The Offshore Voyaging Reference Site

The Importance Of Boat Speed To Autopilot Operation

Modern smart autopilots rely on an accurate true wind calculation to steer well with the wind aft of the beam, particularly when a spinnaker, A-Sail or code is set.

And assuming we have the apparent wind direction wand accurately adjusted for angle on each tack, another important input to true wind direction is boat speed, which pretty much always needs to be calibrated.

This used to be a royal pain in the ass, because it required timed runs over a known distance.

But in these days of super-accurate GPS that outputs speed over the ground, (SOG) it’s pretty easy: just do say four runs of say 1/4 of a mile each way, to account for current, and adjust the calibration until the difference between boat speed (BS) and SOG is the same both ways.

To make this easier, set the damping on both SOG and BS to the high end of the range so that the jitter, particularly in SOG (typically around .3 knots), does not confuse things.

I easily do this in 15 minutes single-handed.

Since I single-hand our J/109 a lot and so rely on the autopilot, I check the speed calibration once a month or so.

I also clean the impeller just before I calibrate.

To make that easier I have added a light line secured with a wire loop as shown in the shot above. This makes it way easier to secure the sensor against backing out without having to screw around with seizing wire, (as the manuals tell you to do) every time we clean it.

I have also found a new (not used for months) stiff-bristled toothbrush by far the best tool for cleaning the wheel.

And, finally, a few thoughts on sensors:

  1. I always select a sensor that has an automatically closing flap to reduce the otherwise-gusher to a trickle.
  2. The new Generation-2 paddle wheels from Airmar are way better, particularly at slow speeds, than the older version. I upgraded this spring and it was worth every penny.
  3. B&G technical support tell me that sensors that provide pulses directly to their system are better, since they update more frequently, than sensors that convert the pulses to NMEA 2000 sentences. Not sure if that’s true, but it sounds right and the source was credible.
  4. Ultrasonic sensors always sound intriguing—I love stuff that gets rid of moving parts—but given that they have been around for over 20 years and still no racing boat I have ever seen uses them, there are probably issues that make them less accurate. I’m guessing turbulence in the boundary layer.
  5. When I was replacing my speed sensor, I was told not to get one with an integrated depth sounder because that usually (always?) results in a smaller paddle wheel. Makes sense.

Further Reading

Inline Feedbacks
View all comments
Arne Mogstad

Hi. I got the NKE sensors, and when I upgraded all my electronics around a year ago, they gave me the electromagnetic(?) version for the same price as the paddlewheel, so I thought why not. It seems to work fine, and it feels immune to growth. There’s no difference in speed reading if it’s quite covered or clean! Only issue I have is that it gives a fairly big difference in speed reading depending on the tack I’m on! This kind of makes sense, since the through-hull is offset about half a meter to the port of centerline, so it doesn’t seem fair to blame the sensor. Annoying, but not really much to do about it on a centerboard boat (OVNI). Maybe at some point I can weld it shut and then make a new one in the bow, but that’s not very tempting either. I’m also unsure how much it matters, I can’t really tell the difference on the pilot, it’ll just slightly shift the true wind reading between tacks. Or maybe it is important? I just don’t know better?

Kindly, Arne.

Jean-Louis Alixant

Hi John,
Several companies explored electromagnetic speed measurements, mainly to eliminate the mechanical concerns around paddlewheel systems and to offer a system that could stay in the water. But an interesting development was the use of more electrodes (five for Airmar’s DX900+) to extract the Leeway component. Unfortunately, they stopped commercializing this device very quickly after receiving negative feedback of the type “it does not work”. From discussions I had with a few installers, I suspected that they had not put in sufficient effort in the calibration. Needless to say that any paddlewheel system that is not calibrated properly will “not work” either.
If anyone has a DX900+ collecting dust, I am interested to experiment with it!

Jean-Louis Alixant

Hi Arne,
nke’s electromagnetic speed sensor has a good reputation. If it has been properly installed and calibrated, I would trust its measurement.
There is no doubt that any boat speed sensor will respond to the fact that it is mounted off-centre, to some degree, because the water flow is even less symmetrical than for centred sensors. As John pointed out, if you have a system that allows for Boat Speed correction as a function of heel angle, you should try to fill it in (takes time to do well). I don’t know whether your nke system allows this. If not, you should consider repositioning the sensor; if you go through that step, you will always get better results from a properly installed sensor than trying to correct for unwanted effects resulting from installation choices or constraints. Some boats install two sensors to ensure that there is always one that sees “proper” flow.
By the way, we are focusing on lateral positioning of the Boat Speed sensor, but the longitudinal position is important too.

Let us know how you get along.

Jean-Louis Alixant

Hi John,
It is worth pointing out that in the case of B&G’s H5000, TWD is also used in the Apparent Wind mode. It is far from trivial, so many users may not realize this. It is due to the fact that B&G back-calculate Apparent Wind after a first iteration to calculate TWD. Thus TWD impacts wind calculations in a second iteration, and therefore Autopilot performance as well, even in Apparent Wind mode. Boat Speed matters doubly.
This is not to say that Boat Speed does not affect the performance of other Autopilots that do not rely on TWD in the same way (such as nke). The algorithms are so complex that it is difficult to say to what degree such measurement affects the outcome.
The only sure thing is to get the best measurements possible out of our sensors: select the right sensors, install them very carefully, calibrate them meticulously and maintain them love. Without that, we can be certain that it not just the Autopilot that will not perform at its best: any indication on (and behind) the screens will be affected, sometimes to the extent of driving us to the wrong decisions.
When it comes to Boat Speed, my limited experience shows that sensors “calibrated” by installers on cruising yachts tend to be off by at least 10%, sometimes more than 20% – and it is always in the same direction! Needless to say that everything that follows is plain wrong.

Jean-Louis Alixant

Hi John,
This may be of limited interest to most, and it is somewhat off topic for this post of speed calibration, but just in case and for our own common understanding…

There is no doubt about B&G’s approach to wind calculation involving back-calculating Apparent Wind from True Wind. It is also spelled out in their documentation, for instance in the online help of the H5000 web socket:
Furthermore, its (sic) important to note that Apparent Wind Angle and Speed (AWA and AWS) are “back-calculated” from True Wind, hence once True Wind is calibrated then these corrections will apply to Apparent Wind also.
That did not change with the latest release, it is absolutely core to B&G. I have argued that they could move on with the use of masthead units high enough above the sails, but I don’t know that they will change it. A good check is to compare Measured Wind to Apparent Wind, you’ll see the difference.

The new software release has just a few key changes:

  1. A key change is the separation of the NAC and Sailing autopilot algorithms. Performance Level 1 remains the “standard” PID autopilot it always was, but it is now completely separate from the sailing algorithm used starting at Performance Level 2. In short, you should now use Performance Level 1 for motoring and preferably start at Performance Level 2 for sailing.
  2. Another important change is the separate treatment of Weather Helm through Autotrim. The resulting rudder offset is not longer part of their Steering Algorithm, which means it will no longer be competing with features like Heel Compensation. As a result of this and the separation of the two pilot algorithms, the use of Expert System features (Heel Compensation, but also Gust, High Wind, Automatic Perf. Level, Recovery) should be much more stable and reliable. We should use them as required without the concerns they may have inspired in the past.
  3. Adapt gave problems in the way the PID parameters were being updated after 30′ and they now recommend not using it at all (OFF) as you mentioned. That makes a notable improvement in the overall performance in certain conditions, especially waves, TWA > 100° and TWS > 15-20 kts altogether. They’re not simplifying, but correcting for a control error in the logic.

I have not tested that in a broad enough range of conditions so far, but the general impression is that the pilot is more stable, and more active than previously.

The discussion on back-calculating Apparent Wind or not gets quite involved as you know, but the apparent (sorry) simplicity of the straightforward extraction of Apparent Wind from masthead measurements hides complexity that is not accounted for by other solutions: that will affect True Wind among many others. There is no perfect solution at this stage, just two approaches that deal with the issue differently. With different results and requirements.

Now we can go sailing ;•)

Dieter Kowalewski

Hi John,

I don’t understand why boat speed is still used to calculate TWS and TWD, no matter how accurately the sensor is calibrated. Often there is some kind of current, and then the calculation with SOG would be more accurate.

Kindly, Dieter