1 2 3 4
wbjones
wbjones PowerDork
10/31/13 7:27 a.m.

I realize that the government requires the airbags, the crash safety tech, the traction control .... but do they mandate the drive by wire ? I thought that was a car co. initiative ...

JoeyM
JoeyM Mod Squad
10/31/13 8:32 a.m.

I don't think the feds require drive by wire, but I think the OEMs adopted it to meet the required fuel efficiency standards

Kenny_McCormic
Kenny_McCormic SuperDork
10/31/13 8:50 a.m.

In reply to wbjones:

DBW is the only way to get those last pesky HC molecules you inadvertently get whenever you snap the throttle shut and the computer has to take a few milliseconds to react, you know, the ones that were probably already there in higher concentrations in the incoming air.

Giant Purple Snorklewacker
Giant Purple Snorklewacker MegaDork
10/31/13 9:00 a.m.
wbjones wrote: I realize that the government requires the airbags, the crash safety tech, the traction control .... but do they mandate the drive by wire ? I thought that was a car co. initiative ...

The level of magic that the Nissan GTR uses to make pigs fly requires every aspect of user input be accessible to programming.

You have to have drive by wire to allow fine grained control of other technology like seamless switching between electric and gas engines, predictive, learning 8 speed automatic transmissions, active stability/traction control, impact sensitive cutoffs, user adjustable "sport modes" etc. It also makes packaging easy - the throttle body(s) can be anywhere relative to the gas pedal.

In truth, it a non-issue. Aircraft and diesel trucks have used it for decades. RTOS is a well known and stable OS. Embedded hardware platforms with overflow protection are commonplace in every factory with any automation. This is a breakdown caused by beancounters, bad engineering or derelict QA. Probably all three.

alfadriver
alfadriver PowerDork
10/31/13 9:24 a.m.

In reply to Kenny_McCormic, JoeyM, and wbjones:

electronic throttle is less expensive than cable throttle (with idle bypass, and cruise control). it also allows much more effective traction control, and lowers the risk of breaking the engine due to over reving.

it's not needed for fuel economy, and it's not needed for emissions.

Kenny_McCormic
Kenny_McCormic UltraDork
10/31/13 9:32 a.m.
alfadriver wrote: In reply to Kenny_McCormic, JoeyM, and wbjones: electronic throttle is less expensive than cable throttle (with idle bypass, and cruise control). it also allows much more effective traction control, and lowers the risk of breaking the engine due to over reving. it's not needed for fuel economy, and it's not needed for emissions.

Duly noted, doesn't change how bad of an idea it is as currently implemented. I wonder how cost effective it becomes when done correctly with true quadruple redundancy and whatnot.

JoeyM
JoeyM Mod Squad
10/31/13 9:38 a.m.
alfadriver wrote: In reply to Kenny_McCormic, JoeyM, and wbjones: electronic throttle is less expensive than cable throttle (with idle bypass, and cruise control). it also allows much more effective traction control, and lowers the risk of breaking the engine due to over reving. it's not needed for fuel economy, and it's not needed for emissions.

Thank you. It is good to hear from someone who works in the industry.

(...and my inner luddite is now happy about traction control not being possible on my datsun.)

alfadriver
alfadriver PowerDork
10/31/13 9:39 a.m.

In reply to Kenny_McCormic:

For who? Are you broad brushing all electronic throttle based on Toyota?

What's part of the idea is "bad" in terms of the current implemenation? In a universal manner- as in all electronic throttles do the same bad idea.

alfadriver
alfadriver PowerDork
10/31/13 9:41 a.m.
JoeyM wrote:
alfadriver wrote: In reply to Kenny_McCormic, JoeyM, and wbjones: electronic throttle is less expensive than cable throttle (with idle bypass, and cruise control). it also allows much more effective traction control, and lowers the risk of breaking the engine due to over reving. it's not needed for fuel economy, and it's not needed for emissions.
Thank you. It is good to hear from someone who works in the industry. (...and my inner luddite is now happy about traction control not being possible on my datsun.)

It is very possible to have traction control on your datsun. Heck, one could even design a mechanical system to regulate the throttle if there are wheels slipping.

You just won't have electronic throttle based traction control.

Kenny_McCormic
Kenny_McCormic UltraDork
10/31/13 9:46 a.m.

I am broad brushing a bit here, if the Japanese are this sloppy at it, its a miracle that say, the GM ones, work at all.

Correct me if I'm wrong here. In the usual setup all the automakers are using , you've got 4 potentiometer readouts, 2 on the pedal, 2 on the TB, and one motor on the TB, tied to a single computer which also tasked with a bunch of other stuff. So any redundancy there is about as good as the programming and hardware, which in the entire history of the digital age, always tends to fall short.

Would you ride in an airplane built like that?

alfadriver
alfadriver PowerDork
10/31/13 9:58 a.m.

In reply to Kenny_McCormic:

Yes, I would. And do.

You see one box and read that as one computer, there are more than one in that box.

And how is redundancy a bad idea? That seems like a very good idea.

Oh, and whatever you do, then the Japanese automatically do it better than you do? Just wondering....

tuna55
tuna55 PowerDork
10/31/13 10:17 a.m.
alfadriver wrote: In reply to Kenny_McCormic: Yes, I would. And do. You see one box and read that as one computer, there are more than one in that box. And how is redundancy a bad idea? That seems like a very good idea. Oh, and whatever you do, then the Japanese automatically do it better than you do? Just wondering....

I wonder how much redundancy is in the "throttle" on my Leaf.

I think that is. How many accidents did this supposedly cause? Now the important question, now that they fixed the floor mats and the maybe-sticky pedals, how many further incidents have there been?

Kenny_McCormic
Kenny_McCormic UltraDork
10/31/13 10:18 a.m.
alfadriver wrote: In reply to Kenny_McCormic: Yes, I would. And do. You see one box and read that as one computer, there are more than one in that box. And how is redundancy a bad idea? That seems like a very good idea. Oh, and whatever you do, then the Japanese automatically do it better than you do? Just wondering....

http://www.davi.ws/avionics/TheAvionicsHandbook_Cap_11.pdf How a 777 fly by wire system works, 11.6 "Fault tolerance" is of particular interest.

When did I say redundancy is a bad idea? I said in the current implementation, it's E36 M3. Thus it needs improvement.

On the Japanese thing, I'm making a lot of broad generalizations and taking things to logical extremes, which is the only way to do things when safety is the topic. Typically speaking they aren't know for putting out cars with crappy engineering, when they do, its usually taken care of promptly.

markgoespop
markgoespop
10/31/13 10:37 a.m.

Does anyone has a copy of Bookout_v_Toyota_Barr_REDACTED.pdf they can send me? That link isn't working. I'm a computer scientist and it sounds like it would make fascinating reading!

Thanks, Mark

alfadriver
alfadriver PowerDork
10/31/13 11:41 a.m.

In reply to Kenny_McCormic:

Ok... I'll just disagree. Having seen the beginning, and noting how it's progressed, it works quite well. We are not flying planes, we are driving cars. The fault tolerance of a car is quite a bit different. To the point that having two sensors that are non contacting, and very reliable, as well as a single extra computer checking the requests and outputs is enough to tolerate a failure.

Typically speaking, I vehemetly disagree that GM has bad engineering. And I very much think that Toyota has questionable engnieering, based on the small subset of engineering that I do. GM's fault is that they (and historicall Ford has) listend to the bean counters over the quality people, and focused on making a car as cheap as they possibly could. Vs. making cars with enough extras that people are willing to pay for them.

You need to tighten up your brushes quite a bit.

Anyway, hate it all you want- it's not going to go away anytime soon.

Knurled
Knurled UberDork
10/31/13 11:42 a.m.
Kenny_McCormic wrote: In reply to wbjones: DBW is the only way to get those last pesky HC molecules you inadvertently get whenever you snap the throttle shut and the computer has to take a few milliseconds to react, you know, the ones that were probably already there in higher concentrations in the incoming air.

Not just that but when you start playing with cam timing or EGR and direct injection's stratified charge awesomeness, the throttle position has little to do with the engine's output anyway. In some cars, basically if the engine is over a fast idle, the throttle is WOT.

It also makes the cars lighter because there's no more cruise control unit, no IAC, no associated bracketry and wiring and cabling, there's one less hole to put in the firewall, etc.

Beer Baron
Beer Baron UltimaDork
10/31/13 11:46 a.m.

All this talk about how simple it is to handle an unintended acceleration situation, look at what happens with even enthusiasts on a track who are trying to train to handle emergency situations:

How often do you repeat the mantra to yourself and students "In a spin, both feet in"? How much do you try to train to focus on your escape route, not target focus on the hazard?

...

How often to people who have been training themselves for these emergency situations end up having the wheels catch halfway through a spin only to shoot across the track and collect the one car, tire stack, or k-wall section along the path?

JoeyM
JoeyM Mod Squad
10/31/13 11:50 a.m.
alfadriver wrote:
JoeyM wrote:
alfadriver wrote: In reply to Kenny_McCormic, JoeyM, and wbjones: electronic throttle is less expensive than cable throttle (with idle bypass, and cruise control). it also allows much more effective traction control, and lowers the risk of breaking the engine due to over reving. it's not needed for fuel economy, and it's not needed for emissions.
Thank you. It is good to hear from someone who works in the industry. (...and my inner luddite is now happy about traction control not being possible on my datsun.)
It is very possible to have traction control on your datsun. Heck, one could even design a mechanical system to regulate the throttle if there are wheels slipping.

My inner lazy butt has decided to ignore ^^^^this post

Kenny_McCormic
Kenny_McCormic UltraDork
10/31/13 12:04 p.m.
alfadriver wrote: In reply to Kenny_McCormic: Ok... I'll just disagree. Having seen the beginning, and noting how it's progressed, it works quite well. We are not flying planes, we are driving cars. The fault tolerance of a car is quite a bit different. To the point that having two sensors that are non contacting, and very reliable, as well as a single extra computer checking the requests and outputs is enough to tolerate a failure. Typically speaking, I vehemetly disagree that GM has bad engineering. And I very much think that Toyota has questionable engnieering, based on the small subset of engineering that I do. GM's fault is that they (and historicall Ford has) listend to the bean counters over the quality people, and focused on making a car as cheap as they possibly could. Vs. making cars with enough extras that people are willing to pay for them. You need to tighten up your brushes quite a bit. Anyway, hate it all you want- it's not going to go away anytime soon.

I think the only real disagreement we have here is where the line should be drawn for acceptable fault tolerance in an accelerator controller. I believe it should be quite a bit better, considering how complicated shutting down a modern car can be, and how fast many of them have become.

There also ought to be more than one instantaneous, one click, guaranteed under all circumstances methods to disconnect the engine from the powertrain. As slapping the shifter forward, or hitting the "off" button doesn't do much anymore, especially if things aren't working as they should.

Perhaps "engineering" was a poor word choice there, "finished product design" a bit better?

alfadriver
alfadriver PowerDork
10/31/13 12:12 p.m.

In reply to Kenny_McCormic:

I could be all smart about it, but in the end, it does not matter. However, if you do really feel that strongly about it, there are real engineering job openings for all OEM's. got to www.(anyoem).com, find the link "carreers" and work on it.

I know we are hiring.

Kenny_McCormic
Kenny_McCormic UltraDork
10/31/13 12:22 p.m.

In reply to alfadriver:

We're debating something on an internet forum, of course it doesn't matter.

As a matter of fact, I need to get looking for a engineering co-op for this summer. Though I'm more into the mechanical side of things.

Sky_Render
Sky_Render Dork
10/31/13 12:32 p.m.

I only read the EDN article, however, this is the most damning thing I saw there:

EDN said: Mirroring (where key data is written to redundant variables) was not always done. This gains extra significance in light of Stack overflow. ... Two key items were not mirrored: The RTOS' critical internal data structures; and—the most important bytes of all, the final result of all this firmware—the TargetThrottleAngle global variable.
accordionfolder
accordionfolder Reader
10/31/13 12:44 p.m.

In reply to alfadriver: I know you work for an OEM from other posts, but in what capacity? I'm spending my spare time on embedded projects ... Maybe one day I can get my foot in the door, I've always wanted to work on firmware (Which is actually why the court doc was so fascinating to me, rare insight w/o being internal to a company). Right now I just do Android.

mfennell
mfennell New Reader
10/31/13 12:44 p.m.
alfadriver wrote: Typically speaking, I vehemetly disagree that GM has bad engineering. And I very much think that Toyota has questionable engnieering, based on the small subset of engineering that I do. GM's fault is that they (and historicall Ford has) listend to the bean counters over the quality people, and focused on making a car as cheap as they possibly could. Vs. making cars with enough extras that people are willing to pay for them.

Thanks for that. Through the Innertubes, I know a former OEM calibrator for GM. He is anything but dumb but he sure as hell had to make compromises for all kinds of management requests.

If we're going to generalize, traditional Japanese deference to the boss is a crappy way to engineer good solutions IMHO. I think American in-your-face culture works better here. When I look back at the best designs I've been involved in, there was always a lot of shouting with no consideration given to the Org Chart. :)

Based solely on reading Lutz's beancounter book, one reasonable fault you could find with GM was they had a long tradition of only hiring the top graduates at the schools they recruited and would miss out on the creative hands-on types who blew off the study session to refine that SAE chassis one more iteration.

""

accordionfolder
accordionfolder Reader
10/31/13 12:49 p.m.

And honestly, while drudging through the source code of this, it all sounds terrible, but Toyota drives cars while QAing them millions and millions of miles and the number of Toyota's on the road boasting these "flawed" ECUs is massive. Things would have to get verrrrrry corner case-y to get to any of these error states to become apparent. Sure it's bad, and should have never been that way in the first place, but are you going to ever see it happen? Probably not.

1 2 3 4

You'll need to log in to post.

Our Preferred Partners
2Xs4SW5ugo8GJWVLYVS7WqJnU0UGaZEXx5vvfprTFoT6e0Ip0N6VwSPfhZNTCNPb