MacArthur Hat issues - power cycling on its own

[jamesbreezing]

Having a problem with my MacArthur Hat.

I have it connected to a RP5 with NVMe bottom / 500GB drive.  The power module is installed and have the IMU inserted on the MacArthur board.  Printed enclosure with lots of openings and no top so is not getting too hot.  Have ethernet (for Halo radar - working).  Have o-charts dongle.  HDMI but normally use it headless.

My system initially was playing well.  Check system passes no issues.  I could see traffic through the board within SignalK. Testing it on a bench before I take it to the boat.  12V into the board.  12V SW with open/closed switch to set the power on/off (like this functionality!).  No power at USB-C input.  MCP251xfd set to SPI0-CE1/200000/GPIO25 (was available) / can0 / can0 when added SignalK connection.  No other CAN0 configurations.

It was ‘working’ where I could save a configuration but then I started having issues where could not get NMEA2000 to see the network.  Shares common 12V power for N2K backbone and power to Pi5 through MacArthur hat.  Impedance for N2K connection is correct at 62 ohms across the +/- for the NMEA2000 connections.  Then the board decided to cycle itself a few times -  it would cycle the whole system.  It is drawing 480mA at full load (variable power supply is set to 12.6V and is 10A so not an issue). 

So thought maybe the power supply board was U/S.  Powered system down.  Took power supply daughter board off.  Rebooted without the power daughter board with only USB-C with a separate 5A power supply directly to the Pi5.  This works for the Pi5 but the MacArthur hat appears to now not be working at all.  The NMEA2000 LEDs are showing steady lights (no blinking) so the network is not working even though the configuration shown above is valid (did try removing / installing again configuration wise).  And no additional lights on the MacArthur board - I would have thought 5V would have lit up.  I know the network is working fine as when I use  Yacht Devices USB device, I can see the traffic (GPS and wind sensor showing)

I can use my Pican-M board which seems to be more resilient but was hoping to use the MacArthur board as it has graceful shutdown, and much more capability with other interfaces.

Any thoughts on trouble shooting would be appreciated or perhaps my MacArthur board is U/S.  My Pi5 / Pican-M / NVMe 500GB drive configuration is working fine.

[Adrian]

The 5V LED will only turn on when the power is provided by the DC/DC power module. It will stay off if power is provided by the Pi.

The HAT/Pi power-cycling could either indicate an overload condition (the HAT has a 2A fuse, which should be plenty of headroom at 12V x 2A = 24W, or about 4.8A to the Pi), or more likely a loose connection of the 12V or GND wire.

If NMEA 2000 no longer works, can you share a close up picture of the board with the power module removed, and the picture taken under good light.

@sailoog Any logs from OpenPlotter or SignalK that show errors during initialization of and communication with the CAN driver?

[Sailoog]

Yes, signal k server log is here: http://openplotter.local:3000/admin/#/se...ration/log

Maybe is just the picture but I do not like the color of that power module.

What exactly "Then the board decided to cycle itself a few times" means? is your Yacht Device getting data from the engine? If yes, look at this: https://forum.openmarine.net/showthread.php?tid=5471

[jamesbreezing]

I have uploaded better complete pictures of the MacArthur hat.  In the process of trying it on a Pi4 running openplotter v3 to see if that will make a difference.
12V VCC and GND are all tight connections.  From a bus bar.  No issues with continuity.

Breezing

[Sailoog]

What exactly "Then the board decided to cycle itself a few times" means? is your Yacht Device getting data from the engine? If yes, look at this: https://forum.openmarine.net/showthread.php?tid=5471

[jamesbreezing]

So, I rebuilt the entire system and it now works! I think there was a conflict on CAN0 that was causing the non NMEA2000 which was a leftover of the Pican-M card versus the MacArthur hat. I don't know why it was power cycling but it is stable now. I have it running on my bench and it has been up and running for 24 hours without a hiccup.
Thanks for all the advice and looking forward to putting the MacArthur hat through its paces! I use Grafana but also looking forward to KIP and Node-RED.

[Adrian]

Happy to hear that

If you encounter power-cycling again, I would inspect the solder joints of the connectors on the HAT and power module for hair-line cracks.

[jamesbreezing]

Thanks Adrian, am hopeful that it will be working ok but definitely will look at that. I have NMEA2k / NMEA0183 simulator so can try out the full functionality and have a Halo 20 radar, and have logging onto the Pi with a TB drive. Definitely stressing it but its working like a champ now.