openCPN course over ground unsteady

[ColinTapp]

I'm using OpenCPN on a powered catamaran cruising about 10 knots running on a raspberry PI 4. I use a Richie fluxgate compass and the OpenCPN dashboard magnetic compass heading display looks steady. I take GPS from a module that uses an MT3333 chip and seems to give very good accuracy i.e. within the length of the boat. The connection between OpenCPN and the GPS is via gpsd. All looks good on the display when following a route except that at times the course over ground seems to change and at some points the image of my boat can be pointing backwards. All other parts of the display are good including the 'road' as part of the active route console window. I run in the 'course up' mode and the display is steady apart from the COG moving around. Its annoying but doesn't affect navigation. Any suggestions gratefully received. Thanks Colin

[rgleason]

You have asked this question elsewhere and not provided enough information.

[ColinTapp]

I'm having engine issues which have stopped me getting the data. I will update as soon as possible.

[ColinTapp]

Sorry about the delay getting back to you but I had engine problems that needed fixing before I could get data. I recorded data from the route and it appears that its a compass problem as shown below: $HCHDM,343.6,M*2B $HCHDM,343.1,M*2C $HCHDM,189.7,M*2E $HCHDM,343.4,M*29 $HCHDM,343.8,M*25 This is very strange as its a commercial compass. I have seen something similar with a compass I built using a bocsh BN0o55 sensor. More investigation needed but its not an OpenCPN problem. Thanks for your time Colin

[rgleason]

Thank you for your report and determination.

You could turn off the compass of course or use the OpenCPN options>connections or nemaconverter_pi to filter out the compass. Have you verified that the checksum is correct for the HCHDM?

Originally Posted by ColinTapp

Sorry about the delay getting back to you but I had engine problems that needed fixing before I could get data. I recorded data from the route and it appears that its a compass problem as shown below: $HCHDM,343.6,M*2B $HCHDM,343.1,M*2C $HCHDM,189.7,M*2E $HCHDM,343.4,M*29 $HCHDM,343.8,M*25 This is very strange as its a commercial compass. I have seen something similar with a compass I built using a bocsh BN0o55 sensor. More investigation needed but its not an OpenCPN problem. Thanks for your time Colin

[ColinTapp]

Thanks for the suggestions and I can of course just turn off the compass but there has to be a better option. I have checked the checksum and its correct. I made my own compass using the Bosch 9 axis sensor and wrote some python code to calculate the heading and convert to NMEA format. I had the same problem of the boat changing direction. I found a Ritchie compass expecting that there was a bug in my code but I'm seeing the same issue. Now that I know about getting a record of the route, I can look further. It sounds like electrical interference or something similar. Nothing is ever simple on a boat :-(. Colin

[ColinTapp]

p.s. I want to use the compass to align the radar overlay with the chart on OpenCPN

[rgleason]

Maybe some clamps in the right places?

[Bill O]

Have you looked at your NMEA log, is your GPS dropping out?

We have two GPS inputs going into our RPi, a primary and back up if the primary drops out.

[ColinTapp]

I have two GPS inputs but filter one out so openCPN only sees one. Doesn't the plotter software get confused with 2 GPS inputs. My boat has several GPS's, none of which put the boat in the same place. Boat position and GPS based course over ground are fine and the GPS doesn't drop out. Its the electronic compass that gives spurious results.

[Bill O]

The 2nd gps is as a back up and only kicks in if the primary drops out. So no it doesn't interfere in this configuration.

[ColinTapp]

How do you have a 2nd GPS that only kicks in if the primary fails? I get my GPS data from the GPS via GPSD

[Bill O]

Originally it was through the multiplexer, now it is through signal K.

Not sure if you are using a RPi, but openplotter gives step by step instructions on how to set it up.

[ColinTapp]

Thanks. I'm not using signal K but its worth thinking about in future. Yes its running on a raspberry pi 4.

[ColinTapp]

Just to conclude this thread, I now put the compass data through a simple software filter that removes single values that are too far away from the previous values. This virtually fixes the problem although the boat icon does occasionally move as before. It appears that I'm occasionally getting two wrong values. The display now looks correct apart from a very occasional glitch. Its definitely a compass problem. I have yet to find the cause but I have fixed the effect.