HIC to Garmin GEA 24

You are correct that most CAN bus cables don’t have the third (ground) wire, but they do have a braided shield. At the stub connections (pass through) and the end terminations the best practice is to solder a wire to the braided shield, in which case this gives you the ground wire. There is a two part video on YouTube that shows how to do this properly. See the link below.

The Rotax HIC connector has the separate ground terminal because the connector is plastic and there is no way to connect a chassis ground at that location. So, your soldered-on shield conductor connects to this third (ground) conductor.  

Regarding where to ground and where to isolate, the Rotax injected engine is a special case. What you are trying to achieve is to isolate the engine ground from the aircraft ground. This is because the engine normally operates fully electrically isolated, and you don’t want to make a ground path between the two with your CAN bus. You can do this one of two ways. You can connect all the shields together at the stub, and then not connect the shield to the chassis at the GEA-24. Or, you can ground the stub shield at the GEA-24, and leave it isolated at the stub connection like shown in the Garmin diagram. You only need one end of the shield grounded for it to drain off EMF. That being said, when I’m connecting multiple devices together that are all on the airframe electrical system, I connect the shield ground at every termination.  I hope this helps.  

https://www.youtube.com/watch?v=nzpFLyrjQhM

Your thinking is similar to mine.  I worked on a fair number of communications systems, from DEC Dataway in the old days up through Ethernet and a few others, but not CAN, prior to retiring a decade ago. I learned that standards aren’t always standard and different vendors interpret standards differently. Often, with plastic connectors, the ground pin provided is a chassis ground similar to what Garmin calls the shield block on their metal connector housings and can thus be used to terminate the shield.  However, this isn’t always the case.  Devices that use isolated communication circuitry often use a signal common that is isolated from the chassis ground and simply provides a reference for the communication chips.  The competent engineers call this signal common rather than ground, but this is hardly a universal practice.

I wish Rotax wasn’t silent on this in their installation manual.  Since they call it ground, I am assuming it really is a chassis ground, but then Rotax calls an alternator a generator so I am always suspicious of their terminology.

Matthew,

I guess the proper name in the case would be “shield drain” since that’s the only function it carries out. The CAN architecture is defined as a two-wire shielded bus. The bus has a 120 ohm resistor at each end giving it a total resistance of 60 ohms across the two conductors. Rotax provides a 120 ohm resistor inside the ECU for each CAN (Lane A and Lane B) and once connected together that gives you the 60 ohm total resistance. The stubs are supposed to be limited to .3 meters and there is a one device limit per stub. The architecture is linear, with star connections not allowed. I think that by Rotax providing the shield drain as an insulated conductor in the HIC harness, makes it much easier to keep the engine ground reference isolated.

Both Rotax annd Garmin seem to have a very clear understanding of the CAN. But I agree that some SLSA manufactures interpret the CAN standard differently, especially on the avionics CAN that connects the remote modules to the head unit. And that’s being generous, the truth is that some manufacturers don’t understand them, and with the ASTM “self-monitored” process this only gets corrected when a problem shows up that forces the manufacturer to dig deeper.  I’ve seen SLSA aircraft with all the CAN wires from the avionics modules simply star connected together at a single screw type termination block. Exactly opposite of the CAN standard.  What’s interesting is that in these cases you see very few data errors. I believe with the short cable lengths in these small aircraft the CAN is very forgiving.  

I am too cheap to buy the standard document at $220, but there does seem to be different opinions as to how the standard is defined.  Many say it is a two wire standard can_h and can_l.  However many others, including a person I saw from Bosch, said that it is a three wire standard with can_h, can_l and can_gnd.  Shielding is optional depending on the environment.

And Rotax and Garmin take different approaches.  Garmin uses just two wires with a shield.  Rotax uses three wires including the can_gnd on all of their CAN channels, but display CAN and maintenance CAN.  That is what gives rise to the unusual wiring diagram on page 28-4 of the rev. AW Garmin G3X installation manual.  It shows a 3 wire shielded cable connecting HIC A CAN to HIC B CAN with the shield connected to ground at both ends.  It then shows a 2 wire shielded cable that taps into the 3 wire cable, but ignores the can_gnd conductor and also keeps its shield isolated from the shield on the 3-wire Rotax cable.

That is the oddity that got my attention and generated my questions.  I notice that Garmin now offers a “B” version of the GEA 24 that uses separate ports for the HIC A and HIC B connections and treats them as fully separate networks.  And with the GEA 24B, they leave the Rotax can_gnd lines unconnected.

I suspect the key aspect of all this is as you stated that in an airplane the distances are so short that it is unlikely that a significant ground potential difference will ever exist between devices (sans something like a lightning strike) and CAN is sufficiently robust that it will easily accommodate any small differences that do exist.