The inner life of an autopilot

Self-steering is a cornerstone of blue-water travel, and it’s safe to say that quite a few long-term sailors would not be voyaging at all if manual helming were the only option. Mechanical wind-vane self-steering systems are traditionally favored for ocean sailing because they operate independently of the vessel’s electrical system and have a reputation for reliability. On the downside, transom-mounted vane systems are heavy, vulnerable and capable only of steering at a pre-set angle relative to the apparent wind. By contrast, electro-mechanical autopilots tend to be smaller, less obtrusive and offer a variety of features and operating modes. Some are also a lot less expensive than high-quality wind-vane systems (although the lower end of the autopilot market can be risky territory for those who are contemplating offshore travel).

Autopilot manufacturers, like Raymarine have made steady improvements in the processing capability, the software and the sensors providing position and attitude information to the control unit.
   Image Credit: Raymarine

In any event, autopilot popularity has skyrocketed in recent years. And while it's easy to recommend that the well-equipped blue-water sailboat carry both a wind-vane self-steering system and an autopilot, there are plenty of voyaging boats both sail and power making trouble-free ocean crossings with only an autopilot aboard. Going " autopilot only " is not a decision to be taken lightly, but it should work out fine if both the autopilot and the onboard electrical system that supports it are properly engineered, installed and maintained. In many cases, the price of a good, heavy-duty autopilot will be about the same as that of a good, heavy-duty vane steering gear. So if cost is a major consideration, there's little to favor one over the other.

An overview of current autopilot technology, as applicable to small and mid-sized seagoing boats, reveals that at the high end, the trend has been toward elaborate computerized autopilot controls that accept process input from a variety of sources beyond the usual magnetic compass and⁄or wind-direction sensors. These advanced systems are particularly desirable for specialized applications like polar expeditions and short-handed sailing in ultra-light sailboats. On the other hand, many offshore voyagers will be well-served by less elaborate autopilots often the straightforward, proven models that are favored for commercial workboats and large coastal motor yachts.

All modern marine autopilots have three main components: a primary directional sensor, a central processing unit (CPU) and an actuator to drive the boat's steering system. Nowadays, the most common primary directional sensor is a fluxgate compass, but conventional magnetic compasses are sometimes used, as are wind-direction sensors and gyrocompasses. The CPU is a dedicated microcomputer that determines an appropriate steering correction each time it recognizes that the vessel has strayed sufficiently far off course. The actuator is ordinarily either an electro-mechanical or electro-hydraulic device that physically moves the rudder blade to apply the steering corrections. Autopilot actuators

Of these three main components, the actuators have changed the least in recent years, so I will discuss them only briefly. For a more comprehensive treatment, see the Boatowner's Mechanical & Electrical Manual by Nigel Calder (International Marine).

Hydraulically-steered boats are often equipped with hydraulic autopilot drivers, consisting of an electrically driven, bi-directional hydraulic pump that operates the normal steering ram via the existing hydraulic lines. In essence, the autopilot actuator becomes the equivalent of a second helm station. This approach is straightforward and inexpensive, but any serious problem with the hydraulics is likely to disable both the manual steering and the autopilot.

As a general rule, it's preferable to have an autopilot actuator that operates independently of the manual steering. That way, at least one system will continue to function while the other is being repaired at sea. If you're seeking the redundancy of an independent autopilot actuator, the choice boils down to hydraulic vs. mechanical. Most mechanical arrangements involve motor-driven jackscrews, gearboxes, chain drives or belt drives. For small and medium-sized boats, electro-mechanical actuators are usually less expensive than stand-alone hydraulics. On the other hand, hydraulic actuators typically provide greater steering power, faster response and, in many cases, superior reliability.

It's common knowledge that belowdecks autopilots are a better choice for offshore use than compact cockpit pilots, which for the most part, are intended for light duty or weekend use. Pedestal-mounted wheelpilots obviously share an entire steering linkage with the wheel itself, eliminating all opportunity for steering redundancy. All the same, many long-distance voyagers and delivery skippers have achieved good results with the popular and inexpensive wheelpilots built by Raymarine/Autohelm and Simrad/Navico. These manufacturers also dominate the market for low-cost, self-contained tillerpilots at prices starting in the $500 range. For steering while under power or sailing in calm seas, even a tiny autopilot is often enough to steer a substantial offshore yacht. Just don't expect much when conditions get gnarly.

Regardless of type, autopilot actuators should be designed to meet the vessel's steering requirements, even in the most difficult conditions. Of course, pinning this down can get tricky, because steering requirements vary radically depending upon a boat's tracking characteristics, steering geometry, speed, rudder size, balance and other variables. Autopilot selection tables based on boat length and displacement are, at best, extremely rough indicators. As a rule, it's safer to buy a size or two bigger than the minimum estimate. Although a heavy-duty actuator may suck up 12 amps or more when it's working hard, most modern autopilots offer an array of settings so the power consumption can be minimized in more benign sea conditions.Autopilot compasses

Almost all autopilots appropriate for small and medium-sized vessels use either a fluxgate compass or, less commonly, a conventional compass to obtain directional information from the earth's magnetic field. This is true even with the growing number of autopilots in this category that are described as gyroscopically controlled. Solid-state rate gyros incorporated in these systems serve to compliment, not replace, the fluxgate compass. True gyrocompasses (let alone inertial guidance systems based on gyroscopic principles) are rarely found outside the realm of large ships and aerospace applications, although this could change.

The basic fluxgate compass is a surprisingly simple electromagnetic device that employs two small coils of wire to directly sense the horizontal component of the earth's magnetic field. To avoid inaccuracies created by the vertical component of the field, the fluxgate array must be kept as flat as possible by mounting it on gimbals or using a fluid suspension system. All the same, inertial errors are inevitable when the vessel is turning sharply or being tossed about by rough seas. To ensure directional readings that are adequately stable, marine fluxgate compasses always incorporate either fluid or electronic damping. For autopilot use, the downside of compass damping is a slower response and less accurate course keeping.

A conventional marine compass consisting of a permanent magnet suspended on a jeweled bearing is not inherently inferior to a fluxgate. However, it has fallen out of favor for autopilot use because it requires a secondary electro-optical reading device to input directional information to the system's CPU. Nevertheless, some authorities believe that for certain applications, notably steel sailboats that frequently operate at substantial heel angles, a high-quality magnetic compass may be easier to adjust and less error-prone than the average fluxgate.

Gyros for autopilots

Incorporating so-called rate gyros into autopilots has brought significant improvements on several fronts. Rate gyros are solid-state piezoelectric devices that generate (or modify) electrical signals in response to linear or angular acceleration. They can be mass-produced from specialized ceramic materials at a relatively low cost. Interestingly, the most widespread application of this technology is for image stabilization in video cameras.

In marine navigation, a rate gyro is most often used to detect and measure yaw (i.e., rotational acceleration in the athwartships horizontal plane). With this extra input, the autopilot CPU can signal a steering response that's proportional to the magnitude of the disturbance that is throwing the vessel off course. In other words, with a gyro involved, a large heading perturbation will elicit a stronger corrective action than a small one. Better yet, the gyro input enables helm corrections to be initiated much more quickly before the fluxgate compass would have even detected the off-course condition. The response time of gyro-equipped autopilots can be as short as 0.1 seconds, allowing the boat to sail a shorter course and reducing rudder drag for extra boat speed. Some manufacturers claim that adding a gyro reduces autopilot power consumption by minimizing the amplitude of rudder movements.

Fluxgate compasses and rate gyros complement one another nicely. The fluxgate provides a directional reference that's stable over the long term, and the gyro adds accurate short-term corrections for acceleration and heeling effects. At high latitudes, where the earth's magnetic field dips downward toward the magnetic poles, the rate gyro data can be used to correct for roll-induced heading errors in the fluxgate output. It can also be used to correct for the roll and heel-induced errors that often plague fluxgate compasses installed on steel vessels.

Rhode Island-based KVH Industries has pioneered the use of three-axis solid-state gyros to control satellite-tracking antennae for shipboard and land-based applications. They are also a major producer of gyro-stabilized fluxgate compasses, so it's no great surprise that the latest versions of their well-known TracVision marine TV/satellite dishes incorporate GyroTrac digital compasses suitable for autopilot use. Recently, a KVH unit aboard Steve Fossett's PlayStation helped this mega-catamaran follow a virtually perfect great circle route as it destroyed the existing trans-Atlantic sailing record.

A spinning gyroscope in a gimballed mount is the heart of a true-direction or north-seeking gyrocompass. Unlike an electronic rate gyro, these precision machines can supply accurate, long-term directional information to an autopilot. Unfortunately, they are too costly and power-hungry for all but the largest yachts.

In the future, a promising approach is likely to be a variation on the inertial guidance theme, using GPS positional updates to periodically correct DR errors. Aerospace inertial navigation/guidance systems use ultra-sensitive rate gyros to measure acceleration and deceleration in all six degrees of freedom. From this elaborate data stream, a powerful onboard computer continually dead reckons the vehicle's 3-D position and specifies appropriate course corrections. A marine autopilot control system would only need to deal with motion in two dimensions and could take advantage of GPS corrections to update its electronically derived DR plot. No fluxgate compass or other magnetic sensor would be required. Comparable military nav systems built by KVH and others are already in use, and their adaptation to the civilian marine sector is likely just a matter of time.

CPU sophistication

Most modern autopilots are designed to operate in several different modes, the most common being "compass," "wind direction" and "GPS." The latter is intriguing because it promises to eliminate the cross-track error caused by leeway or drift and because it enables the autopilot to follow a string of waypoints. However, the course-holding performance that could be achieved using GPS data alone would be unacceptably erratic. Instead, most autopilots operate in GPS mode using a combination of compass and GPS input.

Putting massive computing power into a device like an autopilot is no great feat in this day and age, but the jury is out when it comes to using this power effectively. Quite a few manufacturers have been working to automate the selection of operating parameters (gain, deadband, counter-rudder, etc.), so their autopilot will, in effect, learn to steer more efficiently with practice. For example, almost all contemporary autopilots have an auto-trim feature that shifts the null position of the rudder away from the midline to compensate for an ongoing steering bias, such as weather helm.

The much greater challenge is developing a pilot that can anticipate a course perturbation and forestall it with an appropriate helm action. Even more ambitious is the goal of steering an optimal course (as opposed to a straight-line course) to take full advantage of puffs and lulls, dodge breaking wave crests and so forth. Although genuinely proactive, autopilots are still only a pipe dream; the general approach of adding more sensors and crunching more data appears to be bearing fruit. For example, recent experimental autopilots by NKE Marine Electronics and Brooks & Gatehouse have incorporated the option of steering to the true wind angle (computed from apparent wind angle, apparent wind speed and boat speed). Using an autopilot that's referenced to apparent wind angle can lead to serious problems when a fast sailboat, such as an Open 60, careens down a wave face. By steering to a true wind reference, these experimental autopilots avoid the common problem of bearing away excessively as the apparent wind swings forward, leaving the boat at risk of an accidental gybe as it slows abruptly at the end of the surf. Setting the autopilot to follow a compass course the usual alternative to apparent wind steering has also proven unsatisfactory under these circumstances because it opens the door to a round-up or worse, a crash gybe, in the event of a wind shift.

At the cutting edge of autopilot development is a function known as polar sailing essentially a control program that directs the pilot to seek pre-selected speed or apparent wind angle target values. W-H Autopilots, a small but highly respected manufacturer in the Seattle area, is among the first to offer this feature as an off-the-shelf product.

The W-H system employs software developed by the legendary ocean racing navigator Stan Honey. It requires accurate wind-speed instrumentation and reliable polar data for the boat (derived either by testing and experience or computer-generated performance predictions). Going upwind, the user inputs a target speed appropriate to the wind strength and sea conditions. The autopilot then seeks the target speed, heading up gradually until the speed drops below target, then the unit bears off slightly. Gybing downwind, the approach is similar in principle, but in this case the autopilot keys to an optimal apparent wind angle, heading down in puffs (to reduce distance sailed) and up in the lulls (to maintain speed). As racing sailors know, sailing the polars is dramatically more effective that simply attempting to maintain a constant compass course or apparent wind angle.

The French firm NKE is heavily involved in professional short-handed racing and was among the first to incorporate rate gyros in their CPUs. A recent NKE innovation was the additional speedometer input to reduce autopilot gain (rudder deflection) and hence the risk of oversteering while surfing down waves. As a fail-safe, NKE's current Gyropilot 2 is programmed to disregard excessively abrupt changes in speed- or wind-direction data which might otherwise lead to smash-ups should the speedo or masthead sender fail.

Another NKE innovation of particular interest to voyaging sailors as well as racers is a wireless autopilot remote control that incorporates several man overboard functions. If a single-hander falls overboard with the remote in hand, the interruption of a short-range watch-dog signal will prompt the autopilot to turn the boat head-to-wind. Aboard a crewed boat equipped with appropriate NKE electronics, anyone who goes over the side carrying one of these remotes will promptly activate an alarm, trigger the autopilot-controlled stop maneuver and initiate an automated dead reckoning procedure to facilitate a prompt recovery of the victim.

Selecting an offshore autopilot

Anyone who's read this far will certainly realize that choosing an autopilot is far from a simple task. Luckily, there's a wealth of information available through manufacturers, as well as anecdotal reports from fellow sailors on the Web and in person.

In most cases, the average voyaging boat one that tends to plod rather than skim across the seas will be less demanding when it comes to autopilot controls than an ultra-light ocean racer. But by the same token, it should be obvious that a sophisticated autopilot running appropriate software has great potential for improving the comfort, safety and passage times of almost any boat.

The latest generation of high-powered inboard autopilots is not ideal for every voyaging boat. For one thing, they typically consume anywhere from four to 15 amps at 12 volts (depending upon the particular gear, the boat and the sailing conditions). Sailboats with limited generating capabilities may simply not be able to cope with this sort of draw, day after day, for weeks on end. An autopilot is totally dependent upon a sound, reliable electrical system. If the combined cost of both a good autopilot and the electrical upgrades needed to support it would be unaffordable, then its probably better to spring for a quality wind-vane steering system and perhaps a light-duty cockpit pilot that can steer while motoring in calm weather.

In closing, it's worth mentioning an alternative that can sometimes deliver the advantages of both wind vane and autopilot at a comparatively affordable price. This is the hybrid system typically a small, self-contained tillerpilot that can be mounted in place of the wind vane on a commercial self-steering gear. Using the self-steering system as a power-steering device can be an effective way to steer a relatively large boat using a pint-sized autopilot. Most existing hybrid systems are home-built. However, the notion of using a servo rudder or rudder-mounted trim tab to harness the water flow and multiply steering force is a sensible approach that probably deserves more attention from autopilot makers.

Contributing Editor Sven Donaldson is a marine technical writer and former sailmaker based in Vancouver, British Columbia.

By Ocean Navigator