N2K “Intelligent Gateway”, (updated 10/20)
NMEA 2000—able to bring tons of essential sensor data to a computer in an easy, standardized way—should be a huge opportunity for marine software developers. But there have been two major impediments. One is the lack of enough N2K networks to make its use worthwhile. That is going to change as more major manufacturers get seriously on board. For instance, tonight, when the NMEA Conference exhibition hall opens, I will get to see Furuno’s new range of N2K instruments (above) and also Airmar’s new high spec N2K/0183 GPS compass (yeehaa…and more on those soon). The other, particularly for small developers, is the substantial cost of getting a NMEA 2000 product certified. But I’m hearing about an impending NMEA initiative called “Intelligent Gateway” that sounds like it will largely take care of this problem. The concept is that a N2K gateway like Maretron´s USB100 can act as a firewall insuring that any software on the other side of it can not cause problems on the network, therefore minimizing certification cost. In fact, the plan is to offer a new NMEA 2000 PC “Approval” status for a measly $100. Any company could make and certify such a gateway and Maretron tells me that it might eventually make a second model that would be less expensive because it would not also offer NMEA 0183 translations. I’m hoping that all this means that small operations could, say, develop performance sailing software or conning screens using the same data that’s flowing to those new Furuno instruments…or who knows what!
There’d still be an issue for small developers, though, which is the high cost of the documentation that details the NMEA 2000 standard messages. I get why NMEA charges a lot for certification—it finances the certification tool itself as well as the maintenance and ongoing development of the Standard. But I know I’m not the only person who thinks the whole excellent concept would move along faster if the documentation was freely available. I’m going to try and make that case to some of the folks who make these decisions, but could use your help. Please, if you have an interest, write a comment on this subject. Thanks!
PS, 10/20: Well, I didn’t really look at the documentation prices when I linked to that NMEA page above. They’ve been debundled, and NMEA makes a pretty good case that getting into N2K is feasible for even a small developer, especially now that some testing houses are able to do the certification. NMEA officials also told me that the Intelligent Gateway is for real, and Maretron just submitted the first gateway for final certification. The “Approval” cost will be more than $100 but hasn’t been set quite yet. At any rate a small software developer wanting to explore N2K might only need a Gateway, a $325 NMEA Trade membership, and the $495 electronic Database of PGN details to get started. More on this as I learn more.
NOTE: all the early comments to this entry, mostly posted on Oct. 18 & 19, got accidentally deleted. I’ve tried to restore them below. The order and text are right, but dates and names are dubious. Sorry, all!
Yes I would support NMEA 2000 if the cost were reasonable. So far the demand for NMEA 2000 has been just nil. Hopefully that will change so I can justify doing the development.
I won’t hesitate to chime in loudly that I agree completly with you Ben.
This unique method of charging for the documentation to cover costs of the standard should be abolished. Perhaps a better method is a $5 license fee per unit of N2K product shipped, excluding the first 100 units ?
Dan
Lots of stuff today!
First, those Furuno instruments look very interesting! Hopefully Furuno will put some information on their web site so we can get up to speed before FLIBS.
Second, I don’t know what the cost of certification actually is so I can’t comment on whether it’s prohibitive or not. I do know that it’s critical that any equipment connected to the bus not crash the bus and that anyone
in the hardware business is going to have to be decently funded to develop, manufacture and support a hardware product. Is the cost of certification really holding that back? We need more facts.
That said, NMEA is an industry organization that if it’s like most industry organizations, is essentially controlled by the biggest players. As we know, they have been giving us a mixed message about N2K because they’re afraid it will cut into their proprietary revenues. There is no big secret on how to promote a standard, if they wanted to lower the barriers, they know how to do it. I think the issue is more one of will than way.
Maybe the Furuno announcement will break the dam of market demand open because if they fully support N2K, then the other’s will need to explain why you should invest in their proprietary network.
Just to ensure there is no ambiguity, I think the sooner there are industry standards for all data types the sooner we’ll all be willing to upgrade our systems. It will also make it much easier, and thus more likely, for us to
upgrade in the future because we don’t have to rewire the boat!
How hard can this be? There is an abundance of inexpensive networking technology available off the shelf. If the manufacturers want to grow their businesses, they should really get behind some standards, not give it
the wink and nod approach like they’ve given N2K up to now.
NMEA 2000 is based to a very large extent on other existing standards – DeviceNet (connectors), ISOBUS (a communication protocol based on the SAE J1939 protocol), and CAN (Controller Area Network) Unfortunately, these standards are controlled by SDOs (Standards Development Organizations) who choose to charge for their standards. Nothing NMEA can do about it.
NMEA does publish certain information such as PGNs (Parameter Group Numbers) with descriptions (Latest list is V1.21 August 2007)
One thing NMEA could do is encourage the development a set of high level software APIs (application layer level?) to enable third party implementors more easily access the data on the bus. One possibility would be an abstraction layer similar to the SNIA XAM initiative for storage area
networks.
I’ve just retired after 40 years in the software development business. I cannot recall any successful and/or lasting technology that resulted from this kind of approach. Software developers need to tinker and play to come up with great ideas. NMEA has just shut the door to many great ideas that
will now never see the light of day.
SurlyJoe, how is that?
With the Maretron bridge you have access to everything on the bus from a PC/Mac. You can’t write data back to the bus, but you can certainly do some nice display and computational work.
What is is that the s/w developers need that is not available?
NMEA 2000? Perhaps they should have called it NMEA2015… at the rate they are going it will take 20 years before (if) it get decent use.
API’s ?? Firewalls ??
This is insane. There are much better ways to make it easy for a component developer to reliably incorporate a communication protocol into their device reliably and cheap in most every other industry! How about a cheap marine equivalent of an ethernet NIC card on a chip with a software stack that, in providing an API to accept data for transmission on the bus, becomes “the API”.
Maybe this even exists, I don’t know much about N2K (can’t get the documentation !), but suspect once the will power is there, direct proven solutions to supporting inovative multi vendor solutions reliably should be
the path.
By the way NMEA, how is the API or other solution to meet the desire/need for one vendors N2K display to change configuration parameters on an another vendors sensor ?
I don’t know the spec (to much $$$ for me to purchase) but I suspect that when a standard for this comes out replacing proprietary messages used today, some N2K equipment we purchase now will be left behind. Five years is way to long to wait for a solution to this generic requirement on a protocol intended for this industry.
If there was a wider community in N2K (access to documentation and collaboration initiatives at least similar to what happens in open source projects today in other industries), I bet this problem would be licked already !!
I am also dismayed that ideas I sent to various vendors over the last 2 months (on how to use data from each other to provide a wind speed and angle compensated for sailboat mast movement at low wind speeds or rolling seas) seems so ridiculously futuristic given the issues they are dealing with now.
Rant! rant rant
Ben, take this to NMEA Member’s, (Updated)
In general (even going beyond N2K) I am a perfect example of lost sales to the marine electronics industry in 2007 due to a failure of standards that could have made it easier for me to upgrade my boat this year.
I started the season with a shopping list including an ultrasonic speed sensor, solid state compass, ultrasonic mast weather station, and upgrade from 2kw to 4kw Raymarine Radar, but in the end only purchased the
ultrasonic speed sensor, as I couldn’t get satisfactory answers to integration questions or assurances of compatibility (compass or weather on N2K to Raymarine), and in the case of the Radar and weather sensor,
required a whole new wire run up my mast that could have been avoided if the industry proceeded faster between 2000-2006 around signaling and cable standards that were more universal.
The result is that the current standards and products in the marine electronics industry today (my boat is only 1 year old) caused me to spend only 15% of my original 2007 budget for upgrades to my boat.
… of the four products the only product to make it on my sailboat this year was the Airmar ultrasonic speed sensor, which could plug right into the wiring of the Raymarine component it replaced with the addition of a
12vdc power run.
NMEA 2000 is the light. Lead us from darkness. We need more products, more compatibility, more innovation. That means the barrier to entry should be carefully calibrated for maximum inclusion, while still maintaining reasonably high standards.
I have sent few requests in this regard to NMEA. I can understand how costly is to develop and maintain such standards, but having the proper pricing (maybe tiered pricing depending on sales volume) will let small
companies to enter the VIP room of NMEA 2k.
NMEA 2k is based on CAN bus, same as the one used in automotive industry and manufacturing processes, it is amazing that buying a chip (ELM chip) to talk CAN with multiple ISO protocols (automotive) for $27. Now imagine how many “gateways” exists for these protocols?
Can someone please leak (PDF, scan or photocopy) the protocol documentation and physical interface specs to the hacking community? I’m amazed that nobody has by now. I want to see these materials on bittorrent and for innovative open source projects to start appearing on random eastern European websites. Engineers at Furuno and Garmin are also encouraged to leak details regarding their IP protocols so that we can integrate radar overlays into our projects. Thanks!
This last post was a surprise to read.
Does beg the question, what happens to the NMEA revenue model when their documentation appears on the internet someday !
You are correct, the PGNs are the key. Perhaps users of NMEA 2000 gear would be interested in modifying the NMEA 2000 page on Wikipedia to detail known PGNs.
As noted at the top, all comments up to here were restored from notification emails and are at least misdated, and may contain other mistakes. Sorry! But any comments that follow should be normal.
For a small commercial software development company wanting to explore N2K a startup cost of a gateway, a $325 NMEA Trade membership, and the $495 electronic database of PGN details might be reasonable. But for people who would like to explore N2K in their spare time, e.g. as a hobby, this cost is still likely to be prohibitive. And if you wanted to hook up a CANBus interface from a PC to your N2K network, for example, you’d presumably have to factor in the cost of e.g. the N2K primer, which is yet another $365. Although I suppose such users would be able to apply for associate NMEA membership, which knocks $180 off the price. But it’s still a substantial investment in a hobby where your experiments may lead nowhere.
Clearly NMEA have to generate some revenue to cover their costs. They had a choice of charging a small amount on each N2K-compatible device sold, or an upfront charge on the documentation. The first approach would have allowed small developers and hobbyists to get into N2K development at low cost, and may well have led to lots of interesting, and perhaps unexpected, applications of this technology. I presume such developments would be seen as a threat by the major players in this area, and so, as a trade association representing these interests, NMEA has decided to adopt the second approach. Given the extremely slow uptake of this technology you have to wonder whether they made the right choice.
I can appreciate that for some “safety-critical” applications a CAN/N2K-based solution may be required. But for many situations I can’t help feeling that an NMEA 0183 -> Ethernet gateway, with the gateway converting the closed 0183 sentences into open-source protocols at the IP layer, would be a much more open and attractive solution for many users. And details of 0183 are less of a closed shop to most people. I presume someone will eventually do that for N2K as well. But it’s clearly undesirable to have to develop a parallel set of interfaces and data structures to 0183’s sentences and N2K’s parameter groups, simply to get around a punitive pricing policy. It’s hard to see how it’s in NMEA’s interests for this to happen, but they seem to be forcing hobbyists to go down this route. At one point someone was talking about getting a group of interested developers together to set up a company with each developer contributing a small amount towards the cost of purchasing a set of N2K docs. Does anyone know what the readership restrictions are on docs purchased through the NMEA? Presumably they aren’t limited to a single named individual in a company, are they?
NMEA could also define a “hobby bus” of set aside PCG’s. The other alternative is the vote price (each member that joins NMEA gets a vote) and could lower the entry cost of support.
One could go the (e.g. STANAG NATO Mil spec 4586)) ethernet bust route to define a hobby message bus (run of over IEEE 1588 real time) ethernet (and I mean “hard realtime”). Luminary sells chips and development systems that are far cheaper and do both CAN and 1588 ($17). Good IP68 cable and IP67 routers supporting PoE are available for ship systems. BTW: Can over RT ethernet is common in manufacturing.
Right now I am paying 10K for a 2 helm (6.4 inch) navigation system with 2 engines and sync- Way too much! when the base system electronics are far cheaper. Custom cabling and rewiring made this a pain. Unfortunately I did not do enough homework as a competing system – a better non-NMEA 2000 was cheaper by 1/2.
NMEA 2000 is like early AM radio, it’s all over the map, and low cost FM (IEEE 1588) is around the corner.
I may hack up a hobby Marine message set on 1588 timing and NMEA 183 messages over ethernet. For anonymous devices, there should be support for embedded portlets (as webservice so the display does not have to know the PCG to display an sensor interface).
Also see http://www.lxiconnexion.com/articles/0407/LXI0714Smart.pdf
Ben,
Did this thread get you what you needed ?
Has there been discussion or design of an HTTP over NMEA 2000 protocol? Any engineering reason it cannot be done?
That would allow a component designer to embed a web-server in their device? Is that potentially part of an N2K intelligent gateway?
Support for HTTP/N2K would allow customers & installers to use a web-browser to configure N2K devices, and if the browser was enabled on a chartplotters locked-down O/S, would provide a generic means with which to configure any vendors device on the N2K backbone. For example, you could configure a solid state compass, weather station, etc without needing to by the vendor specific N2K display or use a PC with vendor specific software loaded.
Re: Support for HTTP/N2K would allow customers & installers to use a web-browser to configure N2K device.
It could indeed be done or like some devices an dhtml interface could be sent on initial recognition. That way older gauges could work with newer sensors, by using the sent interface. DVD’s use dhtml that is loaded from the disk. This would support a composable interface in things. As I stated in the IEEE 1588, message I believe that NMEA should set aside a hobby bus.
BTW: the luminary stellaris kit is available for $89 from digikey. I bought the codesourcery kit 726-1086-ND. – Really nice – A 1588 and Can controller with OLED buttons, Discs, cables, USB power cord, and a can sensor board. Haven’t made a stack decision yet.
KIT EVAL ETH/CAN W/TOOL LM3S8000
http://www.luminarymicro.com/products/ekc-lm3s8962_ethernetcan_evaluation_kit.html
At first glance the luminary kit sounds like a good deal at $89. But by the time you add in the $200 to upgrade the 30-day demo of codesourcery to the full version, and then factor in the cost of the N2K docs if you want to use this board for N2K hobby experiments, the cost of the board is swamped by the cost of everything else. Some of the boards made by Compulab (http://www.compulab.co.il) seem to have a similar cost, and run linux which saves having to buy a development kit. But until NMEA reduce the price of the N2K docs to something sensible, this cost is going to be the limiting factor in encouraging hobbyists to experiment with N2K. The costs of boards to base these experiments on is tiny in comparison, no matter which board you choose.