Using it well · 7 minute read
Marine autopilot troubleshooting
Symptom by symptom autopilot troubleshooting: S-turns, sluggish steering, drive alarms, heading errors, network dropouts, low voltage and wind mode faults.
Most autopilot faults are not mysteries. They fall into a handful of patterns, and each pattern has two or three likely causes that can be checked in order without special tools. The trick is to resist the temptation to change six settings at once, because then you learn nothing.
Work symptom by symptom, change one thing at a time, and write down what you did. This guide is organised that way: what you see, what usually causes it, and what to do about it, with a note at the end on the faults that are worth handing to an installer.
Key facts
- First check
- Battery voltage at the drive, under load, not at the panel
- S-turns
- Usually rudder gain too high or counter rudder too low
- Sluggish
- Gain too low, worn drive, or voltage drop in the cable
- Dropouts
- Backbone termination, power injection or a corroded tee
- Rule
- Change one setting at a time and record it
- Hand over
- Hydraulic air, drive alignment and repeated overload alarms
The boat S-turns or hunts around the course
A pilot that weaves side to side is over-correcting. The usual cause is rudder gain set too high, so every small deviation produces too much helm, which overshoots and starts the cycle again. , which is the setting that takes the helm off before the boat arrives back on course, is the second half of the same problem. Too little counter rudder and the boat sails through the course every time.
Reduce rudder gain one step and watch for a full minute before changing anything else. If the amplitude of the weave reduces but the boat still overshoots, add . If reducing gain makes the boat lazy and the weave becomes a slow wander instead, you have gone too far.
Sea state matters. A boat that tracks perfectly upwind and S-turns dead downwind in a following swell is normal behaviour at the wrong settings, not a fault. Most systems have a response or sea-state level for exactly this, and using it is faster than retuning the gains.
- Reduce rudder gain one increment, observe for at least a minute
- Increase counter rudder if the boat still overshoots the course
- Drop the response or performance level one notch in a seaway
- Check for a rudder reference that reads non-zero with the rudder centred
- Confirm the boat is not badly out of sail trim before blaming the pilot
The boat wanders, responds late or will not hold a course
Sluggish steering is the mirror image. Gain too low is the easy case: the pilot applies too little helm and the boat falls off before it corrects. Raise rudder gain one step at a time until the boat holds the line, then back off half a step if it starts to twitch.
If more gain does not help, the drive is the suspect. A slipping belt on a wheel pilot, a worn , or air in a hydraulic system all mean the pilot commands rudder it never gets. Watch the rudder angle display while the pilot works: if the commanded and actual angles diverge, the fault is mechanical, not a setting.
Low voltage produces the same symptom and is easy to miss. Measure at the drive terminals while the pilot is working hard, not at the switch panel at rest. A drive seeing a volt less than the battery has lost a significant fraction of its power.
Drive stop, overload and no-rudder-reference alarms
A drive stop or overload alarm means the system asked for movement and did not get it, or drew more current than it expected. Check the obvious first: something jammed in the quadrant, a line around the rudder, a ram at the end of its stroke before the rudder reached its stops, or the emergency tiller still fitted.
Repeated overload alarms with nothing obstructing the steering usually mean the drive is undersized, tired, or the boat is badly overpowered. A yacht carrying twenty degrees of weather helm will overload a correctly sized pilot. Reef first, then judge the pilot.
A no-rudder-reference alarm on a system that needs one points at the sensor, its linkage or its cable. Check the linkage has not come adrift, then the connector, then the reading on the control head as you move the helm slowly. A reading that jumps or sticks is a failing sensor.
Never silence a repeated alarm and keep sailing
Heading errors and a pilot that fights the plotter
If the pilot holds a steady heading but it is not the heading you asked for, or the plotter and the pilot disagree, the heading sensor is the place to start. Magnetic sensors are affected by anything ferrous or electrically noisy nearby: an engine, a speaker magnet, a battery bank, a steel toolbox, a laptop left on a shelf, or a new cable run past the sensor.
Check what has changed near the sensor, then re-run the calibration swing to the manufacturer's procedure in flat water with room to turn. On most modern systems this is a slow circle or two and takes minutes.
Remember that heading and course over ground are not the same thing. In the East Australian Current a boat can point at 190 degrees and track 197 over the ground, and nothing is faulty. If the two differ by a consistent amount, suspect set and drift or leeway before you suspect the compass.
Network dropouts and data that disappears
Intermittent loss of wind, position or heading data on the pilot almost always comes back to the network rather than the pilot. An backbone wants a terminator at each end and only one, a single power injection point, drop cables kept short, and clean connectors. Extended backbones with two power feeds or a missing terminator produce exactly this kind of ghost fault.
Work through it physically. Check both terminators are present and at the ends, check the power feed, then unplug and inspect each tee. In the tropics, a corroded tee in a damp locker is the single most common cause of a pilot that loses heading for a second at a time.
If the dropout happens when the engine starts or the windlass runs, it is a power quality problem. Look at the course computer's supply and earth, the cable size, and whether the pilot shares a circuit with a high-current device.
- Confirm exactly two terminators, one at each physical end of the backbone
- Confirm exactly one power injection point on the backbone
- Unplug, inspect and reseat every tee and drop cable connector
- Check the course computer's main supply and earth terminals for corrosion
- Note whether dropouts correlate with engine start, windlass or thruster use
- Check that no two devices are sending conflicting data of the same type
Wind mode wandering, and when to call an installer
A pilot steering badly in wind mode when it steers well in compass mode is a wind data problem nine times out of ten. A masthead unit with a worn bearing, a bird-damaged vane, a dirty transducer or too little damping sends the pilot chasing noise. Watch the angle on the display in steady conditions: if it jitters several degrees, the pilot is being told the wind is jumping around and is steering accordingly.
Increase wind damping, service or replace the masthead unit, and confirm the pilot is using true or as you intend. Downwind in a swell, most boats hold better on a angle if the system supports it.
Hand the job to an installer when the fault is mechanical or structural: air in the hydraulics you cannot bleed out, a drive that needs realignment to the quadrant, mounting hardware that is flexing, or repeated overload alarms on a correctly trimmed boat. Those are safety items, and they are also the ones where a wrong guess is expensive.
Questions
Frequently asked
Why does my autopilot steer well upwind and badly downwind?
Downwind in a following sea the boat yaws with the waves and needs quick, small corrections rather than the firm helm that works upwind. Most pilots have a response or sea-state setting for this, and dropping the performance level one notch usually cures it. If it persists, the drive may be too slow for the boat.
What causes an autopilot to suddenly lose heading?
Almost always a data or power fault rather than a broken compass: a corroded network connector, a missing or doubled terminator, or a voltage dip when a high-current device starts. Check the physical network before replacing anything. Heading loss that only happens on one tack points instead at the sensor's mounting or nearby magnetic interference.
Is it safe to keep sailing with the pilot alarming?
No. An overload, drive-stop or low-voltage alarm means the pilot cannot reliably steer. Take the helm, reduce sail if the boat is overpowered, and diagnose before re-engaging. Alarms that are muted and ignored are a common thread in shorthanded incidents.
How do I tell a tuning problem from a mechanical one?
Watch the commanded rudder angle against the actual rudder angle while the pilot works. If the rudder goes where the pilot asks and the boat still steers badly, it is a tuning or sea-state problem. If the rudder lags, falls short or does not move as commanded, the fault is in the drive, the linkage or the hydraulics.
Can low battery voltage make an autopilot steer badly?
Yes, and it is underdiagnosed. Drives lose power and speed as voltage falls, so the pilot commands helm it cannot deliver. Measure at the drive terminals under load rather than at the panel, and check cable size and terminal corrosion before blaming the pilot.
Practise autopilot operation with an instructor
Habits are learned at the helm with someone watching. The training section explains what a good instructor covers and where to find one in your state.