PDA

View Full Version : cosmic rays to blame for Toyota's gas pedal problems?



novaderrik
2010-Jul-31, 05:17 PM
i just came across this article (http://www.livescience.com/technology/toyota-recall-cosmic-rays-100326.html) that says that cosmic rays might be to blame for the problems that Toyota had with their cars deciding to go and not want to stop and figured it might be a good question to ask here..
is there any merit to this, or did someone just catch a show on the Science Channel about cosmic rays and decide to run with it?
and if it is possible, then why does it only affect Toyota vehicles, and not vehicles from every other manufacturer that uses similar components made by the same suppliers?

Rhaedas
2010-Jul-31, 06:05 PM
Why just Toyotas? And why cars, and not other electronics? Anything else, like TVs, PCs, iPods, been wacky lately? Some comments agree that maybe better shielding is one approach, but don't electronics normally have to be shielded from other radiation anyway, like the ignition system?

Seems a bit of a reach for an explanation, when the easiest one is simple user error, combined with a sue-happy society.

mugaliens
2010-Aug-01, 06:18 AM
"or did someone just catch a show on the Science Channel about cosmic rays and decide to run with it?"

Bingo times 20,000,000,000,000 whatever. This is such ridiculuos nonsense I'd be happy to appear in any court of our land to refute any any such claim six ways to Sunday.

John Jaksich
2010-Aug-01, 11:07 AM
Why just Toyotas? And why cars, and not other electronics? Anything else, like TVs, PCs, iPods, been wacky lately? Some comments agree that maybe better shielding is one approach, but don't electronics normally have to be shielded from other radiation anyway, like the ignition system?

Seems a bit of a reach for an explanation, when the easiest one is simple user error, combined with a sue-happy society.



I am certain cold fusion could be added to that list---- of irreproducible mechanisms :wall:

geonuc
2010-Aug-01, 11:08 AM
Dismissing the idea is premature. I note this quote:


The risks are especially high for circuits that are "field programmable," explained Lloyd W. Massengill, director of engineering at the Vanderbilt Institute for Space and Defense Electronics at Vanderbilt University. Field-programmable circuits are systems in which the circuit's function can be electrically altered while it is still in use.

First, that 'sounds' plausible. Second, the speaker has the credentials that at least demand attention to what he's saying.

cjameshuff
2010-Aug-01, 12:58 PM
FPGAs are used elsewhere without problems. And the RAM in your computer is "field programmable" as well and also a large target for ionizing radiation induced corruption. Despite this, random corruption is a rare enough occurrence that error-correcting technologies are often not used.

Automobiles are notoriously high EMI environments, and are also exposed to extremes of shock, vibration, and temperature. While a cosmic ray hit might cause such a problem, singling them out as a cause seems rather silly, especially if the problem occurs multiple times.

neilzero
2010-Aug-01, 01:34 PM
The military has long given serious consideration to equipment failure due to radiation. MOS = metal oxide semi-conductors are especially vulnerable. Typically the civilian industry has given this little attention, as it represents planned obsolescence, that can be beneficial to the manufacture, if it occurs after the warranty period. Most modules have some automatic redundancy, so it is usually the second or third junction failure that creates the safety problem. Likely several percent of electronic module failures are due to radiation. Cosmic rays, more correctly secondary cosmic rays are a portion of the radiation damage. Toyota likely did not sufficiently analyze the possible safety problems connected with the acellerator portion of their electronics. Apparently, Toyota did not provide a sufficient number of viable alternatives, for stopping the vehicle if the accelerator pedal "sticks." ie my 1996 Ford Ranger has an impact fuel shut off near the passenger's right foot which stops the motor if you kick it accidentally or on purpose. Since it is a 5 speed manual, you can also put it in neutral or the lowest speed gear and let the engine over speed. Turning the key off is a very bad solution as it locks the steering wheel on many older Ford models including my Ranger. Other manufacturers have safety problems and they typically give the victims hush money to avoid bad publicity. Neil

Strange
2010-Aug-01, 01:56 PM
And the RAM in your computer is "field programmable" as well and also a large target for ionizing radiation induced corruption. Despite this, random corruption is a rare enough occurrence that error-correcting technologies are often not used.

I don't know the percentage breakdown but in any "serious" computer (from servers to supercomputers) error correcting codes (ECC) are commonly used on DRAM to correct soft errors for this very reason. They are not often used in the home PC market because it is a trade off of cost against the inconvenience of occasional random crashes. Some processors also use ECC on internal storage (such as caches and registers) and on busses, again to minimize the chance of random errors.

mugaliens
2010-Aug-01, 08:30 PM
Dismissing the idea is premature. I note this quote:


The risks are especially high for circuits that are "field programmable," explained Lloyd W. Massengill, director of engineering at the Vanderbilt Institute for Space and Defense Electronics at Vanderbilt University. Field-programmable circuits are systems in which the circuit's function can be electrically altered while it is still in use.

First, that 'sounds' plausible. Second, the speaker has the credentials that at least demand attention to what he's saying.

If cosmic rays were the culprit, they would be affecting all circuitry, not just the ones relating to gas pedal actuation. The lack of issues in the many other circuits indicates this is simply not the issue.


The military has long given serious consideration to equipment failure due to radiation.

We were far more concerned with EMP hardening. :) More seriously, though perhaps one signal (circuit switch) out of a trillion could be expected to change due to radiation, the sheer number of such functions ensures that radiation will affect operation, so various hardening techniques have been incorporated in most electronics over the last decade. See SEE (http://en.wikipedia.org/wiki/Radiation_hardening#Digital_damage:_SEE).


...my 1996 Ford Ranger has an impact fuel shut off near the passenger's right foot which stops the motor if you kick it accidentally or on purpose. Since it is a 5 speed manual, you can also put it in neutral or the lowest speed gear and let the engine over speed. Turning the key off is a very bad solution as it locks the steering wheel on many older Ford models including my Ranger. Other manufacturers have safety problems and they typically give the victims hush money to avoid bad publicity. Neil

My Ford Sport Trac has the same impact sensor. One unwieldy passenger kicked it this spring, and it cost $107 to "diagnose and fix," even though the "fix" is to depress and hold the reset button for five seconds...

Putting it in neutral is fine, as my Sport Trac has an RPM limiter when it's in neutral. Turning the key off is fine, too, as it's an automatic, and can't be put into park (where the steering wheel is locked) while moving.

cjameshuff
2010-Aug-02, 12:58 PM
I don't know the percentage breakdown but in any "serious" computer (from servers to supercomputers) error correcting codes (ECC) are commonly used on DRAM to correct soft errors for this very reason. They are not often used in the home PC market because it is a trade off of cost against the inconvenience of occasional random crashes. Some processors also use ECC on internal storage (such as caches and registers) and on busses, again to minimize the chance of random errors.

I'm aware that ECC RAM is used, this computer I'm using now has it. But as I said, non-ECC RAM is very common, and machines do function with high reliability without it, despite the rather large total silicon area devoted to RAM making it a big target for cosmic rays. A lot of the time the corrupted bits will be in unused RAM or in data, but the same is true of FPGAs.

The bigger issue is that the computer won't react the same way to corruption of different bits in its RAM, and neither will the FPGA. The odds of the exact same failure occurring twice are very remote. EMI is introduced into the system more strongly at certain points, and can cause more specific, more repeatable problems in things like communications hardware. Some parts are more sensitive to thermal stresses than others...again, something that's repeatable. Power supply issues...again, more repeatable results. Etc...

If corruption in a FPGA is what triggers the problem, it's almost certainly doing so through something else that is able to detect such corruption regardless of where it occurs...a self check system of some sort that's supposed to do a clean reset but goes awry somehow, perhaps. If that's the issue, that malfunctioning self check/error recovery system is the problem, not cosmic rays.

Celestial Mechanic
2010-Aug-03, 05:18 AM
"I didn't know the gun was loaded."

"The dog ate my homework."

"The devil made me do it."

And now meet the newest inductee into the Lame Excuse Hall of Fame:

"Cosmic rays zapped my electronics."

:clap: :clap: :clap:

LotusExcelle
2010-Aug-03, 05:43 AM
The problem is that the data does not support the claim. Nor does the mechanism. Throttle-by-waire defaults to closed via a rather strong spring. And so far all of the purported claims of pedal issues are recorded via the data logger as showing the brake pedal being open circuit (not pressed). So the people mashed the wrong pedal.

neilzero
2010-Aug-05, 03:38 AM
"Mashed the wrong pedal" I've done that, fortunately with good results. Good because it warned me that I was not allowing sufficient safety factor, had the senario been different. Good because it reduced my confidence = I might be making an error = this allows me to stop mashing the wrong peddle in about one second.
Yes I kicked the impact fuel cut off, called for a tow truck. While waiting, wife and I read the instuction manual, found the part about the fuel impact switch, found out where it was located, realized that I could have kicked it and that the cut off might have interperted the kick as an impact. I pushed the reset button for 5 seconds and we were on our way. We called the towing company and explained that we had fixed the problem and would not be available to be towed. They were polite and did not suggest that we pay, but they possibly collected from our insurance. Neil

mugaliens
2010-Aug-05, 04:59 AM
I'm aware that ECC RAM is used, this computer I'm using now has it. But as I said, non-ECC RAM is very common, and machines do function with high reliability without it, despite the rather large total silicon area devoted to RAM making it a big target for cosmic rays. A lot of the time the corrupted bits will be in unused RAM or in data, but the same is true of FPGAs.

"Recent tests give widely varying error rates with over 7 orders of magnitude difference, ranging from 10−10−10−17 error/bitĚh, roughly one bit error, per hour, per gigabyte of memory to one bit error, per century, per gigabyte of memory." - Wikipedia, from DRAM Errors in the Wild: A Large-Scale Field Study (http://www.cs.toronto.edu/~bianca/papers/sigmetrics09.pdf)

Interestingly, as memory densities have increased, memory reliability has actually increased.