Tactics Plugin

[FrouFrou]

it works tnx Canne.

I asked NKE on the NMEA output sentence, curious what their respons will be.

Seems like they have 3 different xdr strings for heel
($XDR with 'ROLL' $XDR with 'Heel Angle' and $PNKEP,06).

I'll continue testing

[Canne]

Originally Posted by FrouFrou

...it works...

Thank you for finding the NKE string and for the test!
Thomas, I created a pull request https://github.com/tom-r/tactics_pi/pull/15 for this NKE "feature", quite simple.
Best regards,
Petri

[FrouFrou]

Canne,

One problem, Tactic now sees both xdr strings, the one with ROLL with wrong data and the one with Heel Angle with correct data.

I cant filter XDR because they both come from NKE

[Canne]

Originally Posted by FrouFrou

One problem, Tactic now sees both xdr strings, the one with ROLL with wrong data and the one with Heel Angle with correct data.
...

That is indeed a problem not only for tactics_pi but to whoever receives the two sentences about the same measurement but with different values: which one has the correct data?

Originally Posted by FrouFrou

I cant filter XDR because they both come from NKE

You cannot filter them at the reception side (i.e. in O) because there is not enough information for that in the message identifier. Maybe, if you could change the "II" to something else in one of the data sources? If you cannot modify the source I presume that you would sitll want to keep the NKE black box which provides you other important data you want to retain? I noticed that you have a MiniPlex as a connection point. Let's hope that it is a version 2 or 3. I have version 2 so I would do it like in the attached picture (http://www.shipmodul.com/downloads/m...iPlex-2_EN.pdf p37):



As an example, supposing that you have your e Heel Angle sensor 90-60-343 connected on input channel 3 the Nmea output 90-60-357 in channel 1. After the learning (Capture-button) or by manual entry you would have something like what is illustrated in the attached screen capture, see line 4: you would have, by default, all inputs checked (passed) for sentences IIXDR. Now, you would uncheck all inputs (block them) but for input channel 3 (the heel sensor input). This way, the output (PC) would have but IIXDR sentences coming from the heel sensor exclusively (and with "Heel Angle" identifier - but also with the correct value, hopefully!).

Programmatically, in O side this could be done only by some user set parameters (such as "override ROLL"... how ugly). I don't like this idea too much because it still increases the complexity of the already complex program. Thomas needs to decide which way to go, after all he seems to be equipped with NKE as well, I have but one instrument from them (and it is problematic enough ).

If you get stuck with this, I can compile you a temporary patch to ignore ROLL sentences so that you can work while waiting on them problem getting resolved at the NKE headquarters... But that is also ugly!

[bcn]

Welcome to the XDR hell



Hubert

[Canne]

Thanks, Hubert! I reckon that one cannot check Out
Petri

[bcn]

Difficult to resolve within O, plug-in or core.
As you said: not sufficient data in the XDR sentence to discriminate well.
A NMEA0183 issue with the proprietary XDR sentences. A dead road.

Here SignalK might come handy perhaps, allowing for virtual instruments.

Hubert

[Canne]

Thanks for the SignalK tip! Un bol d'air after some past and present NMEA hell just to read about how things could be what comes to the data model. But for the accuracy of information, FrouFrou is right, only the data producer (here NKE sensor) can resolve the problem by providing unique and accurate information. If not, SignalK would face the same problems with the multiple data sources, perhaps with inaccurate data.

But it would be a great step forward if one could purchase a NMEA/SeaLink/whatsover multiplexer working as SignalK server! It would not be trivial to make O to interface with it - I have noticed how the NMEA-knowledge is deeply rooted both in the core and in plugins. That's why, back to the real life, I use the filtering functions of my MiniPlex to give O only the data I trust (eg. like only one GPS source at a time) and which I know O can digest, and I recommend FrouFrou to do the same

[bcn]

Do you know the OpenPlotter project?
OpenCPN plus multiplexing and sensor integration on RasPI..
SK server included. Provision for NMEA0183 and N2K.
Home | Sailoog

[Canne]

Originally Posted by bcn

Do you know the OpenPlotter project?<snip>

Yes, I've heard about them. One day, not far from now, I will strip down my boat's factory installed mess, keep only the sensors and built everything following my own vision about what is a maintainable, high availability, distributed navigation system. OpenPlotter type of device certainly makes a lot of sense in such a system: one could afford to have two of them, installed on two separate data interconnect bus and with separate power supplies, thus providing a fully redundant system. Oops, we're now a bit off-topic from the NKE-world, sorry!

[tom_BigSpeedy]

Petri,

I merged your pull request #15

I'm not sure if I fully understood, but if you have 2 different inputs for the 2 XDRs you can easily drop the wrong one in the Miniplex.
I'd try to get rid of the wrong XDR sentence in the Shipmodule Miniplex ...
I do the same with a second GPS source in my Miniplex 2 ...

BR,

Thomas

[Canne]

Thanks, Thomas! I'll test your next version with pleasure but it can wait now. Hope you can do some sailing, too.

Originally Posted by tom_BigSpeedy

..I'd try to get rid of the wrong XDR sentence in the Shipmodule Miniplex ...

That was my also my suggestion above. Like you I am using the MiniPlex filtering functions in order not to disturb the old and faithful O with confusing NMEA sentences!

[tom_BigSpeedy]

Hi,
patch 6 uploaded to Ricks file thingie.
Sorry, for being short, I have bad internet connections ...

Since patch 5 there's more NAN value fixes ...

BR,

Thomas

[Canne]

Hello, I have tested successufully patch 6 installer, downloaded using these instructions. Works both with my own old recordings and with FrouFrou's use case XDR-data, either ROLL or Heel Angle (but not both). I recommend FrouFrou to override my temporary version with the production level patch from Thomas. Thank you and good sailing, both of you! (I cannot wait the weekend to try it out again in real...)
Best regards,
Petri

[tom_BigSpeedy]

Use patch 7 please.
Thanks to Petris comments on github I fixed some uninitialized variables in "polar compass" (although I don't see any "long numbers" here").
BR,
Thomas