Building Better Bikes

[Admin]

Kevin Cameron has been writing about motorcycles for nearly 50 years, first for <em>Cycle magazine</em> and, since 1992, for <em>Cycle World</em>. (Robert Martin/)

When I visited its test center in the south of France for a presentation in November of 2019, BMW revealed some of its computer/sensor-based testing methods. Compact data recording/processing units are carried, and such things as elaborately instrumented wheels are employed. The goal is to assure that its products will perform reliably in the actual conditions of their use.

Twenty years earlier, a visit to Minneapolis Testing Service (MTS) in 1999 showed how road testing can be greatly accelerated and made much more accurate by supporting the test vehicles on fast-acting hydraulic struts (“posts” in the language of the testers) that can reduce a brand-new highway diesel tractor to the equivalent of million-mile worn-out scrap in just a few hours. This type of testing, being implemented with machines rather than carried out on the open road, gives repeatable results.

Yet men we know from our own era began their careers as old-time factory road testers. Eraldo Ferracci described his work in the 1960s—being sent out each morning on a prototype Ducati to ride to a distant city, grab a quick lunch, then return in the evening, dodging real traffic and real potholes the whole way.

When BSA was about to introduce its parallel twin in the late 1940s, the company employed that  same time-honored system—riders in all-weather gear setting forth every morning, rain, fog, or sun, to rack up the miles. Often it was a one-way trip, with the bikes returning on flatbed trucks; the engines were breaking their crankshafts in half. Soon the problem was overcome as it had been in the auto business—by placing the fillets that blend the journals to the crank cheeks in permanent, crack-resisting compression by rolling.

Ducati has a horrible frame-breaking machine that basically resembles a large rotating cam to which a bike is strapped. Suspension is relentlessly pounded by this electrically driven torment until something fails—a weld breaks, bolts loosen, a less-than-perfect design gives up.

Test Until Failure

Long ago it was the US Navy’s aviation program that adopted the “test until failure” program that has been a model ever since. (In January 1911 Eugene Ely showed that a Curtis pusher biplane could land and take off from a specially built deck on a ship—the birth of the aircraft carrier.) Failure in flight over the sea has especially grave consequences. (Does the current acronym ETOPS really stand for Extended-range Twin-engine Operations Performance Standards? Or is it shorthand for Engines Turn Or Passengers Swim?) The Navy’s program to improve aircraft-engine reliability led to the development of a tough 150-hour qualifying test as a precondition for flight testing to begin. Former American Honda racing manager Gary Mathers recalled that Honda long used a 2,000-hour product test cycle (the equivalent of 83 days of continuous running). It was surely developed to make sure that Honda products would be trusted by American consumers. They were.

When I visited Kawasaki in 1972 I saw a lighted cabin at the end of each production line. Inside was a double-roller dyno on which each completed bike was given its functional test before being shipped. Fuel and oil were added, the engine started, and electrical tests performed. Then the engine, clutch, and transmission were tested by acceleration through the gears, with separate tests for brake function. Once the cycle was finished, the liquids were removed and the machines were ready for crating. I have since seen many such test cabins in other factories.

The late Don Brown, an experienced motorcycle market analyst, described what can happen when testing is neglected. BSA in the 1960s had adopted what the English called “American-style management.” The central idea was that the less you know about the product, the better you can promote and sell it. Pure marketing unencumbered by distracting motorcycle experience! On this basis in 1966 some 4,000 bikes had to be returned to England to be re-manufactured because they had suffered damage from a new policy of trying to hit a price point by cutting employee pay. Why weren’t the resulting problems detected in testing? Presumably because the Tall Foreheads in management with their marketing degrees had decided testing was too expensive.

Here it must be pointed out that Japanese bikes were by then bringing to market advanced features such as electric start and overhead cams, but at “kickstart prices.” This was not achieved by cheap labor but by use of modern manufacturing and quality-control systems that cut production cost. Remember Dr. W.E. Deming’s famous words: “An increase in quality is an increase in production” (by producing fewer defective parts that must be scrapped). Deming was a quality-control specialist who found his ideas rejected in the postwar US but welcomed in postwar Japan.

In 1971 at our dealership in Arlington, Massachusetts, we received a shipment of 250cc singles badged as Triumphs, arriving with a long list of predelivery warranty work that the dealer was expected to perform. This included removing all the engines from their frames to correct internal faults, plus a head-gasket replacement. The fit of the engine into the frame mounts was awful! Engine bolts had to be hammered out—a shameful process that makes any mechanic wince. What were we to do? File each mismatched hole to perfection by hand? Send the parts to a machine shop? Who’s paying? The dealer? The customer?

More service bulletins came with the new five-speed Triumph twins arriving later that same year. Transmission shafts were being bent and gears broken. Customers with bikes waiting for unavailable replacement parts were unhappy.

The choice is this: Test in your own factory or leave testing to dealers and customers in the field—because one way or the other, such testing is inevitable. Which does a better job of supporting your product’s reputation? The answer is clear: Better to build in the necessary quality during manufacture, rather than shuffle problems downstream.

Automated manufacturing and gaging are not new. Friedrich Fischer conceived his ball grinder in 1883, and 13 years later his plant was producing 10 million balls per week. When in 1944 cast cylinder heads for aircraft engines reached the limit of their cooling ability (fins could not at the time be cast closer than 0.200–0.220-inch pitch)  the manufacturing process was switched to forging, followed by machining not only of the closely spaced cooling fins, but of every feature—rocker boxes, combustion chamber, intake and exhaust ports. This complex machining was performed not by hundreds of kindly old pipe-smoking gents in leather aprons and wire-framed spectacles, but by cam-controlled automatics. That’s right, folks—late-model B-17 bombers were powered by engines whose heads were machined-from-solid, just like the much-admired “billet parts” of the present.

Such expensive parts require many machining operations. Often in such cases, inter-stage quality checks are made so that a defective part can be scrapped before any more process time is wasted on it.

The Myth of Old-World Craftsmen

Once a well-engineered production system has been put in place it becomes possible to make unlimited numbers of reliable products—such systems, like the human body, are self-managing. This is the antithesis of the outworn idea of the “old-world craftsman” who makes each part individually, then by filing, lapping, and hand-scraping fits them together. No two parts are quite the same, so a part from one machine may not fit another. If you need a new part, you can get it only from him. The work is beautiful but also costly and slow.

We can either admire or be horrified by the statement that it took seven hours for a skilled specialist to properly fit the cam drives of bevel-drive Ducatis like the one on which Paul Smart won the 1972 Imola 200 roadrace. But we know the opinion of Ducati’s chief engineer Dr. Taglioni from what happened next. He replaced the time-consuming bevel drive with quick-to-assemble toothed belts and pulleys, introduced on the Pantah in December 1979.

The idea of making expensive handwork unnecessary is not new. In 1798 Eli Whitney accepted a contract to produce 10,000 muskets for the US War Department. In a legendary demonstration, he laid out a quantity of parts produced by unskilled workers using simple tooling and challenged officials to assemble muskets from parts chosen at random. All functioned normally.

Even earlier, German organ builder Arp Schnitger (1648–1719) solved many problems of servicing. He built 150 instruments by maintaining stocks of commonly needed parts rather than having to assign a skilled fitter to perform the work in a distant church. “Ah, you say you need quantity three of 93310-42418-00? We can ship this afternoon.”

Back in the mid-1930s when “the two Phils” (Phil Vincent and Phil Irving) chose the adventure of designing and building their own engines (at first a 500 single, and later a 1,000cc V-twin), Vincent was too small a company to afford anything beyond simple road testing. The company took bikes to race in the Isle of Man with no idea what to expect. When pistons and valves turned to junk it was the chief engineer and colleagues who bodged together improvised solutions on the spot. They were rewarded with a seventh place in 1935—behind a factory Guzzi in first, three factory Nortons, and two factory NSUs.

Meanwhile, Joe Craig at Norton was running long dyno and track endurance tests to guarantee that his factory overhead-cam “Manx” Norton singles had the quality and performance to complete the seven grueling TT laps reliably. Any part that failed was analyzed, redesigned, prototyped, and tested until it functioned properly. This is essentially the same as the 150-hour engine type test that stood behind aviation.

Computers can shorten product development time by finding and addressing problems in the design stage, but physical testing—either by machine or a determined test rider—remains central to reliability.

View the full article