© Copyright Ralph O. Watson, 1/2/2001
(Written by the Editor)
I first became interested in the rotary valve as applied to internal combustion engines around about 1939, after reading an article in a motor cycle magazine describing an Aspin rotary valve four stroke engine. This engine had a capacity of 250 c.c. and it was claimed to produce 29 h.p. at 14,000 r.p.m., using low octane petrol.
At the time, I was living in Nelson and serving an engineering apprenticeship. On occasion I watched a group of engineers, led by the well-known aviator George Bolt, race one metre hydroplanes on the local model boat pond.
These model boats were powered by 30 c.c. engines and ran tethered to a central pole in the pond to provide quite exciting action. Being an enthusiastic experimenter, the Aspin engine came to my mind and I decided that I should give them some competition.
With great, but what turned out to be misguided enthusiasm, I built a model engine based on the Aspin design, which incorporated a cone type valve the same diameter as the cylinder bore, rotating in the cylinder head. The combustion chamber was contained within the rotary valve, which rotated to line up in turn with the spark plug, exhaust port and inlet port.
Full combustion pressure was applied to the valve, forcing it into the taper of its conical seat with the object of ensuring a good seal, but this arrangement could result in the valve seizing in the head due to lack of clearance and lubrication. In order to counter this, the Aspin design incorporated a roller thrust bearing on the valve stem.
I used the same arrangement but could not attain an adjustment whereby the bearing took the load and a satisfactory seal was achieved. When adjusted so that load was on the bearing, the seal leaked and the engine had poor compression and would not run. With load on the cone the valve would seize. After suffering much frustration with broken drive shafts and stripped gears, the engine was eventually run for short periods with load on the cone, thanks to a copious supply of castor oil. This was supplied under pressure to the valve face, by means of a hand pump. My goal of fitting the engine into a model hydroplane came to naught and George Bolt and company remained unopposed at the model pond.
However I was able to test the engine running against a brake and it recorded 1/8 h.p. at 8,000 r.p.m., which was a disappointment when related to the figures quoted in the article which had inspired my efforts.
Many years later the story came out that the Aspin engine was tested by the motorcycle manufacturers Velocette, who found that it produced only half the horsepower claimed, the suggestion being that the original testing had been carried out with a wrongly calibrated tachometer.
Early 1987, some forty years after making the original experimental model engine, the possibility of using the rotary valve principal again, came to mind. I had spent a great deal of time and effort restoring my old BSA special sports racing car, which had come back into my hands as a total wreck. There were indications that the engine had a very limited life between overhauls and each rebuild was becoming more difficult because the parts required had to be made.
The car was based on a 1931 BSA sports car, with 90-degree V, twin cylinder engine, driving the front wheels. The engine had been extensively modified and the power output increased from the original 22 h.p. at 4,000 r.p.m., to approximately 55 h.p. at 6,000 r.p.m., i.e. far beyond the manufacturers expectations and design parameters. Wear was rapid in respect of the large big end rollers and after only 6,000 miles, these could be heard crying out for replacement. The cast iron cylinders, with non-detachable heads, presented a maintenance problem due to valve seat wear.
A spare crankcase and other parts were available and it was time to consider building up another engine, which would be more durable. Over the years I had done a great deal of work on all forms of poppet valve heads and the prospect of further work in this direction was not very attractive. Therefore it is not surprising that my mind turned back a few years to my interest in rotary valves. Experimenting in this direction, could turn the project into an interesting challenge and six months of interesting investigation and learning set me firmly on this course.
Already knowing something of the engines of Ronald Cross, which used rotary valves, research commenced at the Auckland University engineering library and a paper was found which he had read to the Institute of Engineers in England in 1935.
Cross had adapted many engines to rotary valves and had designed several more during the period between the first and second world wars. Several of these were conversions of Rudge motorcycle engines and therefore I made contact with a local enthusiast who was restoring Rudge motorcycles, Mr Norman Maddock. He was most helpful and provided me with road test data covering motorcycles converted by Cross.
Word of my interest soon got about and magazine articles began to arrive to the extent that I compiled a file detailing approximately twenty constructors of rotary valve engines. Mr Lou Aiken, who was making parts for the restoration of vintage and classic motorcycles and had previously been with the Dept. of Scientific and Industrial research had expressed interest. Some of the literature was passed on to him and he commented that each time he read it he found something new, confirming my own experience. It was very helpful to be able to talk the project over with him.
All the engineers mentioned in the data had built rotary valve engines which ran with varying degrees of success. As far back as 1890, a few stationary gas engines incorporating rotary valves were produced and some of these were still running 25 years later, driving workshop machinery and pumps etc.
In the years proceeding the First World War, several car manufactures incorporated rotary valves in their designs. It is worth mentioning Itala in particular who in 1911, produced a well thought out design, implemented with high quality workmanship. Three large, four cylinder, eight litre, Itala cars with rotary valve engines, were entered in the 1913 Grand Prix race. One retired early in the race, “with engine noises,” and the other two at about half distance, i.e. 260 miles, with broken chassis.
During the two decades between the wars Ronald Cross undertook considerable development work involving rotary valves, making or converting approximately sixteen engines. These engines incorporated a cylindrical valve rotating at right angles to the cylinder bores and usually running parallel with the crankshaft. However in some in line engines, including a six cylinder Austin conversion, valves were arranged across the cylinder block and this required a complicated gear drive.
In order to achieve a satisfactory seal, it was necessary to have the cylindrical valves closely fitting in their housings and under pressure. Friction was therefore a problem and an oil supply had to be provided as with crankshaft bearings. In this regard Cross made what he regarded as a breakthrough in design, around about 1934.
He used a floating cylinder which kept pressure on the valve and against the head, this being fixed solidly to the crankcase. It was necessary to provide for only a very small degree of cylinder movement. A clever system of levers held and controlled the pressure to the minimum required to form a seal. Pressure was applied to the closed top of the valve, as well as the ported underside and the area of friction was approximately three times greater than necessary.
An interesting and informative book was published in 1946, written by Marcus C. Inman Hunter, a copy of which I had been able to obtain on loan. This book included an analysis of the desirable features which should be included in the design of a rotary valve. I added my own ideas to the equation and gave considerable thought to all possibilities, in order to decide on a design which would not inherit the limitations of earlier efforts. A cylindrical valve positioned across the top of the cylinder appeared to me the best approach. This arrangement offers several possible variations.
One of which I will call the Cross type, after the original designer Ronald Cross already mentioned. This has separate curved inlet and exhaust passages, which overlap and pass each other to enter and exit at each end of the rotating valve.
A second type incorporates a pocket in the side of the rotor, arranged to connect the cylinder with inlet and exhaust ports in the head. This arrangement has advantages in respect of in line engines, but has a serious disadvantage, in that the pocket is not fully cleared of exhaust, which is carried over to dilute the inlet charge. Furthermore a pocket of inlet charge is wastefully discharged with the exhaust and would make compliance with modern pollution requirements difficult.
In 1957 Dunstan in Adelaide, developed a conversion for the Australian Holden six cylinder in line car engine, using this arrangement. However he ran the valve at quarter engine speed and arranged twin pockets in the valve. In the 1980's the system was again used by Negre in his formula one racing engine. This was a W12 engine i.e. it comprised a formation of three banks, each having four cylinders.
The Cross type arrangement formed the basis of my rotary valve design, but with specific modifications so as to include the following features: -
A cylindrical valve running on ball or roller bearings, with clearance between the valve and housing.
A freely moving port seal, maintained in contact with the rotating valve by gas pressure in the cylinder. I had in mind the effective sealing achieved in respect of piston rings, where gas pressure from behind expands the ring during compression and power strokes. It was appreciated that auxiliary springs would have to be provided to maintain pressure on the seal during exhaust and inlet strokes, as a piston ring in itself provides the spring pressure required.
Port seal and rotary valve to be made from compatible materials, which will run together without wear.
Provision for the lubrication of the valve and seal, with the utilisation of the lubricating oil as a means of cooling the valve assembly. However it was presumed that the close proximity of the inlet and exhaust passages in the Cross type valve, would be advantageous in respect of temperature control.
The valve to be balanced so as to run with minimum vibration.
A limited number of rotary valve engines have been made incorporating some of the above features, but none incorporating all of them.
Lorenzen in 1909 made an in line engine, with a rotary valve running on ball bearings with partial pressure balanced port seals. Dunstan utilised full cylinder pressure on the port seals in his Holden conversion in 1957. Both designs used the pocket type valve previously described, with its inherent disadvantages.
Two successful Norton motorcycle single cylinder rotary valve conversions were completed by Brown Bros. in 1938 and after the war in 1947, and were tested over 16,000 miles of running. Pressure backed seals held in place by a tapered spring ring were incorporated in the design, but roller bearings were not included.
The 1952 Norton Bond rotary valve head was fitted with ball and roller bearings, but in spite of a two-year development period, it was found impossible to keep excessive oil out of the combustion chamber. Wankel type blade seals and piston rings were used, set into the oil cooled rotary valve. The unusually large port area (point four of the piston area) made it impossible to fit the type of seal recommended by Inman Hunter.
In view of the fact that so many previous designs had been only partly successful, I decided that it was necessary to cover every contingency, even if this meant including features later to be proved unnecessary. Prototype testing was a luxury I could not afford. Even after taking everything into account I could still envisage possible troubles, many of which did not in fact occur.
Editor: - A drawing is included below and reference to it at this point will assist in understanding the descriptions of details which follow.
A 36 x 38 m.m. port with 6 m.m. radius in the corners, providing an area of 1360 m.m. squared and being point 21 of the bore area, this was decided on.
Hunter suggested that the flow through a rotary valve would be 20 percent greater than the flow through a poppet valve of the same area. But based on the experience of others and now my own, this figure does not appear to be correct. One would think that the rotary valve would provide superior flow, but the fact that it is only fully open over a short space of time, is a limiting factor.
Information which came to me later showed that Norton used a port area point four of the bore area and achieved a power output which was equal to that of their poppet valve racing engines. This is a design factor which needs more research.
After searching for suitable needle roller bearings, I settled for the nearest sizes readily available, i.e. 72 x 50 x 22 m.m. and 72 x 50 x 40 m.m. The larger size was used on the inlet end, as a simple way of filling the extra space available. The load capacity of the bearings used far exceeds what is required.
The area of the valve containing the ports was made slightly under 72 m.m. in diameter, so that the two parts of the cylinder head could be bored straight through while clamped together, thus simplifying machining.
It was thought desirable to extend the exhaust end cover into the rotary valve as a heat shield. The diameter of the port did not allow space for the bearing inner ring. Therefore the rollers were run directly on the valve and given clearance to allow for probable expansion of the valve.
It was apparent that a rectangular port seal, as suggested by Hunter, would be difficult to make, but that a round seal with a rectangular port through it, as used by Brown Brothers, could be sealed against cylinder pressure with a piston ring and would present few problems.
Brown Bros. used a novel method to maintain pressure between a round port seal and the rotating valve. A spring ring rather like a piston ring, but with a tapered outside face, fitted into a taper in the cylinder head, within which it was constrained. The outward pressure against the taper put pressure upwards on the seal to hold the seal against the valve.
In order to test this arrangement an experimental ring and taper assembly was made up. It was found that the very short distance of travel which could be achieved, was insufficient to cope with the expected change in height due to wear. Thermal expansion effecting clearances was a likely problem and the components involved difficulties in their manufacture. Further experiments followed, none of which indicated a satisfactory solution.
Rotary valve engines made for motorcycles by N.S.U. and for V8 torpedo engines by Junkers, using rotating circular disc nitrided steel valves and steel port seals to which full cylinder pressure was applied, gave trouble due to wear and heat at the seal face. Reference was made in the book by Inman Hunter, to a rectangular seal operating under cylinder pressure. It was suggested that the pressure on the sealing face could be reduced by incorporating two areas in the seal. Application of this principal to a circular port seal could eliminate the fault in previous designs by utilising cylinder pressure to achieve an effective seal and this seemed to me to offer the best chance of success. There was no definitive data available relative to calculating the areas involved and I therefore developed my own theory.
Cylinder unit pressure reaches the sealing face directly via the square port in the port seal unit and assuming there must be some leakage across the face, this pressure must drop to zero at the outside, where it exits the seal face. Taking this into account it therefore seemed reasonable to estimate the average unit pressure across the seal as half of that in the cylinder. Accepting this one can assume that a seal unit pressed towards the valve at half unit cylinder pressure would be in a state of balance and an increase of 12 to 15 percent should result in an effective gas seal.
On the basis of this theory I confirmed practical dimensions as follows : -
1. Base Area (i.e. area under cylinder pressure) :
Diameter 1·845 inches, therefore area 2·673 square inches, less port area 2·061 square inches, gives an
effective area of 0·612 square inches.
2. Seal Area (i.e. area in contact with rotor) :
Diameter 2 inches, therefore area 3·141 square inches, less port area 2·061 square inches, gives an effective area of 1·08 square inches. Half of this equals 0·54 square inches and an increase of 14 percent equals 0·615 square inches, to be compared with 0·612 square inches.
An arrangement of springs and levers was decided upon to maintain pressure on the seal, during periods in the combustion cycle lacking cylinder pressure.
A drawing is provided showing the general arrangement. Some modifications were made after experience was gained over a period running the car, as described later in the text, but the original concept remained unaltered.
Seals were required at each end of the rotating valve as well as at each end of the port area. The valve was held in contact with a fixed bronze ring at the exhaust end, by a spring-loaded ring pressing against the inlet end. Two bronze rings on each side of the port area were held in pressure contact by means tightly fitting O-rings at first, but later wavy springs were substituted as the O-rings did not have the required elasticity.
There was no doubt in my mind that the best material to use for the rotary valve was nitrided steel. This material had been used by Cross and has a very hard surface, which unlike case hardened steel does not soften when heated. However the nitrided case is usually considered too thin to run roller bearings on. The rollers were to be only lightly loaded, but I played safe by using a steel with a high core strength, i.e. a hot die steel, H13 hardened to RC46, prior to nitriding. This may have been overdoing it a little but once again I played safe.
The next problem was to decide on material for the seals, which would be compatible with the nitrided steel valve, against which they would run. Chris Wade of Repco had shown confidence in the project having helped with rough machining of the crankshaft and I sought his advice. He suggested various grades of cast iron and bronze and a test was set up so as to run samples against a revolving gudgeon pin, under a controlled load.
The time taken for each test to seize up was noted and the result was a no contest, in favour of lead bronze. Cross recommended this material and he was again proved correct.
It seemed wise to again be guided by Cross, when selecting material for the cylinder heads. In the somewhat distant past I had used Y-alloy LM14 aluminium, which was then a popular material used for pistons. When asked for in 1987 the usual reaction was, “what is that”. Finally I found Glucina Smelters Ltd, who were most helpful and mixed me up some ingots to order.
Some of the casting involved thick sections, required to allow for the air cooling fins to be machined, after stud holes etc. had been positioned. Progressive Castings found it necessary to do trial castings and place chills in the mould in order to achieve good results.
When it came to heat treating the castings, I gave them a relatively low temperature stabilising treatment. I had heard of castings going out of shape when quenched from a high temperature and chickened out after contemplating the risk involved!
Cross ran his valve directly on Y-alloy with plenty of oil, but my valve tended to seize up if it touched the alloy and difference in heat treatment may have been the contributing factor.
It is probable that the use of Y-alloy was one of my unnecessary precautions and one of the modern alloys available would have done just as well.
Special effort was put into statically balancing the rotary valve. In order to reduce weight in specific places, holes were drilled length wise between the bearings and used for the double purpose of passing cooling oil. As an additional measure the inlet port was set to one side.
At one stage the possibility of drilling a hole to be filled with depleted uranium as a means of adding weight, was considered. Satisfactory results were achieved without such drastic measures.
Oil was supplied to the valve from the main engine system, operating at 50 p.s.i., via a 0·031 inch restrictor orifice and directed to the exhaust end bearing. Flow was then through the valve to the inlet end bearing and back down the chain case. The majority of previous rotary valves had used a system incorporating a small metering pump connected to the throttle, as a means of controlling the quantity delivered. A pump of this type, originating from a Suzuki two-stroke car engine, became available and was procured in case needed, but was never used.
The car was at first run on a two-stroke mixture of oil in the fuel. Over a period of years oil was reduced as I grew more confident in the system provided, to the point where none is added. The crankcase breather is connected to the air cleaner intakes, as is common, but must provide only oil vapour.
The position of the ports was of course critical, but I was able to machine these without too much difficulty, with the valves held in a dividing head and using the milling attachment I have made for my much modified lathe. Finishing around the bend in the ports involved hours of arduous handwork with a flexible drive grinder. Special indicating callipers were made to check the thickness of material, where the ports were in close proximity.
The original BSA camshaft drive might as well have been designed for a rotary valve adaptation and I was able to utilise an old de-lobed camshaft, as an intermediate shaft for two chain drives to the valves. Sprockets on the valves were retained by nuts and 5/32 inch mild steel shear pins, and several positions provided for a vernier timing adjustment. The nut was arranged to unscrew in the event of the pin shearing, so as to prevent major damage should a valve seize. Short chain cases were fabricated from aluminium sheet and welded to the original crankcase.
I had misgivings regarding the possibility of machining rotary valve seals to the accuracy required, but when it came time to assemble everything my fears were put to rest. Valve and seal were checked with marking blue, which showed no high spots requiring hand scraping. Compression when tested using the starting handle, was judged to be equal to that with the original poppet valve set up.
The engine was completely rebuilt in the course of the project and the bottom end modified to incorporate a plain big end bearing in place of the original rollers. This was a major item and required the design of a new crankshaft.
Once again data was researched and concluded with a good hard look at a photograph of a Weslake V-twin crankshaft. A similar component was drawn up to use Morris 1300 bearing shells in new conrods, which were designed and made to suit. Ball and roller main bearings were retained.
I had to decide whether I should nitride the shaft, made from 4340 steel and once again I had to be sure of my facts. At the time a new crankshaft was being made in Christchurch, for a V-twin Cooper Vincent racing car owned by Dave Silcock. The project was being looked at by Murray Jones who had been responsible for the design of modifications for several historic cars. I entered into lengthy correspondence with him. He was of great assistance and provided valuable advice.
Considerable information on the subject was turned up at the Auckland Library and I became quite interested in the scientific aspects of the nitriding process. Nitriding provides a thin hard surface which is under compression, due to the steel absorbing nitrogen as this surface is formed. This surface resists the formation of fatigue cracks. However if the crankshaft flexes enough to relieve the surface compression a crack could soon be on its way.
After many letters had flown frequently between Auckland and Christchurch it was decided to nitride the new crankshaft.
The new shaft and plain big end bearing required a high-pressure oil system, a Hillman car oil pump was adapted by making a special body. The capacity needed to be reduced but I had to decide by how much. The Hillman pump normally supplied nine crankshaft bearings as well as valve gear, say equal to eleven bearings and ran at half engine speed. The modified BSA engine required a supply to only two big end bearings and two rotary valves, say equal to four bearings, with the pump running at quarter engine speed.
Shortening the pump gears and body was a simple way of reducing capacity and was calculated as, original length of 0·844 inch x 4/11 x 0·5/0·25 = 0·613 inch. This worked well and gave the usual drop in pressure at idle speed.
A pair of new cylinders was required as the originals incorporated non-detachable heads and could not be modified. What is more, those available were in poor condition. Steel similar to that used for aircraft engine cylinders was used and purchased in the form of large diameter bar, which I then had roughly bored out on a large lathe. The cooling fins and all accurate turning was completed in my own workshop, including bores to suite 90·5 m.m. big bore Volkswagen pistons.
After two years spent more or less full time, on research, design and construction the engine was in the car ready for the moment of truth. My old friend Herb. Gilroy turned up for the occasion with a tape recorder, hoping there would be pleasant noises in store.
The engine responded by starting without hesitation and a rather sharp exhaust note indicated that improved silencers would be required. The carburettors were adjusted and the engine kept running for eight minutes. Oil smoke in the air increased steadily and oil began to drip from one exhaust connection.
I checked everything over and blocked off the oil supply to the rotary valves. Meantime, Herb. went off looking for higher heat range spark plugs and returned with the hottest ones made. After fitting these we tried again and oil leaks and smoke slowly decreased. I drove the car for approximately quarter of a mile and the engine felt good. The second run lasted about ten minutes.
After things had cooled down, the valve covers were removed, disclosing a lot of oil still remaining on the valves. The port oil seal on the exhaust end, from which most of the oil had leaked, appeared to lack the pressure expected to be applied by the two tightly fitting O-rings which were compressed approximately 0·007 inch. It was expected that this would vary after the cylinder reached operating temperature. Further testing convinced me that a better method of maintaining pressure on the port oil seals must be found.
Murray Jones had sent me a paper covering the design of Belville springs which comprised a cone shaped ring. I had made one of these to fit on the crankshaft, between the flywheel and the main ball bearing and the pressure as calculated was in fact achieved. It seemed possible to design Belville springs to maintain pressure on the port oil seals and 15 lbs be achievable, but the working travel was rather limited at 0·02 inch. It occurred to me that perhaps a stack of four springs could provide the answer. A pair of springs was made up but I found that when compressed together one would snap over, reversing its cone shape, so as to become useless.
Further information concerning the face seals was received from Murray Jones and the most important point raised was that oil working between the seal faces, could exert a separating force of 30 p.s.i.
My next effort involved the use of a circular wavy spring, which was fabricated by bending a normal oil ring rail and welding the gap together. This provided only four waves but a pressure of 16 lbs was achieved with a deflection of 0·125 inch. Elliptical spring formula was used in order to calculate the proportions. It was necessary to make up a new bronze ring for use with the wavy spring, grooved around the outside edge to take an O-ring, which sealed in the normal way. Running one of these in the engine gave encouraging results, to the extent that a full set was made using springs of better quality. Circles were cut from spring steel sheet and five waves pressed in with a wooden die and the formed spring given stabilising heat treatment. The width of the sealing face on the bronze rings was reduced, making the area 0·5 square inches and hopefully the unit pressure 32 p.s.i.
The revised seals proved much better and there were no longer clouds of smoke, but the spark plugs continued to look oily. The car was driven faster and further and one rotary valve seized, shearing its driving shear pin, which proved the practicality of this safety measure. I was some distance from my home and the easy access to components in the rotary valve arrangement was quickly demonstrated, as I was able to make repairs on the side of the road with the limited tools on hand.
After removing the valve I found that the damage was superficial, amounting to a small tear in the bore of the head and a scratch most of the way around. There were signs of a small amount of aluminium adhering to the nitrided surface of the rotary valve. The parts affected were cleaned up and reassembled with a new shear pin and the car was carefully driven home.
It was important to diagnose the reason for the seizing and picking up material from the head. I therefore carefully considered possible causes. There could be uneven expansion of the head affecting the hottest part close to the exhaust port opening. Alternatively, simply the fact that there was inadequate clearance.
The rotary valve had 0·002-inch clearance on diameter. There was also the same amount of clearance at the exhaust end bearing, between the valve and rollers, to allow for expansion which could overload the roller bearing.
An extra 0·002 inch was ground off the circumference of the valve in the port area and 0·001 inch scraped off the bottom of the valve housing in the head. This increase in clearance did not effect compression or the way the engine ran.
At this stage a rather disturbing letter was received from Murray Jones, detailing his thoughts regarding axial movement of the valve, as a result of there being a difference between inlet and exhaust pressures. He quoted figures to support his theory and these indicated that there was a pressure of between 10 and 20 lbs, pressing the valve towards the inlet.
The inlet end seal had been provided with spring loading as previously detailed as the factor now raised had not come to mind. Therefore the valve could move towards the inlet end, opening the fixed exhaust seal and also reducing the loading on the wavy springs and port oil seals. It was now realised too late, that the inlet seal should have been fixed and the exhaust seal spring loaded. As a compromise it was decided to increase the spring loading.
With the oil problems more or less sorted and 250 miles of road testing completed, it was time to try a few laps of the Pukekohe racing circuit. After four laps the ignition was adjusted, then on the next lap the valve seized for the second time.
A smear of aluminium was polished off the valve, rub marks in the head scraped and the shear pin replaced. A further fifteen laps were run at 70-75 m.p.h. and a comparison made between performance with the old poppet valve engine and the rotary valve replacement. It was estimated that top speed would be down by at least ten m.p.h. if the car had been run all out, but the car pulled out of the hairpin corner much better indicating improved torque.
Examination back home in the workshop, disclosed that the high rub spots had become low areas, the evidence being oily soot in some of these. One can envisage that heat from the combustion chamber and valve will cause the bottom of the head to expand upwards, into contact with the valve, thus creating hot spots. These will automatically intensify, sometimes to the point of seizing, where upon metal will be removed. When the area cools the hot spots will become low areas.
A further 0·002 inch was ground off the port area of the valve, increased to 0·003 inch at the exhaust end. Rub spots in the head were scraped so as to be low by 0·002 inch, it being assumed that when hot, expansion was occurring in these places. This work increased valve clearance to 0·006-7 inch on diameter, plus an extra 0·002 inch where areas were scraped.
A second test at the Pukekohe racing circuit took place a month later, by which time the engine had run for some 600 miles. The test provided promising results.
There were no problems involving valve seizures and full throttle produced 90 m.p.h., only 5 m.p.h. down on the best achieved with the old poppet valve engine and there was an improvement in torque at low speeds.
Ignition timing adjustment of between 20 and 28 degrees advance made no difference to the top speed. In retrospect, I think I should have experimented with the ignition retarded further, as most rotary valve engines have run at approximately half the advance normal for poppet valve engines.
Four times, during a total of twenty-five laps there was a sudden, unexplained, cloud of smoke from the exhaust, always at full throttle and with the engine exceeding 5,000 r.p.m. I found the exhaust pipes showing signs of oil, but the spark plugs quite dry. This caused me to recall Murray Jones' theory concerning unequal pressures causing movement of the rotary valve, so as to open the seal at the exhaust end. However the exact situation with an open exhaust system and very little suction at the inlet on full throttle, was hard to determine. If noise was any indication, there were certainly pressure waves up and down the exhaust pipes! I theorised that pressure waves could be resulting in hammer like blows to the end of the rotary valve. A later inspection revealed nothing to support this theory.
Modifications made during further stages of development eliminated all signs of exhaust smoke : -
Valve timing was given careful consideration particularly in respect of the very moderate 14-42, 42-14 degrees providing an opening period of only 236 degrees. It was decided to make adjustments and ten hours with a hand held grinder, so as to remove two m.m. from the ports, resulted in figures of 17-48, 48-17 and an open period of 245 degrees.
The timing chains were originally adjusted by means of shims fitted in the joint between the head and crankcase. This rather arduous adjustment had been required twice so that improvements were called for. Spring blade tensioners were made up and installed with an adjustable stop to limit travel and this arrangement has proven very satisfactory.
It was found that the hardened chain links cut two grooves in the tensioners and rapid wear became a problem.
The chains were polished on the side contacting the tensioners, using emery paper on a flat surface and replacement blades have lasted for 30,000 miles without undue wear.
By early 1989 I had enough confidence in the modified engine to drive the car from Auckland to Dunedin, to take part in the Historic racing series of events held during February.
A second South Island tour was completed the following year after which efforts were made to improve the performance.
The heads were bored out to accept port seals 0·2 inch larger in diameter, so as to allow the ports to be enlarged to 38 m.m. x 42 m.m., resulting in a port to cylinder ratio of 0.25. Viton O-rings fitted in other hot parts of the engine had proved a success. It was decided to use these on the new port seals in place of the piston rings previously used to retain cylinder pressure.
The enlarged valve resulted in timing providing a further increase in the opening period, which was now 258 degrees as a result of 25 - 53 : 53 - 25 degree figures, which can be compared with other engines as follow: -
Cross Rotary Valve 250 c.c. Engine: -
20 - 60 : 62 - 20 = 262 degrees.
Napier Sabre Aero Engine : -
27 - 60 : 60 - 17 = 262 degrees.
Manx Norton Poppet Valve Engine: -
60 - 80 : 80 - 60 = 320 degrees.
Norton Rotary Valve Engine: -
22 - 54 : 54 - 22 = 256 degrees.
Mercedes Desmodromic G.P. Engine: -
20 - 60 : 40 - 17 = 260 degrees.
These modifications provided an opportunity to reverse the face seal spring loading arrangement previously referred to in some detail, so as to eliminate the possible axial movement of the rotary valve and this was therefore attended to. New, Mark 2, Amal concentric carburettors were fitted and set up to suit the engine.
During the return trip from another tour of the South Island including competitive racing events, oil was seen to be leaking from the bottom of the crankcase and a crack was discovered in an area below the rear main bearing. The BSA crankcase comprised a light box, with a support plate for the rear main bearing, located by means of a spigot fastened with studs. The crack had allowed the plate to become loose and without the close fitting spigot the studs were insufficient to retain it securely. It was found that the tracks of the roller bearing were breaking up, probably as a result of the plate moving.
A much stronger crankcase was called for and a replacement was designed, split through the cylinder centre line, as was common with many early motorcycle engines. Aluminium machines easily, so that it was decided to again do the job using my modified lathe, starting with two pieces of solid 6061 aluminium plate four and six inches thick. Actually this is superior to the alternative of making patterns, then castings which still require machining. The Australian manufacturers of light aero engines, Jabiru, at first used castings for their crankcases but later decided to machine them from solid using a C.N.C. machine.
It had been accepted at the beginning that the BSA crankcase was a weak link and as a result the compression was limited to a conservative ratio of 8·2:1. The new crankcase allowed an increase to 8·82:1 which was the maximum attainable without major alterations. The engine capacity was now 1,150 c.c.
After this rather major extra work further testing at Pukekohe proved that all problems had been solved and speeds up to 100 m.p.h. were recorded.
The Thames Valley Car Club held a flying quarter mile sprint event during December 1993 and this provided an opportunity to obtain accurate figures. It was frustrating not to quite make the magic ton, with 99.85 m.p.h. being recorded, on the rather bumpy course, which was a section of flat sealed secondary road, closed for the event.
Editor: - However a year later the club repeated the event using a different section of road and 102 m.p.h. was recorded. During the interim, exhaust tuning had been on the agenda and the exhaust pipes were shortened. This modification secured the desired result.
The engine was now well sorted and the car running well, so tests were carried out on a dynamometer. A flywheel type was chosen, as a power curve can be recorded as the engine accelerates and this required only twenty seconds. This was a considerable advantage when testing an air-cooled engine, especially as the fan available could not produce a blast of air, equal to that obtained at road speeds.
Various ignition settings were tried and different exhaust pipe arrangements experimented with, none of which made much difference. In any event, very soon we received complaints regarding the noise, which were probably justified and had to call it a day. The figures recorded indicated 60 b.h.p. at the engine.
The car has now completed in excess of 30,000 miles with the rotary valve engine and there is no indication that it requires attention, other than occasional decarbonizing. The main port seal, which is the most critical component, has proved to be completely trouble free. There has been no sign of wear and the sealing face has remained partly covered with a black smear of carbon.
Most of the running has involved travelling to and competing in events. On only very few occasions has the car been transported to an event on a trailer. It is both pleasant and practical to drive on the road, lack of weather protection accepted.
Oil consumption amounts to less than a litre in 5,000 miles and multi-grade oil has always been used.
When compared with the original much modified poppet valve engine, the rotary valve engine has slightly more power and this results in an extra three m.p.h. top speed. There was a significant increase in low speed torque and this is of considerable advantage, in view of the fact that the car has only a three-speed gearbox. In view of this, I was unwilling to open the valve timing further, at the risk of spoiling the power curve, which exactly suits the car. The rotary valve engine has proved more economical than the original engine and the increase in torque would no doubt account for this.
A number of fuel consumption tests carried out over varying distances, of from 300 to 1,800 miles, resulted in the following figures:-
Poppet valve engine - 34 to 38 Miles per gallon.
Rotary valve engine - 36 to 40 Miles per gallon.
It has been suggested that my rotary valve engine has proved successful due to the use of modern technology and materials. However the only item not available in the past would be O-rings of improved Viton quality.
On the basis of the experience gained, I now consider that if the travel of the port seal was reduced to say 0·016 inch, the spring loading arrangements could be done without and cylinder gas pressure alone would prove adequate to maintain reliable sealing. In fact on one occasion the engine was inadvertently run without these springs and no difference was noticed. It is worth noting that some desmodromic engines incorporating mechanically closed poppet valves, have no valve springs and rely on gas pressure alone to obtain a final positive seal.
Data covering an engine rather similar to mine, except for the port seal arrangement, provides interesting figures in respect of ignition timing. The most satisfactory advance curve provided four degrees at 1,500 r.p.m. and fourteen degrees at 6,000 r.p.m. under full load. At part load a further advance of 28 degrees was possible. This would indicate that it should be possible to advance the ignition timing of my engine by as much as 28 degrees when the engine is running under partial load.
There were difficulties in arranging a vacuum distributor advance for the V-twin engine and I did not pursue this further. However I am sure that if an effective automatic advance system could be set up, there would be a reduction in fuel consumption.
It has been claimed in the past that a rotary valve engine can run with higher compression ratios, due to the absence of a hot exhaust valve radiating heat. While this is still true, modern combustion chamber design and improved fuels has resulted in this no longer being such a significant advantage. However in the event that an economical, low-grade fuel of approximately 70 octane, was produced by refineries, this could be used efficiently by rotary valve engines and world fuel reserves might be extended.
The rotary valve engine was now running extremely well and had been proven reliable and durable, the car having completed over 30,000 miles. The engine exactly suited the car with a broad power band and further experiment, involving valve timing seemed pointless. Therefore what should be my next project ?
One of several people who had shown interest in my project was Ken McIntosh, a well-known expert in rebuilding Norton motorcycles for classic racing. He suggested that I should give thought to the possibility of converting a Norton engine to rotary valve operation and he provided me with a crankcase, gears and other components to look over.
After examining the Norton lay out, it was apparent that the port seal, would have to comprise the major part of the combustion chamber and that there would be very little space in the head for a spark plug. I therefore proposed to locate a spark plug, in the port seal, which though a moving part, moves only a few thousandths of an inch.
I came up with a proposed design, which resulted in a port to bore area of 0·355, compression ratio 11:1, valve duration 266 degrees and arranged 21-65, 65-21.
The project has not been proceeded with because of doubts as to whether the engine would provide an increase in speed, when compared with the present development of the Manx Norton. What is more the rules covering classic motorcycle racing may exclude the use of a rotary valve conversion, which has not been produced by the original manufacturer of the engine.
A rotary valve arrangement suitable for in line engines was given some thought and for practical purposes a conversion for a small four-cylinder engine, such as a Hillman Imp engine, or a Coventry Climax fire-pump motor, was considered.
In order to demonstrate my thoughts I made up a mock cylindrical valve in wood, designed to run parallel with the crankshaft and positioned above the bores. Ports running through and across the valve were to connect ports in the top of the cylinders, with others opposite in the head. Gas flow was therefore opened and closed simultaneously in two places on opposite sides of the valve and this resulted in a rather obstructed passage.
This arrangement had been suggested by Inman Hunter but after careful consideration I felt that because of the less than ideal gas flow there would be little, if any advantage, over a poppet valve set up. It was interesting to note that Ronald Cross made such a conversion to an Austin seven engine, using his floating cylinder design, for which 30 h.p. was claimed.
Editor : - At the same time as the engine of the BSA Special was converted to a rotary valve configuration, Ralph made 15 X 4 inch wire spoked wheels for the car with great ingenuity and these were fitted with 15 inch X 135 m.m. tyres.
The rims for the wheels, were salvaged from space saver type spare wheels, rejected by owners of imported cars, these being of the desired size. Dimples for the spokes were pressed into the rims, by means of a tool specially made for the purpose. The rims were set up in his versatile lathe and the Watson built dividing head again proved to be a significant asset.
The BSA Special has again passed on and is now being well looked after by Martin Ferner, who is a keen collector of vintage and interesting cars. The car most certainly has not gone into retirement. With his kind permission, a letter he addressed to Ralph regarding the car, is quoted below.
12th November, 2002.
My apologies for not having written earlier. I seem to have been very busy with a number of projects but I am getting to know the car which continues to run very well.
In fact it is a continuing surprise. I am dumfounded at the amazing torque available at low revs. If one parallels it to a poppet valve motor, with what must be regarded as fairly wild valve opening periods, I would expect little torque at low revs and a strong band of torque at higher revs when the gas flow is established. This engine seems to pull lustily from low revs and to do so evenly over a wide range.
I am also amazed at how well sorted the car is. The suspension is extremely comfortable for so light a car and shows your skills in getting the spring rates and shock absorber settings exactly right. I tend to accept the deficiencies built into my other old cars and must pay more attention to these aspects.
The steering is also a delight with no lash and is extremely precise. The clutch although a straight adaptation, is as good as the best available in a contemporary car. I have had no trouble starting and the engine has always started in one or two turns.
My short legs forced me to adjust the drivers seat and investigation disclosed that this had been provided for and I now have the fixing bolts in the most forward position.
I took the car on a short Vintage Car Club rally (sealed roads only) and it created, as you might expect, considerable interest. Following this I gave a talk to the Wellington Branch of the club detailing the origins of the car and the person who created it as a special, as well as the later conversion to rotary valve configuration.
I will be in Auckland shortly and will see you if possible, as I would like to discuss aspects of the workshop manual covering brakes and some small points regarding maintenance.
I hope you are well and send my best wishes.
(Signed) Martin Ferner.
Nearly four years ago, I assisted Ralph with writing and producing a booklet covering the development of his rotary valve engine. Having now read it again, as a result of including the text here, I realise supporting comment is called for.
The preceding article describes the only truly successful implementation of a principle, the application of which has defeated engineers world wide, even though they were all backed by substantial resources. Ralph ranks the rotary valve as his major achievement, for very good reason. Having watched Ralph in his work over a period of fifty years, I am in a privileged position to appreciate the reasons for his success, when so many other engineers have failed.
Ralph always presents a completely open mind and approach to any problem, and at the same time an appreciation of the efforts of others. As a result his research is carried out without indifference, preconception or bias. A very complete picture is in view before a project commences. Practical experience, natural tendency towards lateral thought and an innovative mind, then provides answers not visible to others.
Practical engineering knowledge, dedication and skills of a high order, coupled with the academic, is a rare combination. Many design, but few are able to carry a theory through to practical completion, single handed as Ralph does.
His rotary valve design was not the result of an accidental whim, or a lucky conception. The concept worked, because of applied skill and knowledge. The novel and innovative main seal which he designed is the key to the success of the project. The application of superior materials, not previously available to others, is not a factor which in any way applies.
The surprising aspect must be, that there was not the necessity for several prototypes and endless modification and testing. I was witness to such procedures when visiting in Australia, together with Ralph, an engine development company. They had been working on a rotary valve design for some considerable time, backed by finance from Briggs and Stratton. They were not successful and had given up the project after a great deal of time and money had been spent..
Also surprising is the fact that no commercial motive was involved in Ralph's project. It would not be out of place to describe Ralph as an artist, as is confirmed by any dictionary. In the case of his rotary valve design, others could well reap benefits in the future. The design incorporates several completely novel features and as such would have supported patent protection, but this prospect was rejected on grounds that one could appreciate.
BSA ROTARY VALVE COMPETITION RECORD
Abbreviations : -
S.C.C. - Sports Car Club.
H.R. & S.C.C. - Historic Racing and Sports Car Club.
T.A.C.C.O.C. - Thoroughbred And Classic Car Owners Club.
09/10/88 Meremere Drag Strip Meeting. Retired with clutch trouble.
29/01/89 Ardmore Historic Circuit Race Meeting. Demonstration run.
08/02/89 Queenstown Grass Track Meeting. Racing on grass circuit.
12/02/89 Dunedin Festival Street Race. Circuit racing.
19/02/89 Wigram Country Gents Meeting. Airfield runway circuit racing.
19/03/89 Chelsea Hillclimb, Auckland. Sealed road hillclimb. Time 34.98 secs.
25/03/89 Pukekohe Raceway, Alfa Romeo Car Club. Demonstration run.
02/04/89 Taupo Racing Circuit. S.C.C. of N.Z. Practice meeting. Best lap, 60.03 secs.
09/04/89 Chelsea Hillclimb, Auckland. Sealed road hillclimb. Time 33.35 secs.
16/04/89 Taupo Racing Circuit, Club Meeting. H.R.& S.C.C. Circuit racing.
01/10/89 Meremere Drag Strip Meeting, Standing Quarter Mile. Best time, 16.292 sees. 84.1 m.p.h.
05/11/89 Chelsea Hillclimb, Auckland. Time 33.9 secs.
19/11/89 Taupo Racing Circuit, Club Meeting. H.R.& S.C.C. circuit racing.
21/01/90 Span Farm, Sprint Racing. Auckland Car Club meeting.
03/02/90 Teretonga Circuit Racing, Invercargill. Southland Sports Car Club.
10/02/90 Levels Raceway, Timaru Circuit racing. Best lap, 1 min. 38 secs.
18/12/90 Dunedin Festival Street Races. Circuit racing. Best lap, 2 min. 12.3 secs.
25/02/90 Ruapuna Raceway, Country Gents Meeting. Circuit racing.
03/03/90 Worsley Spur Hillclimb. Christchurch Vintage Car Club.
03/03/90 Southbridge Street Races. Circuit racing. Vintage Car Club.
22/04/90 Ohakea Airfield Races. Circuit Racing. Max. speed attained, 95 m.p.h.
22/09/90 Bay Park Raceway. Practice runs. H.R.& S.C.C.
02/10/90 Pukekohe Raceway. Practice runs. Max. speed attained, 96 m.p.h.
25/11/90 Chelsea Hillclimb, Auckland. Time 32.25 secs.
27/01/91 Pukekohe Raceway, Auckland. Practice day. T.A.C.C.O.C. Achieved 100 m.p.h.
24/03/91 Pukekohe Raceway, Motor Cycle Meeting. Demonstration run.
30/03/91 Pukekohe Raceway. Alfa Romeo Club. Demonstration run, 96 m.p.h. (On 15 inch wheels)
28/10/91 Pukekohe Raceway. M.G. C. C. Test day. 100 m.p.h. (On 15 and 19 inch wheels)
07/11/91 Manfield Auto Course. Three day event. M.G. Car Club. Best lap, 1 min. 47 secs.
24/11/91 Chelsea Hillclimb, Auckland. 32.16 secs
01/12/91 Taupo Racing Circuit, H.R.& S.C.C. Best lap, 58.0 secs.
29/12/91 Nelson Street Races, Circuit racing.
06/02/92 Ruapuna Raceway, Alfa Romeo Club. Circuit racing.
08/02/92 Levels Raceway. S.C.C.C. Three days circuit racing
15/03/92 Albany Sprint Races, Auckland. Timed circuit sprints.
07/11/92 Manfield Auto Course. Three day event. M.G. C. C. Best lap 1 min. 42.54 secs.
29/11/92 Taupo Racing Circuit, Club Meeting. H.R. & S.C.C. circuit racing.
20/12/92 Auckland Domain Hillclimb. Auckland Car Club.
14/11/93 Chelsea Hillclimb, Auckland. Vintage Car Club.
26/03/93 Taupo Racing Circuit. H.R.& S.C.C.
17/04/93 Hamilton Street Races.
21/04/93 Taupo Racing Circuit.
24/04/93 Manfield Auto Course. Circuit racing.
21/08/93 Tahuna Hillclimb. Thames Valley Car Club.
17/10/93 Bay Park Raceway. H.R.& S.C.C. Best lap 1 min. 25 secs.
04/11/93 Chelsea Hillclimb, Auckland. Vintage Car Club.
05/12/93 Taupo Racing Circuit. H.R.& S.C.C. Best lap, 60.1 secs.
18/12/93 Ngatea Sprint. Thames Valley Car Club. Standing quarter mile, 16.51 secs. Flying, 99.85 m.p.h.
08/01/94 Pukekohe Raceway. T.A.C.C.O.C. Meeting. Two days of circuit racing.
05/03/94 Otaua Hillclimb. M.G. Car Club. Best time, 56.2 sees.
20/11/94 Taupo Racing Circuit. Club Meeting. H.R. & S.C.C. circuit racing.
27/11/94 Chelsea Hillclimb, Auckland. Best time, 33.42 secs.
06/02/95 Targa Aoraki Four Day Rally. Christchurch, Queenstown, & Dunedin.
11/02/95 Wigram Wings & Wheels. Airfield circuit races. Country Gents.
01/04/95 Otau Hillclimb, Clevedon. M.G. C. C.
02/04/95 Meremere Drag Strip. Standing 1/4 mile, 15.98 sees.
10/09/95 Pukekohe Raceway, Auckland. Practice runs. T.A.C.C.O.C.
12/11/95 Chelsea Hillclimb, Auckland. Time 33.62 secs.
19/11/95 Taupo Racing Circuit Club Meeting. H.R. & S.C.C.
26/02/96 8/03/96 Fifty Years Anniversary Rally. Vintage Car Club. Route : - Christchurch, Hokitika, Westport, Hanmer, Christchurch. Also included, the Allandale Hillclimb.
20/10/96 Whenuapai Airfield Circuit. Practice runs.
10/11/96 Chelsea Hillclimb, Auckland.
17/11/96 Taupo Racing Circuit. H.R.& S.C.C. Driver, Alan Woolf. Best lap, 54.07 secs.
26/01/97 Whenuapai Airfield Circuit. Practice runs. H.R.S.C.C.
08/03/97 Otaua Hillclimb. M.G. Car Club.
04/05/97 Auckland Domain Hillclimb.
15/11/97 Taupo Racing Circuit Club Meeting. H.R. & S.C.C. Driver, Alan Woolf.
23/11/97 Chelsea Hillclimb, Auckland.
07/12/97 Vintage Car Club Tour. Glen Murray & Rangariri.
03/03/98 Auckland Domain Hillclimb.
15/11/98 Taupo Racing Circuit Club Meeting. H.R.& S.C.C. Driver, Alan Woolf. Best lap, 54.23 secs.
04/04/99 Pukekohe Raceway. T.A.C.C.O.C. Regularity Trial.
02/05/99 Auckland Domain Hillclimb.
19/09/99 Mercer Hillclimb. H.R. & S.C.C. Driver Alan Woolf. Time, I min. 27.03 secs.
04/11/99 Chelsea Hillclimb.
02/01/00 Veteran run. Muriwai.
03/04/00 Taupo Racing Circuit Club Meeting. H.R. & S.C.C. Driver Alan Woolf. Best lap, 53.14 secs.
17/09/00 Mercer Hillclimb. H.R. & S.C.C. Driver Alan Woolf. Time I min. 13.4 secs.
19/11/00 Chelsea Hillclimb, Auckland. Driver Alan Woolf. Time, 32.09 secs.
25/11/00 Bright Road Hillclimb. Driver Alan Woolf.
10/05/01 Thames Valley Car Club, Flying Quarter Mile Sprint. Car run with open exhaust pipes. With 15 inch wheels, 100 m.p.h. and 19 inch wheels, 102 m.p.h. Driver Alan Woolf.
--/09/03 Te Onepu Hillclimb. Driver, Roger White, 1 min. 11. secs. Driver, Martin Ferner, 1 min. 13.64 secs.
Previous Contents Next