I am very interested in the N2K network for my boat but it seems it was designed for powerboat usage where everything is always on and thus consuming amp-hrs. I have seen people suggesting using 2 networks; one for normal use and one for power saving mode as when sailing. I have looked at the wiring diagrams of the split powertap. It seems that all the network data goes straight through while the power is split into a left and right side with positive and negative wires to each. If this was to be placed between a Tee in the backbone and a device (or a multioport box with several devices such as tank senders) and join the negative wires together and the positive to a switch then that limb of the network could be switched on or off as desired. Has anyone tried this? Ben?
Thanks, Ross
Ross, you can definitely power and use sections of an N2K backbone independently. I've done it a bit without problems. I have heard that sometimes when you power up a section, already running devices may hiccup a bit while the new addressing is sorted out, but worse case is that you have to power down running sections before turning on the whole backbone.
Just wanted to make a clarification for anyone skim reading this.
Powering the N2K backbone with more than a single supply can lead to ground loops that can potentially blow all the transceiver chips of connected devices. This would have a very expensive refit bill to replace everything, still you might be lucky.
The point is to have power, data and shield common throughout the network.
If you just want to turn the power off to a section then this is acceptable.
The Actisense QNB-1 can be used to disconnect power to a section of the network. In this case removing one of the protection fuses would do this quickly and easily.
http://www.actisense.com/HTML/Products/NMEA%20Interfaces/Quick_Network_Block_1/index.php
Thanks, Actisense; I should have mentioned that. To further clarify, is it correct to say the data and shield wires should be continuous in an N2K network, but the power +/- wires can be discontinuous and switched, AS LONG AS they originate from the same battery or other power source?
Discussion of split power taps here:
https://www.panbo.com/archives/2009/03/nmea_2000_power_problem_part_2.html
Data, shield and ground lines should be continuous throughout the N2K network. The power line may be split (at a single point) to provide 4 amps onto each of the two sections running from the split.
The power and ground lines should have a single entry point onto the N2K backbone.
The point of the bus is to have the data lines referencing from the same ground line. This is only achieved from a continuous ground line with a single entry point from the supply. You also need a continuous shield to ensure the noise on the power and data lines is the same throughout.
Regarding the link you posted: The left side of the first image is not correct. For a split power tap: THE GROUND LINE MUST NOT BE SPLIT.
Dosn't sound right to me. If that is the case, then you can never have a three or more segment NMEA-2000 backbone for purposes of handling additional power.
I agree, Dan. I think Actisense is talking about basic networks. I know for certain that the NMEA 2000 Installation Standard discusses multi-leg power setups in depth. For instance:
"One advantage of multiple power legs is that each leg can be constructed of the type of cable appropriate for that leg; for example the center power leg in Figure 24 may be constructed of heavy cable and use an 8 amp fuse, while the legs on either end could be constructed of light cable and use a 3 amp fuse for each."
Isolated power supplies are allowed and so is breaking both Net-S and Net-C lines (power +/-), but a common ground reference is really important apparently. I don't think I'd mess with multi-legging without the Installation Standard, or guidance from someone who understood it thoroughly.
Yes, I wasn't getting into the realm of floating supplies for this thread. You can only have two segments from a split using a single power supply.
Bringing floating supplies into the equation you can have more than two segments. At all times the ground, shield and both data lines must be continuous. The floating supplies still need common ground references and the data lines also need a common ground. The CAN transceiver chips that are connected to the data lines use the NMEA 2000 ground as a reference to the chip, if they don't have a common ground you will have problems, as not all chips will be using the same ground and so the bus may start showing more bus errors than is acceptable. Most CAN transceiver chips also have a maximum safe difference between ground and the data lines, so if one part of the bus has a different ground potential to another part, there could be some risk of damage to the transceiver chips.
The subject is surely quite complicated, you need to avoid ground loops, you need a common potential, etc. I am not sure I follow every nuiance carefully enough to generalize it (but I can sure draw a picture !), and very strongly disagree with what's written above.
A single supply can absolutly supply multiple segments, and really careful thought should be given before not electing the option to seperatly lead each taps ground lead the whole distance back to a common point for all the nmea-2000 segments in a single power supply. Making the ground continious between segments invites the possibility of exceeding the 4 amp (or 8amp) maximum of the segment.
I'm not sure I understand the difference, from an electrical perspective. If the grounds from separate segments are connected at a common ground point, doesn't that make all grounds operate at the same potential? And isn't the effect the same as a continuous ground?
I will draw some scenarios and explain what works and what doesn't and post the link here when ready.
In the mean time, you are right Dan, using a single supply and having a continuous ground can lead to more than 4 Amps in a section of the ground wire. This is why you cannot use a single supply for more than 2 segments on a bus and why a single entry point is required for the 2 segments.
It is a requirement of the NMEA 2000 specification that a continuous ground be used, quoting from section 2.0 of the specification:
"Differential signalling indicates powered interface circuits and a signal reference common to all nodes on the network. A single-point common reference is specified in order to avoid radio interference caused by ground-loops and to maintain control of ground-voltage levels between nodes such that they remain within common-mode range (approximately +/-2.5 Volts) of the network transceiver circuits.".
If that isn't enough to convince you then maybe this will:
Using some simple values for demonstration.
You have a single power supply at 16 Volts powering 2 segments of a bus from the far end of each bus with a break in the ground line where the two segments meet.
Section 1 has devices drawing 4 Amp of current on a power cable run that has 1 Ohm of resistance down each wire. This then drops 4 Volts on each wire making the ground at 4 volts and the power at 12 Volts.
Section 2 has devices drawing 1 Amp on a power cable run that also has 1 Ohm of resistance down each wire. This then drops 1 Volt on each wire making ground at 1 Volt and the power at 15 Volts.
At the break between the grounds where the data lines join the two segments together you now have 3 volts of difference, breaking the spec and causing data corruption.
Of course this doesn't factor in that, typically, power cables are unshielded leading to the taps. Now you have some VHF interference on one segment of the bus but not the other causing even bigger random voltage swings.
That was quite an extreme situation but even with smaller voltage differences you are reducing the ability to filter noise from the system. The +/-2.5Volts is the tolerance from interference. The nearer the limits of this tolerance the greater the chance that noise will cause bit errors. Then you get increased bandwidth usage as error messages and re-sent sentences start appearing on the bus.