Drift doesn’t just depend on wind speed. Still depends on the displacement of the boat, the size of the keel and much more. But on any yacht all the dependencies can be reduced to two-Wind and Roll!
ah, I think I understand.
For calculation of the true wind angle and speed the vehicles velocity vector is used so this will take into account any side slip.
Your suggesting to monitor the roll angle as a way to try and check if the sails are filling or not, and to check the side slip to see if we are too tight to the wind.
These situations are very tricky for the autopilot to spot, just as arduplane stall prevention has to be very conservative. Also they have to be written such that they will apply to any sailing vehicle in any conditions.
The longer term plan is to define a polar the boat will sail to. This would allow the boat to sail at the best true wind angle for each wind speed and this would hopefully avoid any issues with too much side slip. The polar would give the ideal heal angle for each wind speed and heading.
Hopefully we can add a auto tune style optimization algorithm to generate this polar on board.
Thanks. No big trouble, let’s wait.
I just don’t know what means “a polar the boat”
It a record of the best boat speed at a given true wind speed and direction, see http://www.ockam.com/2013/06/03/what-are-polars/
This allows selection for the best heading to more towards your destination as fast as possible.
Polars, of course, it’s cool.
But they must be calculated separately for each boat and separately for the storm sail.
How many races do you need to make to have enough material for their calculations?
If in another place there will be a significant flow of water, then Polars will be incorrect?
Taking into account the angle of the roll, you can optimize the course in the first race
I was thinking that you would run a auto tune style mode for abit, probably 30 mins or so for a couple of different wind speeds. The boat would then use a maximizing algorithm to optimize its speed for each heading, the best speed would then be recorded as a point the polar. We would also record the heel angle this speed is reached at. As you say it would take some time to build a full polar for each boat as I guess we would have to average boat speed over tens of seconds to get a reliable reading. As you say if there is flow they will be wrong, but until we support a direct boat speed sensor we cannot tell if there is any flow in any case.
The advantage of the polar is you can use it to select tacking angles for best VMG.
I am not familiar with your roll based method, do you have any more information? I guess it is a method to optimize boat speed on a single heading rather than picking the heading in the first place? It would be nice to support vehicles with movable ballast, it sounds like this roll method may not apply in this case as the boat would always be upright?
My roll based method of determining the course is purely empirical (informal ). Based on personal experience sailing without any navigation equipment, only windvane. Always choose a course with a minimum angle to the apparent wind. When the speed falls and the roll (“heel angle”) is large, then slightly increase the angle. So constantly lie on the steering wheel and keep in mind: a sharp course is more important than speed
But at the same time speed is needed to complete a tack and not get locked up in irons. It’s all about VMG but when its time to tack often a boat will fall off the wind a bit to build up speed and then tack.
That’s right, you need a certain speed to turn the boat. So before you turn increase the speed to the desired. Speed sufficient to turn your boat should be remembered
I’ve just loaded the binary. Still don’t have the wind speed sensor on the boat yet. The sail parameters were all mixed up and had to be adjusted. Windvane parameters changed also with the max saying 3.3 when mine is 5. All the ATC stuff seemed to remain the same. Will give it a go tomorrow or Wednesday.
So SAIL_MAX_XTRACK is now WP_OVERSHOOT and SAIL_LOITER_RAD is now LOIT_RAD? I’ve just been reviewing the parameters again.
correct, there was some duplication of function so the parameters were consolidated
Latest update for what it is worth. Yesterday I ran a 6 Kilometer run around the waypoints in my last video. It was like 11 or so runs around the pond. I was fortunate to have wind initially that was very lite and then it picked up a lot over the two hours. My tuning seemed to handle both conditions very well. since then I updated to Randy’s code. I think I have the change of parameters understood. I was going to let the boat loose in the Gulf of Mexico tomorrow but I am thinking that doing another test in the pond with the new code is probably safer. Although my windvane seems to be fairly accurate I again tried to calibrate it voltage wise and nothing happened. It’s spot on when pointing directly forward and agreeing with the heading of the boat but is off about 15 degrees when running down wind with bench testing. Also redesigned the wind vane to be more effective when the boat heels.
I keep thinking of other questions. I can’t search them because this is so new…If WP_OVERSHOOT is 0 then will the boat just do one big tack? Can WP_OVERSHOOT be greater than 10 Meters. MP say’s 0 - 10 but I’m guessing it can be greater. The only problem I have really seen is what the boat does with a failed tack. I have seen the boat tack just when a wind shift comes and fails. The boat falls off the wind and keeps trying to tack with the rudder hard over. I don’t know if there is any way to realize a failed tack, straighten the rudder to build speed and try and tack again
exactly right, setting WP_OVERSHOOT to zero is the equivalent of setting it to infinity. You can go as big as you like. Yeah when it tacks and fails is a know issue. Tricky to tell when its happening. I guess if its light you could increase the tacking angle so it will be going abit faster. It also struggles to bear away in strong wind.
What is the weight and size of your boat, the weight and size of your wind vane? You may have reported it, but I can’t find it. And it is still unclear why the efficiency has improved when the boat heels after redesign of the wind vane.
It’s 5 foot long and 24 pounds or so. Total weight of the wind vane is just under 3 oz. and its about 12 inches. My old vane design was being effected by the mainsail air flow I think and when the boat heeled I don’t think it had enough surface area perpendicular to the wind to point properly.
This sensor Hall-type non-contact sensor
seems better than Bourns 6630S0D. He has a Rotation torque: <5nN.m and Resolution: 0.088 ° (12-bit ADC)
How does that compare with .4N-cm of torque. I’m not good with some metric stuff.