The Offshore Voyaging Reference Site

Autopilot Buyer’s Guide

While writing our recent article on fore-reaching as a heavy weather strategy, I got thinking about how smart autopilots are, or, in the majority of cases are not.

And that reminded me that a couple of years ago, when replacing our vintage, and pretty stupid, Simrad autopilot brain on our then boat with a new B&G NAC-3, I found that setup and tuning settings were little changed from the old.

Yup, I think we can be pretty sure that the new pilot is using 25-year-old software, albeit upgraded some. So it’s probably not much smarter than our old Simrad, which was pretty darn stupid.

Which kinda makes sense, at least from the manufacturer’s point of view, particularly for a company like B&G that makes a much more expensive and truly smart racing autopilot, and so has no incentive to improve the software in their less expensive offering marketed to cruising boats.

Don’t get me wrong, the old pilot served us well, with near 100% reliability—it was a bad and no longer available control panel that finally did for it—but it could not even steer halfway as well as an even moderately skilled helmsperson.

And the much ballyhooed claims of auto-learning turned out to be auto-stupidity.

Why Good Steering Matters

So why am I writing this, other than to beat up on B&G? Because good autopilot steering matters, particularly on a shorthanded cruising boat where we don’t have enough crew to take over when the autopilot struggles.

But surely most autopilots steer well? Sadly, no, although the common perception is that they do.

I can’t tell you the number of times cruisers have said to me:

I love my autopilot because it steers way better than I can.

Sorry, the next part of this is going to sound harsh, but it needs to be said so we don’t make the mistake of thinking our autopilot is better than it is.

Unless we have a full-on smart racing autopilot—more on how to recognize one in a moment—that has been manually calibrated and tuned for the specific boat, all the above statement means is that we don’t steer very well, not that our autopilot does.

No shaming here. In most cases this is because we shorthanded cruisers don’t get enough practice, particularly offshore in waves, to be good at steering.

This was brought home to me last summer when we started sailing our new-to-us J/109 and I was shocked to find, despite having been a pretty good helmsman back in the day, how much my steering skills had deteriorated over nearly 30 years of relying on autopilots and vane gears.

The point being that most of us cruisers should not use our own steering skills as a benchmark to evaluate those of an autopilot (or vane gear).

By the way, the other day I was at a virtual meeting with Stan Honey, one of the world’s leading ocean-race navigators, in which he confirmed that even the best autopilots in the world (think tens of thousands of dollars) can’t steer as well as a skilled helmsperson in daylight, although said autopilots get competitive in the black-dark when the human can no longer see the waves.

Stan also opined that the next big advance will be when cameras, like those being used for collision avoidance by racing boats (OSCAR), are interfaced to autopilots, and that will result in automation being able to beat a human helmsperson in all conditions, because the pilot will be able to see the waves, even at night.

Probably not applicable to any of us, but interesting.

Anyway, back to why good steering matters.

The poor steering abilities of most autopilots on cruising boats contribute to:

  • Higher fuel burn when motoring.
  • Substantially higher electrical use—my guess is as much as double what a really smart race autopilot would use, once offshore in big waves.
  • Seasickness. Yup, good steering can make all the difference to how quickly and how badly the dreaded malady hits us.
    • I would always take the wheel while Phyllis was below doing anything sick-inducing, and she always said she could immediately feel the decrease in motion, even with my somewhat rusty steering skills.
    • And she could do the same for me, despite only coming to sailing as an adult and never having been a race-boat driver.
  • Falls on deck or below due to increased and unexpected motion.
  • Broaches when running off.
  • A lot more green water breaking aboard when sailing upwind.
  • Sailing much more slowly than the boat’s potential.

Bottom line, it’s worth investing some time to make our autopilot steer better, as well as shopping for one that will steer as well as our budget will allow.

And this in turn got me thinking about three important things to know when shopping for autopilots or thinking about how to make the one we have do a better job:

  1. The claims of smarts and auto-learning for autopilots are often more marketing BS than fact, something that I think the manufacturers get away with because most potential buyers really don’t think much about good steering capability.
  2. That said, it’s actually pretty easy to check if an autopilot is stupid or smart by understanding a bit about how they work and checking the manual. More on that in a minute.
  3. Even if we end up with a relatively stupid autopilot for economic reasons—the super-smart ones are silly-expensive—we can improve its steering a great deal, by first being realistic about how badly it steers using the default settings, and then learning how to improve that by hand-tuning it for the conditions.
    • More on that in the next chapter coming soon.

By the way, autopilot steering smarts is top of my mind because, much to my surprise (not listed on the inventory), our new-to-us J/109 came with all the expensive stuff—computer (unlocked), sensors, etc. to support a really smart autopilot: B&G H5000, same as many pro shorthanded sailors use—so I’m just in the throes of adding the comparatively inexpensive drive computer.

The result is that by late summer I should be in a position to provide a first-hand report on this racing autopilot and how much better it is than the less expensive alternatives.

The Difference Between Smart and Stupid

So what’s the difference between smart and stupid autopilots?

Well, the first thing we need to understand is that the autopilot itself only stores, and varies its steering on, four fundamental parameters, plus a few secondary ones.

We will get into what those parameters are and how they work in the next article on autopilot tuning, but, for the purposes of this chapter the key differentiators are:


The parameters are all varied in concert, continuously and automatically, depending on wind direction and speed, boat speed, and wave state, generally (always?) by a separate computer commanding the autopilot drive controller.


These variables, with one or two exceptions, are only set once, when the pilot goes through automatic sea trials under power, and then are not automatically changed ever again, and that sucks for good steering.

An Autopilot IQ Test

But how can we check which type we have or which type a model we are thinking of buying is?

Let’s dig into that:

Login to continue reading (scroll down)

Inline Feedbacks
View all comments
George L

There’s also Madintec – competing primarily with NKE. They focus their efforts on computer, controller, motion box (gyro), rudder sensor and interface to the wind instrument. For the wind interface they use B&G (WS310 and 700), Garmin 24XD GPS, B&G precision9 compass. Same price range as NKE (that is eye wateringly expensive) but they now offer a couple of cruising packs that are a bit discounted.

There is also the GHP Reactor from Garmin, but I cannot find any info about its quality. Same for Raymarine. Both have the advantage that they aren’t part of Brunswick Corp; though considering your Canadian abode, I am surprised about your concern 😉

You are spot on – a wind pilot could be quite an attractive alternative.

Denis Foster

Hello John,

Great teasing, I am impatient to read the rest of your book.

As a “mature” cruising couple we realize how dependent we are on autopilot, we have invested in two independent autopilots that has everything in double and can be chosen at steering station by the flip of a switch. The only item shared is the mast head wind indicator. I was wondering if mounting an ultrasound wind sensor on the aft radar pole would make sense?

As you can see we spent more money on reliability and redundancy than “smart” performance was it the right choice?

Best regards.


George L

Had a long talk with an NKE engineer, who had tested just about every unit on the market and he said the ultrasonic units didn’t stack up.

Brian Russell

Our Maretron WS100 ultrasonic has performed admirably for 20K sea miles, including 2 trans Atlantics. Seagull proof.

George L

the wind calculations are independent of the sensor; they are done in the computers or dedicated boxes.

Paul Clayton

There’s also Pelagic which seems to specialize in tiller pilots. I want one, but will have to save my quarters for a while longer. Not that they’re especially expensive, compared to some of the other kit out there.

P D Squire

The mention of polars brought to mind a question I’ve long had: Do polars take sea state into account?

Colin Speedie

Having sailed quite extensively with NKE autopilots aboard various Boreal models, I know how good such systems can be. Reliable, too, to my astonishment….

But a decent vane gear in the hands of someone who has taken the trouble to learn how to get the best out of it will work wonderfully well. I’d argue that a well-matched vane/boat combination will sail nearly as well, at a fraction of the cost (see John’s excellent post above before even thinking of spending) and many other good things.

No electric demand, quiet, low cost crewman – what’s not to like?

Marc Dacey

Nothing, really, that’s why we have one. Even you mentioned Tony Gooch’s thesis:

George L

I think this is spot on.

While on a boat that has electricity and electronics only, the cost isn’t nearly as high as John writes, the marginal cost for high-end self-steering gear is still eyewatering.

If nearly as well is good enough, then the Windvane ought to be considered – at least for the backup system.

Philip Wilkie

Without wanting to pretend to be smarter than I am, the other core functionality I would be thinking about is the style of the control algorithm implemented. My background is industrial automation and I have lost count of how many control loops I have worked with, but 99.9% of them would have been implementing the classic PID algorithm, and I am aware that while for the most part it works, it has many limitations and compromises. And most of them would apply to autopilots.

Well at least everyone knows what PID is – but if know that if your autopilot is using it (and the clue will be the presence of Gain, Integral/Reset and Derivative/Rate parameters) – then know that by the standards of modern control theory it is a dumb controller indeed.

Algorithms such as LQR (Linear Quadratic Regulators) and MPC (Model Predictive Control) are way more sophisticated than your grandfather’s PID loop – and frankly for $20k worth of autopilot I would expect to see vendors offering systems that belong in this century.

Andre Langevin

I agree with you Phil. To me boat autopilot are the last bastion of corporate owned software. They run their software on cheap processor that sell for thousands of $… At that price you can mine crypto or run OpenAI siblings of AI. I would love to see an open source project coming up for developing the best autopilot software compatible with NMEA2000 sensors.

Look what happen to OpenCPN. We just need OpenPilot 🙂

Philip Copeland

Agree 100%.

Star Tracker

What about PyPilot? I’m considering building one at the moment. Or going raymarine, then using an H-bridge to let me get away with a cheaper ACU. Side note: Raymarine used to have a waterproof version of the rotary sender. It was never listed commonly and often had to go through the hassle of finding a dealer who would punch in a part number and just order the result. Much Much more reliable than the stock Raymarine sensor they shipped with. Only thing worse than the un-reliable one was the people Raymarine has been telling that it is so smart it doesn’t need one. Apparently there is a guy 3d printing an upgraded waterproof one on Saltspring Island in Canada.

Trevor Hope

Does PyPilot meet the Open parameters?

Philip Wilkie

Here is an example of the kind of techniques I was referring to above:

Our main goal is to design an energy aware autopilot for a heading and foil dynamic control. To answer this problem, in this paper, we propose a new control method which is aware of the energy consumption. This control method is based on a MPC controller which minimizes a cost function taking into account two criteria: the square error and the energy consumption. It makes a prediction of the system behavior from an electric actuator model and a load model. This control method is then applied on a rudder control. In fact, this control loop is the first step to completely manage the autopilot system and its energy. For the same energy consumption as a PI controller, we increase by 45% the precision level or we are able to reduce the consumption by at least 25%. This controller optimizes the use of the energy onboard to steer the boat

Stein Varjord

Hi John,

In my serious racing days we mostly used NKE, as we sailed much in France. It was great build quality and ahead of it’s peers at the time when it comes to smartness, but user friendliness was ridiculous. You needed to dig into software programming for the more advanced functions. All the advanced stuff was software only.

That brings me to my main point: Are the extreme price levels really related to actual cost, or is the prices just there because they can? What you mention about having to pay to unlock functions that are actually already in your unit, support that notion.

The top notch items will always use top notch materials and production methods. That can’t be done for free, of course. Still, how much of the cost is from that? There are only so much data to collect and only so many tillers to turn.

The data we can possible collect is, (if I haven’t forgotten anything):
– Wind direction
– Wind speed
– Boat speed through the water
– Compass heading
– Boat orientation and rate of change around 3 axes.
– Rudder position

Actually knowing what to do with the data isn’t rocket science. The calculations needed, the stuff going on between collection of info and executing a movement, are very simple. Any smartphone can do them without breaking a sweat.

The answer is always super simple: Should the rudder be pulled or pushed? That’s it.

So, why do we need 2-3-4 different “brains” or computers to do that? It seems like the “brains” don’t really deserve their name and these companies might be living in the past.

I’m fine with having to pay the actual value for whatever I need, but if prices are kept artificially high, and it can be done way better, I get annoyed and look for alternatives. I wonder how long it takes for some independent maker to grab that opportunity…

George L

you have outlined the sensors – high end and connected on NMEA2000 (mostly – i haven’t seen a rudder sensor yet), will set you back a few thousand.

for calculation any computer will do – we use Intel NUC; cheap, small, quiet and unbelievably powerful.

then you have the actuator unit, but you probably can hijack a commercial one.

Rest is software, but anywhere, software and support is 80 to 90 percent of the cost nowadays. (trust me, that’s what I made my living with …) And software nowadays is not straightforward – in the days of C and the like you had the cost of really learning the language, but once you understood it you could build stuff fast and reliably. Now there is always something new and a lot of it is interconnected with other pieces and the web. The systems I could build twenty years ago with 5 people in 1/2 year now take 30 people and two years and they still don’t work as well. Just think about the many forms on the web that break down. The other day I tried to send a parcel from the NL to UK and the dropdown and the form check were incompatible ….

I do believe the companies are living in the past – lot of legacy stuff.

“unlocking”, though annoying is legitimate – somehow there needs to be a pricing model to make ends meet.

I have visited NKE and from the looks of it, nobody there is getting filthy rich. If you need to attract good engineers with a low-volume product, it is always going to be a struggle.

Matt Marsh

The labour involved in selecting, obtaining, learning, installing, and tuning one of these things seems rather daunting.

I am not sure how many labour hours John will have put into this AP setup by the time it’s all done. 50 is probably a lower bound and it might end up being over 100? And that’s assuming serious DIY, electronics, and embedded computer skills to start with.

Something to keep in mind if you’re the type of person for whom free time is limited and precious.

Colin Speedie

Good point, Matt. Every time I get on board a new boat these days it’s like the bridge of the Starship Enterprise with kit that is (on paper) capable of just about anything! Endless facilities that I would never use, even if I could work out how to.

If you’re prepared to settle down for weeks reading some unintelligible online handbook, then maybe. Even then I can’t work out the half of it, let alone when woken by some calamity in the middle of the night to buzzers, alarms and utter confusion.

Someone like John who has an innate sense of this stuff can work with it, but not the average dimwit – like me.

Matt Marsh

I feel like I could *probably* get a smart autopilot installed and working well, given enough time.

But…. I spend my days designing equipment of considerably greater cost and complexity. I know what it’s like to hire and train people to install and service the equipment my company builds. And I know what standards of workmanship to expect from contracted 3rd party technicians at various levels of training and salary.

If I were contracting someone to install such a smart autopilot system, I feel like I’d have to budget 50-70 hours of my time for project management (writing specs, tendering, evaluating candidate contractors, supervising, QA) in addition to the 50-100 hours of installation labour.

Philip Copeland

Thanks John – excellent article and will be really interesting to follow your progress with the H5000. I have a Seawind sailing catamaran. We used our B&G NAC3 Autopilot to steer upwind last year on our return from Lizard Island down to the Whitsundays. I think the more basic NAC3 works much better on a cat as you are not dealing with heel issues. Its ok in wind mode – but could be so much better.

B&G is a very very very frustrating company to deal with – and I doubt any of the other nav companies are much better. Its pretty increadible that they have not made many improvements over all this time.

Stan Honey is correct – vision based AI systems will take over in the next 4-5 years – similiar to what we are seeing happening on cars. I just drove my Tesla on a 3000 mile road trip a few weeks ago where the camera based (8 cameras) AI drove all the way. Its not perfect yet – but getting very very good – and safe as it never gets sleepy or distracted.

Lets hope some system comes along and gives B&G a big kick up the backside – they deserve it.

Rob Gill

We have a Raymarine autopilot with their ACU controller and control head. We also have the Raymarine Evolution autopilot EV-1 sensor that has “3-axis digital accelerometer, 3-axis digital compass, 3-axis gyro digital angular rate sensor”. We also have Raymarine log and wind instruments networked in using SeaTalk.

The system has worked without any technical faults coming up 10 years, so no complaints from me there. It also provides good alarm functionality around wind shifts, failure to hold a course, or in the event of the actual steering gear failing (like the hydraulic pump electric motor).

In other respects it seems to have been designed to the extremes of the KISS principle. Having chosen a boat type (sail, motorboat etc) in setup, we only have three helm settings we can toggle between whilst the autopilot is operating; Leisure, Cruise and Performance.

Upwind and reaching we use Leisure mode, otherwise the autopilot tries to oversteer. Downwind in waves we need Performance mode (until we start surfing). Especially nearing hull speed, the autopilot struggles to steer anywhere near as well as any half-decent helms person would in big waves – it just doesn’t anticipate the boat being thrown off course by the next wave and is usually slow to react, then often oversteers.

I can’t help thinking Raymarine have taken things too far, and in this case KISS stands for “keep it stupid, stupid”.

But this article (great topic by the way John) makes me wonder if part of the problem is where we originally mounted the autopilot sensor – up in the bow, away from all electromagnetic interference.

Having the sensor in the bow in big waves, the sensor will be moving rapidly in three dimensions. Should we be moving this into the back third of the boat where the motion is easiest?

The Raymarine manual and install guide make no mention of sensor positioning, only that it should be aligned with, or parallel to the centreline, and away from interference.

Any insight would be really helpful – hoping to avoid re-running the sensor wires unnecessarily to test a “half-baked” theory.

Many thanks, Rob

Dan Tisoskey


I have the same setup and I just started testing. I have read articles or other forums that did give advice on moving the compass in order to achieve better performance. I installed our EV-1 in the center under the floorboards, in the middle of the boat. I hope to achieve good performance from the compass in this location.

Rob Gill

Thanks Dan, unfortunately we have a large cast iron keel in the centre of the boat and behind that, two of our main water tanks. Finding a place free from electromagnetic interference is challenging hence our initial setup.

Mathieu Fortin

Just playing couch general here but have you considered going up? Even if it doesn’t fit between the liner and the ceiling, a nice little box on the forward cabin at the top of the bulkhead should not be too noticeable.

Too much interference from the mast/wires?

Rob Gill

Thanks Mathieu,

I will be trying to site aft of the keel, because that’s where I think the boat pivots about, when turning.

I think we will have to go up vertically to find a spot, but just as we have our sensor mounted too far from the longitudinal pivot point, we need to take care it is placed close to the vertical pivot point of the vessel also, which for a yacht with its heavy keel is usually just above the floor level of the cabin I believe.

In this way I can try and minimise the 360 degree motion the sensor experiences in a seaway, and hopefully provide the best input, based more on the actual pivot point of the vessel in 360 degree orientation.


Rob Gill

Thanks for the confirmation John – winter job for me…!

I believe gain is controlled through hard over to hard over time, which can be adjusted in setup, but not in AUTO.

Dead band I believe is controlled by the three modes; Leisure, Cruise and Performance which can be changed in AUTO.

I have just found deep in a Raymarine tech forum that there is another hidden setting, Rudder Damping, to increase or decrease how much the autopilot hunts. This is normally hidden by the system after setup, but can be toggled on/off in the setup menu.

Not sure that will help with our issue though, but look forward to following this series.

Rob Gill

Thanks John, understood.

George L

If I recall correctly, it should be close to the center, pretty much at the mast foot.


I have an Alpha 3000 autopilot and Hydrovane both off course totally independent from each other and the rest of the electronics. Sometimes I wish I could have both operating together. Sounds like what these sophisticated systems described above do. But people on convoy rally’s arrive across the Pacific kind of bragging about how many Raymarine autopilots they destroyed. Does not sound very Attainable Crusing to me.

Matt Marsh

Indeed. If you fit your autopilot with a basic Raymarine Type 1 linear drive, well, that thing can exert 2.9 kN (650 lb-force). Do you have hydraulic steering? A Raymarine Type 3 hydraulic pump hooked up to a 2″ ram makes about 17.3 kN (3900 lb-force).

Are your autopilot mounts strong enough to jack up the weight of a small car or truck, every six seconds, all day, every day?

A couple of 1/4″-20 screws through a two-inch aluminum angle isn’t going to cut it for long.

Marc Dacey

Interesting comment to me as this week we are modifying our Marol ram (which weighs a metric you-know-what) to be raised six inches. It is bolted with four 5/8″ SS bolts to a 1/2 inch steel plate tied into the frames with multiple L-angle steel bars. It’s all hydraulic, pushed and pulled by an Octopus 2000 pump. It will create some interesting carpentry jobs for me, but will, we hope, improve our autopilot’s performance.

I was helping to deliver a Bristol 45.5 about 14 years ago that had a Raymarine unit on a 1/2 inch plywood mount with 1/2 inch bolts. It tore itself off its mount in seven metre swells and the Dyneema line controlling the Monitor windvane chafed through the coaming woodwork. It was…instructive.

Dick Stevenson

Hi Ian,
That is, indeed, unsettling news with regard to Raymarine autopilots. Early days, cruising friends had much trouble with RM autopilots, but recently their QC seems to have improved.
I also have an Alpha 3000. You do know that Alpha has closed its doors: if not, I am sorry to convey that news.
My 3000 carried me many miles and numerous ocean passages over 15 years without a hiccup. When returning to North America, I felt, the 3000 deserved a rest, I bought a whole new 3000 (a clone of the first) installed it and put the old, but still working one, in storage (so, I should be good for as long as I have my boat).
It is a shame to see such a quality piece of kit close its doors: not sure why it was not better known except that Alpha seemed to largely resist the bells and whistles, interfacing and what not, that makes some products so attractive.
My best, Dick Stevenson, s/v Alchemy


Yes Dick, I am aware Alpha has ceased to service and sell parts. I do keep a spare control unit and Chris (Alpha owner) told me that a good electronics technician would be able to facilitate repairs. I am wondering if the fluxgate compass can be replaced with another manufacturer. The oil has spilt out a couple of times. I think that the linear drive is good tough component with an integrated rudder position sender.

Marc Dacey

We have an NAC-3 to a pair of Vulcan displays and have that B&G rudder sensor on, to be honest, a rather lumbering motorsailer we hope to take across the Atlantic in some nine weeks. I concur with you on much of your estimation, but I do “feed” the B&G system with yaw and heel information and the results are…well…pretty reasonable. I use a Lars Thrane LT-1000 electronic compass, and it’s a back-up to the Vesper XB-8000 GPS function. I notice that and playing around with the settings on the autopilot module itself that we do pretty well.

That said, we have a Voyager windvane for under sail. I’m a bit Scottish with amperage.

Marc Dacey

Thanks for checking my work, John. We, in a fit of both belt and suspenders, have the LT-1000 and a Precision 9. I’m going to have to fire up the device list to confirm what is talking to what and in the meantime, there’s construction at that end of the boat and the AP’s not really doing anything. We bought a Vulcan 9 for the pilothouse in order to get more screen area there and the Vulcan 7’s going to the “outside” helm, so the device list may not be what I think it is.

It’s quite possible that the heading sentences for the AP are originating from the Precision 9 and the GPS co-ords are going to the plotter. We generally only use the AP while motoring or motorsailing unless we are making a lot of amps.

Emile Cantin

The setup you mention in conclusion is the one I have on my boat: An ancient Raymarine wheel pilot, and a Cape Horn windvane.

After nearly 4000 miles on this, I have to say I really like it. I mostly use the pilot when motoring or in very light winds, and the Cap Horn when sailing.

There’s a bot more of a learning curve with the latter, and sometimes it zig-zags a lot (pretty sure it’s a sail trim issue, I’m pretty bad at it), but it’s a very solid piece of kit particularly if there’s any kind of swell going on.

If I get another boat without a wind-vane it’ll definitely be on my list of upgrades.

I can probably hand-steer better than both systems (a pretty low bar, honestly), but my main issue is that I get distracted.

Brandon Reese

Any thoughts or experience with any of the Raymarine units? (e.g. EV200 Sail Pilot)

Pete Running Bear

Interesting article. I must admit I’ve never played with the settings on mine through fear of making it worse so will look forward to the next instalment. One thing to note if two units are being installed, make sure there is a power switch that simultaneously powers up one and powers down the other so there is no way for them ever to be on together. Although I’m sure most of the readers here are sensible enough to realise that.

Kevin Dreese

Question regarding the B&G chartplotter setups… is it possible to integrate or transfer PC based navigation plots to the B&G multi-function-display MFD?

The Furuno units ($$$) seem nicely integrated with the TimeZero (PC, Ipad, etc.) and their MFD’s. I have searched but haven’t been able to find an easy way for a B&G plotter to upload (without using a card) plots from a PC to the MFD.

My plan would be to do the weather and route planning on a PC (laptop or ideally with a larger monitor) and transfer the route to an MFD(s) on deck.

Also, it would be really great to see more articles about your J109 fitout for cruising… especially the anchor setup, windlass, navigation equipment, sail choice, and any other mods. Inquiring minds want to know *cough* copy what you do.

Kevin Dreese

Thank John. That confirms my research. Good to get a real-world point of view.

Kevin Dreese

Awesome. Looking forward to it!

Eric Klem

Hi John,

This series has been interesting and I have to say that I haven’t seen as good info on how to tune published before. When I first started using autopilots, I was hopeless at tuning until I watched one of our test engineers at work tune a PID loop for the test stand of a new type of machine we were designing.

What surprised me most by these articles is the actual cost to get a smart autopilot setup and running. I have used at least 2 that I can remember but they were on other people’s boats. While at times their steering drove me nuts as they don’t have eyeballs to see waves or gusts before they hit, overall it was quite good. Now that I know the cost, I see them a bit differently, I used to think the solution on a lot of boats was just to use a smart pilot but maybe that was a bit flippant of a response. On the other hand, I share your dislike of the steering capabilities of a dumb pilot, it only takes a mediocre helmsperson to be better.

I wonder if someone has ever really quantified energy usage of these different systems. If you could install a much smaller battery bank, have less renewables and even maybe ditch a generator, that could really impact the cost equation. I don’t know if the usage is that different though.


Vasja Zupan

Hi, any thoughts on NAC-3 + Precision 9 + Edge Processor combo upgrade, from AC42 with RC42? Zeus 3 and HS60 are also in the system.

The idea was to upgrade, but not go “all in” with H5000. H5000 might be soon updated to Hxxx as Edge just came out. Plus H5000 is much more complicated to tune, while Edge seems to offer reasonable results via a more “mickey mouse” approach (at least that is my hope :-).

Gust response and polars seem to be key missing things. But I’m not looking for performance. Primarily For the pilots “help” for short handed sailing on longer cruises. Gust response is the only one I am aware of and sorry to miss.

The Boat is a 40″ Elan S5 (slightly lighter version of E5)…

I started with adding Edge and Precision 9 to the system, but later learned from B&G support that AC42 will always calculate true wind by itself, therefore Edge would only add apparent wind correction tables making much less of an impact.

Vasja Zupan

I’ve just finished first short sail after upgrading to NAC-3 + Edge Processor, from “only” AC42.

Nothing extreme, only 12kn and small 1m waves, sailing downwind.

However, already massive difference after basic calibration and auto trim compared to only AC42. Less difference upwind and reaching, as expected.

I think there is a lot of potential when properly tweaked and calibrated.

Vasja Zupan

Yeah, AC42 is good but the difference to NAC-3 + Edge seems to be quite big. I expect even more results after tweaking wind correction tables etc.

I’m also wondering what difference is between DST210 and DST810 and a new wind sensor. (old 508 vs new wifi ws308) More frequent dates etc.

George L

Some updates, since a lot has changed since March …

Concerning the wind instruments, the WS310 is identical to the WS700 with the exception of the carbon mast extension for a fraction of the price. Hence, adding a really good wind instrument can be done for a few 100 USD. With the new Hercules unit just released, up to 3 can be connected, allowing for a port and starboard instrument to take advantage of the more accurate one at windward – redundancy for a very modest cost.

Concerning Brunswick, I have been _very_ pleasantly surprised about the interactions lately – at least, as far as I am concerned, it seems like a new company. They have developed very positively, with very interesting new releases, e.g., four new AP units (Triton/NAC3, Hercules, adv. Hercules and Hercules WTP). If one is willing to live without polars and wind correction tables then the Hercules will be a very nice unit for cruisers – at any rate, that’s our choice.

For the budget outlined in the article, a solid AP can be sourced. In this bracket, B&G seem to have the edge. At the high end, NKE, Madintec and the top offerings of B&G are quite a bit more. At the low end, I’d exclude Garmin and Raymarine: Just spent 500 NM on an all Garmin ship – we ended up steering almost all by hand. The owner has Raymarine on his other ship and does long-distance double-handed racing with it; he feels hohum about it – they helm whenever possible.

On the B&G, just as was the case with the H5000, the Hercules units seem to be more or less the same boxes, with features unlocked as the price goes up. Hard to go wrong here considering that the Hercules is a little over 3K and the WTP (as has just been installed on Pip Hare’s foiling TP60 Medallia for the next Vendee Globe) is about 8.5K.

Concerning AP vs. human – that’s really a moving target, depening on how good the helm is, how tired and how much money one is willing to sink into the AP.