Weird Network Problem

[Dockhead]

I recently added two N2K devices to me network:

1. An N2K depth sounder

and

2. A Maretron tank sender.

And I started having problems, as if devices were conflicting with each other.

N2K network problems can be maddening, because the testing required to isolate a problem can involve a huge number of combinations.

The two new devices are connected at the forward end of the network, with depth being the very last device forward, with the resistor on it.

The last set of network T's have the following devices:

1. depth
2. boat speed
3. H2183 compass
4. H5000 computer
5. tank sender

I could not get all these devices to work at the same time. I tried all kinds of different thing, including even disconnecting my satellite compass to reduce the bus load.

Nothing. One device or another wouldn't work, and sometimes all these 5 devices would drop out.

I got some funny feeling about the one Maretron network T connected with 3 black plastic B&G T's. And swapped it for another black plastic B&G T. And lo and behold, all 5 devices are now working right.

How could that be, really? Incompatibility between NMEA2000 certified devices, such simple ones?

Anyone experience anything like this? Weird.


Incidentally, I noticed that the bus load was really high -- 40% to 50% -- as long as the Furuno satellite compass was connected, dropping to low 20's when I disconnected it. This device gives a ton of data at very high refresh rate. I was advised by someone that the 0183 version of this device, directly connected to the H5000, is much better, because the N2K network will introduce a lot of latency which messes up the wind calculations.

[cram]

Have you calculated the LEN and where is the power inject?

[Dockhead]

Originally Posted by cram

Have you calculated the LEN and where is the power inject?

I thought about that, but taking other devices off the network with equivalent LEN to the ones giving problems didn't help.


Power injection is a two-meter cable away from this group of T's, and the network is fed from a stabilized power supply at 13.6v (I'm a 24v boat).



And now, with the B&G network T instead of the Maretron one, all devices are working OK, even with the other devices re-connected.

[stevead]

Could be a dodgy T,-piece which is not unheard of.

With the "suspect" T inserted, did you measure both the resistance across the data pairs blue/white (should read 60 Ω) and the voltage across the DC pairs red/black (should read your input voltage of 13.6V).

Bus load is just a ratio of the number of messages (essentially number of bits) sent across the network compared to the bit rate of NMEA 2000 (250 kb/s).

What affects latency is a combination of bit rate, the length of the network and bus load. At a basic level, how long will it take an electron travelling at 250 kb/s to travel the length of your piece of copper. WRT to bus load, as the CAN bus uses "Carrier Sense, Multiple Access" (CSMA) with collision detection it means that devices must wait until there is no activity on the network before they can transmit.

In the case of a collision, the arbitration scheme is designed so that messages with a higher priority are retransmitted before messages of a lower priority. On a heavily loaded network, there is a possibility that there is some delay before a device senses no activity on the network, and in the case of collision, lower priority messages may have their transmission delayed. I'm not sure if these stats are captured anywhere. Someone with more knowledge may be able to say whether bus loads of 40-50% are going to negatively impact your network. I personally don't think it will.

[Dockhead]

Originally Posted by stevead

Could be a dodgy T,-piece which is not unheard of.

With the "suspect" T inserted, did you measure both the resistance across the data pairs blue/white (should read 60 Ω) and the voltage across the DC pairs red/black (should read your input voltage of 13.6V).

Bus load is just a ratio of the number of messages (essentially number of bits) sent across the network compared to the bit rate of NMEA 2000 (250 kb/s).

What affects latency is a combination of bit rate, the length of the network and bus load. At a basic level, how long will it take an electron travelling at 250 kb/s to travel the length of your piece of copper. WRT to bus load, as the CAN bus uses "Carrier Sense, Multiple Access" (CSMA) with collision detection it means that devices must wait until there is no activity on the network before they can transmit.

In the case of a collision, the arbitration scheme is designed so that messages with a higher priority are retransmitted before messages of a lower priority. On a heavily loaded network, there is a possibility that there is some delay before a device senses no activity on the network, and in the case of collision, lower priority messages may have their transmission delayed. I'm not sure if these stats are captured anywhere. Someone with more knowledge may be able to say whether bus loads of 40-50% are going to negatively impact your network. I personally don't think it will.

I was advised that latency goes up rapidly and exponentially over a bus load of 25% -- 27%.


Everything is working now but I'm afraid to reconnect the satellite compass.


I might sail for a couple of days like this before I try that.

[RickVicik]

I had intermittent NMEA2000 issues after adding an SCX-20. Bus utilization was hitting 50%. After disabling some PGNs and reducing the frequency of others, bus utilization dropped to ~30% and the NMEA2000 issues disappeared.

Controlling SCX-20 output PGNs can be done from a TZT chartplotter settings screen or by using the Furuno "SC Setting Tool" that runs on a PC. A NMEA2000-to-USB gateway such as an Actisense NGT-1-USB is needed when using the PC app alternative.

The attachment is a screenshot of the SC Setting Tool.

---rick

[tanglewood]

Originally Posted by RickVicik

I had intermittent NMEA2000 issues after adding an SCX-20. Bus utilization was hitting 50%. After disabling some PGNs and reducing the frequency of others, bus utilization dropped to ~30% and the NMEA2000 issues disappeared.

Controlling SCX-20 output PGNs can be done from a TZT chartplotter settings screen or by using the Furuno "SC Setting Tool" that runs on a PC. A NMEA2000-to-USB gateway such as an Actisense NGT-1-USB is needed when using the PC app alternative.

The attachment is a screenshot of the SC Setting Tool.

---rick

Do you know if the SC setting tool works with the SC70 as well? It being able to control PGNs is one of m few gripes with the SC70.

dockhand, you might see if any of you connected devices keep track of network errors. The Navico gear mostly does. Fast packet errors seem common, but you should have zero send or receive errors. Zero. It sounds like you just encountered a defective tee.

[RickVicik]

Tanglewood, the "SC Settings Tool" manual lists only SC-33, SCX-20 and SCX-21.
---rick

[Dockhead]

Originally Posted by RickVicik

I had intermittent NMEA2000 issues after adding an SCX-20. Bus utilization was hitting 50%. After disabling some PGNs and reducing the frequency of others, bus utilization dropped to ~30% and the NMEA2000 issues disappeared.

Controlling SCX-20 output PGNs can be done from a TZT chartplotter settings screen or by using the Furuno "SC Setting Tool" that runs on a PC. A NMEA2000-to-USB gateway such as an Actisense NGT-1-USB is needed when using the PC app alternative.

The attachment is a screenshot of the SC Setting Tool.

---rick

Thanks; hot tip. I am going to cautiously reconnect the SCX-20 and try that. Thanks!

[Dockhead]

Originally Posted by tanglewood

Do you know if the SC setting tool works with the SC70 as well? It being able to control PGNs is one of m few gripes with the SC70.

dockhand, you might see if any of you connected devices keep track of network errors. The Navico gear mostly does. Fast packet errors seem common, but you should have zero send or receive errors. Zero. It sounds like you just encountered a defective tee.

I have been checking that and not seeing any errors, just high bus load.

[stevead]

Indeed a weird one...

The B&G stuff do show errors on the diagnostics screen, but I am not sure whether these are higher layer errors more related to NMEA 2000, or whether they also include the low level CAN bus errors (bit error, start & end of frame error, ack error etc.) which are what you might see if there was a cabling or terminator fault.

I guess an interesting experiment would be to remove a terminator and see what, if any errors are displayed. Can't expect you to carry a TDR on board, but at least being able to easily diagnose the resistance across the data pairs and voltage across the power pairs never goes astray.

Good luck.

[Dockhead]

I've sailed a few hundred miles since first posting this and the network has been fine. With the SCX-20 disconnected.


I see bus load of 25%-29%, zero Rx overflows, overruns, rx errors, tx errors, but 1331 fast packet errors.


I will try reconnected the SCX-20 at some point and see what happens.

[FlyingScot]

We run 40-50% bus load on our N2K as a normal condition now.
The SCX-20 does stream a lot of data and is approximately 15-20% of overall bus load.

We get very very few errors and everything works as expected. The Fast packet errors climbs slowly over time. I mean slowly. Single frame, small fast packet and fast packet overflows all remain at zero.

I do not use any single equipment supplied T connections.

I found the best value to be Ancor brand 2 and 4 port connectors sourced from Amazon. I use some of the equipment supplied drop cables. All of the cables between groups of items are Garmin. I found these to be of high quality with good pricing off Amazon.

Spaces between groups is where the cables connect between the 2 or 4 way Tee connectors.

LEN MFG ITEM Description
1 Maretron TLM100 4 Stbd. Fuel Tank
8 Airmar UDST800 Stbd. Fwd. Transducer Depth/Speed/Temperature
1 Victron VE.Can Lnyx500
SPARE SPARE SPARE

3 B&G Triton-2 4 Generic Display above throttles
3 B&G AP Keypad AP Keypad
3 B&G Triton-2 AP Display
0 B&G Triton Edge Sailing Processor
POWER INSERTION POINT 2
2 B&G WR-310 AP Remote Key Fob
3 B&G Triton-2 4 Gerneric Display
1 B&G Zeus3 12" Helm MFD
1 B&G Zeus3 7" Helm MFD
4 Furuno SCX20 or Precision-9 Satellite Compass

1 B&G V90s 4 VHF
0 DigitalYacht AIT5000 AIS
1 B&G Zeus3 7" Nav. Stn. MFD
SPARE SPARE SPARE
0 Predict Wind Data Hub 4 Tracker / Polar
1 Victron VE.Can Victron Interface
1 Maretron TLM100 Port Aft Fuel Tank
1 B&G NAC-3 AP Computer

1 Maretron TLM100 4 Port Fwd Water Tank
1 Maretron TLM100 Stbd Fwd Water Tank
1 B&G WS310 Wind Sensor
2 Airmar 220WX Ultrasonic Wind Sensor + Baro/GPS/Etc

[FlyingScot]

Did you change the tank # on each TLM? Default is Zero with 0-15 as choice. This presumes you had more than one on the network.

[sailingharry]

I had horrible problems with my N2K that came with my boat. Turned out the Garmin power injection cord had a short between two pins (between a power pin and a data pin). Finding it was a real challenge, because none of the normal test recommendations look for this. My solution was to grab the guilty data pin with a pair of pliers and yank it out like a troublesome baby tooth. That was years ago.


Point is that it wasn't a "compatability" problem, but a "problem" problem.