Rolls-Royce’s first successful cast-block engine, the Kestrel, was so successful in airplanes such as the Hawker Fury biplane that it sold the British on the water-cooled inline engine format. A scaled-up version of the Kestrel became the Buzzard, and was used in the Schneider Trophy racing seaplanes, producing up to 2500 horsepower (but not for long!) in 1931. The dimensions of this engine were later revived in the Griffon, designed to produce more power than the Merlin while having approximately the same outside dimensions. Although very successful in certain applications, however, the Griffon never managed to outshine its smaller brother.
The Merlin was another scale-up of the Kestrel design, intermediate in size between the Kestrel and the Buzzard. When originally conceived, it was intended to produce 750 horsepower, but by the time it was introduced, it gave 1000. The version powering most later Mustangs gave 1650, and there were versions rated at over 2000. Much of this improvement was made possible by the rapid advances in bearing design and fuel knock-resistance of the time. The Buzzard racing engine achieved very high specific output, but with a maximum TBO (Time Between Overhauls) of 10 hours. Of course, fuel dragsters of today produce more horsepower than the Buzzard out of far fewer cubic inches, but their TBO is measured in seconds!
The original Merlin used one-piece head/cylinder blocks, with the sleeves clamped down onto the crankcase with the cylinder blocks, which required that all sleeve recesses in each block be exactly the same depth, and all sleeves on a side be exactly the same length. This system proved to give somewhat unreliable sealing; the worst result was leakage of water into certain cylinders, causing damage on startup. An interim fix was provided by the addition of pull-up bolts at the tops of the sleeves, and the ultimate solution was separate cylinder heads and blocks, with the tops of the sleeves clamped between them. In hindsight, this was the obvious solution, but the original design was based to some extent on previous inline aircraft engines, which typically did not have separate cylinder heads.
Another feature of the Merlin, found on many V-type aircraft engines, is the 'fork-and-blade' style of connecting rod. In most V-type automobile engines (and a few 30's aircraft engines), a pair of identical connecting rods sit side-by-side on a single crankshaft journal. In order that each piston be centered over its rod (to avoid uneven loading), the cylinder banks must be offset. In this system, the connecting rod bearings are the same width as the rods, which is to say fairly narrow. With the advent of steel-backed bearing inserts, towards the end of the 30's, narrow rod bearings became practical, and today's engines can endure truly amazing bearing pressures (witness the BMW Formula 1 turbocharged 1.5 liter four-cylinder engine of the 80's, which could produce in excess of 1200 Horsepower!). When the Merlin was designed, however, babbitt bearings were the standard. Babbitt is a very soft metal, usually an alloy of tin and lead, which makes an excellent bearing material on steel surfaces, even soft ones. However, since babbitt is so soft, it cannot withstand very much pressure. In the original Merlin's fork-and-blade connecting rod design, a babbitt-lined split steel shell fit over the entire journal, with the fork rod bolted to it, and the blade rod sliding on a bearing on the shell's outside. Thus the two rods share a single large bearing surface, and since their power strokes do not overlap, they very effectively share a single bearing with large area. The Quarter-Scale Merlin also uses fork-and-blade connecting rods, with a slightly simplified design.
Although the Quarter-Scale Merlin is intended to look as much like the real Merlin as possible, it is not intended to be an exact scale model, right down to every external and internal detail. This would be a worthy goal, and would produce an extremely interesting engine. However, it would greatly increase the amount of time required to produce a complete engine, since every detail requires time to implement. Simplicity is favored to further the goal of producing a running, functional engine as easily as possible. For example, the present design uses two valves per cylinder, instead of four. The usual reasons for using four valves per cylinder are: to maximize valve area for good flow at high rpm, to allow better cam profiles because of lower valve weight, and to avoid the problems of valve cooling associated with large valves. If the scale Merlin were intended for all-out performance, these factors could possibly (but not necessarily) become important; however, it is primarily intended to a scale airplane with realistic sound; this means it must operate efficiently as close to the real Merlin's rpm range (a normal maximum of around 3,000 rpm) as possible. For an engine of this size, 3,000 rpm is very low, and using four valves per cylinder would give absolutely no advantage.
Another area of simplification on the prototype was the use of glow plugs (instead of spark plugs), for the simple reason that that was the easiest way to get it running. However, the glow plug/alcohol-fuel method brings its own set of problems. The prototype was equipped with an oil pump, which drew oil from the sump, filled with castor oil. Used and pump bypass oil returned to the sump. After a few minutes running, the oil pressure started to drop, and it was obvious the oil was becoming contaminated. Surprise, surprise, there was a great deal of blow-by, with 12 single-ring pistons and plenty of excess fuel. It seems that most engines which run on alcohol have this problem. In racing, the solution apparently is to change the oil after each run, or to run what is called a 'total loss' system, where no attempt is made to reuse the oil pumped into the engine (great for the track!). The latter was a temporary solution to the prototype Merlin's problem; the oil was in a separate tank, with the oil pump bypass returned to it, but the oil which was actually pumped into the engine drained into a container. In regular four-stroke model engines, the ring-bypass effect works to our advantage, because of the large amount of oil that can reach the crankcase. Actually, there is probably no reason why the quarter-scale Merlin couldn't run this way (using conventional oil-mix glow fuel instead of having an oil pump) since it seems to work well with large four-stroke model engines. For maximum longevity, an oil pump is probably best, but oil mist lubrication may be satisfactory. Another solution, of course, is to set the engine up to run on gasoline, with spark ignition. This will be attempted on the current model. The real Merlin has two magnetos, one on either side, with one firing the inside plugs, and the other the outside. It would be possible to build functional magnetos for this engine, although they would almost certainly require coil boosting at low rpms because of the inherent ineffficiency of small magnetos at low rpm. Since this is the case, it would seem logical to just use the magneto housings as distributors, and skip the extra complexity of the magneto windings.