One purpose of dynamometer testing is to measure an engine's horsepower and torque curves. This kind of testing is a long tradition at Cycle World. We use our own dynamometer rather than simply accepting figures published by motorcycle manufacturers. Why? Because then our readers can rely on the fact that all the test results we present come from the same dyno and the same testing technique.
Another and arguably more important purpose of dyno running is to develop an engine's reliability and endurance. The US Navy was the first organization to require the aircraft engines it bought to pass rigorous dyno endurance and dive testing. Racing engines, too, are developed through full-length race simulations on dynos. One of the first to do so was Norton's famed engineer Joe Craig, who regularly ran 2-1/2-hour Isle of Man TT simulations on the dyno to be sure that engines and their components could stay the course. Run it 'til it breaks, analyze the failure, design better parts, and test again. I learned from former American Honda race manager Gary Mathers that Honda for years required new products to pass a 2,000-hour life test sequence.
There are different test methods. In early days, engines were step-tested at constant rpm, increasing by 50- or 100-rpm steps, taking data at each step, with that data being graphed out to give horsepower and torque curves. Some testers were even more picky, insisting that results from step-down testing be identical with step-up testing. Some dynos absorbed engine power by hydraulic means and others employed electrical absorption. Horsepower was calculated from observed torque going into the “brake” (the proper name for a dyno) and engine rpm. Step-testing is very hard on engines because they must spend long periods on full throttle and high revs. Lots of parts!
A big change in the cost and availability of dyno testing occurred when Dynojet began to offer its acceleration rear-wheel roller dyno for sale at a widely affordable price. A great many shops and individuals bought such dynos. In this kind of testing, the bike's rear wheel is strapped down against a heavy roller which the bike's engine must then accelerate. Cycle World's dyno is of this type, as you can see from our test videos. This test method simulates the acceleration of a motorcycle on a road. Horsepower is calculated by knowing the rotational inertia of the roller, its speed, and how long it has taken for the engine to accelerate it to a given speed. If horsepower and rpm are known, torque can be calculated from them as well.
Purists instantly objected to this kind of testing, saying it had to be inaccurate because of possible variable wheel slip, because the test did not take place in a proper laboratory setting, and so forth. But then some sensible individual asked this question: What aspect of motorcycle performance really interests most of us? We love motorcycles mainly for their vivid acceleration so doesn’t it make best sense to test them that way? Yes, an engine dyno is better for serious, detailed engineering development, but a rear-wheel-roller acceleration dyno does a fine job of simulating what we want our motorcycles to do: accelerate.
Three basic kinds of horsepower figures may be found in the literature: crankshaft horsepower, horsepower at the gearbox sprocket, and rear-wheel horsepower. For the very same engine, all three will be different. Coupling the test engine’s crankshaft directly to a dyno eliminates the friction and oil churning losses of primary gear, gearbox, and rolling losses of the rear tire. Running a chain from the engine’s output sprocket to the dyno eliminates rear tire rolling loss but includes primary gear and gearbox losses. Rear-wheel horsepower includes all three kinds of loss.
Once again, the voice of common sense: Every real motorcycle, being ridden in the real world of street, highway, or racetrack, also includes all three kinds of loss.
Now the purist’s objections are heard: How can you hope to learn anything useful with all those losses? All over America, riders like vintage racer Todd Henning have run thousands of hours of comparative testing on rear-wheel dynos and have won races and championships with the power they found. Okay, fine, if you need to measure the difference in friction loss between two types of top piston ring, a rear-wheel dyno is not the best tool. But rear-wheel dynos have vastly increased the number of people doing useful dyno testing, an overall gain.
An expression often heard is “dyno correction factor.” What is it? Because engine power depends on atmospheric density (engine power is proportional to the mass of air it burns), power varies with the weather. When the barometer goes up or the temperature comes down, air density increases (and vice versa). If I test at my local altitude of 278 feet above sea level, how can I compare results with a dyno operator in Denver, whose 5,400-foot altitude reduces local atmospheric pressure to 86 percent of sea level? If I find my 600cc Supersport engine is making 126 hp at 278 feet altitude, then send the engine to Denver to be tested there, the Denver dyno operator will find only 86 percent as much, or 108 hp.
To bring such different results together, we must correct for altitude and changing weather. This is what the correction factor is intended to do. In NASCAR, where tiny differences can add up to a winning lead at the end of 500 miles, dynos are operated in sealed test cells at controlled standard atmospheric pressure and temperature.
Now we know why an engine, tested at Bob’s Cycle Shack on Tuesday, a rainy, muggy low-pressure day, makes less power than if we test on a cool, clear Friday with a high barometer over at Dyno Dave’s shop.
If we have suspicious minds, we might imagine that an unscrupulous individual might “cause” his dyno to give attractively higher numbers than competing dyno operators. Let the buyer beware.
Another point that can lead to variation is temperature rise in the dyno cell. Some test cells have big fans to keep temperature constant, but other shops just tuck the dyno in wherever it will fit, and if engine heat happens to drive room air temperature over 100 degrees, it’s not even worth a shrug. Yet that high temperature will result in a lower dyno result.
Engines are variable as well. Let’s say we build up a hot 600cc Supersport engine and break it in. We do a static compression or “leakdown” test and find only 2.5-percent leakage straight across the four cylinders. Really good. And so are the dyno results. This engine is sealed up tight.
Now we go on the circuit, running the bike in several races. If we run the leakdown again, we’ll get bigger numbers, showing that rings and valves are no longer sealing as well as they did right after build and break-in. We dyno our now “baggy” engine and find it’s down several horsepower from its best. This isn’t weather, tire slippage, or “optimistic” dyno operators. It’s just the natural result of engine operation. Nothing—least of all horsepower—is forever.
And there’s more. The hotter an engine is, the more the air entering it expands, loses density, and makes less power. So drag racers frequently run their engines cooler than a streetbike’s operating temperature to take advantage of the temporary ability of a cool engine to make more power than a hot one. After a run, bikes return to their pit where they are deliberately cooled off to the desired temperature before the next run.
Another case of temperature effect was shown me by dyno operator Jim Czekala of DynoTech Research. After lunch break, we started up a Yamaha TZ250 two-stroke twin and made a pull, noting that the power dropped off quickly after peak.
“Now watch this,” Czekala said as he made a second pass while the engine’s two fat exhaust pipes remained very hot. This time, power remained high for 200 to 300 revs more beyond peak. Why? Because the hot pipe “acted shorter” than the same pipe when cool. This is because the speed of sound rises with gas temperature. Two-stroke GP bikes made use of this fact to increase over-rev power (power given beyond the peak) by retarding ignition timing, thereby dumping more heat into the pipes. Subtle, these humans.
If a certain bike makes 132 peak horsepower on the Cycle World dyno, that says nothing about what power might be measured if a second bike of the very same model were tested as well. Back in the 1920s, Harry Weslake bought and dyno tested three Rudge racing motorcycle engines. He found they made 25, 26, and 29 hp respectively. He wondered why? His hypothesis was that the differences (16 percent from least to most!) might be caused by differences in the airflow resistance of their ports. Upon constructing a flow-measurement rig, he found that this was substantially the case. This is the basis of modern engine flow testing.
But why should the airflow be different from one engine to the next? The answer has to do with how accurately the casting cores for the ports can be located in the mold; the port in this head might be nicely centered on the valve-seat ring, but on the next engine it might be offset a millimeter or two, creating a flow-resisting step there. What if there was a chip of metal under a machining fixture that held the head as it was finished? That could offset how the intake system lines up with the port in the head. This is why racers running in production classes try to get their hands on as many cylinder heads as they can find. They flow-test every one, keeping the best-flowing and returning the duds.
Yet another source of horsepower variation is the manner in which an engine is broken in from new. A good break-in achieves an intimate seal between piston rings and cylinders, but if the parts fail to bed in, the wear surfaces of the rings just show a few shiny spots of true contact with unworn areas of leakage between. Factory-built engines today require less break-in than formerly because prelapped piston rings and high-finish cylinder walls are now common.
But when older engines are rebuilt, break-in remains an issue. Is it better to “baby” a new engine, relying on steady but light friction to accomplish the last manufacturing process, bedding the rings to seal against the cylinders? Or is it better to apply heavier pressure in a series of applications, with periods of relief between? This is complicated by the fact that modern oils contain such effective anti-wear additive that a “babying” break-in can result in poor sealing that never improves. The smart money is on the practice of applying substantial throttle for a time, then allowing an interval in which the oil system washes away wear particles, and repeat. The equivalent for a dyno break-in is to make a series of pulls. For an effective, well-sealed break-in, substantial throttle has to be used in a series of applications, for otherwise the anti-wear additive may effectively prevent break-in.
What’s the bottom line? It is that horsepower is not a fixed, inherent fact about an engine but an actively varying quantity that depends on a great many variables.
Another time, we’ll discuss what we can learn from the shapes of power and torque curves.