Cheap rate compass

[adwuk]

It was surprisingly labour intensive. A fair bit of soldering, vero board to cut, fitting the connector etc. Even without the code, there is about 3 hours of time to build the thing.
Anyway, it was a good intellectual challenge, and I think it has the potential to be more accurate than the B&G compass I have - with the added advantage that it provides roll, pitch as well as heading and rate of turn. So I can see myself making one for my Nic 35 at some stage!

 

[GHA]

Great stuff - well done.

Thanks for taking the time to post an update.

 

[adwuk]

No problem GHA. Its now in the post. The next instalment hopefully comes from Javelin!

 

[Javelin]

Well I plugged it into the backbone and after persuading my zeus2 to use it all seemed fine except for some reason the triton display didn't tally with the Zeus2.
That turned out to be that the Triton had an option hidden that uses the cog as the compass heading.
So that was fixed.
Calibration was a little tricky as we had a 20knt onshore today so I didn't venture out.
Moving the unit by hand certainly showed the rate of turn working and the heel but the heading didn't change as much as I would have expected but this may have something to do with the fact I wasn't actually moving and tied up to my pontoon.
Also tapping didn't seem to have any effect but to be fair I'm not entirly sure what to look for.
Using the nmea0183 output on the plotter I wired it up the ST2000 but it didn't seem to make any difference.
It wasn't picking up wind either so it must be a plotter to st2000 problem.

But anyway the bottom line is that it works
Thanks Adwuk, hopefully I'll have more to tell once I can get out of the harbour.

 

[adwuk]

I wouldnt trust any of the heading figures until it is calibrated. What to look for when calibrating: as you are motoring on each heading, eg 0000, watch the triton. About 2 seconds after a double tap, the heading will move closer to your actual heading without any change in course. The controller will blink, but you will only see that if you take the lid off the compass. There are 4 screws on the back, and it will come apart. One 4 pin connector connects the nmea cable in the top to the controller on the bottom.

 

[Javelin]

I spent a lovely few hours sailing round in biggish circles off Southwold in the sun.
I found a good location for the compass however I found that steering, setting the auto pilot, confirming direction using had bearing compass and then dashing down below to double tap was not totally successful.

At first sailing out of the harbour and turning right into the tide at 180, the compass was reading 260.
When I had enough sea room and headed north the compass was reading 026, dash below, double tap, come up and it still read 026 give or take 5 either way due to the swell.
I went east, South and then West double tapping each time.
I did this three or four times.
North is now around 10 degrees out.
East and West appear to be around 20 to 30 out and South is around 40 degrees out.

It was quite lumpy with up to 16 degrees rolling under power so maybe I need a much flatter day and take a friend so they can tap.

 

[adwuk]

I guess this is why other systems go for a more simple calibration routine. If we need to, I can produce a version of code which simply calibrates from sailing in a circle, but I think the results from calibrating to known headings will be better. Flatter water and a friend will certainly help. Also, work out which way 'out' the compass is for each heading. For example, if 3 headings read 10 degrees less than they should then the other heading will most likely read 30 degrees more than it should. It means that 1 heading needs to be calibrated to bring them all into line. I hope that makes sense.

 

[skipmac]

Very interesting project. Even more so as my son-in-law and I both just installed identical Simrad APs in our boats (his power, mine sail) with the new Precision-9 compass which advertises it outputs heading, rate of turn, pitch and roll at a cost of only ‎£500 (each). Would be very interested to compare the performance of your home built to the expensive professional model.

Any plans to make more or publish construction instructions, etc? Would love to do a side by side test.

 

[adwuk]

Once we have it working properly I will certainly be looking to make one for myself. I have the old B&G rate compass, which isn't that accurate, and doesn't provide pitch and roll which autopilots certainly use. If anyone else wants one then I will happily make a few more.

I will take a look at providing the code open source. I make use of quite a few libraries, and so will need to read their Ts&Cs. I have a NMEA2000 to Wifi gateway also working which may be published as well at the same time.

 

[Playtime]

. . . . . . If anyone else wants one then I will happily make a few more. . . . .

Excellent thread.

I have a Simrad AP24 autopilot with AC12 computer and FC40 compass. I have been considering the Simrad rate compass (RC42) for some time but the £500+ price has ruled it out. I would therefore be very interested in one of your units.

One thing I have noted - the RC42 apparently transmits PGNs 127250 and 127251 but the spec doesn't mention PGN 127257: presumably, the AC12 doesn't need/use the attitude data.

Please PM me if you are considering producing further units.

 

[adwuk]

One thing I have noted - the RC42 apparently transmits PGNs 127250 and 127251 but the spec doesn't mention PGN 127257: presumably, the AC12 doesn't need/use the attitude data.

I have the RC42 and you are right - it only outputs 127250 and 127251 - it also isn't that accurate despite numerous attempts at sailing circles. That isn't to say that the AC12 won't use attitude data if it is available though. The most important thing though is rate of turn.

Hopefully we get this thing working well for Javelin and then I will happily make a few more. Will keep you informed.

 

[Playtime]

. . . . Will keep you informed.

Thanks

 

[adwuk]

A quick update:

Javelin had a go with the compass, and was pleased to report that it connected to his NMEA 2000 network and reported heading, heel angle and rate of turn with no fuss. There were two pretty major issues though:
- it was really hard to get the calibration right, which resulted in headings being way off
- the compass stopped reporting data after 1 hour of use

I have done a load of work on the code, and have managed to get the calibration routine working whereby you simply sail in circles for 2 minutes, ideally on flat water. The calibration is started by knocking twice on the compass cover, and in the 2 minutes you really want to try and get in a couple or full turns.

I have also ported the worldwide magnetic model calculations so that the compass can compute the magnetic field, variation and inclination at the current position (assuming there is a GPS on the network), which aids the calibration. I'm really pleased with this part, as there are a few more possibilities as a result. It means that the compass could now report True headings and variation values onto the NMEA 2000 network if needed.

So once I have worked out why it stops after 1 hour, then the compass will be heading back to the East coast for another test.

 

[Javelin]

Its great you're still willing to nail down the issues.
Happy to give it another try when ready though flat water on the North sea has been a little rare over the last couple of weeks