Robot Throttles
(Excerpt from Cycle World's new SportBike special edition)
When I was building racing motorcycles, the things that interested us—items we believed would give us quicker lap times—were things like ports and pipes and cams. But in the 1980s it became clear that without adequate tires and suspension, more power was a joke. The emphasis of tuning shifted to tires and chassis setup. This forced the development of more durable and cooler-running semi-radial tires, and pushed rapid suspension experimentation that also brought results.
Today, control of the very nature of engines is falling into the hands of software writers, so that there is a need for people who understand riding, engines and code writing. Usually, little good comes from long committee meetings among specialists in these separate areas, so we have to develop people whose understanding covers them all. At first, changes to engine powerbands could only be made by specialists back at the factory; people with degrees only reluctantly trust those who lack them, and vice-versa. But results were not forthcoming. The language and social barriers between trackside and R&D personnel were too great.
Honda took a leap and began trackside engine mapping. The rider would come in and with the usual impatient arm gestures explain to his crew chief the troubles he was having. Changes would be made to the engine powerband on the spot, and the rider would go back out in just a few minutes, while all the questions were still hot in everyone's minds.
Hey, where's the GPS guy? Multimillion-dollar F1 budgets mean teams can explore all manner of speed tricks, including satellite technology to pinpoint track position. There are shortcomings in this way of working. Softening a region of the torque curve to get off Turn 3 may rob torque that would be useful somewhere else. Over how wide an rpm range should the smoothing be applied? Riders have their own preferences here. Biaggi likes a wide brush, but Sete prefers a limited treatment. And if torque is taken away at some lower rpm, won't it rush back in like the tide as the engine revs up?
These are some obvious problems with
"passive" engine mapping, even though it is currently dominant. How
about an "active" system, one that knows where it is on the track and
can apply remedies only where they are needed? This has been the direction
chosen by Ducati, which has developed its engines to make high power and then
applied active systems to make that power usable. Up to now, this has not
brought success, but early 2006 testing saw Ducati at or near top lap times and
lead rider Loris Capirossi won the opening GP in
Reportedly, Formula One has used GPS data for some time now, and Ducati's system is said to combine GPS with angle-of-lean and acceleration/deceleration data from an on-board stable platform—an outgrowth of the ultra-precision guidance systems that made ICBMs possible. Now the computer knows where the bike is, how far over it is leaned and what its state of acceleration or braking is. But even this is not enough, for every corner has camber that varies, surface grip that varies and possibly change-of-elevation, too, as in Laguna Seca's famed Corkscrew. All this can be mapped—probably Ferrari has 3D maps of all the circuits on which it runs.
The rumour mill has Ducati also experimenting with GPS as an aid to traction, though as Loris Capirossi demonstrates on the '06 Desmosedici, MotoGP will never be easy.
Why go to all this trouble? Wouldn't simple anti-spin do the job, either comparing front and rear wheel surface speeds to control slip rate, or using tachometer data to limit rate of rear tire acceleration? An obvious problem is that tire rolling radius varies with lean angle—and because front and rear tire profiles aren't the same, this doesn't automatically factor itself out. The second problem is fuel tank size limit. Anti-spin or accel-limiting systems may have to make deep power cuts if the rider grabs a handful, and that wastes fuel.
Initial power cuts are made by retarding the ignition, which makes the engine less fuel-efficient. The limit to power reduction made in this way is exhaust valve temperature—operation with retarded ignition pushes valve temperature up rapidly. If a deeper power cut is needed, fuel mixture may be leaned out. Both of these processes are smooth and do not upset tire grip as cutting out spark or injection events do. Leading every lap but the last one is hard to take, so fuel economy is a serious concern.
This means it would be nice to be able to control torque with only small power cuts. That would be possible if our on-board system could continuously calculate approximately what the back tire's grip is at all times, feeding appropriate throttle, leaving only small torque errors to be corrected electronically.
There is a lot of discussion and experiment going on as to how rider and computer should share power. Honda has its differential throttle system, in which the rider's and the computer's inputs are summed and then sent to the throttles. In this system, either can, if necessary, overrule the other.
In other systems, the rider controls two cylinders and the computer controls two.
Chips ahoy! Yamaha's 2006 YZF-R6 ushered in "fly-by-wire" for street bikes. That's just the beginning of what next gen on-board computers might do for motorcycles.
How about failure modes? If the computer suddenly decides "Drive C does not exist" or water condenses inside a connector, can the system produce an uncommanded full throttle? Early systems would do just this, so you can be sure a lot of effort has gone into preventing it. Once a rider is spooked by an electronic system, he may very well stay spooked for days or weeks. Racing contains enough danger without adding new kinds. This means the system people must get it very close to right before the expensive star rider gets on the bike.
Will auto-throttle bikes more complicated than Yamaha's R6 be in showrooms soon? Probably not because of the difficulty of making such systems fail-safe or even fail-passive. But a lot will be learned and some factory racers may already be able to just pin the throttle and "let the computer do the rest." It will be fascinating to watch these systems evolve.
Kevin Cameron