Furuno NavNet 3D Crashes?

[DotDun]

Originally Posted by CarinaPDX

I think that may be damning with faint praise... Certainly the slow startup is a function of Windows. I don't know why the initialization is so modal, and requires rebooting, but I really dislike that aspect. Why can't I make a few small changes without spending so much time? Bad design - I just don't know whether that was necessitated by Windows. At least with Linux they would have had all of the code available to make patches, so could fix any software problem (and wouldn't have to pay royalties); OTOH porting would have been an immense PITA.

IIRC, WinCE is in ROM and is copied to memory on boot up, hence the name 'embedded'. The hard drive contains just the charts and config. But, you are right, the architecture causes the slowness. With my comment about 'not bad', my intent was that compared to a PC where there is multiple fingers in pie making Windows unstable, embedded in a single use machine makes it pretty stable.

Originally Posted by CarinaPDX

Inevitably a custom machine like this will start with technology a bit behind the computer market, and then stay there until a major change., thus falling even further behind. It seems to be powerful enough to do the job; its zooming is impressively fast. I suppose it has an early Intel Atom chip inside, or the equivalent from AMD. Given the additional power required for a more potent CPU I think they probably made the right decision, at least for the MFD8. It might have made sense to put faster CPUs into the MFD12 and certainly the MFDBB as it is safe to assume that vessels that carry them can afford a bit more electrical draw. Still, I don't see the hardware as a major issue. However, we are getting to the point that Furuno should be considering replacing the HDDs with SSDs.

I have all the 0183 & N2k connections to a MFD12 inside and it's struggling to keep up. Example: My MFD8 at the helm is connected via Ethernet only, so all sensor data is repeated from the MFD12. I witness regularly AIS targets that appear on the MFD8 but not the MFD12. The only explanation is that the MFD12 has cpu to repeat the AIS data onto the Ethernet, but doesn't have enough power to render to it's own screen. Also, those adhoc 'heading missing' errors - I've monitored the N2k bus when those happen, the heading data is there but the MFD12 simply misses them. It's running very much on the edge of it's capability to handle all the interrupts. Yeah, I've been thru the tech support trip of reload combo, reset, yada, yada.

Originally Posted by CarinaPDX

Until the TZ came along Furuno was making frequent updates to the software, adding a lot of nice features as well as bug fixes. It is understandable (if frustrating) that resources were diverted to get the TZ products out. The real test is what happens next: how long will Furuno continue to update the NN3D products? If we have seen the last update, then I will not be a happy camper.

Most companies won't completely abandon the existing product line when building the new one. They invest $$ to run dual engineering teams.

Originally Posted by CarinaPDX

I don't see the TZ as replacing the NN3D, at least in small boats. The idea of putting my salt-soaked fingers on an LCD is a complete non-starter. I operate the MFD8 in my very wet cockpit, so salt crystals and smearing are already a fact of life. As long as Furuno keeps selling the NN3D products, it should be updating them, and I should be satisfied; otherwise...

The TZTouch isn't going to work in 40kts winds and 10' seas where you have to walk your fingers up the bezel to find a button/knob.

Originally Posted by CarinaPDX

BTW, other than the slow startup and modal initialization, I don't really have any software complaints. At least not in the sense of unresolved bugs (perhaps ignorance is bliss here). There is always room for improvement. The GUI could have been better, but the capability is crazy good, so I'm willing to give it a pass. YMMV

Greg

Agree, functionality trumps UI!

Back to the OP - turn off tide/current overlays, it'll work!

[Paul L]

You might think a loaded Navnet3D is slow, but you have to compare it to whats out there. When I scan and zoom on my MFD8 it seems to me as adequate - not screaming, but not slow. But when I scan and zoom on the Raymarine setup on the boat I just delivered the Ray seems glacial. To the point that you have to have work arounds to avoid scanning. Made my Furuno feel way fast. (Granted the Ray setup was older)

[colemj]

Originally Posted by CarinaPDX

BTW Furuno came out with a firmware update to the radars in June.

Greg

Where can I find this? On the website, I find a DRS 1.16 update for the ztouch only. They don't mention the NN3D.

Mark

[Paul L]

Originally Posted by colemj

Where can I find this? On the website, I find a DRS 1.16 update for the ztouch only. They don't mention the NN3D.

Mark

My understanding is that all the update does is make the DRS compatible with the ztouch.

[CarinaPDX]

DotDun - Good stuff.

Yes, one would expect that an embedded system would be a lot more stable than a desktop with who-knows-what software running on it. And CE does seem to deliver that. Plus compatibility at the API level which was the attraction here. But Windows is still bloated and slow, at least compared to Linux. I suppose at some point a decent hibernate plus SSD will make starting up less painful.

Interesting, and disappointing, about your MFD12 not keeping up. I suspect that it wouldn't require a lot more CPU power to have done the job. I have a little sympathy for Furuno here, as allowing consumers to create complex plug-and-play networks is like writing blank checks - at some point there isn't going to be enough left to do the job.

A year or two ago on another thread someone was trashing N2K, which I strongly disagreed with. Their "evidence" was from posters on Panbo with 50+ node, multiple subnet systems. Again, push a p'n'p design far enough and problems will arise. More basic systems are actually pretty robust.

What I think is missing is the stress testing to determine the limitations, with those limitations published for all to see. As it is we are going on faith, and some of us are going to hits some limits at some point.

My experience working in new product development in the electronics industry is that current product support often suffers as the need to get a new product to market intensifies. It doesn't always start out that way, but it often ends up that way. In this case there were a number of competitors bringing touch screens to market, and the market is seasonal. Miss the boat show circuit, miss the year, miss the early adopters. Of course they cannibalized the current product support to make that happen - I would have too. The real test is whether they turn resources back to the older product; we'll find out soon enough.

As for the radar update, I have no idea what changed. AFAIK Furuno does not publish changelogs for public perusal. And they are not publishing the update for installation by NN3D MFDs - you only get it by first putting an SD card into the MFD and request an upload, which puts a file on the SD which must be sent to a Furuno web page in MyNavnet, after which you can download an updater folder that contains any new firmware. Not the easiest approach...

Greg

[DotDun]

Originally Posted by CarinaPDX

DotDun - Good stuff.

Yes, one would expect that an embedded system would be a lot more stable than a desktop with who-knows-what software running on it. And CE does seem to deliver that. Plus compatibility at the API level which was the attraction here. But Windows is still bloated and slow, at least compared to Linux. I suppose at some point a decent hibernate plus SSD will make starting up less painful.

Not sure a SSD would help with startup. WinCE is stored in ROM/flash and is copied to RAM for execution. A SSD would be faster loading charts, but that doesn't seem to be an issue.

Originally Posted by CarinaPDX

Interesting, and disappointing, about your MFD12 not keeping up. I suspect that it wouldn't require a lot more CPU power to have done the job. I have a little sympathy for Furuno here, as allowing consumers to create complex plug-and-play networks is like writing blank checks - at some point there isn't going to be enough left to do the job.

I've pointed out a problem that I think is wasting cpu. I finally convinced Furuno about an issue with their N2k stack, but they tell me it isn't causing a problem. If you monitor the N2k network, the MFD will get stuck in a mode of sending spurious pgn 59904 & 126208 to adhoc addresses on the network, it'll pick one and just keep hammering it. It sends a poll every 10-20ms. If the sensor answers, that's an additional 50-100 interrupts/second that are unnecessary. It took them a year to admit to the issue, then they told me it isn't causing any problem. They also claim the extra 20kbps on the N2k bus isn't a problem either. I'll bet a heavily loaded bus would care.

Originally Posted by CarinaPDX

A year or two ago on another thread someone was trashing N2K, which I strongly disagreed with. Their "evidence" was from posters on Panbo with 50+ node, multiple subnet systems. Again, push a p'n'p design far enough and problems will arise. More basic systems are actually pretty robust.

I'm not a huge fan of N2k, I think they missed an opportunity to do something really good. I think the NMEA members choose a path to protect their proprietary products vs. true innovation.

Originally Posted by CarinaPDX

What I think is missing is the stress testing to determine the limitations, with those limitations published for all to see. As it is we are going on faith, and some of us are going to hits some limits at some point.

My experience working in new product development in the electronics industry is that current product support often suffers as the need to get a new product to market intensifies. It doesn't always start out that way, but it often ends up that way. In this case there were a number of competitors bringing touch screens to market, and the market is seasonal. Miss the boat show circuit, miss the year, miss the early adopters. Of course they cannibalized the current product support to make that happen - I would have too. The real test is whether they turn resources back to the older product; we'll find out soon enough.

Yep, 4 things I'm tracking -
Fix tide overlay issue
Fix current overlay issue
Enable AIS on N2k
Fix the N2k stack problem

Anything less is falling short after waiting over 18 months for a fix.

Originally Posted by CarinaPDX

As for the radar update, I have no idea what changed. AFAIK Furuno does not publish changelogs for public perusal. And they are not publishing the update for installation by NN3D MFDs - you only get it by first putting an SD card into the MFD and request an upload, which puts a file on the SD which must be sent to a Furuno web page in MyNavnet, after which you can download an updater folder that contains any new firmware. Not the easiest approach...

Greg

[CarinaPDX]

Good point on the SSD - no easy fix that way.

The N2K stack issue is a bit odd. There is no way Furuno can know all of the impact of flooding N2K with spurious PGNs. Agreed that it will probably not cause problems, but there is no guarantee that some device from another vendor won't get tripped up. There is no excuse for not fixing it. And even if it proves harmless it is wasting resources in the MFD. Fixing it may not be enough to cure your speed problem, but it would be a start and a lot easier way to improve performance than to deal with hardware.

We have beaten the N2K quality issue to death on other threads so no point rehashing here. It is a good, workable standard that is a big improvement over 0183. It could so easily have been far, far better - even great. And every major marine vendor has had a hand at one time or another in keeping it from being great. Yep, I want to scream too.

Furuno's stand on the AIS issue is very strange. My Simrad AIS is connected to N2K to receive heading and ROT info (the only way it can, as Simrad didn't implement receiving the sentences on the 0183 bus) so my N2K bus is flooded with AIS PGNs, like it or not. N2K seems to have survived the experience. And I still am required to connect using 0183. Very lame.

I am unfamiliar with the tide and current overlay problems you refer to. I wasn't impressed by their implementation last time I tried to use them however.

I have made a few suggestions for improvement on the Furuno forum - it will be interesting to see if anything comes of it. Getting input from customers is a smart move - as long as it is followed up. We'll see what happens...

Greg

[DotDun]

Originally Posted by CarinaPDX

I am unfamiliar with the tide and current overlay problems you refer to. I wasn't impressed by their implementation last time I tried to use them however.

If you leave the tide and current overlays on, the MFD will reboot sporadically.

Furuno USA Community • View topic - Navpilot drifting and no nav data

[CarinaPDX]

Now that is amusing, dumb, or disastrous depending on your point of view. Why, oh why, would they have the system calculating tides/currents for stations that were not being displayed? Sure, calculate nearby stations to be ready if the user pans the display, but ALL of them? Sheesh.

It seems there is a common thread here: the programmers do things without concern for limited resources. It is a bit of a change to go from over-powered desktop systems to limited-power embedded systems, and it seems some of them didn't get the message. To be fair, programmers don't do that much optimization anymore - they rely on the compilers to do that. Still, some things are pretty obvious...

Greg

[DotDun]

Originally Posted by CarinaPDX

Now that is amusing, dumb, or disastrous depending on your point of view. Why, oh why, would they have the system calculating tides/currents for stations that were not being displayed? Sure, calculate nearby stations to be ready if the user pans the display, but ALL of them? Sheesh.

It seems there is a common thread here: the programmers do things without concern for limited resources. It is a bit of a change to go from over-powered desktop systems to limited-power embedded systems, and it seems some of them didn't get the message. To be fair, programmers don't do that much optimization anymore - they rely on the compilers to do that. Still, some things are pretty obvious...

Greg

I would guess it's not a cpu issue, but a memory leak.

[CarinaPDX]

Yes, the crashes would be something like a memory leak. But tying the cpu up in knots doing unnecessary calculations is just going to hasten the point where the MFD runs out of gas. Your problem with the MFD12 not keeping up may be a limitation of i/o handling, not cpu bound, but still the waste of resources offends. I have a long memory - I still remember delaying a project while the code was compressed/swapped to fit in a 64k address space...

[Paul L]

What I don't understand is how long they have left this on the bug list. It is easy to repro the problem, so it would be relatively easy to fix also.

[CarinaPDX]

I wonder if it still exists in the TZ software. I would guess that the base software is the same, with *just* <grin> the GUI changed.

A memory leak can be tricky to track down - after all, this crashes the system only after a long time of a lot of repetitions, so the leak could be quite small. Still, agreed this could have and should have been fixed a long time ago. And the N2K stack should have been cleaned up. And ...

[CarinaPDX]

When I bought my NN3D system 3 years ago it was streets ahead of Raymarine's offerings at the time, at least in terms of features. I just downloaded the manual for the new Raymarine MFDs and boy has that changed. They do everything Furuno does, and with a slick GUI. And the 9" c-series is HALF the MSRP of my MFD8 (and the 9" e-series is half the price of the TZ9). I do like that Furuno stores a lot of charts internally - including free US charts, but for that kind of saving bring on the charts on SD cards. Time for Furuno to step it up a notch.

[DotDun]

Originally Posted by CarinaPDX

When I bought my NN3D system 3 years ago it was streets ahead of Raymarine's offerings at the time, at least in terms of features. I just downloaded the manual for the new Raymarine MFDs and boy has that changed. They do everything Furuno does, and with a slick GUI. And the 9" c-series is HALF the MSRP of my MFD8 (and the 9" e-series is half the price of the TZ9). I do like that Furuno stores a lot of charts internally - including free US charts, but for that kind of saving bring on the charts on SD cards. Time for Furuno to step it up a notch.

Yep, a friend buying a new boat ask for my opinion on the e97, basically I'm jealous of all the features for the price. With the e97 at the helm and an iPad, he is going to forego a chartplotter at the nav station. I was really surprised when Furuno announced the pricing for TZTouch, competitors offering more functionality for the same or less $$, Furuno raises prices 40-60%.