Shortly after I arrived home from Davenport, I took the suggestions offered by the experts at the show and began production on a working motor. What follows is that process.
It is now Oct 31, 2006 and I have been back from Davenport for two months. These eight weeks have been so very different from the eight leading up to Davenport when the shop was humming at warp speed and the chips were flying.
Autocad is now in constant use, along with internet, sketch pads, catalogs, emails, phone calls, visiting people, and more phone calls. I have fabricated almost nothing.
The mockup cylinder heads (below) were completed by Lance Hayward by finalizing the port shapes. I really believe Lance is one of the best. The tan bondo is mine, and the pale blue filler is his. The heads and barrels will be cast in 356 aluminum. I had originally wanted to use ductile iron to be true to the original, but the casting industry has changed greatly, and no local foundry was willing to take on this complex job. They want big jobs that aren’t too fussy . . . lots of “tonnage”. Casting in aluminum means lighter weight, better cooling, and a hi-tech Nikasil liner.
The valve train design came together in August when the mockup heads were built. The focus now has shifted to the bottom end, and it has been much more of a challenge. Originally, these engines produced about 8HP for the street models, and the 1919 OHC racer may have had 20-25HP; a lot has changed in 85 years to say the least, and I’m told to expect 50HP with a more modern design, materials, and fuels.
Finding space for the larger bearings required by the increase in HP has been difficult. In some ways it would have been a lot easier to start with the proverbial “blank sheet of paper”, but I had to stay true to the original form on the outside.
And so, the bevel drive had to be redone yet again; this will be version #3. Pinion side bearing will now be a 62mm double row ball bearing, exactly the same as a Norton Manx. It took weeks to come to this decision . . . because of the bevel gears, the flywheels have to be located on the right side, and there was no narrower bearing that would do the job without having them custom made, and that didn’t make sense at all. This process of engine design calls for a multitude of correct choices before embarking on the expense of pattern making and foundry work. I’m very aware that one poor decision can mean a weakness in the motor that could be VERY expensive to re-engineer. You know how when you do something for the first time you can ALWAYS do it better the second time? In a sense I’m trying to do the impossible by getting it right the first time.
Using Autocad to define the bevel drive. The bevel gears are 47.5 degrees, and I started with 17 teeth on each. This placed the gears too low, so I tried 18 and 19 before settling on 20 teeth. To design a bevel gear is a full page of formulas from Machinery’s Handbook. If you get one equation wrong, it affects formulas down the page. Took me right back to Physics 12 that I wasn’t very good at . . . thank you Scott Taylor for correcting my errors. At the bottom of the bevel drive case are the two oil pumps. The yellow lines show where the case needs to be enlarged slightly to make room for gears and bearings.
The bevel drive itself is a modular system. The three gears — with no teeth — are aluminum blanks machined along the pitch angle as a mockup. The bearing holders will be CNC machined from 7075. The two arcs at the top are for location of the assembly inside the case.
Here it sits in the case. The lower bearing support is missing in this photo.
December 6 / 06. The snow is piling up, but there’s a lot going on inside the shop (the peacock hut is on the right).
For the first time in almost two years I sense that I have turned a corner in this project. The mock-up crankcases are almost complete, and I have the JE pistons on my workbench. Carrillo rods are ordered, and the flywheels will be shipped tomorrow from Truett and Osborn. Everything is custom. Rims and spokes are also ordered, and I now have a pattern maker in Missouri for the heads and barrels. The pattern maker for the cases is in North Vancouver, and the foundry in Burnaby, all close. Rocker arms are being CNC machined from 4340 HT/SR, which means heat-treated and stress-relieved. This shop will also machine the crankcase castings, as they have experience with radial 2-stroke motors for ultra-lights. It’s a very exciting time!
New piston / old piston. What a difference 85 years can make! The Excelsior piston is made from cast iron and has three compression rings, no oil ring. The JE piston is forged from high grade aluminum, CNC machined, and weighs 321 grams.
New flywheel and old flywheel. The new one arrived from Truett & Osborn. Basically, it’s a high quality Sportster flywheel with a stroke of 3.46″. With a bore of 85mm I get 998cc. The sprocket shaft is from a JD Harley, and the pinion shaft will be custom to suit the bevel gear. Balance factor will be 52 percent.
Webbing on the LS case got re-done for the fourth time. It was inspired by a photo of a 1914 Excelsior ball bearing race motor. Only a few of these motors exist, and you won’t find any photos of them in any motorcycle book. The owners are very secretive about them, and rarely show them in public (no pictures please). The circle at the bottom of the case is for a “sight glass”, like on the Cyclone. The big webs are still a little long and will be shortened 1/4″ to the red felt pen marks.
People sometimes come into our lives at exactly the right time. Kevin Holland is one of these people. Pattern maker, motorcycle enthusiast, and needing to be involved in a project. He saw this website and gave me a call. I emailed him a drawing of the cylinder, and this big lump of urethane plastic is about to get a lot smaller.
This is one half of the cylinder pattern. Kevin said he switched on the machine and then went home for the night, at least a couple of times, so a lot of machine hours went into this. The cylinder will be cast with the base up / fins down. The big ring next to the cylinder base is the riser. This should give a very clean casting because all the gas and any debris will go into the riser. After machining, all that’s needed is a light sanding then off to the foundry . . . thank you Kevin!
Here the patterns are mounted inside boxes, and ready for the sand to be packed in.
A very good casting — high quality! Note how the fins line up perfectly where the two patterns meet.
I turned my attention to finishing the patterns for the head tube, engine mount, and cam support (lower left). Because they will be cast in steel, I made them 1.021% larger than the actual part, to allow for shrinkage. Now they go to Pacific Pattern for mounting on boards, gating, and other necessary details as required by the foundry.
For the mockup bike, one grip came from Indiana and the other from Oregon — they were both soft rubber, and they matched! The folks that operated that company sold it, and the “new” folks have all the equipment, but aren’t making any grips. I can buy grips from “another” source (below), but they’re much more like plastic. In fact, they shine! The company says they last forever, and I believe them!
These are the engine patterns, made by Pacific Pattern in North Vancouver. I was impressed by the quality — very nice! The big lump where the mag drive goes is the pattern for the core box. A special sand is packed into the “core box” — then it is set, or hardened, by exposure to carbon dioxide. The “core” is then removed, and used in conjunction with the “green sand” in the regular mold.
Patterns for the left case, and inside of the RS case. Notice the cast in oil scraper?
Feb 24/07. After returning from Davenport last year, the next four months were spent designing the bottom end and the bevel drive. I wouldn’t have thought it would take that long, but it did. Around New Years, I decided to embark on another tight deadline: build a running OHC Excelsior for the Legend of the Motorcycle, May 5, California. I knew the planets would all have to line up for this to happen. Shortly after, my sister read my horoscope, and told me the planets are all lining up! I will do my best, and put my trust in the universe.
So, I have about nine weeks left to make it all happen. I have half a frame built, all the parts to build wheels, two foundries are casting engine cases and frame parts this week (hopefully), and I have almost enough parts to build the forks. The bevel gears and shafts are being machined. I have rocker arms, pistons, and flywheels. Rods are promised in three weeks, so now I know it takes five months to get custom rods from Carrillo (this time anyway).
After I get the cases they have to get machined. Southern Cross is handling that, and also machining the bevel gear module and the aluminum/bronze oil pump. The head pattern is barely started, so a little concern there. I need a custom pinion shaft for the flywheel, and a couple of hardened spacers. Have to weld up the gas and oil tanks. Cam blanks need to be machined and Ron Lacey tells me that because I have a long rod motor, and the pistons spend more time at TDC and BDC, that I need to look at his books that deal with cam timing specifically. . . there are a lot of details with a project like this.
Here’s a box of front hubs. Material is 303 stainless. Machined by Southern Cross in Surrey. I’ll drill the spoke holes on the rotary table, in the mill, then polish them.
Rocker arms machined by Southern Cross from 4340 HT/SR. Nice! These are already 32 Rockwell, so no further heat treatment is required. They do need shot peening and nickel plating before the two needle roller bearings are pressed in.
And so, the upper bevel cover got re-designed for the fifth time. I really liked the fourth version, but there wasn’t enough room inside for the gears. I figured out the new shape on Autocad, and machined six pieces of aluminum; holding them together with fast-setting epoxy glue. I scaled them up (1.015%) to allow for shrinkage, as this will be the pattern when it’s finished.
The finished pattern . . . and I like the shape as much as #4. I filled in the aluminum gaps with bondo, primed, sanded, then spray painted. Now it’s sent to Pacific Pattern for getting mounted on boards, and having a core box made to keep the inside hollow.
These are all the pieces necessary to make the head tube casting. I made the initial pattern from steel and bondo (inside the top, middle box), and Pacific Pattern did the rest. They contacted the foundry to find out how to mount it, and what sort of gates and riser(s) are required. Each foundry has their own way of doing things. All this for ten castings for ten bikes. The two red and yellow halves are the core box. The “core” is made in there, often from a chemically based sand that hardens when exposed to CO2. The core is broken out after the casting is poured, and I get a hollow head tube casting. The foundry only gets the red and red/yellow boards. After the castings are made, everything shown here will sit under a bench (collecting dust) for many years to come.
Fresh from Globe Foundry, cast in 356 Aluminum, and heat treated to T6 (sure hope I got all those dimensions right). The two main cases were poured from the middle (more or less) and the circular protrusion (eg; middle of the bevel case) is the “gate” where the molten aluminum has been poured in. After cooling, the excess metal is cut off and re-melted for the next pour.
Close-up. Right now these two cases weigh 21 lbs, which is a lot. I wanted to go strong and machine them lighter, rather than discovering flex in cases that were too light-weight.
This is the mockup camshaft. The five pieces are held together by the 6mm allen screw on the right. It allows the individual lobes to be “tuned” to the correct opening and closing specs. The lobes are copied from an early Sportster motor with roller rockers. Nothing super radical.
This is the fixture that allows one lobe to be moved at a time, independently of the other lobe and shaft.
I setup a bicycle chain and sprocket on the mockup head and barrel, and used a degree wheel and dial indicator to check settings. It was a brain teaser, partly because one cam turned clockwise and the other was anti-clockwise, plus the rocker arms were reversed because of the V-twin configuration. Here I’m using carb slide springs in place of the regular valve springs.
Four camshaft blanks were machined manually from 4130, to be sent out for heat treating and then grinding at Colt Cams.
These are the upper bevel drive housings in their raw state, fresh from Globe Foundry.
In all, five fixtures were needed to fully machine these housings. Notice the counterweight used to keep the lathe spinning freely at about 700 rpm.
Back to the frame. The head bearings used on the the mockup bike were a copy of the ball bearing originals, and not very accurate, so all ten production bikes will have Timken tapered roller bearings.
Here is the fixture used to hold all the fork parts together for the brazing process. This truly is the most complex front fork I have ever made, consisting of bent, swaged, tapered and ovalized fork tubes, five castings, plus a leaf spring.
Ten custom 4140 crank sets from Jeff Willis. Jeff looked at this site and saw the hole in the mockup’s bike crank casting, and wanted me to tell you that he stands behind his work, and that any casting with a defect will be replaced free of charge.
Frame casting from Robar Foundry. Material is 8620 in a “normalized” state.
Learning to machine the head tube casting. 8620 can be tough to machine, and this wasn’t working very well.
The fixture was beefed up dramatically, with much better results.
Slightly different fixture to bore the holes for the frame tubes.
Holding down the front engine casting. It’s not always easy trying to figure out the best way of clamping a casting.
The fixture is clamped sideways to drill the engine bolt hole. Note how two C-clamps hold the fixture to the cast angle plate. The fixture is easily rotated 180 degrees to drill the other side. The trick is to join the two holes perfectly in the middle. Note the block and allen screw under the casting to stop it from moving down from the force of the drill.
Things are taking shape. Notice the huge difference in the “rake” or bend of the two forks. Remember the problems I had in getting the mockup forks to bend? Plus the S-bend for the linkage clearance was too low; I ended up ordering another batch of blades from Taiwan, moving the S-bend, and paying another mold charge for them to do the raking. They did a wonderful job, and the new blades match the Excelsior originals perfectly. Material is 4130.
Autocad drawing of the oil pump. This was also a brain teaser, trying to fit everything into a very limited space.
Southern Cross CNC machined it (beautifully) from 954 Aluminum bronze, and each pump has two sets of spur gears to keep the oil moving. I didn’t find out until later what a huge error I had made in the design.
New rods, old rods. New are made by Carrillo, basically a Sportster rod with a custom length.
Here’s the flywheel assembly. Sportster flywheels by Truett and Osborn, with a 3.46″ stroke and custom mainshafts.
Sitting in the right side case, oil scraper in plain view.
This is the fixture for drilling the cylinder barrel. Drill one end, flip it over, drill the other end.
Cylinder head studs from ARP. Beautifully made and perfectly straight.
Barrels ready to go to Mongoose Engineering for boring and Nikasil plating.
During this period of time I had been working basically 8am to 8pm six days a week. I left the shop to buy metal and food, and hired a house cleaner and someone to mow my lawns. I used a lot of couriers to deliver and pickup parts from various suppliers. Kevin Holland was still working hard on the patterns for the all-important heads, and I still felt confident that I could have a working engine for Legend of the Motorcycle on May 5. It was a delicate balance — working so hard, building a bike that was truly a labour of love, and trying not to get stressed out over a self-imposed deadline.
CNC filler caps from Pacific Bending and Machine.
Front rockers for the forks. The grease cups are a standard feature, and a nice touch on this bike.
Checking to see how everything fits. I know my time is running out, so I hired my ex bicycle frame welder to come in and finish welding the tanks, and my ex painter sprayed the tanks and cycle parts. Very few friends stopped by without being put to work in some fashion. I was now working 8am until midnight 6 days a week. Sundays were still reserved for riding my trials bike.
Meanwhile, down in Missiouri, Kevin Holland was working on the heads. He sent me this drawing over the internet, and I could rotate, cross-section, and look at it in other ways as well. Very cool.
I may be the main character in this story, but Kevin is the hero here. A volunteer, someone I had never even met at this stage, he worked his ass off to help me complete this project. At one point, he was working three CNC machines simultaneously at 3am to get these patterns made. Plus working sixty hours a week at his regular job.
Here’s part of the CNC pattern for the front head. Kevin would go home only to mow his lawn and do laundry, then go straight back to the shop.
I don’t fully understand this, but it’s the parting line for the molds, and the core patterns for the ports. Kevin said this project was more complex than he had imagined, and that it really stretched his limits as a pattern maker.
This is what it takes to pour front and rear cylinder heads. After it was all done, Kevin went grocery shopping for the first time in three weeks. He’d been living off McDonald’s burgers and peanut butter for that time. Like I said, a hero.
The heads showed up five days before the bike was to leave for the May 5 California show. There simply wasn’t enough time. Plus I couldn’t install the flywheels into the cases because of an unexpected main bearing problem. It was now a race just to get the bike assembled for the show, and in the last five days I slept an average of three hours a night. There was an energy of excitement and adrenaline, and despite the sleep deprivation, I never yawned once. My body ached all over, and the skin on three knuckles cracked from contact with oils, grease, solvents, hand cleaners, and over-use. My hands hurt like hell. My eye-sight also took a hit from the extended use and concentration. I knew I was pushing my body to the limit, but I never got sick.
Legend of the Motorcycle, May 5, 2007. Basically complete, except for the linkages and fuel line. You can’t tell the engine is empty, and that the heads are still from the mockup bike. It was well received, and a LOT of pictures were taken.
Back at the shop, I was disappointed I had missed my deadline, and my energy was at an all-time low from pushing so hard. It was easy to get distracted and work on other things. But, I knew if I was to sell Excelsiors and pay off the line of credit I had been living on for the past year, I would have to focus and finish the motor! So, I needed to finish the heads, which meant building many fixtures. This one holds the head to machine the exhaust port threads. I extend the shaft that’s held in the lathe chuck, and put it in my wheel balancer to figure out the weight of the required counter-balancer.
Fixture and head are held in the lathe chuck, and you can see the counter-balancer. The flange on the head will get machined off for the thread. The flange will only be used on my race motor because I prefer the 2-stud system to hold on the header pipes.
This is the fixture for the intake thread. One fixture for the front head, one for the rear. Four threads, four fixtures.
Exhaust nuts, manually machined from aluminum bronze. Nice to machine.
Intake nuts, manually machined from 303 stainless. Quite a bit of work.
Fully machined heads, finally. Sparkplug threads are 18mm. Included valve angle is 52 degrees.
Did I mention there was a bearing problem? I had chosen straight roller bearings for the sprocket side, and never realized the outer race had been designed to have a “size-on-size” press fit, which doesn’t work in an alloy case that expands when the engine gets hot. Here I’m boring out the case to get less of a press fit, but I also have to hone out the outer races to make this system work.
The engine sprocket has six tapped 5mm holes.
I can bolt on a hub with a 3/4″ drive for the electric starter. A black delrin “nose” will hide the allen screw heads.
Or, bolt on a degree wheel to setup cam or ignition timing.
Machining some of the linkage parts. The complete linkage turned out to be a LOT of work, easily over one week. Parts are small, and tolerances have to be tight, otherwise you’ll have SLOPPY linkage, which is not good! These parts were investment cast by Jeff Willis and were well made.
Machining the valve cutouts in the piston. Fixture set at 26 degrees in the mill vise, and a carbide fly-cutter is used. Intake clearance is .050″ / exhaust is .080″.
The bevel drive for the valve gear is designed as a modular unit, and it all comes off like this. It makes assembly and disassembly very easy.
Scale is showing 91 pounds, but take off 3 pounds for the degree wheel, and total engine weight is 88 lbs. Not bad!
Engine is finished, and ready to be installed into the frame.
The engine is installed. Excitement is building . . .
Electric starter. Comes from a Sportster. Plugs in to the countershaft sprocket .
June 30, 2007
Gagan and I got the motor into the frame, and hooked up the gas and oil lines. He timed the magneto, and discovered that one of the 80-year-old spark plugs had no spark, so it was taken apart and fixed. We took the Excelsior outside and put in fuel and oil. The 1920s Schebler carb was leaking gas out of the float bowl, so that needed attention too. I found a fire extinguisher, just in case. Gagan operated the throttle and mag advance, and I plugged the hi-torque Sportster starter motor into the crank. This was the moment, and I pressed the button. The starter motor was really straining, the big V-twin barely turning, and then it suddenly spun a few revolutions and fired up! It is loud! Doesn’t sound like a Harley either.
The motor ran, but the linkage still wasn’t finished. These little pieces were very tricky to machine from 303 stainless. Hard part is figuring out how to hold them.
Here they are, in place, with the rest of the linkage.
After startup, the motor had to be pulled apart to see what was going on inside. It looked like the piston clearance was too small; the pistons were showing signs of seizure, and the Nikasil needed honing. It didn’t look like the oil squirters were working either, so that needed checking out. The oil pump was setup in the milling machine to check oil flow, and I discovered my BIG mistake. I had copied an Aermacchi oil pump which was fine, but I had got the rotation backwards so the Excelsior oil pump was running backwards (horror of horrors)! I had already sent out the invitations for the official startup party on Sunday, July 8, so there was very little time to solve the oil pump and piston problems. It was back to working 8am until midnight. The barrels were honed out .002″ for a total of .006″ clearance. There was no bronze (the right size) available in town, so I couldn’t machine up another oil pump, I would have to modify the existing one by plugging existing oil galleries and machining new ones. The oil squirters were now working, but the pump was cavitating badly, so I had to go online and learn about machining anti-cavitational grooves to solve the problem. The engine was reassembled and installed into the frame at 11pm the night before official startup. There was no time for additional testing.
Startup went very well, with a crowd of about 100 people, plus CBC radio. Watch a video of the startup.
The motor was pulled apart the following morning. The bores were scored again, so it wasn’t a clearance problem. We discovered it was an oil ring problem. The oil ring expander was putting too much pressure on the walls, and scoring the Nikasil. The solution was to shorten the expander (45 bumps instead of 46) so there was less pressure on the cylinder walls. Future pistons will have a different oil ring. So, new barrels and rings, put it all back together, and use 5/30 oil instead of 10/40. We ran it for a good five minutes, and found the oil pump modifications worked very well. The oil was being scavenged well, with the oil level in the middle of the sight glass, just as planned! Tore the motor down AGAIN, and the bores looked good. Nothing else to modify, so put it all back together again.
Excelsior #001 was officially finished Wednesday, July 11, 2007 at 10:20pm. Gagan picked it up the following morning for its tour to Waseon, Mid-Ohio, and Davenport.