Slashdot Mirror

Maximum spark temperature is about 500C not 50,000C
see e.g. http://rb-aa.bosch.com/aa-th/en/static/produkte/zuendkerzen/zuendkerzenkunde/waermewerte.htm

"in which case 1000C is extremely hard to manage in something the size of a normal motor."

I have not seen details of the size and weight (power density) of this electric generator. And it will also need an electric motor if it's going into an automobile. OTOH a temperature of 1000C is not so hard to manage (with e.g. ceramics) if the stresses are not high.

"your(sic) not comparing apples to apples here". Good point.

Did you look for statements from Conti or Bosch?

First, All ABS systems are supplied to the OEMs by many suppliers (two of which are listed above -- also see TRW, Nissin, Aisin, Sumitomo, Delphi...). So it is likely you will not find statements by the OEMs, who typically do not design or develop such systems. The suppliers listed above do, and sell their latest tech to the OEMs.

Second, you misunderstand some basic tire physics, as do many other posters. Google for the "mu-slip curve" to find out why ABS will shorten stopping distance for most drivers -- the system tries to maintain a slip that delivers the most friction. For most tires this slip is a velocity around 10-12% lower than the true vehicle speed.

For continued reading, look for information on "Kamm's Circle", which shows the relationship between lateral and longitudinal friction.

It would be interesting to take the most experienced driver you could find, and place this driver into a vehicle that he has never driven, without ABS. Take a second identical vehicle (but equipped with ABS) and an inexperienced driver who has also never driven the vehicle. Have both vehicles perform the same maneuvers.

Straight line stopping performance: I'm betting on the inexperienced driver with ABS. As I mentioned above, the ABS system will prevent locking, but automatically keep the wheel operating at peak friction.

Hard braking in a turn: The inexperienced driver will stop sooner *and* maintain his or her intended path. The non-ABS vehicle will unfortunately start to understeer as soon as the driver unintentionally brings the front wheels to lock.

If the choice is between burning biodiesel or burning petroleum-derived diesel; the biodiesel is much cleaner. In any case, there's all manner of funky new anti-pollution gear for diesels coming on line soon; one of the more important types are particulate filters.

But, in any case, you sound like a global warming denialist, so having a rational discussion on pollution with you is pretty much impossible anyway...

This area happens to be within the scope of my research, and I think it would be next to impossible to go overboard with the hardware.

If your research is somehow connected to the auto industry, it sounds like you need to visit Detroit.

I can think of several applications where having a "full-powered" computer in a car would be useful, not least of which is if you are on an automotive network (which doesn't really exist yet) and the environment requires high-grade encryption or other measures to provide the appropriate security. If the automotive network is P2P, lots of resources might be needed to make it work well.

Almost every vehicle produced today uses some sort of network to allow all of the modules to communicate together. The most common is CAN. Please list three more applications where this type of computing power would be useful.

Overkill? If anything, the PPC 603e is way undershooting what I think could/should be appropriate for automotive computing. Why not a 3GHz Pentium?

What types of computing tasks do you think are necessary inside an automobile, and what types of processors do you think are employed today? The ABS controller that I work on utilizes a 33MHz Motorola (proprietary) micro. The code that runs it is highly optimized, and every byte of RAM is scrutinized.

...but why not provide more than what's needed...

A ten-cent change will be discussed for literally hours on high volume platforms, because that $0.10 * 250,000 cars/year = $25,000. You're talking about hardware that will cost significantly more than what's employed today. And the benefit? None.

The problem here may not be just getting car manufacturers to divulge engine codes, but rather the maker of the ECU itself. For instance, Bosch produces ECUs or other engine components for virtually every car maker. Last year, Bosch had over $23x10^9 in sales in the automotive sector alone. Since nearly all European manufacturers rely on the Motronic ECU for higher-end engine applications, I'm sure that many Engine Control and Error Codes are similar amongst many different models and manufacturers.

I'm not entirely sure about whether or not the problem is the manufacturers' reluctance to supply codes. Perhaps they have a contract or something with the supplier of the ECUs to keep things hush-hush. Not only do the manufacturers make out by requiring service at their garage with their advanced tools, but the ECU supplier stays safe from the competition by not having all the details of their system easily divulged.

Seeing how it's been argued that TCP/IP isn't very good for high-speed wireless, and that AX.25 (never heard of that before) doesn't cut it either, left me with one question - what protocol would work? I'm pretty much a newbie in terms of wireless, but for embedded systems an industrial network called CAN (Controller Area Network) is widely used because of resistance to noise. While having numerous features that link it to twisted pair wire, parts of it are about protocol design, and the ideas in that area might work very well with wireless. Any ideas? Am I completely off track here?