It's finished! Peter Rooke completes the restoration of his John Smyth engine.
1914 John Smyth engine.
A long run of pipe was needed to from the fuel tank at one end of the cart to the mixer at the other. Steel pipe with a nominal size of 0.25 inch was used. The pipe was secured to the mixer by what looked like a 0.5-inch pipe thread, but the thread pitch was 18 threads to the inch (TPI) rather than 14 TPI. Nothing to match this was found so this clamping nut had to be made, but it’s not easy to turn a taper thread without the correct attachment for the lathe.
To make the nut, a piece of hexagonal steel was mounted in the lathe chuck then a hole bored to a diameter of 0.718 inch and depth of 0.5 inch. The lathe was set up to screw cut 18 TPI and a parallel thread cut to fit a minor diameter of 0.84 inch. Pipe taper threads normally contract by 0.75 inch over 12 inches, translating to 0.0035 inch per revolution for an 18 TPI thread. One of the flats of the hexagon was marked with a daub of red ink as a reference point. A very slow feed was used to give complete control and enable counting the number of turns. It was then a case of increasing the cut of the thread by 0.0035 inch then cutting on this setting for eight threads deep, increase the cut again for seven threads and so on, reducing the number of threads cut until the start of the nut was reached.
A spigot to fit the internal dimensions of the inlet was made, with a shoulder and fiber washer that would be pressed against it when tightened. The new nut was put on the spigot and then a pipe elbow was brazed to the spigot. Elbows were used to align the mixer inlet with the pipe from the fuel tank, and a tap was added.
To make assembly and disassembly easier, a pipe union was fit between the pipe from the fuel tank and the tap assembly. When tightened to the cylinder head inlet pipe, the mixer was twisted too far around to align with the fuel pipe. Gasket paper was cut to fit over the connecting thread of the mixer, proving sufficient to correct the alignment. With the tank centered on its mounts the outlet from the fuel tank was slightly out of alignment, so the pipe was bent just a little to pass down the side of the engine to the mixer.
The governor collar was cleaned before refitting, and it was then noticed that it had a brazed repair and the gap between the collar shoulders varied. The original roller fitted to the end of the governor lever had a step worn in it from this defect. To correct this, the shoulders were filed true and a new roller was made from brass, fractionally undersized so that it was an easy fit. The edge of the catch plate was trued up and made square on the bench grinder.
A new spring was found for the speed adjustment knob, although without the original, size was used as the parameter, not wire thickness. I knew this might be changed later when the engine was actually running. The collar was then tested, but when moved by the governor lever it tipped to one side and locked up. Further examination showed that the internal hole was not round, being oversize in places. Given its rough condition and the repairs, I decided to make a new collar.
A length of 3-inch-diameter steel was put in the lathe chuck. This was drilled then bored to be a close fit on the worn crankshaft, with new channels machined. After drilling a 0.125-inch-diameter oil hole through the side wall, the collar was fitted and its action tested.
The mounting plate for the governor weights was fitted to the flywheel, one new screw being made to replace a missing one. New spring adjustment screws had to be made to replace the severely rusted originals, and the pivot pins for the weights were also replaced.
Finally, new springs were used on the weights and the assembly was fitted. When put on the pivot pins, there was a gap between the weights and the ears of the bracket approaching 0.188 inch. Two pairs of washers were made to fit between the weights and the bracket. The weights themselves had holes through them for adjustment bolts that had been filled. These holes were similar to those seen on Amanco engine weights, suggesting they were not original.
The trip assembly was taken apart and cleaned and the square trip finger itself examined. Only one of its four sides was fit for duty, the others being badly misshapen, so a new finger was made. Some 0.375-inch-square bar was chucked in the lathe and 1.25 inches turned to 0.375-inch diameter ready for threading 0.375 inch UNC. The original had a hardened surface, so after machining and cutting the square section to length it was heated before using case hardening compound to toughen the surface. A suitable spring for the lift finger was found to replace the missing one.
The trip assembly was then bolted to the exhaust pushrod. After making a new gasket the igniter was fitted and the meshing of the igniter and trip checked. The timing would be adjusted later once the flywheels were in place.
In typical fashion, this engine came without a crank guard. Reproductions are available, but shipping from the U.S. to the U.K. is cost prohibitive, so I decided to make one. The first step was to examine some photos before cutting out some 0.25-inch-thick plywood, drilling two 0.4375-inch holes in it matching the mounting holes in the engine base.
This board was cut then filed to shape before using it as a template to cut out a steel mounting plate for the guard. First, a piece of 0.125-inch-thick steel was drilled, cut and filed before cutting two more small pieces to fit along the edges and around the holes for the bolts. These lifted the center of the plate away from the upward curve of the main casting. These two pieces were then brazed to the main piece.
The cast guards appeared to be approximately 0.25 inch thick, so some 4.5-inch-diameter steel pipe of this wall thickness was slit down its length to get a section with the outside edges 4.25 inches apart. There was not a suitable length in store so the guard was made from three pieces.
To get the curve right, a plywood template was made with a radius of 10 inches to check the bend to each section. At 1-inch spacing, a disc grinder was used to cut slits to within an inch of the middle from both edges of the pipe. A vise bending tool was then used to create the curve, pressing at each 1-inch spacing. It was necessary to re-cut the end of each slit another 1/2 inch before bending a second time to get to the profile to match the template.
The ends of the three segments were then ground to fit each other before tack welding the first section to the base plate, fitting to check for clearance against the connecting rod greaser and for alignment. It proved necessary to redo this a number of times to get it right. The remaining pieces were then added and clearances checked again.
No catalog photographs examined showed a battery box. To make moving the engine to shows easier it was decided to fit a box between the skids of the cart, under the cylinder head. This meant fitting an extension pipe to the cylinder drain so that draining water fell outside the skids. With this in mind a box was constructed 12 inches long, 5 inches tall and 4.5 inches wide. A flap was positioned at the cylinder head end, opening down. The box was made with simple butt joints, nailed and glued, then varnished.
The homemade coil and a small sealed 12-volt battery easily fit inside and a knife switch was screwed to the top of the box. The earth wire was run through the side of the box, then to a cleaned engine front mounting bolt hole. This gave a good contact for the earth, with the wire clamped between two washers under a nut.
The front of the box rested on the front axle mount and a short piece of angle iron was fitted to the rear of the box and screwed to the skids as a support. The hole for the drain plug was cleaned out using a 3/8 NPT tap before fitting elbows and a short length of pipe, plus a replacement tap, for draining cooling water away from the battery box.
Putting everything together, the main engine casting was first hoisted onto the cart and bolted in place before fitting the piston connecting rod and the crankshaft. Care was taken fitting the gears to ensure the cam gear meshing marks were aligned with the center of the keyway. A piston ring compressor was used to fit the piston, the big-end bearings having already been correctly shimmed when the bearing was repaired. The crankshaft bearings still had plenty of metal, so new shims were made so that it turned freely.
The cylinder head was fitted with a new gasket. To ensure that the rocker arm post was square a washer was made. First the head was fitted and the stud nuts, including the one that is part of the rocker arm post fitting, were fully tightened working in a cross pattern. The rocker arm post threads onto the lower right stud and should be vertical. The pitch to the 0.625-inch bolt thread is 11 TPI, meaning a distance of 0.09 inch per thread and 0.0075 per 30 degrees (or hour). The washer was therefore made 0.01 inch thick to allow for some further adjustment when the cylinder head nuts were retightened after running in. The pushrod guide near the cylinder head was well worn, but as there was no excessive movement it was left untouched and the pushrod and governor latch were fitted.
The old flywheel keys were cleaned up with a file to remove any burrs before checking their fit by coating them with engineer’s blue, lightly fitting them and checking for high spots. They were then filed where blue was removed from a high spot, with the process repeated until the majority of blue was removed when fitted. The keys were then firmly, but not excessively, driven home, taking care on the governor side to allow enough clearance for the governor collar to move.
The water hopper gasket appeared to be made of leather, so a new one was made from some tanned leather. This was cut into two strips roughly 0.5 inch wide, each strip cut long enough to go around the hopper mounting base. The leather was soaked in neatsfoot oil to soften it and also provide a degree of waterproofing.
The first strip was fitted around the perimeter of the hopper base, instant glue being used to temporarily hold it in place and the ends trimmed to give a close butt joint. The second strip was fitted inside the first, the joint being in a different position to the first strip.
The original securing bolts were destroyed when removing the water hopper, so replacement 0.375-inch- by 1.5-inch-long stainless nuts and bolts were used to bolt the top of the hopper in place on the cylinder.
Once the mixer and pipe work was fitted some fuel was poured into the tank and allowed to run through to the mixer. It appeared that the compensating valve was not working properly, as fuel was leaking from the mixer. The mixer was stripped again and the seating of the compensating valve checked by coating it with engineer’s blue and fitting it. There was an area around the hole for the fuel adjustment needle that was not touching the valve, so grinding paste was used to bed it properly. The igniter and trip were fitted, ready to adjust the timing.
The generic timing for this type of engine is for the exhaust valve to start to open 35-45 degrees before bottom dead center and close 5 degrees before top dead center. Spark should occur at a further 5 degrees before top dead center for each 100rpm of engine speed. The manufacturer’s speed setting for this engine was believed to be 350rpm, but for shows it will run with the governor set to 250rpm. This meant that the timing should be set to 12-1/2 degrees.
To bump start an engine, turn a flywheel (hold the rim, never the spokes) to go through the priming stroke while choking the mixer: The engine timing needs to be set for the normal “run” setting. Then, turn the flywheel backwards and let it “bump” against compression. The igniter should then fire and set off the engine.
Good in theory, but it was found that with good compression coupled with an effective compensating valve in the mixer the engine refused to turn past the normal ignition point. There was no decompression release to assist and extra hands were not available to hold valves open to get beyond this point. It was a case of reverting to the use of a starting wrench to get some momentum, holding open the exhaust valve then switching on the coil at the same time as releasing the exhaust valve.
It took some time to adjust the mixer settings, and it was found that even running at just over 300rpm the air intake needed choking. Putting a lighter spring on both the compensating valve and inlet valves did not help. Choking would be expected if the engine was set to run at low speeds, but not at a “normal” operating level. Further, fuel was leaking where the pipe to the cylinder head fitted the top of the mixer. This was a tight fit and had not been removed during restoration in case it was damaged. To remove it, heat was applied and then penetrating fluid. Using the vise and a pipe wrench the pipe was removed so that both threads could be cleaned. With tapered fittings there should be no need to use anything to seal the threads, but to be sure some fuel-resistant PTFE tape was used to ensure a perfect seal.
To make choking easy, a small, adjustable choke plate was made to fit a flange screwed into the mixer, enabling, subject to governor and timing changes, a wide range of engine speeds from 150rpm up to the normal running speed of 350rpm.
A further problem was found with the gasket on the igniter blowing, twice. As this occurred in the same place and the igniter was bolted tight against the engine block, it looked as though some remedial work was required. A piece of perfectly flat steel was coated in engineer’s blue and rubbed against the mounting boss on the engine block where the igniter fitted.
This showed high spots, so a scraper was used to remove them and the process was repeated. Once the igniter boss was flat it was coated in blue and the igniter held in position and twisted slightly so that high spots on the igniter body could be identified. The igniter body wasn’t flat, so the scraping process was repeated until this was a good fit. Another gasket was made and fitted and no further problems were encountered.
The governor weight springs were slackened and a lighter spring fitted to the adjusting screw on the latch, and after a bit of trial and error the engine speed was set in the 220-250rpm range, suitable for ticking over at shows. At this speed the timing was set to approximately 10 degrees before top dead center.
After a further wiping over with automatic transmission fluid, the John Smyth was done. It was another interesting restoration, and certainly one of the most complex I’ve undertaken in view of the number of problems that had to be resolved.