1 2
ignorant
ignorant SuperDork
3/4/10 4:25 a.m.

http://www.popularmechanics.com/automotive/how_to/4347704.html

hmmmmm

NYG95GA
NYG95GA SuperDork
3/4/10 5:06 a.m.

That was well explained and sounded very convincing.

Still, I don't trust them;

just something else to go wrong.

ignorant
ignorant SuperDork
3/4/10 5:12 a.m.
NYG95GA wrote: That was well explained and sounded very convincing. Still, I don't trust them; just something else to go wrong.

The problem is that all manufacturers are going to them. So in 5-10 years we'll all be driving them. They are part of the emissions controls now.

btw.... kid got up at 3:30 today. I am happy I can type semi coherent sentences.

vibiant
vibiant New Reader
3/4/10 6:36 a.m.

It's too easy to blame the car, and too difficult to admit when you're wrong. People weren't saying "I never checked the presure in my tires, and then totally overloaded my Ford Explorer, which then suffered a blowout, I panicked and rolled it on the highway." They blamed bad tires and a poor SUV design.

Raze
Raze HalfDork
3/4/10 7:23 a.m.

I still have to interject that while the Popular Mechanics issue has valid points talking about mechanical wear and electronic interference, the real issue no one including Toyota wants to talk about is a possible electronic control design flaw (read software) which under a very narrow set of circumstances could cause unintended results. Everyone thinks electronic controllers are so much better than mechanical, and in many ways they are, but they present their own unique challenges which must be just as carefully approched as a mechanical system design and implementation.

I know this is possible even on in service engines, I'll say it again, I worked for a major aircraft engine manufacturer in the engine control software group doing testing for FAA certification, design, etc. These engines and their electronic controllers went through rigorous FAA certification and yet still there were instances years later where the plane was placed in an unusual set of circumstances no designer had forseen and the result was either engine damage or failure.

Popular Mechanics is only grasping at the fundamentals of the design architecture and is very misleading. Sure there are no 'traditional' parts to wear out and the hall effect throttle position sensors recalibrate every time on POST and that there are 2 fully independent paths with limp home mode thus doubly redundant and therefore hard to fail. Sure it could be driver error. But what about the software itself? As Toyota, it would be far cheaper to blame a pedal manufacturer than reveal it made a software error. Ignorant, you posted yesterday how GM is blaming Toyota for it's problems, yet Toyota is blaming it's supplier, funny isn't it?

OH and BTW, my wife's Camry is under both recalls, and they say the computer needs a software update, I wonder why?

I'm not saying Toyota hasn't solved the problem yet, I'm saying they really need to ensure there are no 'cliffs' to fall off in the software.

foxtrapper
foxtrapper SuperDork
3/4/10 8:02 a.m.
ignorant wrote: hmmmmm

Hmm indeed. For the basis of the article is that software is inherently perfect, and redundancy prohibits error.

I well remember when computers were the new toy, and it was believed that they were indeed perfect. Today, we all know better (except Popular Mechanics apparently).

Equally, I've had many a double redundant system fail. True, they are less likely to fail than a single system, but they are not absolutely prevented from failure by virtue of being doubly redundant.

Popular Mechanics looked at the pedal assembly, saw two separate circuits, and made the leap that the car is therefore flawless. That is a leap I will not make with them.

eastsidemav
eastsidemav Reader
3/4/10 8:17 a.m.

Raze has a very good point. I've spent more than a decade doing QA on software and electronics. A few lazy developers, and not enough rigorous testing, and bad things can happen around the margins where you aren't planning on the system working in.

Minor case in point, thankfully nothing that would cause a safety problem: In my (now departed) 2001 Saturn, if you started the engine, put your seatbelt on, drove somewhere (say to a post office box or ATM), shifted into park, and took your seatbelt off (to get out of the car quickly to get something, or get a little extra reach), then put your belt back on, the seatblet light would start flashing most of the time, even after you put the belt back on. I could never figure out the exact parameters under which it did this, but as far as I'm conerned, that was a software glitch, and a fairly simple one, at that.

Dr. Hess
Dr. Hess SuperDork
3/4/10 8:37 a.m.

When this started, I asked if these cars were fly by wire and suspected a computer control issue, if y'all will recall. Being a software developer myself, and having to fix a lot of other people's bad code, I see two potential issues:

Bad code, as stated. A buffer somewhere gets over run, writes to the memory where the throttle position to set at is stored, bingo, open throttle. Or instead of writing to the data part of memory, writes over the program part. Buffer overruns are a main reason that microsoft products are so virus prone. Sloppy coding. It takes a lot more work to write: "Here's 64 bytes of RAM. Load a byte into that block until there are no more bytes to load and don't load more than 64 bytes, ever, even if there's more there which there should never be" than to write "Here's 64 bytes of RAM. Load a byte into that block until there are no more bytes to load because there's never going to be more than 64." Besides the buffer overrun, there are other likely bad coding problems that could creep in, but that one is the hardest to figure out after the fact. The compiler could even be part of the problem. And the ECU now controls just about everything, so any one little piece, like the seat belt light on the dash, could wipe out something else, like the ABS or the throttle control.

The other problem I see is just stray electromagnetic interference. You can shield the crap of out everything and reduce your problems to one in 100 million, but then when you have 7 million cars on the road every day, well, there's a significant chance of a problem. A spike comes down the line, not filtered out or the filtering circuits are shot from the last million spikes (it's a car, remember, lightening is around, there's 4000 * # cylinders/2 20KV spikes just a couple feet from the computer and electrically connected to the computer every time it's going down the highway, etc.), spike changes something in the ECU RAM, zoom.

Raze
Raze HalfDork
3/4/10 8:58 a.m.

In reply to Dr. Hess:

I work in embedded real time software so I hear everything you're saying and totally agree. For this application you would hope that Toyota and it's Engineers would be smart enough to isolate a chunk of memory specifically for each input, throttle, brake, steering and associated processing, thus building redundancy into the software/hardware architecture, but you know that's a much more expensive route through either fabrication of the PCB or through software. Case in point, I am a relativley young engineer (by engineering standards) but I notice many persons my age not worrying about memory, space, size, or timing requirements when writing code for real-time systems, or the target compiler and options set during compilation for that matter! It's shocking but the 'oh well it works on my emulated test bench so it must work perfectly in the real world' mentallity is unbelievable to me and when we get to integration testing and whole sections of code blow out becase they miss their window for the above listed reasons it ends up taking them 3 times longer to write, integrate, and fix the code to run fast enough/ without memory leaks than it would have been to do it correctly the first time and build and test the extra complexity into the code in the first place!

iceracer
iceracer HalfDork
3/4/10 9:05 a.m.

Now it is reported that some are still having problems even after the recall/modification. I like my old fashioned throttle cable and hydraulic PS.

Junkyard_Dog
Junkyard_Dog HalfDork
3/4/10 9:10 a.m.
iceracer wrote: Now it is reported that some are still having problems even after the recall/modification. I like my old fashioned throttle cable and hydraulic PS.

I like my old fashioned throttle lever and bell crank linkage and armstrong PS

Raze
Raze HalfDork
3/4/10 9:12 a.m.
iceracer wrote: Now it is reported that some are still having problems even after the recall/modification. I like my old fashioned throttle cable and hydraulic PS.

That's why I was alluding to the fact that Toyota is quietly software patching the cars when they take it in for the 'pedal' repair.

Here in Atlanta there have been reports on the news of people complaining the issue is either still happening or, they got a software patch applied and now the car response is 'weird' with CELs going on and off...

rogerbvonceg
rogerbvonceg New Reader
3/4/10 12:28 p.m.
Raze wrote: In reply to Dr. Hess: I work in embedded real time software so I hear everything you're saying and totally agree. For this application you would hope that Toyota and it's Engineers would be smart enough to isolate a chunk of memory specifically for each input, throttle, brake, steering and associated processing, thus building redundancy into the software/hardware architecture, but you know that's a much more expensive route through either fabrication of the PCB or through software. Case in point, I am a relativley young engineer (by engineering standards) but I notice many persons my age not worrying about memory, space, size, or timing requirements when writing code for real-time systems, or the target compiler and options set during compilation for that matter! It's shocking but the 'oh well it works on my emulated test bench so it must work perfectly in the real world' mentallity is unbelievable to me and when we get to integration testing and whole sections of code blow out becase they miss their window for the above listed reasons it ends up taking them 3 times longer to write, integrate, and fix the code to run fast enough/ without memory leaks than it would have been to do it correctly the first time and build and test the extra complexity into the code in the first place!

Closed - Could Not Duplicate.

or,

"Works as Designed."

eastsidemav
eastsidemav Reader
3/4/10 12:39 p.m.

In reply to rogerbvonceg:

"But it works on my machine!"

"Okay, lets send your machine to the customer"

Yes, that was my reply once. The software engineer was not amused.

nderwater
nderwater Reader
3/4/10 12:41 p.m.

In reply to eastsidemav: lmao

hamburglar
hamburglar New Reader
3/4/10 12:58 p.m.

I still don't get this whole thing. I mean, electronics aren't my strong suit, but I have had the throttle pedal stuck because of the floormat issue before. The quick fix has always been bump it to neutral or depress the clutch, if applicable. This is regardless of how the pedal systems works. Sure, there is a certain amount of reaction time involved, especially when this is unexpected, but it's not a big deal.

I understand that Toyota has recognised the issue and they are fixing it. What else are they supposed to do? Internally I mean they can sack whoever they want for screwing up, and apologise to customers.

If people feel this is such a big issue why don't they just vote with their wallet and let the government stay out of it?

Dr. Hess
Dr. Hess SuperDork
3/4/10 1:02 p.m.

That's a good one 'mav. I'll have to remember that one. "But, it works on my machine." "OK, box up your machine and put it on the next boat to the client."

I am currently battling AT&T with my internet dongle. You would not believe the incompetency. After a week of no call back, I called to see what was happening. Oh, they closed that ticket. Turns out they never called me, made something up for their log saying they did and complete fiction as to what I told them, then closed the ticket. I was not amused. They have been getting daily calls from me since. I'm going to cut them some slack today as I got a month free yesterday. Tomorrow I start in again.

cwh
cwh SuperDork
3/4/10 1:15 p.m.

Please correct me if I got it wrong, but I believe that on some of these cars with the push to start, push to stop engine controls, you must push and HOLD the button to shut down. At the same time, the electronics prohibit you from shifting into neutral. You repeatedly jab the button, nothing happens, your power brake booster runs out of vacuum, at wide open throttle you are not generating vacuum, and it all goes to hell. Is this possible? I'm liking my B-3000 more every day.

Dr. Hess
Dr. Hess SuperDork
3/4/10 1:58 p.m.

4 throttle bodies, 2 return springs, 1 cable:

friedgreencorrado
friedgreencorrado Dork
3/4/10 2:06 p.m.

I think y'all have just about talked me out of "fly-by-wire" until the next generation of controls comes along.

Raze
Raze HalfDork
3/4/10 2:13 p.m.
hamburglar wrote: I still don't get this whole thing. I mean, electronics aren't my strong suit, but I have had the throttle pedal stuck because of the floormat issue before. The quick fix has always been bump it to neutral or depress the clutch, if applicable. This is regardless of how the pedal systems works. Sure, there is a certain amount of reaction time involved, especially when this is unexpected, but it's not a big deal.

That's part of the problem, if you have an automatic transmission built in the last 20 years it is most likely electrically controlled, when you shift the lever it's just like pushing a button, there's no actual mechanical motion from your end. Depending on design and ECU layout most modern transmissions depend on the software that's running the controller, the gear select just decides what logic to run.

A perfect example was my 98 Cadillac Eldorado. It would alter the shift algorithm and line pressure when you left it in lower gears, 1 having the most line pressure, 2 having the next most, 3 having less and Drive having grandma levels. However, if you redlined in 1, you would bounce the limiter, if you redlined in 2 it would automatically shift you to 3, not bounce you off the limiter but would bounce you on the limiter in 3 when having 2 selected, having it in 3 would automatically shift you to 4 (4 was at 120MPH).

What the above explanation is to show is simply that your transmission is just as electronic as your throttle in newer cars, what's scary is that if the memory space is shared among the transmission controller, throttle, and steering and there is either a code error 'cliff' or buffer over run you could end up screwing up the values in place in your transmission logic, thus preventing you from shifting to neutral!

It's all in the design, but for cost, if I was a penny pinching company and employed an army of Industrial engineers to figure out a cheaper/faster assembly method for my FWD vehicle, it would be to integrate the trans and engine controllers all into one unit you would simply plug both into, instead of having 2 or more individual computers.

alfadriver
alfadriver Dork
3/4/10 2:28 p.m.
friedgreencorrado wrote: I think y'all have just about talked me out of "fly-by-wire" until the next generation of controls comes along.

You do realize that there IS a world outside of Toyota, right? With that VW representing you, I'd figure you would have heard about Bosch NOT being Toyota/ Denso....

There are a lot of examples of ETC out there that are very, very good. Don't let what Toyota is not getting right ruin it for everyone.

Eric

Bobzilla
Bobzilla Dork
3/4/10 2:39 p.m.

Hell, my GMT800 TBW is just fine....

spitfirebill
spitfirebill Dork
3/4/10 2:45 p.m.
Raze wrote:
iceracer wrote: Now it is reported that some are still having problems even after the recall/modification. I
That's why I was alluding to the fact that Toyota is quietly software patching the cars when they take it in for the 'pedal' repair. Here in Atlanta there have been reports on the news of people complaining the issue is either still happening or, they got a software patch applied and now the car response is 'weird' with CELs going on and off...

There was also a TSB (maybe a recall) where the car would have very sluggish to no acceleration. The fix was a reflash of the computer that was done free under the emissions warranty. My wife's 07 Solara sure doesn't fell very peppy and was considering taking it in when all this hit. Of course all this carp hit a few months after we bought the car.

And a couple of wekks ago, one of the national news shows reported that the throttle was blue tooth connection, not wired. Now that almost freaked me out until I remembered the source.

And yes, as I and a lot of others predicted, the metal shim fix has nothing to do with the acceleration issue.

WilberM3
WilberM3 None
3/4/10 3:25 p.m.
Raze wrote: That's part of the problem, if you have an automatic transmission built in the last 20 years it is most likely electrically controlled, when you shift the lever it's just like pushing a button, there's no actual mechanical motion from your end. Depending on design and ECU layout most modern transmissions depend on the software that's running the controller, the gear select just decides what logic to run. What the above explanation is to show is simply that your transmission is just as electronic as your throttle in newer cars, what's scary is that if the memory space is shared among the transmission controller, throttle, and steering and there is either a code error 'cliff' or buffer over run you could end up screwing up the values in place in your transmission logic, thus preventing you from shifting to neutral!

i'm not very familiar with a lot of other manufacturers but BMW has still been using a shift cable that mechanically moves the manual valve and parking pawl into this decade, though the last few years they may no longer. but even on a lot of electronically shifted transmissions once in D or lower 'gears' the computer takes over but it first requires the mechanical movement of the manual valve connected to the shift lever to put line pressure to the right places, which would still allow a mechanical neutral.

if all these new trannies are fully electronic that driver-disconnect kind of scares me...

1 2

You'll need to log in to post.

Our Preferred Partners
SiMIubIkZaMASXF3rh8adDUqw1EyLVHfnU44JxBge0hCEPKn56GxqBcyPiiJ1NmN