Unintended Acceleration...

NASA from what I understand, got all their information from what Toyota chose to furnish. In the Oklahoma trial, expert witnesses actually got source code to analyze.

"Jim Thompson" wrote in message news: snipped-for-privacy@4ax.com...

I'm well aware of your involvement in the auto industry Jim. Didn't know that about you and limp-home though. Got to respect someone that stands by their principles and refuses to cave. Does help to be a certifiable genius and irreplaceable.

On a sunny day (Thu, 31 Oct 2013 15:24:09 -0400) it happened "Oppie" wrote in :

OK, I even have a pdf, and I gave a link to it here last year, of an internal report that describes the bug, google toyotamemoenglishtranslation1.pdf

I have just read these links you gave, and indeed from Barr's POV it is crap. But we must not forget ECC RAM and stuff like that is not that common in embedded systems. It must be clear to anyone that 'electronics' will not make a car safer (than simple mechanics). In airplanes you may have 3 the same systems and an arbiter so to speak, if any system disagrees... majority vote, but even then, no 100% guarantee. If Mr Barr looked at my latest code he would probably die drowning in tears, my PICs have only limited stack space, and lack some of the features he considers necessary...

Where is the limit? I do not want to justify Toyota, neither the F*ckupshima nuclear plant design, but hey, life goes on. Buy an old Mercedes diesel and you are OK...

If it is anyway like the environment I used to work in, then there is an incredible time pressure to get things working, not sure how many man hours Barr spend testing all that stuff, but think about it.

We can learn from Mr Barr's report though, I often worry about stack depth. But really and honestly, I would drive a car that was controlled by a controller I designed without all those features Barr says it 'needs'. Something is up to the driver too. For sure cars should not start accelerating by themselves, and coffee should not bee too hot. (Mc Donalds lawsuit, milions?) But would a decent driver not know a few tricks how to stop it? I have driven cars that .. well you do not want to know. US lawsuits... come to think of it : Should not have NSA been listening to the controller? Now you get internet, maybe remote software updates by the factory while you are driving? Do you think things will get better? LOL :-)

I'm certainly certifiable... not sure of which categories... though I've been a certifiable asshole since early Motorola days >:-}

I even forced out a Motorola VP when I was in my mid-twenties. ...Jim Thompson

Modern ignitions know when to fire the spark (and turn on the injector) because of the crank position sensor. If you lose that, the computer has no idea where the crank is.

Trying to "free run" the ignition would soon lead to it getting out of sync with the crank, meaning that before the engine died entirely you'd get some very interesting pops and bangs and possibly tongues of flame out of various orifices before the inevitable happened and the engine stopped turning.

Modern electronic ignitions for racers are more sophisticated than that

-- they stagger the miss so that it's not always happening to one cylender or set thereof. They also cut in more and more misses as the RPM goes up, until ultimately there's no ignition at all.

Modern bracket racers* use this to the point where a nicely-working RPM limiter is pretty much required for winning. They just floor the throttle off the starting line, and the RPM limiter keeps the car going at a set speed all the way down the track.

• Bracket racing is a form of drag racing where you declare the time you'll take to go down the track. You get scored by how closely you approach that time from above, and disqualified if you go under. Consistency is everything with bracket racing.

It's been 30 years since I last did any serious ignition stuff, but there used to be a coarse position indicator on the flywheel to handle such events.

Also, are you forgetting that the distributor (if there is any such animal in multi-coil designs) knows where you are (roughly). ...Jim Thompson

You still have pressure in the lines. And, a carbureted engine still has fuel in the bowl! A *second* of continued operation is enough to drive you into that wreck!

I've gone through a similar exercise with "theft prevention" in mind (aftermarket being far preferable to factory installed as it would be less ubiquitous -- less likely to be a "known" to a would-be thief),

In your specific case, "neutral" is your friend! :>

The biggest risk associated with uncontrolled acceleration, et al. is that it can *move* the vehicle without your control. You want to disconnect the engine from the drive train without losing everything that the engine also provides (brakes, steering).

Drop car into neutral (which, IME, is possible on *any* vehicle at any speed) and let the engine race. Worst case, you throw a rod, need a new engine -- but walk away from an *avoided* accident!

And, it doesn't cost you a penny to implement! ;-)

I know some engines have more than one sensor for timing. There is a crank position sensor for TDC and also a valve timing sensor.

If one fails, it is still possible to fire the plugs in a usable manor.

My attitude as well. ...Jim Thompson

Seems like a parts-per-billion risk. You'd waste hours of your life for, likely, seconds of probable return.

And, if you don't panic, can't you just shift into neutral?

Den torsdag den 31. oktober 2013 22.30.21 UTC+1 skrev Jim Thompson:

the is only one sensor hall/reluctance looking at a wheel with teeth, the Bosch timing wheel is 60 teeth with 2 missing, japanese i think

so with out the software continuously looking at the timing of the teeth and finding the missing ones it doesn't even know where the crank is

doubt there is any distributors left, everything is wasted spark or coil on plug

some have a cam sensor because for some number of cylinder or odd cylinder angles it is important to know which in which stroke to fire, and it can be good know to do injection in intake or exhaust stroke at some loads one might be better than the other

I believe some ECU do with out a cam sensor and find the stroke by looking the varying speed of the crank at low rpms

I don't know how they handle keyless but with Bosch has always been KL30 is permanent power, KL15 is ignition key switched power and that is what powers ignition coils and injectors

-Lasse

Absolutely.

I have had cars with mechanical failures of the clutch, throttle linkage and accelerator pedal springs at various times. Even one with a gear-box that decided to be extremely random about which gear it would let you have. All still limped home for remedial work OK. All of the recovery actions that should be common sense to all drivers (sadly it does not seem to be) always worked well enough for me. This was, however, before the ECU became an addition to the vehicles. Only had a failed hydraulic clutch system since then and still managed to get that home as well.

Now consider the emergence of the Electric Vehicles into the general populace. Having loked for the current "Construction and Use Regulations (UK), there was only mention of the charging points for such vehicles. Thankfully the United Nations has something (see UN100): "5.2.2.3 Unintentional acceleration, deceleration and reversal of the drive train shall be prevented. In particular, a failure (e.g. in the power train) shall not cause more than 0.1m movement of a standing unbraked vehicle."

Perhaps this should apply to all vehicles anyway.

Does the transmission selection go through the computer on some of these vehicles?

It always amazes me how easily people seem to panic/lose all sense of reason in these situations! :-(

I once lost a wheel bearing in one of the front rotors (disintegrated). The wobble in the rotor was enough to forcibly drive the calipers apart rendering the hydraulics useless.

I discovered this in rush-hour freeway traffic while driving in the left lane (65+). And, still quick-witted enough to think of stomping on the "parking brake" before ass-ending the bloke in front of me!

I drove the 20 miles home "hunched over" so I could keep my hand on that release to "unapply" the brake each time I stepped on it! :-/

Thankfully, didn't mangle the spindle so it just cost me a rotor and bearing (and an hour of my time)! Though, in the process, discovered that the shop I had brought the car to many years before had neglected to re-install one of the components that allowed the "parking brake" to work correctly -- which accounted for its poor performance in this situation! (maybe a GOOD thing?)

That was the event that convinced me not to let anyone work on my vehicles -- regardless of how "easy"/routine the task! :<

Doesn't say anything about how you can "enforce" a "brake". Or, disengage the drive if it doesn't want to! (would they call that "unintended ignorance of intended operations"?? :> )

When things *work* is not when you have the problems! :-/

Ouch! That would be unfortunate!

"Open the pod bay doors, HAL."

Could you do something similar to the EU politicians, they are certified idiots.

Den torsdag den 31. oktober 2013 23.55.07 UTC+1 skrev Don Y:

haven't a dual diagonal brake system been mandatory since for ever ?

-Lasse

would be the obvious solution, the engine would stutter at the revlimiter and not much more

and I don't think there's a car where the brakes cannot over power the engine

look at the statistics

it's not like it started going up after computer were invented

-Lasse