What's new
Van's Air Force

Don't miss anything! Register now for full access to the definitive RV support community.

Garmin CAN Bus - single point of failure?

V111Pilot

Active Member
Here's an observation/question:

Garmin G3X (370) with 400WAAS, GNC255, ADHRS, Eng monitoring, G5.

I have noticed from the beginning that the CAN bus can be finicky, i.e. you have to have the boxes all grounded together, early on Garmin said only ground the shield of the CAN bus at the two terminating ends, now their most recent guidance is to ground the shield at every CAN connection. And so on. Once you figure it out, it is simple though :)

When it is working all is good, and I have never had an issue.

Having said that, after installing a G5, I was not sure if the system would still see the G3X LRU as the CAN bus termination point (as it initially was, with pin 28 jumped to CAN LO), or if the G5 should be the new termination point as it's CAN bus connection was just off the G3X connector and the lead is about 10" long.

The system did not like leaving the G3X as the termination point for the CAN bus. Easy fix, remove the pin 28 to CAN LO on the G3X LRU and install the CAN termination adapter on the back of the G5. Everything works.

This brings on the question, when the CAN bus was inoperative, I lost ALL CAN connectivity, engine monitoring, ADHRS, everything. So what would happen if in flight, you lost a ground on the CAN bus or the CAN bus was compromised i.e. cut, burnt, loose?

Would you lose everything? Even if you had a three screen system, with three ADHRS, they are ALL relying on the single CAN bus are they not??
 
There is an optional serial port connection between the ADAHRS and the G3X that provides a backup connection in case the CAN bus fails. Do you have that connection hooked up?

Jerald
 
(Edit: Jerald beat me to it.) You can wire an RS-232 backup connection between the GDU and the engine monitor and ADAHRS. Page 24-8 of the current install manual shows how to wire the GEA 24. You'll just need to remember to program the GDU's serial ports.

Which brings me to a question, what data format do you use for the backup RS-232 for the ADAHRS? The link in the manual for the configuration brings me to the GEA 24 page, so I wonder if it's "Garmin Instrument Data" like the GEA 24. I can assume, but we all know what that makes...
 
.... so I wonder if it's "Garmin Instrument Data" like the GEA 24. I can assume, but we all know what that makes...

Page 24-5, Configuration Guidance (B) indicates "Garmin Instrument Data" is correct for the GSU 25.
 
Hello,

We have had backup data bus support for the GSU 25 ADAHRS for quite some time, but providing backup data bus support for the GEA 24 engine monitor is a rather new feature for both G3X and G3X Touch systems.

Even in the unlikely event of a total loss of the CAN bus, primary flight and engine data is still available to the pilot via these backup data paths.

As Katie pointed out, page 24-8 the new Rev. Z G3X Installation Manual provides new guidance on wiring and configuring a backup RS-232 data bus from the GEA 24 engine monitor to any available serial port on a GDU 37X or GDU 4XX display.

For those using GDU 450/455/460/465 displays, there is also some interesting information on page 23-19 of the Rev. Z G3X Installation Manual (which isn't new information).

RS-232 port 6 on these displays is located on the P4501/4601 9 pin rear connector and provides a pre-configured (Garmin Instrument Data) serial port for connecting the backup ADAHRS data path. Using this serial port to connect the GSU 25 keeps the user configurable serial ports 1-5 open and available for other uses.

V111Pilot, we wouldn't read too much into the shield ground changes you mention. We started out recommending the more conventional CAN bus shielding method which specifies that the simple twisted, shielded pair braid be grounded at only the extreme ends.

At one point we discussed this with the engineers here who spend all their time testing and studying RF emissions and RF susceptibility, and they advised us that it would be slightly better to ground the shield at the backshell of each device, so we made a running change to start recommending this method of shield braid grounding.

Both methods of shield grounding seem to work equally well, so we tell everyone who asks that it is NOT necessary to change an existing installation to ground the shield braid at each device, but we do recommend this method for new installations.

Thanks,
Steve
 
Thanks for the responses. I was not aware of the RS-232 backup option and will be wiring that in this weekend.

And yes, I agree with Steve regarding the grounding of shields. This system seems to prefer being grounded only at the extreme ends. Next step is to double check that all of the power grounds are at the same spot to ensure no future issues.

B
 
Kindof off-topic, but since we're talking CAN Bus... What exactly happens if a node is too long?
 
Here's an observation/question:

Garmin G3X (370) with 400WAAS, GNC255, ADHRS, Eng monitoring, G5.

I have noticed from the beginning that the CAN bus can be finicky, i.e. you have to have the boxes all grounded together, early on Garmin said only ground the shield of the CAN bus at the two terminating ends, now their most recent guidance is to ground the shield at every CAN connection. And so on. Once you figure it out, it is simple though :)

My experience with the CAN bus has been just the contrary, I've had zero CAN bus issues both with my own aircraft and with a fair number of RV installs now. I've also added a few short "nodes" for G5's and G20A installs to existing networks with no ill effects.

A few years ago when the GI 106 AOA indicator came out there were some early units that were causing CAN bus errors (which have since been modified to correct the issue), you didn't mention that in your euipment list but thought I'd throw that out there in case you have one.
 
To clarify: The only issue I had with the CAN was the initial install and it had to do with the shield grounds. No problems at all since then (120+Hrs).

Did have the small hiccup w the G5, solved by moving the CAN termination.
 
CAN bus basics

Kindof off-topic, but since we're talking CAN Bus... What exactly happens if a node is too long?
I have not worked on the Garmin implementation but in general, the speed of the CAN bus determines the maximum cable length. This is nessisary for CAN to resolve communication conflicts if more than one device tries to use the bus at the same time (auto negotiate without any loss of data).

If the wire is too long, bus conflict errors can result. Garmin most likely has some secondary error correction on top of this.
 
I have not worked on the Garmin implementation but in general, the speed of the CAN bus determines the maximum cable length. This is nessisary for CAN to resolve communication conflicts if more than one device tries to use the bus at the same time (auto negotiate without any loss of data).

If the wire is too long, bus conflict errors can result. Garmin most likely has some secondary error correction on top of this.

If this conflict scenario happens, can it show up as an intermittent network error, or would the system just not work at all?
 
Kindof off-topic, but since we're talking CAN Bus... What exactly happens if a node is too long?

This is a simple question with a can of worms answer.

Longer nodes degrade the signal quality and if it degrades enough there will be transmission errors. These errors will not sneak through unnoticed, the CAN bus will detect them and tell the sending device to try again. As long as there are not too many errors you will never know there was a problem.

When you start getting too many errors then good data is not getting through, you just have a lot of failed retries. The CAN bus never gives up, it keeps trying but eventually the Garmin software says "enough, this isn't working" and gives you the "red X".

So how long is too long? The early G3X manuals specified the maximum node length at 1 meter and say "shorter is better". The latest versions say the maximum is 1 foot. I will bet there are many 1 meter nodes out there working just fine, and a few that didn't.

It is tricky because the signal integrity is effected by the entire network, not just one node. If all your nodes are short except for one you might be fine. If they are all 1 meter long then maybe not. I made all mine about 6 inches.
 
Tom, thanks, that helps me imagine what's going on.

I work with the G3X Touch system every day at my day job. We haven't had any known problems with the CAN bus so far after 6 completed airplanes, but I know there is still a lot to learn. So, I've signed up for the G3X Touch installation class and looking forward to it very much!
 
Hello all,

The title of this thread is a bit off the mark... although CAN is the primary interface to connect devices in the G3X system, of course we provide the most critical components in a G3X system (namely the ADAHRS, and optionally the engine interface) with redundant mechanisms for communicating with the pilot's display screen.

The CAN network interface has been around for more than 30 years, and has been used in millions of automotive and industrial applications. If you drove a car today, you probably used a CAN bus. In a proper installation, which is easily achievable by following a set of simple principles, CAN is highly reliable and fault-tolerant.

In the latest revision of the G3X installation manual (Rev Z), the section on CAN network installation and troubleshooting (section 2.3.1.3) has been rewritten to be more user-friendly, including plenty of examples. We encourage anyone with questions about the G3X CAN bus to check it out.

- Matt
 
Last edited:
In my day job as an embedded programmer I've worked with CAN bus for almost 20 years. It's exceedingly robust, and you can violate lots of rules and it still works (I've seen nasty unplanned "star" networks in piles of wires on the floor - also at a NMEA2000 "plug fest" where a dozen companies plugged into the same bus at the same time, physically strung out in a hotel ball room, I've seen huge numbers of devices communicating simultaneously). Although obviously doing it by the book is best.
At a hardware level, CAN devices track message failures and the hardware will stop attempting to transmit after a certain number of failures. Sometimes network protocols will tell you to restart the device (like ISOBUS, the CAN network for farm tractors). But that has to be a deliberate software effort. I've seen CAN networks function with missing terminators, or terminators in the middle of the network, or extra terminators. If you wire it as Garmin tells you (ISO 11898 standards), it's nearly impossible to kill it, outside of major wiring failures.
 
...... it's nearly impossible to kill it, outside of major wiring failures.

As long as you define getting one strand of the wire shield contacting the bus wire as a "major wiring failure". Since fixing that issue it has been rock solid.
 
As long as you define getting one strand of the wire shield contacting the bus wire as a "major wiring failure". Since fixing that issue it has been rock solid.

I believe that just about anyone with electrical knowledge, and even most without, would consider introducing an un-specified ground into a circuit as a MAJOR wiring failure. One strand or 100 strands, a ground is a ground.

Larry
 
It is also called a "short circuit". Fortunately, this was not on a power cable. Building a CANBUS harness and checking the hi and lo and shield connections from node to node is quite simple. In fact, checking all your wiring for continuity and shorts is pretty simple. A short circuit is the exact opposite of the goal to isolate and insulate individual wired paths so I'd say it was more of an error by the builder and not a failure where something broke.

Yes, I've had CANBUS problems when I first started. I stripped the signal wires using 24 slot and I had 22 wire and I also had an old batch of solder sleeves that did not flow properly. it did not pass the first power-up. :(

CANBUS is great on the G3X when built as directed.
 
Can you wire a CANBUS through disconnects like at a wing root or is everyone running one-piece wires from each equipment's connector?
 
Kindof off-topic, but since we're talking CAN Bus... What exactly happens if a node is too long?

Could be a lot of different things. We've typically seen a high network error rate associated with random autopilot disconnects.
 
Intermittent CAN fault - please help

Hi All,

I have a G3X system in my RV10 and have developed an intermittent fault in the system. I wired it myself and have previously done a similar system before that had no snags for my Kitfox 7.

90% of the time everything works just fine, especially on the ground when network error is 0% and every box is ticked green, when in still air, and for no apparent reason, I get intermittent autopilot faults and red crosses that affect mainly the equipment in the back of the airplane, namely the remote radio, remote transponder, ADHRS and AP servos. The GEA24, Audio panel, MFD all work fine in the front of the airplane and are on the CAN bus. However the AOA indicator is in the front also and that crashes at the same time. a few seconds after being off, everything is back working fine, then a few minutes later its crashed again. Some flights I can go 2 hours and not a single error, sometimes its every 5 minutes.

When there are red crosses on the screens the LEDs for power are still blinking on all the devices so it must be a CAN bus problem. I have checked that these failures are not related to power distribution either from the VPX.

I have made the node lengths and whole bus as short as possible and have checked all the connections and backshells for short circuits. The resistance is correct for the CAN bus. Because everything is fine on the ground it makes it very hard to troubleshoot anything. I can re-create the fault on the ground by manipulating the wires or bunched harnesses and I must have opened and checked everything nearly 100 times now. I have also had professional airline wiring people come and look at the system and they can't find any faults. They are also lost as to where the error is.

What am I missing? Any tips would be much appreciated. I am at the point of just removing the whole system and starting from scratch.

Thanks

Amer
 
When I built my system, I tested each node and isolated the rest until I confirmed the integrity of the signal and the grounds and wiring. Then I added the next node. In your situation, you can isolate the servos with the adapter that comes with the servo kit as one step. You can also power down and trace all wiring from each can-bus node to each node. There should be no shorts and no open connections and continuity on the high and low wiring with no resistance and no changes as you move connectors. With the meter set up to measure continuity, the harnesses disconnected from your boxes, I'd then start wiggling connectors. Your can-bus nodes should be rock solid on the meter.

If wiggling did not highlight the error I would simply open the connectors and start diagnosing each can-bus node for bad connectors or possibly broken wire during stripping (and other errors in assembly). I did that and found my earliest nodes were the culprits. It is not hard to find the fault in the sense there are only a few wires to each node and they have to be continuous in your can-bus diagram. Of course, having that diagram available and correct is the main check item. Post it if you have doubts.
 
Can you wire a CANBUS through disconnects like at a wing root or is everyone running one-piece wires from each equipment's connector?
Absolutely you can use connectors. I used AMP CPC connectors that were previously installed with an Avidyne system. I removed the old and installed the G3X and I passed the CAN-Bus through three pins (shield, hi, lo) and it works perfectly. My panel comes completely out of the airplane and fully attached with AMP circular connectors and I wanted to preserve ability to remove it without cutting wires.
 
Many thanks

When I built my system, I tested each node and isolated the rest until I confirmed the integrity of the signal and the grounds and wiring. Then I added the next node. In your situation, you can isolate the servos with the adapter that comes with the servo kit as one step. You can also power down and trace all wiring from each can-bus node to each node. There should be no shorts and no open connections and continuity on the high and low wiring with no resistance and no changes as you move connectors. With the meter set up to measure continuity, the harnesses disconnected from your boxes, I'd then start wiggling connectors. Your can-bus nodes should be rock solid on the meter.

If wiggling did not highlight the error I would simply open the connectors and start diagnosing each can-bus node for bad connectors or possibly broken wire during stripping (and other errors in assembly). I did that and found my earliest nodes were the culprits. It is not hard to find the fault in the sense there are only a few wires to each node and they have to be continuous in your can-bus diagram. Of course, having that diagram available and correct is the main check item. Post it if you have doubts.



OK thanks, will have another go at it. I think I need to open each backshell and go systematically through the whole system from end to end. I also have only terminated the shielding of the bus at the 2 ends (as per the old manual), but I understand the new G3X manual asks you to terminate the shielding at every back shell. Perhaps this is also the issue?

Best

Amer
 
Just start by ringing the connections, you don't even need to remove all the shells. You could find the problem quickly with a few male/female pins and the ohmmeter. I'm betting it will surface quickly. If you find you nicked insulation or some other problem, then maybe start reviewing all the other shells. Good luck.
 
Not sure what generation GI260 you have but some of the early ones would cause can bus faults. If you have an older one try disconnecting or powering off the GI260 and see if that helps.
 
OK thanks, will have another go at it. I think I need to open each backshell and go systematically through the whole system from end to end. I also have only terminated the shielding of the bus at the 2 ends (as per the old manual), but I understand the new G3X manual asks you to terminate the shielding at every back shell. Perhaps this is also the issue?

Best

Amer

The old guidance works in most cases. The newer guidance just gives more overhead according to Garmin.
 
OK Thanks

Thanks All,

Will have a go this weekend and work systematically.

Fingers crossed it will get sorted easily.

Best wishes and happy landings

Amer
 
GEA Backup Data to PFD or to MFD

Does anyone know if it matters which GDU the GEA24 Data Backup line is connected to?

I am planning on two GDU 460 units; one as PFD and the other as MFD. I initially have run the Data Backup RS232 connection from the GSU25 to the PFD and the Data Backup RS232 connection from the GEA24 to the MFD. I did this thinking that the attitude information would be used on the PFD and the engine information on the MFD. But I have seen others connecting the GEA data backup to the PFD.

Thanks to anyone who can shed some light on this question!

Leonard Westermeyer
RV-7A “almost” done.
 
EIS Backup Datapath

Does anyone know if it matters which GDU the GEA24 Data Backup line is connected to?

I am planning on two GDU 460 units; one as PFD and the other as MFD. I initially have run the Data Backup RS232 connection from the GSU25 to the PFD and the Data Backup RS232 connection from the GEA24 to the MFD. I did this thinking that the attitude information would be used on the PFD and the engine information on the MFD. But I have seen others connecting the GEA data backup to the PFD.

Thanks to anyone who can shed some light on this question!

Leonard Westermeyer
RV-7A “almost” done.

The EIS information will only be available on the GDU to which you ran the backup RS232 line. If you would like the EIS backup data path to be wired to the PFD, that will work just fine, just keep in mind that information will only be available on the PFD.

Thanks,

Justin
 
Last edited:
Back
Top