Drop Lengths Section 2.1

Martin Perlot of Silverleaf propose the following:

Edit Table 2.1.1a:

Allow the Main Trunk to use wires of 15 AWG – 24 AWG.

Add to Section 2.1.1

It is acceptable for the drops to be of a different gauge than the main trunk.

New Section 2.1.2

The network topology is a linear bus with drops of limited length. Each drop must connect exactly one node to the main trunk. Both ends of the bus shall be terminated with a 120 Ω resistor, and these shall be the only terminating resistors on the bus. Table 2.1.2 sets limits on the number of nodes (devices) on the network, the bus length, and the length of drops. The value in the table for the maximum node-to-node path is longest distance between any two nodes – typically the length of the trunk plus the length of the drops nearest the ends.

New Table 2.1.2
Bus Limitations

AWG (Signal) 16ga 18ga 20ga 22ga 24ga
Number of Nodes 172
Maximum Node-to-Node Path 660m 425m 265m 160m 105m
Longest Drop 6m
Total Drop Length 30m

Attached to this submission please find a study that provides the calculations that provides the reasoning.

AttachmentSize
CAN_Topology_Study.pdf145.19 KB
Section 2.1 Bus Length edits.doc30 KB
Section 2.1 Bus Length edits - Revised.doc14 KB

Note the Revised document. I

Note the Revised document.

I have incorporated Daniel Allen's feedback into the submission, in preparation for incorporation into the protocol document.

Proposed: Change Section

Proposed: Change Section 2.1.1 from "The physical medium shall be unshielded twisted pair wire; shielding is optional. The wires are defined in 2.1.1a. " to "The signal medium shall be unshielded twisted pair wire, the characteristics of which are defined in Table 2.1.1a. A shield is allowed. Optionally, power may be provided in the same cable. The characteristics of the power supply and the transmission medium are defined in Section 2.3, and there is no requirement for the power medium to be twisted or shielded.

Justification: This clarifies certain ambiguities in the section regarding whether the power medium has to match the signal medium.

The following notes are from

The following notes are from Daniel Allen, Tiffin Motorhomes:
---
* Allow drops to use 15awg like main trunk?
* Standardize to one term used for the backbone of the CAN; main trunk, bus, backbone, etc.
* Can spare drops without a node be installed for future use? Proposed new Section 2.1.2 sentence 2 does not allow for spare drops already installed.
* Proposed new Table 2.1.2 replace “AWG (Signal)” with “CAN_H, CAN_L signal(or data) wires gauge”.
* Proposed new Table 2.1.2 replace “16ga” with “16 AWG” to be consistent with other tables in section. Do same for remaining entries in table.
* Section 2.3(f): How are network power supplies expected to limit the current? Setting? This should probably read something like, “Network power supplies shall be sized correctly so they do not exceed the current allowed based on network cable conductor size and length, as shown in Table 2.3.”

In response: - 15 AWG drops?

In response:

- 15 AWG drops? Why not? Actually, I thought I had put that in there, but I guess I didn't. There is no technical reason not to allow it.
- Agreed. I tried to be more consistent within this text, but the terms are spread around the entire document. We should simply direct the Administrator to scan the document for the term "bus" and replace it with "trunk" when it refers to just the central cable (and leave it when it refers to the trunk AND the drops together). The work "backbone" should just be replaced with "trunk". "Main trunk" can be replaced with "trunk".
- Empty drops are harmless as long as the overall length doesn't exceed the threshold. I would not object to changing the wording to "no more than one node".
- If that's more clear to others, I have no objection.
- And this is embarrassing, as I made a point elsewhere of avoiding using "ga" in favor of "AWG".
- Agreed. David Bailey is working on this particular item for the task group - we'll make sure that he sees this.

I think this language could

I think this language could be interpreted to imply prohibiting termination inside a device. Is that intended?

" Both ends of the bus shall be terminated with a 120 Ω resistor, and these shall be the only terminating resistors on the bus"

Perhaps add a sentience talking to device termination, either clarifying it is allowed (and maybe some labeling / indication?), or prohibited.

" In lieu of bus termination, termination may be placed at a device/drop. In such cases the device shall be located at one end of the bus and clearly labeled as containing integrated termination"

Or

"Termination shall be located only on the distribution bus level, termination inside devices or on drops are prohibited"

My sense from the task group

My sense from the task group was that we should prohibit internal terminators altogether, so that's how I wrote it up.

Personally, that's my preference, too, as I see OEMs continuing to struggle with "hidden" resistors in their RVs. What makes the problem nasty is that when the network is loaded to 40 ohms, things "mostly" work. CAN transceivers are usually rated to work down to about 45 ohms, and of course most of them can do better than that - at least, most of the time. So it's easy for systems to get past QC and into the field. And once in the field, the customer complaint is of intermittent failures of various network processes. It's a serious challenge for troubleshooters - who don't usually think of checking the network quality right away - and once they find the problem often the only solution is to completely replace the offending product.

In the Zoom meeting, I recall some suggestion that products might be allowed to have a resistor if they were clearly labeled as such, but there was some questions about how useful/practical that really was. I also recall a suggestion that we require the connector itself to be distinguishable by color, which drew similar criticism. My sense of the debate was that the OEMs have been stung by this and want to get the problem under control, and any vendor that has an internal resistor in their product can simply remove the resistor from the programming of their pick-and-place machine.

(I'll note that many of the products from my company have a resistor available, but require adding an internal jumper to use it. We simply never ship these out with the jumper installed. When we have an aftermarket customer that wants to make a mini-network, they can open the box and add the jumper.)