NMEA N2K - What's Wrong With It ?

[DotDun]

Originally Posted by s/v Jedi

We don't need NMEA for Ethernet access to devices. Check out Furuno's AIS and WeatherFax NavNet devices. Just start your browser and access them. Especially the weatherfax is a very good webserver based product.

I don't want radar over NMEA. Same for sonar, video-cameras etc. So the only wish would be for compatibility among components of different manufacturers, like connecting a Furuno radome to a Garmin plotter. I can tell you tht this is not going to happen because of our lobby. It's gonna happen when they become the same company

ciao!
Nick.

Nick,

Let's take a look at your Furuno AIS example. The Furuno Class B AIS (FA50) with an Ethernet interface is $1800. The Em-Trak without Ethernet is $549 (street prices $1550 & $480 respectively). Em-Trak produced their unit based on standards based interfaces (0183, 0183HS, & N2k), obviously they had to as they have no market share that would support something proprietary. Is that Ethernet interface worth $1250? Is the Furuno unit worth 3+ times the price? If there was a NMEA standard for an Ethernet interface, I'll bet Em-Trak would have it, and maybe it would raise the price $50-$100 (for $5-$10 cogs).

[DotDun]

Originally Posted by twistedtree

Go for it. That's how I retired early and bought a boat.

I don't that much ambition! (and too close to retirement)

Originally Posted by twistedtree

- Power. Presumably this would be powered with POE (power over ethernet), but I'm be worried that each node would use a lot more power than the N2K equivalent. A typical hub or switch draws in the 2-5 watt range (just an estimate on my part). In contrast, a 1 LEN N2K device only draws 0.6W, and a large portion of the devices in my boat (just using it as an example where I've made a point of getting everything over to N2K) are 1 LEN devices. At least on my boat it's critical that I be able to leave the N2K network alive while at anchor for GPS, weather, and anchor alarm, so power consumption really matters. Trippling it or more would be a real problem.

Agreed. Starting with a clean slate would allow things like 802.3az/Green Ethernet. But you make a good point, a complete Ethernet solution would probably raise the power draw by 10-20 watts overall (just a wild guess on my part - more engineering required).

Originally Posted by twistedtree

- Cost. I can't argue the low cost of consumer ethernet products, but a big part of that is driven by the massive volumes involved. Some of that volume benefits a comparatively small, niche market like boat instrumentation, but it won't be the same as consumer electronics. Parts pricing, even for the same parts, varies by volume. And other product specific components like plastics, assembly, packaging, have costs highly dependent on volume. Many N2K sensors probably have nothing more than a PIC inside them, where a device that also includes a web server will need more compute power and more memory, both of which cost $$, space, power, and cooling. I'm sure it could be made to work, but if it were me I'd want to study to cost structure VERY carefully.

But that's kind of my point. If today's N2K sensors only have a PIC, why does the N2K version of the Airmar transducer (in the example above) cost $130 more than it's analog cousin? More power draw, probably. More $$, no, less - it would be off the shelf parts.

Originally Posted by twistedtree

- Topology: Maybe it's just me, but the thought of home-running all my device wiring would immediately end any further consideration of adopting an ethernet-based replacement for N2K. To over come that issue, you presumably would build mini hubs (or something along those lines) to be used in place of the N2K drop Tees. That way one could create the moral equivalent of a backbone by daisy chaining these hubs together, then connecting in devices at various convenient points around the boat. Or would to do it otherwise? Either way, I think it aggravates the power and cost issues.

These are probably all solvable, and that's the fun of a start-up.

[twistedtree]

Originally Posted by DotDun

But that's kind of my point. If today's N2K sensors only have a PIC, why does the N2K version of the Airmar transducer (in the example above) cost $130 more than it's analog cousin? More power draw, probably. More $$, no, less - it would be off the shelf parts.

I think the issue is that COGS doesn't necessarily translate into selling price. In fact, vendors actively seek products/businesses where they can command high margin on their products, and avoid areas that have been forced into low margins. COGS sets a floor for selling price, but by no means does it set a ceiling. Smart vendors will price as high as they can for as long as they can without losing market share or otherwise leave an opening for someone to come in and steel their business.

Using your example of Furuno AIS, I'd expect them to maintain their 3x price as long as enough people like me are willing to pay extra for a well integrated product. But as soon as Garmin or RM does a comparable job utilizing standard PNGs over a standard interface for 1/3 the price, and that starts to cost Furuno business, you'll see their prices come down. But in the mean time they will make a good buck on that product.

Where companies get into trouble is when they become too enamored with their proprietary system and start to think it's invincible. That blinds them to what customers see, and creates enough of an opening for someone else to get their nose in the tent.

It makes for great drama.

[hellosailor]

Tree-
"the thought of home-running all my device wiring would immediately end any further consideration of adopting an ethernet-based replacement for N2K."
You don't have to home-run anything for ethernet. There are 'star" and "ring" variations, and "token-ring" configurations were largely dropped because anything that has to do with rings (or daisychains) will have to deal with collisions, which in turn slows down data and devices.
NMEA has to deal with collisions and IIRC does so by having devices randomly shut up for a short period and then try again, randomly hoping for a clear shot. When you've got only a few "talkers" that works, but as the data gets busier, that falls apart.
So you either run a "star" configuration, with separate runs to an intelligent hub or router, or you compromise in other ways.

There's a reason for proprietary systems: Profits. No one has an incentive for open standards, and the market is too small to "command" them.

If you want "open"...you'd add bluetooth or WiFi to each device (which is feasible) and then haha, deal with all the damned boat stuff being networked and suffering all the usual network communications problems again anyway. Neither BT nor WiFi is perfect, I'd hate to see two dozen boats in the same place trying to fight over that.

[s/v Jedi]

Eh guys, token ring was 100% collision free. Ring topology is fuelling the world data networks today (SDH fiber rings) and collision free. It is Ethernet that has (can have) collisions.

[DotDun]

Originally Posted by s/v Jedi

Eh guys, token ring was 100% collision free. Ring topology is fuelling the world data networks today (SDH fiber rings) and collision free. It is Ethernet that has (can have) collisions.

Full duplex Ethernet is collision free, CSMA/CD is turned off. The switch arbitrates (stores or drops packets) if there is more traffic than a link can handle.

[hellosailor]

"Eh guys, token ring was 100% collision free. "

OK, who's got the Talking Stick? <G>

I've got a bucket of plenum-rated thinax that I really don't want to throw away, but i'd swear I haven't touched that bucket in over a decade.

[twistedtree]

I guess I'm not making my point clear about wiring. It has nothing to do with CSMA/CS, or token passing, QOS, or any of that. It's just about how you have to route the wires through the boat. When I say "home-run" I mean pulling wires from each device (or a group of devices) back to a single point where they connect to a switch.

Running one backbone cable with drops along the way to pick up various devices I think is easier on a boat than pulling cables from 8 devices back to a single location and connecting them to a switch. If you've ever been the poor schmuck pulling wires through a boat in places that only a trained monkey should be able to go, trying to fit them into small wire chases, and generally struggling to do a neat wiring job, you'll know exactly what I'm talking about. If you've just written checks so someone else who's done it, then you might not appreciate it as much.

[DotDun]

Originally Posted by twistedtree

I guess I'm not making my point clear about wiring. It has nothing to do with CSMA/CS, or token passing, QOS, or any of that. It's just about how you have to route the wires through the boat. When I say "home-run" I mean pulling wires from each device (or a group of devices) back to a single point where they connect to a switch.

Running one backbone cable with drops along the way to pick up various devices I think is easier on a boat than pulling cables from 8 devices back to a single location and connecting them to a switch. If you've ever been the poor schmuck pulling wires through a boat in places that only a trained monkey should be able to go, trying to fit them into small wire chases, and generally struggling to do a neat wiring job, you'll know exactly what I'm talking about. If you've just written checks so someone else who's done it, then you might not appreciate it as much.

I think your point may be boat specific. I've got 40+% of my wire chases still empty.

My issue with N2k is more about the lack of functionality than it is wire topology. Since I don't ever see IP on N2k, a change of layer 1/2 would be required.

[AuroraGH]

Main problem we already have many device interconnection protocols. TCP/IP and UDP are very solid protocols used to run this interbox we're yammering on. There are standards to run it over USB, Twisted Pairs, BlueTooth, and WiFi. It can be made peer to peer or support other control architectures, it runs on processors the size of a stick of gum, it is very well understood, you can buy these little hubs and routers for a few bux anywhere they sell gear and so we need a new way to hook things together why?

The sentence protocols etc are all fine - they can run very nicely over UDP on top of all those communications buses so why there is a new serial port protocol in town completely eludes me. I want to plug in all my sensors to an off the shelf processor, a build it meself computer, and I want it to work like stereo components, like X-10 switches, like USB mice and keyboards.

I'll give all my business to the first company who starts acting like an open standards component manufacturer and makes it easy to hack new systems out of parts.

As a computer guy, NMEA 2k just looks backwards and dumb.

[DotDun]

I printed this thread and handed it to an interested person at the NMEA booth at the Miami Boat Show. We had a good chat about what is wrong with NMEA 2000 and he agreed with most of the claims in this thread. In fact, he was a recently hired by NMEA for the sole purpose of advancing the standards. He promised we would see some action soon on these issues.

Time will tell.

I have his contact info, but I'll reserve posting it in public. If you are interested in contacting him directly, PM me and I'll pass the info along.

[DotDun]

Originally Posted by DotDun

I printed this thread and handed it to an interested person at the NMEA booth at the Miami Boat Show. We had a good chat about what is wrong with NMEA 2000 and he agreed with most of the claims in this thread. In fact, he was a recently hired by NMEA for the sole purpose of advancing the standards. He promised we would see some action soon on these issues.

Time will tell.

I have his contact info, but I'll reserve posting it in public. If you are interested in contacting him directly, PM me and I'll pass the info along.

No need to protect his identity, it's on the NMEA website in public. I spoke with:

Mark Reedenauer, Documentation Specialist

His email is on their website at:
NMEA

[goboatingnow]

Originally Posted by twistedtree

I guess I'm not making my point clear about wiring. It has nothing to do with CSMA/CS, or token passing, QOS, or any of that. It's just about how you have to route the wires through the boat. When I say "home-run" I mean pulling wires from each device (or a group of devices) back to a single point where they connect to a switch.

Running one backbone cable with drops along the way to pick up various devices I think is easier on a boat than pulling cables from 8 devices back to a single location and connecting them to a switch. If you've ever been the poor schmuck pulling wires through a boat in places that only a trained monkey should be able to go, trying to fit them into small wire chases, and generally struggling to do a neat wiring job, you'll know exactly what I'm talking about. If you've just written checks so someone else who's done it, then you might not appreciate it as much.

Yes busses are great until you get a short or a damaged or loose connection. Depending on where it is you could loose huge parts of the network and often the whole network ( as in the case of a short) now you have to access the whole bus cable to find the problem, selectively re terminating increasing bigger lengths until you find the broken part ( of course you might have a TIme domain reflectometer ) and you then find the short you are faced with fixing it or pulling new cable or running a parallel bypass cable. Yuk.

Contrast in a active hub, star system, firstly only that section goes down. With proper planning you have critical devices on different and possibly unique spurs. So you only loose a device or two. The hub will tell you the spur has gone off line and you live with the failure as your critical systems are working. worse case you string a wire across the boat to the critical node and plug it into the hub.

There's a reason Ethernet dropped bus configurations a very long time ago. Inflexible, costly, prone to failure, hard to expand and sensitive to node numbers.

For me I'd pull a few more cables. It's what we're all doing now anyway and it's not difficult to pull cables in a boat as the distances are very short.

Bus based NMEA2k has no future other then simple "behind the panel" instrument busses. Expand it through the boat and watch the trouble begin.

I mean people are actually running the NMEA 2k up a 60 foot mast to an unreachable terminator at the top( which NMEA wouldn't certify unless it was separate , see the airmar Furuno debacle). Can you imagine after a few years how much trouble that mast cable will generate for the network. ...madness really.

Can bus was designed for a closed controlled networks, set up with close cooperating devices, it has no place being used as a general purpose network interconnect amongst disparate devices. it has met with significant success in the car and heavy vehicle marketplace, characterised by high quality controls, consistent manufacturing standards, careful wiring loom design, one company (manufacturer) control of what is added to the bus and is backed up by industry standard diagnostics protocols( OBD) ( now mandatory under legislation) and extensive support tools both own brand and third party.

None of this exists in the marine space.



The mistake was that NMEA invented this protocol in the mid 90s ( when there was a CAN-fest about) then they butchered the only useful protocol J1936 to make a proprietary design ( why you ask?) they then used a fee structure to ensure that it effectively remained a closed proprietary network. Further to that, because they are a very weak industry body ( its not even a worldwide body, it's a US body) they allowed the major players ( who happen to fund NMEA and second key personell to develop the standard ) to build it further proprietary measures to further render the network closed. ( I mean there isn't even a standard connector amongst the major 4 players , all who sit on the committees )

The up shot of this was a network designed in the mid 90s, when Ethernet was still expensive, was not brought to market until really 10 years later and arguably its only now (15-18years later) we are getting a reasonably comprehensive product range. This has meant it has been completely surpassed by technology

It's worth noting that car companies have moved on in that time developing LIN, MOST and other lower and higher speed networks.


Dave

[DotDun]

In case you haven't seen the latest NMEA discussion on Panbo.

Panbo: The Marine Electronics Weblog: The N2K WiFi gateway issue, is NMEA stifling innovation?

[hellosailor]

On reading the discussion on Panbo, I have to ask, has anyone in NMEA ever had any experience in the software industry? In the computer industry? In any industry at all?

So NMEA wants to charge an arguably reasonable (more than it costs, but within reach, and no doubt "supporting" the NMEA operating budget itself) fee to test stuff before it is "certified" by NMEA. There's nothing new about a program like that, for years and years Microsoft has had the "Certififed Logo Complaint" program, where vendors can pay a testing fee (I've heard $50,000, similar to UL fees) and if their product, hardware or software, passes the test suite it is awarded the right to use the MS logos and trademarks as "Designed for xx OS".

If you fail the tests, that's confidential. You pass, you get awarded the right to use the marks. But in no case would anyone ever be dumb enough to say "You can call it compatible without our permission." Hello, 1st amendment, freedom of speech, simple fact, unless you are outright lying and defrauding the public when you know it is NOT compatible...

I can sell scraps of endangered-species leather and call them "Windows-compatible Mouse Pads" and no one can stop it from being called that. I can make them out of human flesh cut from child labor, and Microsoft still can't stop it being called "compatible".

NMEA needs to wake up and smell the long-standing roses. They can make everyone spend lots of money enriching the lawyers, or they can license out an "NMEA Certified" logo program and educate consumers as to the value of it. Otherwise, it is just a question of time until NMEA gets taken to task, and they WILL lose this one. They don't own the rights to the word "compatible".

This iceberg IS compatible with that steamship. Abusing the way they meet, that's another story.