New technology is to replace mechanical controls with electronics. Now understand the problem. Every digital logic circuit is logic One when above a certain voltage. And logic Zero when below. Between those two levels is an undefined reason (because everything in the world is ternary). So that noise does not cause confusion, any signal in 'no man's land' means logic stays at its old value. Then noise does not cause failure.
Well, sort of. All noise is statistical. Therefore all noise will eventually cross that 'no man's land' threshold. We calculate the probability. So that your PC is reliable, that noise threshold might be achieved once every (maybe) 1000 or 10,000 years. No problem for you. A major problem for Toyota with so many millions of cars.
Described earlier is one method to avert disaster - see watchdog timer discussed earlier. And a problem. There really is no way to test for proper implementation of that watchdog timer. Quality (as taught by W E Deming) says quality must occur at the design level. That means top management must implement, enforce, and understand how that can be accomplished OR if it is being accomplished.
Mechanical designers did not have to be so careful. Now that brakes, accelerator pedals, windows, power steering, etc are all partially or fully electronically controlled, then software interfacing with hardware means everything Deming taught is even more important. Means that understanding even that above statistical fact is essential to reliable design.
How does one test for a failure that happen almost never - such as the throttle pedal? That is what every automaker has confronted. Every automaker has suffered the same problems in other variations before Toyota. But this time the problem was bigger news - probably because one event was so newsworthy.
Ford's problem with Firestone (now Bridgestone - they changed the name so that you would forget how many they killed) was so trivial and obvious compared to Toyota's. Ford eventually discovered Firestone knew they were killing people, lied about fixing the tires, and then denied the problem existed. In that case the problem and its solution was trivial simple.
Well, Toyota's problem has a similar story line. Toyota knew the problem (and some others) existed. But top management stayed in denial mode. We are now seeing new Toyota management trying to deal with a problem that existed maybe in 2003. And have only been at trying to eliminate it (and other problems such as the Prius braking software) for a year.
Trying to identify the transistor among millions that might be too susceptible to statistical noise failure when it only happens to a few of millions of vehicles. And the failure leaves no electrical history of the failure. Toyota has a major challenge. If I understand the problem from details in so many stories, I believe the accelerator pedal will not be (or may be only one minor) reason for failure. Appreciate why bad management means a statistically rare problem can make the solution almost impossible to find. And why missions (such as Apollo) did not have such problems because management understood what the engineers were saying. Therefore could avert problems before death resulted (ie Apollo 13).
Engineering and management techniques once good enough for mechanical brakes or Bridgestone tires is no longer sufficient in today's world of Deming quality, six Sigma, millions of transistors in each car (every car currently has about 50 motors), and world wide news.
|
