Test One Thing at a Time. Or Not?

Technical Editor Kevin Cameron shares his wealth of motorcycle knowledge, experiences, insights, history, and much more

Kevin Cameron's Top Dead Center banner
Technical Editor Kevin Cameron shares his wealth of motorcycle knowledge, experiences, insights, history, and much more.Cycle World

Throughout my life I have been told over and again – usually by old-timers – that in engine development it is essential to test only a single change at a time. For if there is an improvement, how else will you know which change is causing it?

This seems as smugly self-evident as a “Ted Talk,” but it’s important to note that it was also old timers who placed Galileo under permanent house arrest for daring to say that the earth travels around the sun.

Therefore it was a breath of fresh air to read the words of Jack Williams, the man who developed the classic AJS 350 single, in Institute of Mechanical Engineers paper C142/78, discussing that lengthy and productive program;

Concluding Remarks, item (2);

"It was shown also that the results given by single experiments in isolation are often disappointing – and that it is only when a number of potentially improving features combine that the individual effects can be fully appreciated." (My emphasis)

In other words, the very opposite of the “test only one thing at a time” idea.

Why, for example, would you test a camshaft intended to give improved power at higher-than-normal rpm with the exhaust pipe whose length was chosen to work best with the previous, lower-rpm cam?

The classic case of 'test only one thing at a time' supposedly occurred in California during the 1970s, when a well-to-do racing enthusiast decided to get to the bottom of the Big TZ250 Question; is there or is there not a way to improve on the performance of the stock Yamaha 250 production racer? At the time, the official and widely-believed explanation of why factory bikes were faster than privately-entered stockers was "careful assembly." What could this mean? That Kel Carruthers (then road race tech chief for Yamaha US) took, say, 30 minutes to ever-so-gently lower a new crank into the upper crankcase (surely in a Class 1 cleanroom), while devil-may-care privateers tossed the crank into the case from across the shop?

The wealthy sponsor ordered thirty cylinders from YPDI and when they arrived, each was slightly modified in a different way. Then each was built up on a dyno engine with new pistons and rings, broken-in, and then run to produce full power and torque curves. All were inferior to the stock curves.

Excellent – the principle is proven?

Yet at the same time, a select group of experimenters here and abroad knew that if the exhaust port of a C-model 250 were raised one millimeter, 20-mm of length taken out of the exhaust header pipes, .025” milled off the cylinder head (and water seal groove re-cut), ignition timing retarded to 1.6-mm BTDC and the final drive re-geared, the resulting combination would run away from a stocker like it was stopped.

It is only when the concerted effects of several changes, all working toward a single result, are combined, that a useful improvement is achieved.

Raising the exhaust port enabled spent combustion gas to leave the cylinder more quickly – as it must do if the goal – making power at higher rpm – was to be achieved.

Twenty millimeters of length was removed from each header pipe so that the returning positive exhaust wave could arrive properly timed to block the exhaust port at higher-than-stock rpm.

Raising compression by itself reduced exhaust gas temperature, causing the exhaust pipes to resonate 500 rpm lower than stock (many a privateer tried this only to achieve this disappointing result – and go back to a stock head).

Retarding the ignition timing simply did what later electronic ignitions would do automatically – give the engine the timing that would produce peak combustion pressure just as the piston was about to descend on its power stroke. Stock timing was a compromise, but the modified 1.6 timing was chosen to work with the other three changes to boost high rpm torque.

Making each change by itself would have had little or no result, for the rest of the engine’s details would still have been optimized for droning stock performance.