Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

Re: [802.3_10SPE] number of nodes for the multidrop channel



Kirsten-

I don't understand your assertion that:

Of course delays increase with the number of nodes. However, this is also a job for the network designer to decide on acceptability.

Delay across a multi-point CSMA/CD segment is only dependent on the propagation delay between the transmitter and receiver and any backoff delay caused by other offered traffic (if any) from other points on the segment (this can be eliminated with the TT (Timed Token) protocol).

Delay only increases with the number of nodes if:
- you have a (3 port?) bridge at each node.  Then there is a delay of a little larger than one packet duration unless it is a cut-through bridge.  Cut-throughs have their own set of problems and haven't been favored by the standards community but could possibly be OK in an automotive situation.
- you have a repeater at each node.  In that case the delay per repeater is (if I remember correctly) about 9 symbol times (i.e. it would longer for 4B5B than for Manchester)

My impression of automotive SPE is that it was to be used only for items that are not time critical (e.g. locks, latches, dash switches, environmental sensors and controls, lighting, window lifts, mirror controls, turn signals, tire pressure signals etc.) so none of this should be an issue anyway.

I hope this helps.

Geoff

On Jan 17, 2018, at 9:08 AMPST, Kirsten Matheus <Kirsten.Matheus@xxxxxx> wrote:

Dear all, 
 
I have a comment/question concerning the 15 participants in the multidrop channel.
 
In the measurements of the multidrop channel we presented, we had 8 participants. In this set up a change of the MC to 43dB (which in our opinion would be an important change) is (just) feasible. Who is going to present measurements of the channel in Geneva that show, whether 15 participants yield the same results? What is the motivation behind 15 participants? Is there a specific application for which 15 units on 25m is necessary?
 
In my understanding, it is always allowed to use the multidrop link with more than the specified number of units (technical feasibility might be possible e.g. because you have a significantly shorter link). However, if 15 is defined for 25m, then it needs to work, even if in 99.9% of the cases it is either 15 units or 25m but not both at the same time.
 
Of course delays increase with the number of nodes. However, this is also a job for the network designer to decide on acceptability.
 
Any comment?
 
Kind regards, 
 
Kirsten
 
--
BMW Group
Dr. Kirsten Matheus
EE-322
Knorrs
traße 147
80807 München 
 
Postanschrift:
80788 München 
 
Telefon: +49-89-382-12635
Mobil: +49-151-601-12635
Mail: kirsten.matheus@xxxxxx
Web: http://www.bmwgroup.com
<image001.gif>
----------------------------------------------------------------
Bayerische Motoren Werke Aktiengesellschaft
Vorstand/Board of Management: Harald Krüger (Vorsitzender/Chairman),
Milagros Caiña Carreiro-Andree, Markus Duesmann,
Klaus Fröhlich, Pieter Nota, Nicolas Peter,
Peter Schwarzenbauer, Oliver Zipse.
Vorsitzender des Aufsichtsrats/Chairman of the Supervisory Board: Norbert Reithofer
Sitz und Registergericht/Registered in Germany: München HRB 42243
----------------------------------------------------------------