So far we have taken a look at the two main types of autopilot: super-smart and ultra-configurable, but complex to get working right and silly-expensive; and the rest.
And then we dived even deeper with a two-part guide to tuning autopilots while offshore to make them steer better.
So now let's end this—if you are not ready, I sure am—with some usage tips and hacks that I wish I had known when I started using autopilots (and vane gears) some 40 years ago.
You understand the dynamics of fast boats well, from your experience with them, and you mention several places in the article how that influences steering and autopilot behaviour. However, I assume that a significant number of the readers here have less experience from truly fast boats, so they might underestimate the actual importance of this factor. Some might think “yeah, interesting to know, but this isn’t really relevant for my old heavy cruiser”, (wrong), or perhaps “I get the point, but still don’t really have a grip on how to use this info on my boat”. An actual quote from a fairly experienced monohull cruiser/racer: “Relative wind direction and strength will only change a tiny bit from the real wind, and then stays about the same, so I don’t bother with digging into this. I just sail.” No matter what boat, he’s flat out wrong.
I think this issue is important and and not obvious. My way of getting this issue through the acknowledgement barrier is to be a “besserwisser”, which I must admit I enjoy more than I should. 😀
1. Quantify, to show that it matters, and then,
2. Insult the majority, 🙂 and then,
3. Suggest some solutions.
I watched Sail GP in St.Tropez on location some years ago. The boats would pass us 15 meters (50 feet) away at over 30 knots, while the sea was like a mirror and a lit candle would have had zero problems staying lit. This means their wind speed over deck can be many times the actual wind, and the angle they feel has changed well over 100 degrees. That’s kinda noticeable….
The old turbo cars when in lower RPM would feel as there was nothing special. As the RPM passed a certain level, say 3800, the turbo would kick in and the engine suddenly had double power, which would just keep increasing. On really fast sailboats, also far from the Americas Cup extremes, using relative wind feels exactly like that. The word “turbo” is even used sometimes. On downwind legs, one will steer significantly higher to “fetch power”, which really kicks in like a turbo. Then bring it dramatically lower at double speed or more, with sails sheeted almost all the way in. In gusts we just steer a bit deeper and in lulls a bit higher, but the speed can be kept up pretty well through almost zero wind.
The AC boats etc are as wild as it gets on water, but shows that relative wind is what we sail on, and actual wind is just a useful reference. For way less extreme boats, even heavily loaded cruisers, the only difference is quantity. The degrees of angle change and the change in wind speed is far less, but the principles are exactly the same. Any boat will experience some of the “turbo” effect, even though it’s way more subtle. If we race our cruiser, we must learn to use the “turbo”, or we’ll certainly be beaten by one who does. We only sail on apparent wind, not on “actual” wind. This matters for safety too, big time: If our wind angle and power suddenly change significantly, which it will on any boat downwind in heavy weather, we need to understand it and know what to do. If we don’t respect that, we’re looking for a hard punch in the face.
Every single exceptionally good sailor on the planet has much experience from fast dinghies. Downwind in heavy weather, all boats behave much like dinghies. The logic from the AC boats becomes really relevant. Surfing on waves and change in relative wind angle are game changers. Then, anyone without significant experience from fast dinghies or very similar, is an incompetent and a dangerous helmsperson. The steering and trim actions will often be the opposite of the correct ones, which will usually result in wipeouts. No, you cannot learn this well on anything but dinghies, or really extreme pure racing boats. The only people who actually do get insulted by these claims, luckily quite few, are the ones who don’t understand the issue. That probably didn’t help… 😀
3. Solution 1.
Go sail fast dinghies! It’s the best fun you can have! A one person boat like a Laser is ideal. Simple, not very fast, but enough to show what it’s about, and ubiquitous. An old one can be bought very cheap. Fast dinghies explain things by dumping you in the water when you did something wrong. 🙂 Still, most anyone can do it. No need to be an athlete, unless it’s quite windy. I very strongly recommend this solution.
Know your limitations. Before the boat gets too powerful, reduce sails, or change course, or pick a heavy weather strategy suitable for the boat and situation, like heaving to or JSD, early enough to avoid too much challenge. Learning the first steps of high speed sailing with your home in a storm isn’t ideal…
In the context of this article about autopilots: Even without any experience from any fast boat, most of the time you can steer far safer than close to all autopilots. That counts triple for downwind courses. If you feel the edge of your comfort zone is approaching (coping ability, not actual comforts), the autopilot will not help. It totally depends on your input and help, and it will fail to cope far more often than you.
Great comment. To me the key part, and something I did not make clear enough, is that an autopilot will not get us out of a bad steering situation, particularly if we don’t understand how it works.
I also agree on the benefits of sailing a Laser. Although I was primarily a 505 sailor, I did sail Lasers for a season and that little simple boat taught me a huge amount. I can still remember the lesson where I got a bear off wrong on a wave when running in big breeze and death rolled (capsize to windward) for the first time! Really made me think about what I had done wrong and how to improve. Lesson: don’t be still bearing off when the bow hits the trough and starts up the next wave. The bow locks in and the bear off increases at just the time it should end. Wipe out.
This is just the sort of mistake an autopilot can make if steering to apparent wind in big breeze and waves because the apparent wind goes forward as we accelerate on the wave face a lot, but the pilot does not know that the next wave face is coming. When steering to true wind, it won’t bear off so aggressively so no wipe out. But if the true wind is not correct…
Have you ever tried actively playing with controls as you sail on a seconds type time scale as opposed to an hours one? The Dashews wrote about actively adjusting the gain for big waves. I have never tried it and wonder whether the mental energy required makes sense.
I think your advice to test is crucial, it seems like an awful lot of abandonments are on recently purchased or refit boats.
Yes and no. Yes, in that we adjusted gain a very frequently on our old Neco, particularly, when racing to Bermuda, because it had knob to do that. No, on the Simrad because the gain was buried deep in the menus.
(Steve always used WH autopilots on his sailboats and those also had a gain knob.)
And good point about abandonments on new boats. I wonder how many of those were at least contributed to by an autopilot that was not functioning once the weather got even a little challenging.
Hi. You mention that using GPS (Speed Over Ground) won’t cut it for true wind, and that you need speed through water. Can you elaborate on this? My understanding is that speed through water is mainly interesting for performance polars, but the autopilot does not really care about that. The autopilot is more interested in frequent and accurate updates, which it can easily get from a high performance GPS (10 Hz or even more). I doubt speed through water will give a higher update frequency, or at least that the autopilots will have much use for even faster updates than 10 Hz? I am currently installing a fairly fancy full NKE system (sans the external processors). The manuals state that you can use SOG as a true wind reference, and further I understand it that the autopilot (Gyropilot for the NKE), does not prefer one over the other. In fact they even imply the opposite due to the fact that the sensors will very often be inaccurate (like you say), due to a lot of factors (growth, boundary layer and so on). See the fifth topic in this link.
Anyway, this topic is incredibly interesting, and really opened my eyes about how to use the autopilot!
That’s a really interesting question and one I have been thinking about since Svein brought it up. My gut reaction is “of course we must use speed though the water”. And that’s what all the B&G manuals seem to assume we will do. So that’s why I wrote it. But more I think about it, the more difficulty I have in coming up with a reason for that, particularly with, as you say, a GPS compass which will be dead nuts accurate and frequently updated.
I still have a nagging gut feeling that speed through the water is the way to go, but I can’t justify that with a logical argument, so quite possible I was wrong.
In fact, the more I think about it, at least for those with a sat compass, speed over ground might be better: Imagine that we are broad reaching at 6 knots but there is a 2 knot current against as. So we are actually moving over the ground at 4 knots, and therefore that’s also the speed at which the rig is going through the air. QED SOG is will yield more accurate true wind.
…OK, I’m changing it in the article.
Thanks for the correction to you and Svein
Of course I’m still open to correcting the correction if someone shoots holes in the above.
Anyone else have a logical argument to support either speed over ground or speed through the water?
Hi John and Arne,
It’s a little tricky to get into this conversation through a comment but given the interest I will give it a try anyhow.
To answer the question regarding which speed to use to derive True Wind requires agreeing on definitions first. Apparent Wind as we can all agree is the movement of air (speed and orientation) measured with an instrument fixed to a boat (let’s forget about vertical variations for now). Likewise, Ground Wind is the wind measured from a station that is fixed with respect to the Earth. True Wind is often mistaken for Ground Wind, but it is different: it is the wind expressed with respect to the surface of the water, as if the anemometer was drifting with water at the speed of surface current. Remember that the sailing yacht has no knowledge of Earth-bound references: she only feels and responds to relative motion of air on her sails and water along her hull, so True Wind is directly relevant to her performance, rather than Ground Wind. Sadly, most sailing manuals call “True Wind” what is actually Ground Wind. Be assured that sailing instruments (such as those developed by B&G and nke, I can’t comment on Raymarine or others) actually calculate their best estimate of True Wind as defined here.
Therefore, if you are interested in Ground Wind, for instance because it is relevant to compare to weather forecasts, “you” (that’s the onboard computer really) should use Apparent Wind and SOG/COG to back-calculate Ground Wind. However, if you need True Wind (for example to have the autopilot properly steer on a reach, as explained in the article), the triangle you must use involves the speed of the boat through water, let’s call that SOW. Bad news: SOW is not the measured Boat Speed. Boast Speed as derived from a paddlewheel only provides the axial component of SOW, not the lateral component, which is Leeway. More bad news: there is no commercially available sensor for Leeway.
So, what to do? For True Wind, always use Boat Speed (provided that has been thoroughly calibrated, this is fundamental) and consider including an estimate for Leeway. Advanced autopilots allow for this, either as a fixed value (an angle), a computation based on empirical relationships (B&G uses this) or a lookup table based on VPP (recent nke calculators allow for that).
Each method has its limitations, but as soon as there is Surface Current and Leeway, using SOG and COG defeats the work gone into calibrating the sensors and tuning the autopilot meticulously. I would compare this to using a compass reading to figure out True North accounting for Variation only and without correcting for Deviation. Using SOG and COG, you will calculate Ground Wind, not True Wind, and you will be off by the Surface Current (“Tide”). It may not seem like much, but it would introduce a systematic error that is also variable. To make things worse, you’ve just instructed a system that is trying to compte True Wind that Surface Current and Leeway are both nil, so don’t expect any accurate Wind output. Given all the trouble we are going through to get a properly tuned autopilot, we should stick to Boat Speed together with some estimate of Leeway.
Yes, the bottom line is that we should measure Leeway. It is the missing link in having our expensive gear deliver what it is meant to provide reliably.
Hope this helps, JL
Great explanation. I totally get that we must NOT use COG from the GPS, my confusion was more about the relative benefit of using SOG in the true wind calculation. I think you have me convinced that the answer is no, mainly based on your very good point about ground wind as against true wind.
That said, to really get this clear in my head I’m going to need to do a vector diagram with say a boat on a broad reach and say a 1knot current on the beam, which I will do in the next few days. Being an old time navigator vector diagrams really help my aging brain get these things clear!
Thank you for taking the time.
People looked at me strangely when I had a shape 3D-printed to correct for the Rudder Feedback Unit level, it’s good to know that I am not the only one! I prefer your solution though, because you can easily adjust to obtain the exact angle, I had to print the piece twice to get it right.
Regarding rudder angle, I recorded a curve of actual rudder angle (a linear function of the rotation angle of the helm) vs. instrument indication after Dockside tuning. Before fitting the angular correction for the Rudder Unit and correcting the geometry of the quadrangle, it gave a curve with up to 4 degrees offset in places. Once both corrections were implemented, the curve became perfectly linear. Producing this curve gives an indication as to the magnitude of the issue. I don’t know to what extent that correction made a difference to the overall stability and performance of the system, but at least it is one cause of concern that disappeared.
To visualize weather helm, I taped two markers each side of the aligned rudder mark on the wheel, corresponding to 4° rudder angle that everyone can see. I have found this very useful to draw our attention to the need to trim especially when conditions change slowly.
Kind regards, JL
That’s really interesting that the offset was as much as 4 degrees. Like you, I don’t know for sure what the effect on autopilot performance would be, but given that the algorithm must be assuming a linear return from the feedback unit I would expect some degradation, perhaps significant.