A sailing story for nerds

Driving to the boat today I have electronics on my mind. I want to capture the output on the victron ve-direct interface, to be able to build a sensesp unit for it, I have an oscilloscope with me to check out the nmea2000 signal from the emtrak, as my first canbus exposure, I want to measure the dBm of the marina wifi with different positions of the wifi dongle, and I want to test whether the esp32-cam (future mast-cam) has enough wifi power to reach the openplotter access point from 10m above the steel cabin.


When I drive over the causeway dam, that blocks the sea from the IJsselmeer fresh water lake, there's hundreds of metres of drifting ice on the lake side. For Dutchies, that's becoming rare. As I make a stop at the lock through the dam, the ice turns out to be all small bits and pieces. The marina, 20 minutes under the lock, is free of ice. Once at the boat, I see a police vessel on the AIS pass through the icy bit and I decide I can give it a go. It's exhilariting to see the ice line when I exit the harbour; it sparkles in the bright sun! This is what it must be if you go far north. Slowly I motor into the ice and it gives away easily. I have a steel hull and I don't care about the paint. Close to the lock I call them and they say they are willing let me out, only if I'm not planning to go back in. Weird condition, but I agree. The ice gets denser as I push in. In some way the ice cubes have a festive ring to them. With a big grin I moor off in the lock. The police vessel moors behind me and the deck hand loves the conditions as well. He's never sailed through ice either. 

There's no wind on the salty side. My mast-top unit is broken: it gives me wind angle but no wind speed. The anemometer cups are still there and spinning. I noticed it earlier this week from home, as I started synching the signalk database with the cloud, so I can read my instruments at home. For that, I had to keep my instruments and openplotter running over the last week. Has that been the cause of the wind meter failure? Unlikely, however another weird thing has happened. The water temperature, that comes from my log instrument, reads 37C on the display unit, which is obviously wrong. But the nmea message, and thus signalk as well, shows the right temperature (I measured it with an infrared thermometer). So the error has to be in the raymarine display unit. That's also weird.


Anyway, the electronics can't fix the missing wind, so I motor through the Boontjes, on, of course, the Pypilot, compass mode. This will take at least an hour, so I ty-rap the esp32-cam with a battery to a block of wood and hoist it into the mast with the topping lift - the lazy jacks can hold up the boom for a while. It works brilliantly! Even at 1600x1200, it has a rate of 5 frames per second, which is good enough really. A big boat passes by and now I can look at it from above, which feels like a kind of victory on more than one level. The camera swings about obviously, but the ultimate plan is to integrate it inside the anchor light. I hope the masthead won't shield the signal, but that's what we will find out in the intermediate phase, when I glue the camera onto the top of the anchor light. I have to climb the mast anyway to fix that stupid wind unit.


Anyway, at Harlingen there is still no wind, but navionics estimates Vlieland well before the evening curfew, and I have ample diesel, so I turn into the Pollendam. There's another boat the opposite direction with its sails up, so I furl out the jib as well - don't want to look like a woossy! But the jib hangs straight down and looks pathetic, so I roll it back in. I eat a sandwich and upgrade my layers of clothing. This a windy channel, and I hate it when I forget to turn and have to be reminded by the depth alarm. I program a route in opencpn and set pypilot onto it. The wifi gets a bit sluggish so I turn the camera resolution down. 

I have a new set of pypilot gains, but I forgot to set the FF (forward feed) gain, so at each waypoint the boat turns excruciatingly slow. It's not really a problem because there's no-one around and the sea is as flat as a pancake, so it's easy going. When I want to change the FF gain anyway in the opencpn plugin, it turns out the wifi connection with pypilot (tinypilot, really) is broken. This never happened before - I guess it must be the camera. I restart the pypilot, which reinstates the wifi connection, but after that I forget to set the FF gain. I have a 30-minute straight end before me, so I open the propane tank and heat up some ready-made babi pangang. I can't see the food in the bowl when sitting in the cockpit because of the glare of all the screens from inside, but when I lean over the hatch I can see it in the moonlight. It's 2 days after half moon, neap tide, so we have moonlight right after sunset and the least tidal current. Still, it's good for 2 knots current in our favour. When I rinse the dishes the North Sea swell makes itself felt. We're nearing the gap between Vlieland and Terschelling and I have to pay attention, because when that swell rolls over the shallows you can enter a rodeo ride within a few boat lengths. I've been there.

Orion made a photo bomb. Note the reddish color of Betelgeuse.

I've tweaked pypilot to ask for confirmation before turning at a waypoint, and so far this is what I have been doing this trip. But now I'm on it, I decide I can cut through to the next waypoint. I right-click the route in opencpn and advance to the next waypoint, and weirdly enough, pypilot turns the other way! Now I have seen this kind of stuff before in more challenging conditions, at which points in time I have cursed the whole universe whole-heartedly, including, I must admit, pypilot. But now I'm in a relatively happy mood, not soaking wet and just properly fed, so I watch and see what happens why. And slowly, that green arrow with the white dot at the end, turns toward the waypoint I wanted to advance to. And it seems to coincide with the XTE going down... So here I had, what the germans would call an 'AHA!-Erlebnis'. The moment I advance waypoint before the current waypoint is due, I immediately inherit a huge XTE, and pypilot alters heading to dissolve that XTE, which might be the other way! So, wiser now, the next waypoint I advance as well, and immediately after that, I select 'Zero XTE'. This time the green arrow with the white dot properly flips into the direction of the new waypoint. Sweeeet, mystery finally solved, workaround acceptible. It would really be great if the autopilot-route plugin would be properly debugged, because that would really solve this type of problems.


At this point I also find out why the direction where my boat icon in OpenCPN points to is wrong. Since I've switched to SignalK as multiplexer, I've selected the raymarine compass heading as primary Heading Magnetic. This is because I have integrated my pypilot IMU in the tiller unit (st2000 conversion). The pypilot-provided compass heading, being constantly autocalibrated and heel-corrected, may be better than anything else in the world, but when I detach it from the tiller and swing it away, that compass heading is useless. Which is why, with grinding teeth, I selected the raymarine compass heading as primary heading, and it is this heading that dictates the orientation of the boat avatar in opencpn. I've done numerous calibration rounds to linearize that raymarine thing, but without an imu, on a steel hull, there is only so much you can do with an old-fashioned (read: WW2-style) fluxgate compass alone. Knowing it's the raymarine that's wrong makes it a lot more bearable having this crabbing avatar, but I just might consider replacing it with a tinypilot on a pizero with an imu that's fixed to the boat.


In the approach of Vlieland, and in the throes of the waypoint advance conundrum, I see a green buoy light coming straight towards me. The buggars must have moved a buoy again, and I haven't updated my buoy file recently. I do have a crontab running every night that creates a GPX file with all the buoys from the Dutch coastal service data hub. All I need to do is copy that file to the layers directory of opencpn. But I forget. So here's this buoy coming at me in the dark, and I know it's very hard to judge its distance. So I put pypilot in compass mode and click '-10' 3 times. But because I haven't changed the FF parameter yet, it goes terribly slow and I freak out, jump into the cockpit, yank off the tiller pilot and take the tiller in my own hand. I normally create my opencpn routes so I stick starboard side of the channel, like any good seaman would do, but this is quite a-relaxed. I go down the cabin and drag the remaining waypoints mid-channel. 


10 minutes out of the harbour, I make my approach checks, hanging out fenders and the like. I realize I might trip over those lines that are still holding up the esp32-cam, so I hurry to pull them in. As I do, there's this huge ferry boat at a few boat lengths distance overtaking me. That's what you get when you drag your waypoints mid-channel! Normally you hear them coming from a distance, but not when your stumped off by 6 hours of diesel engine noise. You know that feeling when you switch off your engine and proceed under sail? That feeling is what I missed today. Nevertheless is was great fun. Hope there's some wind tomorrow. 






Postscript: The wind unit appeared to have a defective reed switch. The miniature glass switch (9.6mm length) just had an ohmic resistance, not on, not off. Replaced it with a reed switch I had around, but  I had to hotglue a bigger neodymium magnet into the rotor for it to work. Got a mobile setup and checked it from the car window, then counterbalanced the magnet with an M5 nut. The windset has a 4 wire waterproof plug, that you forgot was there, and which you taped off with self-vulcanising tape, so what you need to bring with you into the mast is just a knife and a PH1 philips screw driver for the M3 screws that hold it in place. More photos here.



Comments

Popular posts from this blog

Various tricks

Stellendam v.v.

Singlehanded 2022