. The excess will be trimmed with a diamond wheel and the edges sanded smooth. Molds need to be at least 1/4″ thick for good strength and durability.
Building the heads is one of the most exciting stages. As a kid out of high school I worked in an automotive machine shop and dreamed of making my own engine. Now is the time! Below is the fixture for locating the valves, which is crucial. The top plates hold the valve stems, while the heavy round stock further down the stems holds the valve seats in position for TIG-tacking. The 1/8″ steel in the shape of a head gasket was laser cut.
The combustion chamber is taking shape. The five years I spent developing my Aermacchi road racer has turned out to be invaluable. I kept notes on all the engine modifications and specs, and that knowledge is used here.
Perforated steel has filled in the shape of the combustion chamber, and the valve guides are fixed in place. Note that bondo has been used as the glue (3 places) to fasten the guides. TIG-tacking always pulls a little as it cools, but bondo will setup fast and NOT pull; leaving the guides exactly where they should be. The hole in the middle of the aluminum plate is the position of the camshaft. What’s missing now are the camshaft supports, cylinder head mounts, rocker studs, and (of course) the intake and exhaust ports.
The exhaust ports are held in place. Everything about this engine is big, compared with others of this era. The stubby exhaust pipes will be 1.75″ in diameter, and the bronze exhaust nuts shown below are almost 2.5″ OD.
Here the intake port is mostly finished. Port size is 1.375″ which matches the carb, a Schebler deluxe. This carb is not entirely correct, as it came out in about 1924, but will do for the mockup. The Schebler Model H is what I’ve been looking for, but they seem to be a little hard to locate. Something like hen’s teeth . . .
All ports are now located, and the bevel drive, cams and roller rockers need to be finalized. Yes, you read correctly, “roller rockers”. The 1916 Cyclone engine (from which this was modeled) had 3/4″ bearings in the rockers; I have a set of the patent drawings. I think it’s pretty cool . . . roller rockers in 1919.
On the left is the fifth and final mold for the seat. It’s used to make the seat pan, shown on the right. The mounting bolts have been Tig-welded onto 18-gauge sheet metal, bonded to the base with epoxy, and fiber-glassed. I’m glad my part of the seats is over. Now it’s up to Walter to work his magic with the leather.
The bicycle cranks had to be faced, drilled and tapped for the pedals. It’s important the pedals don’t attach at strange angles, so a little fixture was made and bolted to the side of the mill table. The cutter you see has two carbide inserts. Material of the cranks is 4140.
This project has taken over my life. I feel like a hermit. My phone has stopped ringing (almost) and hardly anyone stops by. I wake up in the morning and head out to the shop. I haven’t worked on anything except the Excelsior for some weeks now. I am living off my line of credit. I made a list of all the things that need to be finished and estimated the number of days for each. It totaled 29 days. It’s now the end of July ’06, and I have only 25 working days left before I leave for Davenport, Iowa.
These are the billets for the camshafts. A day’s work to draw them up on Autocad and machine from solid 4140 on the lathe.
Start of the cam towers. Material is mild steel with a bit of Tig-welding. They’re pretty ugly and sure could use a bit of shaping.
A fixture was made that bolted to the faceplate of the lathe. I can change cam supports and they all end up in the same position. Here the bearing support has been bored to size, and now a parting tool will cut the circlip groove.
The shape is starting to develop. These are for the left side of the engine; the “ears” bolt to the lower top tube of the frame to help the engine act as a “stressed member”.
The shorter ones attach to the bevel drive, on the right side of the engine.
With a bit of bondo, the shapes resemble castings, which is what they would have used in 1919.
Figuring out the rockers. That’s the little pile of metal above the calculator, to the left. The rollers are 3/4″ ball bearings, and the rocker ratio is 1.57:1. Everything is taking longer than planned, and I can feel my stress level rising.
More mild steel with a bit of Tig-welding.
Little bit of micro-fillet-brazing and careful use of a small file.
I started working in the evenings, trying to get more (and more) hours out of a day. It didn’t really work; I ran my battery down and got sick. It’s a guy thing!
The cylinder heads needed some fins, so a couple of chunks of aluminum ended up in the lathe chuck.
Four fins done, and four holes bored to fit over the stud towers on the right.
The middle of the fins was bored out, then cut into five sections to be glued around the stud towers with bondo. For some applications it really is a good glue.
The combustion chamber is taking shape. The sparkplug angle had to be changed to accomodate the cam towers. Not a lot of finning for a 500cc cylinder.
Head fins all in place, and rocker shaft position has been finalized.
Middle support between the valve guides was taken out, a “valley” was needed between intake and exhaust to help define the final head shape. The mesh is the start of the shaping of the ports.
And the days just go by. . . never done anything like this before, and the first time always takes longer. I want to get it “right” – the shapes, dimensions, and functions. I want it to be simple and effective, and easy to work on. 1919 on the outside and modern inside.
Shape is finished, and a flycutter in the mill does the hole for sparkplug clearance.
Head gaskets were water jet cut from .045″ copper. A little deburring is needed.
A few parts have returned after nickel plating. It’s a pretty simple bike, but there are a lot of pieces.
Assembly took place August 12, 2006. That was only a full six days behind schedule.
View of the bevel drive, cam towers, roller rockers, and sparkplug location.
Ever wondered what the left side of an OHC Excelsior motor looked like?
Hint: you are not alone!
The rear stand was made from ten pieces of steel, and held together with rivets, the kind you hammer, not pop.
Remember these patterns that went off to the foundry? Well . . .
They were used to cast these parts; the rockers and struts for the front fork. Material is 4140 steel. Stock parts are in the foreground. Do you see the red arrow pointing to the crack? Even foundries have bad days. The “riser” (the big lump in the middle) was too small where it entered the mold, and metal couldn’t flow back into the part as it cooled. I can weld it up and use it on the mockup bike, only because it will never get ridden.
A little drill fixture to get the rocker holes in the correct location.
Rockers waiting for the rest of the forks.
Checking to see that everything fits well before brazing.
Trying to rake the forks. In the motorcycle world “rake” is your steering head angle, while in the bicycle world “rake” is the amount of offset in your forks: the centre-line of the steerer tube to the centre of the front axle. I see the Excelsior as having a heavy duty bicycle fork, so here the forks are being “raked”. The problem is the strength of the 4130; they kept springing back into shape and I didn’t have nearly enough rake. The solution was to . . .
Heat them up to red hot, and then let them cool slowly. Little smoke in the shop, but no big deal. It worked.
Fork jig used to set the position for attaching the dropouts. This needs to be precise, otherwise your front wheel will be crooked a little or a lot! This fork jig was used in the early mountain bike days, and just got modified for Excelsior duty.
Here you can see the shapes of the fork crowns. This is a bit of a complex fork to make. During this process it was discovered the “S-bend” in the main fork tubes was too low, not leaving enough room for the linkages to work. I had copied the wrong fork, and will now have to re-order slightly different fork blades. Not enough time to get down about that one; there’s now only ten days left to finish the bike before I leave for Davenport.
I did a trade with my fried Dan: A few hours of welding on his project in exchange for a CNC die for the seat logo. I know I got the better deal. For Dan, it was the “job from hell”. A full day of programming, with each letter needing its own program because Mastercam (the software) got overloaded and overwhelmed with the details of the pattern. The cutter was only .010″ wide, so Dan would set up the machine, get it cutting, go home for the night and come back next morning to see the results.
On the left is a standard Excelsior engine bolt. In the middle is a grade 8 bolt, currently available. I cut the head off, machined a 3/8″ NF thread, took some 1144 steel, tapped a 3/8″ thread, screwed it on the bolt, TIG-welded the end, milled the hex, and domed the end. It’s strong, and less work than starting from scratch.
The start of the gas and oil tanks. Material is 18 gauge steel. You can see how the (5) point mounting system works.
Eighteen pieces finished; ready for bead blasting and a quick trip through the paint shop.
The TIG-welding around the head tube area was built up with bondo to emulate a casting. Future models will have castings . . .
Getting closer. Managed to get the engine into the frame without losing any paint.
I chose grey as the basic color for the bike. Yes, I know grey was a few model years earlier, but the green they used in 1919 is quite pukey and not for me. The factory racers were finished in 1919, but debuted in 1920 when the official color scheme changed to a royal blue with gold pin-striping. You could have a good debate over what the color REALLY should be.
A real Schebler racing carb was obtained at the last minute. The mock-up bike is done – I just didn’t have time for control linkages, oil and fuel lines.
Next milestone is a running motor and a bike on the track!
Here’s my baby . . .
I finished my 1919 OHC Excelsior as a “mock-up bike” and headed for Davenport, Iowa. Bike in van, lots of insurance just in case, and three days to get there. I should have allowed four, because it was 41 hours of driving each way. You cross 2 time zones, and drive through Washington, Idaho, Montana, Wyoming, South Dakota, and finally Iowa. Round trip was 6860 km, or 4286 miles. I might have been hallucinating after all those hours on the road, but I swear there’s wild horses out there.
Also saw a guy leading a dinosaur of some kind . . .
Here I’m leaving the “Badlands” – it’s 5:40 am and time to leave before Ranger Bob gets to his command post. The Badlands are just east of Rapid City, South Dakota, and quite a fantastic place. The wind blew the van around all night, but it was warm.
Driving into the smoke of a forest fire. Montana and Washington each had their own blazes, and the air quality was not good.
At Davenport, the largest annual antique swap meet in the US, you can find many things. This model 9-cylinder radial aircraft engine has over 600 pieces, each made by hand.
This is a 1928 BAC, designed by Louis Bac, an aircraft engineer for Peugot. It’s a pacer for bicycle racing (on a boardtrack). The engine is a 2400cc twin cam OHV V-twin. They tried to get it running on Friday night at the races, but the belt (to the rear wheel) kept falling off.
Very nice Harley board tracker owned by Mike Lange.
Pope board tracker.
I know where it is, and I’m not telling . . .
Getting ready for the big race on Friday night.
The sun had just gone down and the big overhead lights were on.
Mike Lange and his wife Linda. Mike was in second place when his engine didn’t want to go anymore. Mike has raced for 27 years, I believe . . .
1919 Excelsior on the Davenport track. Maybe next year that big engine will be running?
Spectacular skies coming home.
You have to be careful at the pumps. If you rush in for the regular, you’ll pay too much!
Getting ready to do a little night driving.
Obligatory landscape shot of better times in the past.
Do I need to explain this?
Thanks for sharing the memories.