NMEA 2000 Spur Cable Length

[lpdsn]

The general rule of thumb is that an NMEA2000 spur should not be longer than 6m. I'm trying to find out what the principles are behind this. Obviously it's a rule that comes from DeviceNet originally.

The obvious things I can think of are that it is based upon are voltage drop and timing of messages, but neither seem to be key.

For example, the 6m is measured from the bus not from the power supply. OK, bus cables are supposed to be thicker so will have a lower voltage drop, but if this were the main driver I'd expect a rule that combines spur length and bus length from power supply to the junction. (I have found formulae that calculates expected voltage drop but my point these aren't reflected in the 6m spur limit.)

Similarly, NMEA2000 devices take the timing from the last transmitter to synchronise their own transmissions. But a quick back of the envelope calculation shows that shouldn't be a problem (as you might expect). And anyway, if this were the reason I'd expect there to be a rule of thumb about how far apart the two most distant nodes in the network are not a rule about spur length that doesn't account for the actual backbone length.

I'm looking at putting a 6.5m spur in at the end of the backbone where that end of the backbone is 5m from the power and total backbone length is 14m. My hunch is to just go for it, but I'd prefer to understand the principles behind the limits first. Any info?

 

[PaulRainbow]

I'm not an expert, but... if you are using light cable the backbone length can be 100m, nothing to say where the power has to be connected, can be at one end. You could remain technically correct if you remove the terminator, extend the backbone by 1m in the direction of the device, fit a tee and refit the terminator, then connect the device with a 6m drop cable. In practice, i very much doubt that a 6.5m drop cable will give you any problems.

 

[David2452]

In practise it is not the principle that is important but the actuality, one of the best tools I have used (and I found it very useful during my NMEA 0400 exam too) is the Maretron planner which can be download for free if you enter your design into the planner it will give you all the voltage drops and other information you need for each node before you start. I have to say that on size of network usually installed on boats of the size discussed on this forum almost any configuration of backbone and drop cables seems to work, though for masthead nodes I generally err on the side of caution and design that as one end of the backbone using an in line terminator.

 

[Norman_E]

Download Maretron N2KBuilder software here http://www.maretron.com/products/N2KBuilder.php
Draw your system design using it and add the lengths of all cables. Once you have completed your design, with all devices added the program will calculate the voltage at each device, based on receiving 12 volts at the input point.

EDIT: David beat me to it, but I have added the link.

 

[lpdsn]

Thanks David & Norman,

Sounds like you both think the reason behind it is voltage drop. I've already got a formula for calculating voltage drop in the network based on LEN and the length of cable. Thanks for the link though as that makes it much easier.

The rest of my network, backbone length, spur length, total LEN values is way below the limits I've seen expressed, so if it is voltage drop I'm happy an extra 0.5m in a spur isn't going to be an issue.

 

[Boo2]

The general rule of thumb is that an NMEA2000 spur should not be longer than 6m. I'm trying to find out what the principles are behind this. Obviously it's a rule that comes from DeviceNet originally.

The obvious things I can think of are that it is based upon are voltage drop and timing of messages, but neither seem to be key.

...snip...

I'm looking at putting a 6.5m spur in at the end of the backbone where that end of the backbone is 5m from the power and total backbone length is 14m. My hunch is to just go for it, but I'd prefer to understand the principles behind the limits first. Any info?

My understanding is that DeviceNet (a variant of CAN bus) is indeed limited by timing. The issue is that if two bus masters try to issue commands at the same time the conflict is resolved by one of them noticing that the other has overridden its signal on thew bus. In order for this to happen the signal has to propogate from the overriding controller to the overridden controller in the space of one bit time on the bus. The signal propogation speed for any particular cable type is constant and so the bit time determines the maximum cable length between any two transmitters on the bus.

A 6.5m spur will not give any problems on its own, timing wise, but if you have a total length of 14m then that might well do.

The power issue will depend on the cable diameter and the current draw, without knowing what you intend to connect at the far end of the spur it's impossible to say whether it will be OK or not.

Hth,

Boo2

 

[lpdsn]

Just an update. I played with the Maretron tool, but in the end wrote a spreadsheet to calculate voltage drop in about an hour. Just a matter of using a formula I got from the NMEA website and the downstream LEN to calculate the drop one cable at a time. Even with a spur length of 6.5m I have in mind the voltage drop at that transducer is less than that at some of the cockpit displays where none of the rules of thumb are broken.

Curious that the Raymarine i70 has a LEN of 3 and the Garmin GMI20 has a LEN of 6.

Thanks Boo2. I was aware of the potential for timing issues, but as you can have a backbone of 200m in DeviceNet (if using mini cables rather than micro) I'm hopeful I'll be fine with a 14m bus.

I also calculated the time for the signal to transit the network using the speed of light as an approximation, just to be sure.