New Py Zero 2 W is Out!

[seandepagnier]

what if you just reboot one or the other? Do you know which one has to be rebooted? Does it require both?

[hans-martijn]

what if you just reboot one or the other?  Do you know which one has to be rebooted?  Does it require both?

It requires both, if I reboot only one nothing works, only if I reboot both at the same time the connection works.

Currently the connection is broken again, the last time I rebooted was just before I sent the last message…

[seandepagnier]

if you have to reboot both then there is a bug in both pypilot and signalk server in my opinion.

Whichever one has to be rebooted to fix it is the one with the problem.

Long long before it stops working on average? Did you enable debug in signalk.py (uncomment line in debug function) and check the log?

I would like to reproduce it. I could at least make it not have to restart pypilot with some kind of timeout and monitoring of the signalk connection or a way to detect the server restarted.

Do other sensors and does other data in signalk continue to work?

[hans-martijn]

if you have to reboot both then there is a bug in both pypilot and signalk server in my opinion.

Whichever one has to be rebooted to fix it is the one with the problem.

Long long before it stops working on average?  Did you enable debug in signalk.py (uncomment line in debug function) and check the log?

I would like to reproduce it.  I could at least make it not have to restart pypilot with some kind of timeout and monitoring of the signalk connection or a way to detect the server restarted.

Do other sensors and does other data in signalk continue to work?

I just rebooted the signalk server, but there nothing happened. After enabling debug in signalk.py, I rebooted the pypilot, and I got my connection back. N2K still worked, which mainly delivers AIS and tank status. Even my victron Venus connection keeps working, although that doesn't deliver much data, since my solar panels don't seem to work.
Setting the date obviously didn't solve much, since the connection crashed, so I'll leave it being Sunday November 21, 12:04 UTC 2021.

I'll check tomorrow again and send you the logs.

[andrewAB]

Where would I find the latest / best Tinypilot image for Pi Zero 2W?

This one in Downloads seems a little old.
>> Pi4/Pi zero 2 (stelian's image) tinypilot13-20211126.img.xz

Is there a single Tinypilot image that would work on both Pi Zero W and Pi Zero 2W?

I have been using tinypilot_2020_10_27.img.xz for the last 3 seasons and it has worked great.
However, I've got a Pi Zero 2W so are thinking to upgrade the HW along with the SW

Many Thanks!
Andrew

[hans-martijn]

AFAIK that’s still the last image.You might try it though.

-Hans

[seandepagnier]

The problem is I have too many half-finished projects to get to this...

Basically there is no advantage to the pi zero 2 over the pi zero at this time for using pypilot.

[andrewAB]

The problem is I have too many half-finished projects to get to this...

Basically there is no advantage to the pi zero 2 over the pi zero at this time for using pypilot.

Sure, totally understand and will stick will use the latest SW on the Pi Zero standard version this season.

I thought I read somewhere upthread that boot time with the Pi Zero 2 would be much faster?

I have a Shit-Hit-The-Fan  "master" 12V supply that goes to both the Pi DC-DC regulator and the motor controller (and hence clutch on my Jeff/Garmin below deck linear drive).  So if autopilot misbehaves/crash, sudden wind shift or course change required etc I flick the autopilot off (to release helm) and then back on again (to restart in standby). Meanwhile I hand steer until the pilot has rebooted. So faster reboot time would be nice to have.  I have had pilot go wonky once in about 100 days use over 3 years.

Thanks
Andrew

[seandepagnier]

That is actually true about boot time.

Rather than rebooting on "misbehave crash", consider a switch in series to the clutch output to gain manual control. Further, it would be better to actually determine why the autopilot misbehaves or what occurred, as this should in theory never happen. Was it with older software? Rebooting fixed it? Could it have "updated" the calibration? Was the calibration locked? Can you reproduce it? Perhaps the problem will go away with newer software anyway.

I would rather determine the actual problem.

[andrewAB]

Hi Sean
I agree a switch in series with the clutch would be nice, but from under-deck it would have to route the cables 4 meters each way back to the chart table and adds extra connection that can corrode or come loose. I already had one problem with a dry joint connection that could carry voltage but not current.

As following Murphy's Law the one "misbehaviour" happened when the wind and waves had rapidly increased while I was weaving between island and boat was over-canvased and surfing at 8+ knots. After I took over control I reefed the sails, got a distance offshore and put the rebooted Tinypilot back in control at a nice 6~7 knot boatspeed and it was good for the remaining 20 miles of my journey. I will double check the calibration was locked. No further problems have ever been seen. I have an updated a Pi Zero with latest beta image installed and the new gain menus look good and I will swap this in and recalibrate in a few weeks time.

I now have an old android phone with bicycle mount on my binacle so I can see pilot status over wifi so I can see how far off course the pilot if I suspect misbehaviour. I also have a RF garage door controller relay that goes to the "standby" so these are my first point to take back manual control. I use the IR aimed down the companionway to make the course adjustments and have the phone at the wheel as feedback that the changes have been seen.

All up I am a very happy sailor and looking forward to cruising the Greek Ionion islands in June! You are welcome to join me if you are ever in that part of the world, and enjoy Greek food and wine as much as sailing.

Many thanks
Andrew