Searching \ for '[EE] Why not Ethernet in Vehicle' in subject line. ()
Make payments with PayPal - it's fast, free and secure! Help us get a faster server
FAQ page: www.piclist.com/techref/index.htm?key=why+not+ethernet
Search entire site for: 'Why not Ethernet in Vehicle'.

Exact match. Not showing close matches.
PICList Thread
'[EE] Why not Ethernet in Vehicle'
2007\11\19@224048 by Xiaofan Chen

face picon face
en.wikipedia.org/wiki/Industrial_Ethernet
It seems industrial Ethernet is getting more and
more popular.

However it seems that Ethernet is not so popular in
the vehicles. CAN is dominated now and FlexRay
seems to be the future. I also hear about MOST.
http://www.automotivedesignline.com/howto/197004873

Why is like this? The manufacturing lines of automotive
industry are certainly going to be Ethernet based.

Regards,
Xiaofan

2007\11\20@011856 by M. Adam Davis

face picon face
Horses for courses.

Automobiles are incredibly cost sensitive, and when you're sending
data mere meters you don't need some of the ethernet things such as
transformers, the jacks are too flimsy, etc, etc.  When you throw out
everything you don't need, and then add stuff on top that you do need
that ethernet doesn't have natively (real-time capability, etc) then
you end up with something more like CAN and Flexray than ethernet.

-Adam

On Nov 19, 2007 10:40 PM, Xiaofan Chen <spam_OUTxiaofancTakeThisOuTspamgmail.com> wrote:
{Quote hidden}

> -

2007\11\20@195834 by Morgan Olsson

flavicon
face
Den 2007-11-20 04:40:46 skrev Xiaofan Chen <.....xiaofancKILLspamspam@spam@gmail.com>:

> Why is like this?

CAN have bit by bit collision detect and arbitration with priority on the fly.
Thus getting small control messages through using CAN can be granted within a smaller time frame than an ethernet link at much higher speed.
Also needing a single pair is cheaper on conductors and connectors etc
CAN also incoporates lot od the lower layer protocols in the hardware.

For even simpler protocol and hardware such as to the car door or chair, look for LIN, the "little sister" to CAN

For High reliability i have seen a double system RedCan for terrain work vehicles that thus can take much damage.
--
Morgan Olsson

2007\11\20@213341 by Xiaofan Chen

face picon face
On 11/20/07, M. Adam Davis <stienmanspamKILLspamgmail.com> wrote:
> Horses for courses.
>
> Automobiles are incredibly cost sensitive, and when you're sending
> data mere meters you don't need some of the ethernet things such as
> transformers, the jacks are too flimsy, etc, etc.  When you throw out
> everything you don't need, and then add stuff on top that you do need
> that ethernet doesn't have natively (real-time capability, etc) then
> you end up with something more like CAN and Flexray than ethernet.
>

This is kind of interesting. On one hand I agree that CAN is cheaper
(CAN MCU is cheaper). On the other hand, Ethernet in the industrial
automation world is actually going to Ethernet more and more. On the
lower level sensor/actuator, ASi, DeviceNet (Can based) or other
simpler protocols will still be dominant. But the upper level communication
(with the central control unit/units) is increasingly based on Ethernet.

So I am thinking that CAN/LIN will stay but maybe Ethernet will be the
backbone in the future and maybe it will replace MOST/FlexRay.
For example, now there are Ethernet based Safety protocol. It
can be the backbones for CAN/LIN through gateways just like
FlexRay.

Another thing is development tools. Ethernet is ubquitous on the
PC. Ethernet sniffer (bus analyser) is free. CAN bus analyser
and software are actuallly quite expensive. For Automotive industry,
maybe this is not the problem due to higher quantity though.

Xiaofan

2007\11\21@021710 by William \Chops\ Westfield

face picon face

On Nov 20, 2007, at 6:33 PM, Xiaofan Chen wrote:

> On the other hand, Ethernet in the industrial
> automation world is actually going to Ethernet more and more.

Ethernet tends to be fundamentally statistical rather than
deterministic, and it tends to make people who deal with
real-world THINGS very nervous...

BillW

2007\11\21@024621 by Xiaofan Chen

face picon face
On 11/21/07, William Chops Westfield <.....westfwKILLspamspam.....mac.com> wrote:
>
> On Nov 20, 2007, at 6:33 PM, Xiaofan Chen wrote:
>
> > On the other hand, Ethernet in the industrial
> > automation world is actually going to Ethernet more and more.
>
> Ethernet tends to be fundamentally statistical rather than
> deterministic, and it tends to make people who deal with
> real-world THINGS very nervous...
>

That is why protocols based on standard Ethernet are
popular now. Ethernet/IP is one of them. It uses standard
COTS hardware components but adds something on the
firmware side to achieve better deterministic result.
http://www.odva.org/

More examples here:
http://en.wikipedia.org/wiki/Industrial_Ethernet

Last week, some of my colleagues attended the Automation
Fair in Chicago and clearly Ethernet/IP is one of the
highlight. Cisco and Rockwell Automation are actually
coming out "Reference Architectures" guide to
integrate manufacturing and business systems utilizing
Cisco's "Ethernet to the Factory" and Rockwell Automation's
"Integrated Architecture".
http://www.controleng.com/article/CA6497093.html


Xiaofan

2007\11\21@051738 by wouter van ooijen

face picon face
> It uses standard
> COTS hardware components but adds something on the
> firmware side to achieve better deterministic result.
> http://www.odva.org/

'better deterministic'? as opposed to plain deterministic?

Wouter van Ooijen

-- -------------------------------------------
Van Ooijen Technische Informatica: http://www.voti.nl
consultancy, development, PICmicro products
docent Hogeschool van Utrecht: http://www.voti.nl/hvu



2007\11\21@131254 by William \Chops\ Westfield

face picon face

On Nov 20, 2007, at 11:46 PM, Xiaofan Chen wrote:

>> Ethernet tends to be fundamentally statistical rather than
>> deterministic, and it tends to make people who deal with
>> real-world THINGS very nervous...
>>
>
> That is why protocols based on standard Ethernet are
> popular now. Ethernet/IP is one of them. It uses standard
> COTS hardware components but adds something on the
> firmware side to achieve better deterministic result.
> http://www.odva.org/

The way I read it, Ethernet/IP (who was stupid enough to
name it THAT!) might improve determinism over something like
telnet, but it doesn't look like it does anything at all about
the inherent non-deterministic nature of ethernet.  (Now,
some of us might be pretty convinced that a non-deterministic
100Mbps pipe is going to beat a 9600bps deterministic CAN
network ALMOST all the time, but it's that ALMOST that causes
the nervousness.)

BillW

2007\11\21@132657 by peter green

flavicon
face

> The way I read it, Ethernet/IP (who was stupid enough to
> name it THAT!) might improve determinism over something like
> telnet, but it doesn't look like it does anything at all about
> the inherent non-deterministic nature of ethernet.  (Now,
> some of us might be pretty convinced that a non-deterministic
> 100Mbps pipe is going to beat a 9600bps deterministic CAN
> network ALMOST all the time, but it's that ALMOST that causes
> the nervousness.)
>
>  
No system is completely deterministic.

I see no reason why fully switched full duplex ethernet with qos is any
less deterministic than say CAN.

I suspect the real reason you don't see ethernet used much in cars is
because it is simply an insane level of overkill for the jobs being done.


2007\11\21@194310 by Herbert Graf

flavicon
face

On Wed, 2007-11-21 at 10:04 -0800, William "Chops" Westfield wrote:
> The way I read it, Ethernet/IP (who was stupid enough to
> name it THAT!) might improve determinism over something like
> telnet, but it doesn't look like it does anything at all about
> the inherent non-deterministic nature of ethernet.  (Now,
> some of us might be pretty convinced that a non-deterministic
> 100Mbps pipe is going to beat a 9600bps deterministic CAN
> network ALMOST all the time, but it's that ALMOST that causes
> the nervousness.)

Ethernet is a brute force design, throw as much bandwidth at the problem
and see what sticks.

This makes it great for certain things, but real time reliable CONTROL?
No, it's completely the wrong technology and I can't ever see it being
used for things like the buses in modern cars (at least not the buses
where determinism is important, i.e. transmission module talking to
engine module).

It's like trying to unscrew a Robertson screw with a Phillips
screwdrive, if you push hard enough and use enough force you might get
the screw out, but much more likely you'll strip the screw head...

TTYL

2007\11\21@203937 by Xiaofan Chen

face picon face
On 11/22/07, Herbert Graf <EraseMEmailinglist3spam_OUTspamTakeThisOuTfarcite.net> wrote:

> This makes it great for certain things, but real time reliable CONTROL?
> No, it's completely the wrong technology and I can't ever see it being
> used for things like the buses in modern cars (at least not the buses
> where determinism is important, i.e. transmission module talking to
> engine module).
>

Modern vehicle is more than engine/sensor control (which CAN/LIN
or FlexRay may still hold on). But MOST(Media Oriented System
Transport) is a serial communication system for transmitting audio,
video and control data via fiber-optic cables. Why not Ethernet?

And Ethernet based protocol is well suitable for many applications,
including control for medium cycle time I/O data. Now there are even
functional safety (IEC61508) protocol based on Ethernet.

Eg 1: Profisafe on Profinet
www.profibus.com/pall/events/press/article/00813/
Eg 2: CIP safety on Ethernet/IP
http://www.controleng.com/article/CA6405521.html

Xiaofan

2007\11\21@210946 by Nate Duehr

face
flavicon
face

On Nov 21, 2007, at 5:43 PM, Herbert Graf wrote:
>
> It's like trying to unscrew a Robertson screw with a Phillips
> screwdrive, if you push hard enough and use enough force you might get
> the screw out, but much more likely you'll strip the screw head...



Heh... what's a Robertson screw?

Is that anything like a Posidrive screw?

I have some radio gear here (GE MASTR II) that someone who'd worked on  
them for years kindly pointed out to me had posidrives, complete with  
little triangles etched into their tops (to tell you they're not  
normal Phillips screws) which work MUCH better with the right  
screwdrivers.

Finding posidrive drivers for my Xcelite handles was a pain, and they  
were more expensive, but they work oh-so-much-better on those radios  
when disassembling them for "surgery".

--
Nate Duehr
natespamspam_OUTnatetech.com



2007\11\21@211253 by Xiaofan Chen

face picon face
On 11/22/07, Xiaofan Chen <@spam@xiaofancKILLspamspamgmail.com> wrote:
> And Ethernet based protocol is well suitable for many applications,
> including control for medium cycle time I/O data. Now there are even
> functional safety (IEC61508) protocol based on Ethernet.
>

Some more examples:
1. http://www.automationworld.com/view-3099
"Schneider Electric, has become one of the principal members
of the Open DeviceNet Vendors Association (ODVA), joining
other principal members Cisco Systems, Eaton Electrical,
Omron Corp. and Rockwell Automation. Schneider Electric's
increased participation in ODVA coincides with ODVA's plans
to extend the CIP Network specifications to provide
compatibility of Modbus/TCP devices with networks built on
the Common Industrial Protocol (CIP)."

2. ProfiNet
http://www.profibus.com/pn/technology/description/

3. Industrial Ethernet
http://en.wikipedia.org/wiki/Industrial_Ethernet

4. CISCO white paper: it can be deterministic
http://www.cisco.com/application/pdf/en/us/guest/products/ps628/c1244/cdccont_0900aecd8013313e.pdf

Xiaofan

2007\11\21@223057 by Herbert Graf

flavicon
face

On Wed, 2007-11-21 at 19:09 -0700, Nate Duehr wrote:
{Quote hidden}

I'm not that familiar with posi-drive, a quick google shows it's a
trademark for snap-on? Pictures look like it's just another form of
phillips with four additional "points", I suppose this reduces the
chance of stripping the head? I have seen them, but they've never stuck
out as being much better then Phillips.

Robertson is "square", comes in a variety of standard sizes, and the
colour of the screwdriver handles is standardized.

Most common sizes are #2 (red handle) and #1 (green handle). #0 (yellow
handle) is very common in the "small world", and #3 (black handle) is
the largest. Never actually seen a #00 (orange handle). The standard
colours for handles makes it VERY convenient to use the screws since
just by looking at the screw I can choose the exactly correct driver.
That's rarely the case with Phillips.

The beauty of Roberson screws and drivers is you don't need to hold the
screw on the driver, the design of the driver and head lends to the
screw just "staying" on the head, far better then Phillips screws.

For more info:
http://en.wikipedia.org/wiki/Robertson_screwdriver

TTYL

2007\11\22@045509 by Alan B. Pearce

face picon face
>I'm not that familiar with posi-drive, a quick google shows it's a
>trademark for snap-on?

I thought Posidrive was a company in itself, that owned the trademark.
Posidrive tools are certainly made under licence by many companies, of which
snap-on is probably one.

>Pictures look like it's just another form of Phillips with
>four additional "points", I suppose this reduces the chance
>of stripping the head? I have seen them, but they've never
>stuck out as being much better then Phillips.

The angle on the side of the flutes is different (steeper) and the point is
sharper. IIRC the flutes are also thinner. This is why Nate found getting
the correct tool for the screw made them much nicer to use. The extra 4
points are not actually functional AFAICT, but are there mainly to
distinguish the screw and tip from Phillips ones.

I find the Posidrive is nicer to use than the 'ordinary' Phillips because of
these features, even when using the correct tip for the screw.

2007\11\22@082744 by Tony Smith

picon face
{Quote hidden}

Robertson screws are popular with woodworkers, well, Canadian ones at least,
you never saw them anywhere else for a long time.

The 'square peg in a square hole' means they are unlikely to slip, as
opposed to the more common slotted type.  Slipping is bad of course, you
scratch the wood, the screw itself, the bench, your arm and the cat.  The
second factor is that the hole in the screw itself won't get damaged
(distorted) from repeated use, so it still looks nice after you've
re-assembled your clock for the 23rd time.  Allen head screws are
more-or-less the same idea.

Phillips head screws were never intended to be removed, they were designed
for quick assembly.  The general idea is that they are self centering; if
you get the screwdriver tip reasonably close to the screw hole, it'll drop
right in.  You can do it blindfolded.  With other screw types you have to
pay a bit more attention, so assembly takes just that little bit longer.
More so when using power tools, hands up all those who have the bit already
spinning before fitting it into the screw head?  Tsk, impatient sods.  :)

Tony

2007\11\22@155644 by peter green

flavicon
face

> I'm not that familiar with posi-drive, a quick google shows it's a
> trademark for snap-on? Pictures look like it's just another form of
> phillips with four additional "points", I suppose this reduces the
> chance of stripping the head? I have seen them, but they've never stuck
> out as being much better then Phillips.
>
> Robertson is "square", comes in a variety of standard sizes, and the
> colour of the screwdriver handles is standardized.
I've seen them once fitted on a fence post that was put up by an
american neighbour, ended up cutting them off (it was the boundry fence
between us and they had moved away by the time it needed attention). I
think mecano may have used them at some point too (though most meccano I
have seen is flat blade).

here in the UK flatblade and phillips/pozi (which will just about work
in each others heads provided the screw is not stupidly tight) seem to
be the most common. allen keys (hex) are often seen on bycycles and
flatpack furniture but rarely anywhere elese. Torx is relatively common
in electronic stuff but rare anywhere else.

One thing about hex, torx and presumablly robertson is they all require
exactly the right screwdriver size or you can't generally get any grip
at all. This can be both a blessing (lots of screwhead fowling is caused
by poor choice of screwdrivers) and a curse (when you really need to get
a screw out ASAP and you don't have exactly the right size handy.



2007\11\26@152627 by alan smith

picon face
besides...if it ain't broke....don't fix it.  CAN and LIN gets the job done, the industry is tooled for it...why change?  Same reason RS485 is still used it alot of cases.

     
---------------------------------
Be a better sports nut! Let your teams follow you with Yahoo Mobile. Try it now.

2007\11\26@214042 by Xiaofan Chen

face picon face
On 11/27/07, alan smith <KILLspammicro_eng2KILLspamspamyahoo.com> wrote:
> besides...if it ain't broke....don't fix it.  CAN and LIN gets the
> job done, the industry is tooled for it...why change?  Same
> reason RS485 is still used it alot of cases.
>

Ethernet is not poised to replace CAN/LIN/RS485, but it
will be the backbones of information transfer on top
of the existing device level networks (DeviceNet, ASi,
Profibus DP, etc).

2007\11\26@222338 by Stephen R Phillips

picon face

--- Xiaofan Chen <RemoveMExiaofancTakeThisOuTspamgmail.com> wrote:

> Ethernet is not poised to replace CAN/LIN/RS485, but it
> will be the backbones of information transfer on top
> of the existing device level networks (DeviceNet, ASi,
> Profibus DP, etc).

In simple terms Ethernet has it's place, CAN (which is the low level
interface for the latter network protocols you specified) has it's
place. Ethernet requires a complicated physical interface with certain
limitations, which are bad for anything that requires fast
deterministic behavior. It's the difference between a tractor trailer
(Ethernet) and a bicycle or motor bike one is definitely more suited
for carrying large amounts of data, but not just a small packet like
CAN tends to carry.

Stephen


     ____________________________________________________________________________________
Get easy, one-click access to your favorites.
Make Yahoo! your homepage.
http://www.yahoo.com/r/hs

2007\11\27@002337 by Vasile Surducan

face picon face
On 11/19/07, Xiaofan Chen <spamBeGonexiaofancspamBeGonespamgmail.com> wrote:
> en.wikipedia.org/wiki/Industrial_Ethernet
> It seems industrial Ethernet is getting more and
> more popular.
>
> However it seems that Ethernet is not so popular in
> the vehicles.

Because in the vehicles there is already WIMAX2006 (portable wimax)
see 802[1].16e - 2005 "Air interface for fixed and mobile broadband
wireless access systems."

2007\11\27@014321 by Xiaofan Chen

face picon face
On 11/27/07, Vasile Surducan <TakeThisOuTpiclist9EraseMEspamspam_OUTgmail.com> wrote:
> On 11/19/07, Xiaofan Chen <RemoveMExiaofancspamTakeThisOuTgmail.com> wrote:
> > en.wikipedia.org/wiki/Industrial_Ethernet
> > It seems industrial Ethernet is getting more and
> > more popular.
> >
> > However it seems that Ethernet is not so popular in
> > the vehicles.
>
> Because in the vehicles there is already WIMAX2006 (portable wimax)
> see 802[1].16e - 2005 "Air interface for fixed and mobile broadband
> wireless access systems."

Hmm, is Wimax so fast to catch up in the real world?

For industrial automation, wireless is not really popular. Yes
there are some wireless applications (eg: remote RTU)
and some new initiativew (eg: Wireless HART).

Inherently wireless is even difficult to guarantee deteministic
behavior.

But it might be a good idea to replace MOST with wireless.

Xiaofan

More... (looser matching)
- Last day of these posts
- In 2007 , 2008 only
- Today
- New search...