RE: different cables in the same network

Hi Tamas,

While the specification of the cables obviously has an impact on the
performance of a CAN network, the design of most FMS networks is so limited
in bus length, number of nodes and bus load that you can get away with
nearly everything as long as you have at least one terminator somewhere on
the network. Iveco is using the more or less the same basic installation you
just described: FMS gateway under the fuse board on the passenger side, and
a stub cable running up to the driver side overhead console as to
accommodate the future installation of a fleet management system, while all
termination is performed below.

I would venture that for all vehicles equipped with an aftermarket node on
their FMS network that I've encountered, about half of them had potential
issues with the network due to anomalies with the (physical) connections.
Yet it is pretty rare to find a situation where it actually malfunctions due
to the robustness of CAN combined with the simplicity of the FMS network
itself.

The actual things you should be worried about are (in that order):
connectors, terminators and 'configurable devices'. 

The single most common source of problems I've encountered is in badly
specified and/or badly installed connectors. There are those that use the
cheapest self-splicing connectors one can imagine, preferably suited for a
wire gauge completely different of the CAN wiring in the vehicle itself.
There are the soldering fanatics that have the iron, but not the skills nor
the proper heat shrink tubes to stabilize the connection afterwards. There
are the enlightened souls who did buy real, suitable automotive connectors,
but were out-of-pocket afterwards so they couldn't buy the matching crimp
tool and use needle-nose pliers to manhandle the wires into the contacts.
And then there are those who think that screw-style connectors are a swell
idea in an environment that's vibrating more than the QA department of an
adult-toy production facility, and who happen to have misplaced their bottle
of Loctite just when they were performing the installation. So: good,
qualitative, easily manipulated connectors will go A LONG WAY(tm) in
ensuring your stuff does what it's supposed to do. I like Tyco/AMP MCP28 a
lot for these particular reasons.

In case the connectors did not kill the network, chances are that it's the
termination (or lack thereof) that's doing you in. Nearly all FMS
implementations in vehicles provide at least 1 terminator, and most even
two, so over-termination of the bus is far more likely. Especially because
for some reason a lot of the device designers think it's a brilliant idea to
perform the termination on the PCB of their node. Invariably, each device
has its own way to enable/disable termination (jumpers, jumper wires that
need to be installed, jumper wires that need to be cut, or even fancy
'configurable' termination that's switched in software), and each device has
its own default whether the built-in termination may be already activated or
not, given the phase of the moon and its inclination in relation to the
ecliptic plane on the date the PCB was assembled. As you can imagine, a lot
of installers don't even bother to change anything with the termination
unless their gizmo isn't working after installation, after which they
"invert" the termination setting on the device (whatever it may have been)
and hope for the best. After a couple of nodes like this, your CAN network
is operating pretty close to continual short-circuit. A real short-circuit
to e.g. VBAT, may or may not fry those freebie terminators, depending on how
well they are spec'd. It shouldn't be surprising that I'm a big fan of
plug-in terminators on both ends of the network (or the longest drops,
whichever those may be). They are easily counted, can usually take a beating
and if blown are far simpler/cheaper to replace than the node itself.

In the off chance that your FMS doodad still isn't working, and it's not one
of the above, you're probably USING IT WRONG(tm). E.g. configured the
controller to be passive (to be on the safe side -- we don't want to cause
perturbations), while there's only one other node on the bus (it happens
more than you'd like to think), or plainly misconfigured the bitrate. Unless
your bus is in a really bad shape, sample point issues almost never crop up,
as they are more likely with long bus lengths (not common for FMS) and a lot
of nodes (also not common for FMS). The biggest source of these problems are
installers that know what wires to connect to where, but really don't have
any way of validating that the installation they just performed is actually
working, even the dealership where you bought the vehicle with the FMS most
of the time hasn't got the proper tools to validate if it's actually
functioning, other than what their diagnostics doohickey is telling them. 

So if this all fails, you look at the cabling, find the place where that
other guy that was working (or the marten that was gnawing) on the truck
nicked or pinched the cable and fix it. 

Cheers,
Bram
-----Original Message-----
From: Hadarits Kft. [mailto:[email protected]] 
Sent: maandag 24 november 2014 13:00
To: 'Bram Kerkhof'
Subject: RE: [CANLIST] different cables in the same network

Dear Bram,

thanks, it's good news. We don't have much practice in this area so it isn't
easy to measure the importance of different specifics.

Regards: Tamas

-----Original Message-----
From: [email protected]
[mailto:[email protected]] On Behalf Of Bram
Kerkhof
Sent: Monday, November 24, 2014 9:53 AM
To: [email protected]
Subject: RE: [CANLIST] different cables in the same network

CAN is pretty robust, and the great majority of FMS networks as implemented
by the OEMs are nowhere near the limits of CAN. I wouldn't worry too much
about the type of cabling, let alone the stub length one particular node as
long as it's not measured in the tens of meters. I'm pretty sure that you
could connect most FMS nodes using a pair of metal coat hangers, and still
have it work.

And having seen a lot of installations of telematics equipment in HGVs, I
would even think that the metal coat hangers wouldn't be that bad a choice
in comparison with what's already out there ;-)

Cheers,
Bram
-----Original Message-----
From: [email protected]
[mailto:[email protected]] On Behalf Of Hadarits
Kft.
Sent: zondag 23 november 2014 21:54
To: [email protected]
Subject: RE: [CANLIST] different cables in the same network

Dear John,

Thanks for your detailed answer. As I know at the J1939 network the max.
allowed bus lenght is 40m, so I hope our newly installed plus 4m stub lenght
will remain within its tolerance. But I'm afraid this newly installed stub
(without termination resistor) will be longer than the existing FMS CAN
network (with the two 120ohm resistors). As you adviced I will check it with
scope.

Regards. Tamas

-----Original Message-----
From: [email protected]
[mailto:[email protected]] On Behalf Of John
Dammeyer
Sent: Sunday, November 23, 2014 6:44 PM
To: [email protected]
Subject: RE: [CANLIST] different cables in the same network

The short answer is yes.  You can mix it.  I just has a similar discussion
with a client yesterday.
There's nothing magic about CAN bus wiring. Although CAN isn't a truly
differential signal, since it's referenced to ground, the principles are
exactly the same.  A signal change somewhere on the bus propagates in both
directions.  When it reaches the end of the bus if there isn't any
termination or in other words if it hits a really high impedance it
reflects.  Like throwing something at a wall.  It bounces.   A change in
cable impedance by changing cable types can also do the same thing but to a
lesser degree.  
The reflection which, is lowered in signal level goes back down the bus
until it hits the other end and then bounces back. Each time with less
energy.  On the way back it runs into the other reflection and either sum or
subtract creating a larger or small signal.  As long as that point isn't at
a tap for a CAN device it doesn't matter. 
Only if there's a CAN device at that point does it become a problem.  If
this signal happens to coincide with a new bit in the same message or the
next message and it's strong enough to change the signal then you get a bit
error in the message.  CAN handles this by retransmitting the message so
it's transient.    And the next message may have a different value in the
data or CRC part and the reflections do no harm.  So the errors are very
transient and incredibly hard to track down.
That's why we have termination resistors at each end.  These are matched to
the impedance of the cable.  You can use 45 Ohm cable if you want and have
22.5Ohm resistors but that pushes the drive capability of the CAN interface
so people stay with 100 to 120 ohm and live with minor reflections.
The faster the bit rate and the longer the cable the more likely breaking
the rules will result in peculiar errors.
So the advice is adhere as best as possible to standard engineering
practices but if you can't at least scope the bus and look at the signal.
CAN is pretty robust.  Odds are switching cable types won't make a
difference if the network is short.
John Dammeyer

> -----Original Message-----
> From: [email protected] [mailto:canlist- 
> [email protected]] On Behalf Of Hadarits Kft.
> Sent: November-23-14 9:05 AM
> To: [email protected]
> Subject: [CANLIST] different cables in the same network
> 
> Hi all,
> 
> we deal with installation of fleet management devices into trucks. One 
> of our latest work was to make connection between the fleet management 
> device and the digital tachograph to reach the possibility to download 
> the data from the tachograph via CAN network. In this case the fleet 
> management device was already connected to the truck's FMS CAN network 
> and we had to connect to this network the tachograph too: FMS
> gateway+fleet management
> device+ tachograph. I hope it is understandable till now.
> In the trucks I found only unshielded twisted pair wires for CAN purposes.
> Now we got from the fleet management device's producer shielded 
> twisted pair wires to make the connection between the fleet management 
> device and the tachograph (the required wire lenght is appr. 4-5m).
> 
> My question: is it possible to use shielded and unshielded wires 
> together
in
> one network?
> 
> Thanks in advance: Tamas Hadarits
> 
> www.tachografszerviz.hu
> 
> --
> Archives and useful links: http://groups.yahoo.com/group/CANbus
> Subscribe and unsubscribe at www.vector.com/canlist/ Report any 
> problems to <[email protected]>

--
Archives and useful links: http://groups.yahoo.com/group/CANbus
Subscribe and unsubscribe at www.vector.com/canlist/ Report any problems to
<[email protected]>

--
Archives and useful links: http://groups.yahoo.com/group/CANbus
Subscribe and unsubscribe at www.vector.com/canlist/ Report any problems to
<[email protected]>


--
Archives and useful links: http://groups.yahoo.com/group/CANbus
Subscribe and unsubscribe at www.vector.com/canlist/ Report any problems to
<[email protected]>



--
Archives and useful links: http://groups.yahoo.com/group/CANbus
Subscribe and unsubscribe at www.vector.com/canlist/
Report any problems to <[email protected]>