When the evidence is hiding in plain sight
When Honda first went GP racing in the early 1960s, its four-cylinder 250 was built just as Honda’s production twins were—with a horizontally split crankcase having main bearing saddles bored half in one case, half in the other. When the crankshaft with its ball or roller main bearings was set into the upper case and then the lower case was bolted into place, the crank was well-supported by the entire structure. Honda’s early fours also included both gearbox shafts in this structure, the saddles for their bearings likewise being bored half in the upper, half in the lower case.
For its racing engines, Honda soon gave up this apparently rational and robust construction, switching to one in which crankshaft main bearings, in the form of “pillow blocks” (a bearing in a footed housing that can be bolted to a flat surface), were bolted to the underside of the cylinder block. The whole idea of strongly supporting the crank bearings between upper and lower case halves was given up, and the lower case served only to contain oil, with no structural function. Its material was changed from the aluminum it had been in earlier designs to lower-strength magnesium. This is the way the RC165/166 in-line six-cylinder engines of 1964 to ’67 were built.
This puzzled me, but I somehow dismissed the “why” question by noting that Honda, at the same time, gave up the use of heavy, large-diameter ball main bearings (on production twins of the 1960s, these are really hefty) in favor of much more compact small-diameter needle rollers. This saved both weight and friction. Yet that did not explain what made Honda engineers (probably none other than Shoichiro Irimajiri himself) willing to give up the structural robustness of clamping crankshaft bearings between crankcase halves.
A red herring also entered my thinking. When Gilera built its first air-cooled 500cc four in 1947, it was given pillow-block-style main roller bearings that bolted to the underside of the cylinder. This is similar to how countless millions of auto engines have been and still are built. Half of each main bearing housing is machined in the underside of the cylinder block (look at any traditional American V-8), and the other half is machined into a bolted-on main bearing cap. It’s not a pillow block but is the next thing to it.
It’s no mystery why the early MV four-cylinder racers were designed the same way—count Agusta had hired away Gilera’s designer Piero Remor. But when the MV Triples appeared, Remor had left the picture. The first Triple was a 350, but in rapid increases of bore and stroke, it became the 500 that would be Giacomo Agostini’s favorite machine. The Triples had full pillow-block
Those jewellike sixes peaked at 18,000 rpm...
construction using small-diameter split-outer-race needle/roller main bearings—very similar in effect to what was coming from Honda’s race department.
A simple-minded theory would propose that one company copied the other, but it’s far more likely that both were responding to common problems of high-rpm engines.
In recent years, Japanese new-model brochures have made much of saving horsepower by putting vent holes through main-bearing saddle-support webs so air being pushed into the crankcase by a descending piston can more easily and with reduced pumping loss reach and follow the nearby rising piston. Without the holes, this back-and-forth pumping process is very much like those screen-door closers that hiss as they close because they keep the door from banging by pumping air through a restriction. At high rpm, several horsepower can be saved by providing cylinder-to-cylinder crankcase vent holes.
Yamaha in Motogp has used one Formula 1 approach to solving this pumping-loss problem: It provides a crankcase evacuation pump that pulls the pressure of crankcase air and vapor low enough to nearly eliminate pumping/windage loss. In the 1990s, when we cheered favorites in AMA 600 Supersport racing, we had no idea that similar crankcase evac pumps had been cleverly hidden in the gearboxes of certain brands. Very hush-hush.
The other approach is to isolate each cylinder’s (or each V pair’s, in the case of V engines) crankcase, provide each with its own scavenge oil pump, and just let the piston(s) compress and expand the air beneath without loss because it is not being forced through orifices or confined spaces.
Honda engineers were well-aware of this kind of loss. A look at the lower crankcase of its RC-161 four shows fair-size holes to allow oil drain-back, and easy cylinder-to-cylinder crankcase air pumping.
By adopting pillow-block construction, Irimajiri eliminated nearly all resistance to cylinder-to-cylinder crankcase airflow. I was embarrassed at how obvious it was.
Those jewel-like sixes peaked at 18,000 rpm, so their potential for losses from crankcase pumping/ windage was huge. The pillow-block solution, by eliminating the traditional obstructive bearing saddles and their supporting webs between cylinders, must have saved a bunch of horsepower that would otherwise have been consumed huffing and puffing that air back and forth through restrictive holes, slots, or other half-measures.
Then I remembered another thing. The V-5 four-stroke that Kenny Roberts’ GP team designed and built in the early to mid-2000s never came close to its horsepower goals, and the back chat was that the problem was “something to do with the crankcase.” Could it be that the missing horsepower was “gone with the wind,” blowing rapidly from cylinder to cylinder at 16,000 rpm?
I believe that Honda’s present RC213V V-4 Motogp engine employs the loss-free F1 separateand-sealed crankcase system. Instead of pushing crankcase air back and forth through restrictions, each V pair of pistons just compresses and expands the air under them—without loss. Each sealed crankcase has its own scavenge oil pump. Honda coyly calls this system a “semi-wet sump” because, although the two crankcases (one for each crankpin) are separated from one another, there is a common underengine oil sump below.