8 things I wish I’d known about the B&G H5000 Autopilot

Me during the Bermuda 1-2, just south of Nantucket. Tricky conditions for any autopilot.

After more than 7 seasons with the H5000 across two boats, countless hours on Sailing Anarchy threads, B&G’s own advanced training materials, and conversations with people who set these systems up for a living, I’ve collected a set of insights that would have saved me a lot of time. Many aren’t obvious from the manual. A few I had to learn the hard way.

Written from the perspective of doublehanded racing on a 33-36 ft performance yacht (J/111, now J/99), but the principles apply to any H5000 installation where the autopilot matters.

Before anything else: update your software. Running the latest H5000 Pilot Computer software (currently version 2.0.0.2) is essential for getting the best performance out of the system. Many of the issues people still describe online were addressed in this release. Download it here, and read point 3 below carefully – there’s one setting you need to disable after the update or you won’t see the improvement.

1. There are actually two different pilots inside your H5000

This is the single most important thing to understand, and it isn’t always obvious from the documentation.

Perf 1 is essentially the same algorithm as B&G’s standard cruising pilot, the NAC-2. No sailing-specific intelligence. It behaves like a classical PID controller, which means three things happen at once: Rudder Gain (proportional) sets how much rudder per degree of heading error. Big error, big correction. Counter Rudder (derivative) watches how fast the error is changing and brakes the turn before you overshoot. Think of it as anticipation. AutoTrim (integral) slowly learns a steady-state rudder offset to compensate for persistent weather helm. Cruising Speed scales all of this by boat speed (see point 4). That’s the whole algorithm. And that’s what AutoTune calibrates.

Perf 2-5 gradually blend in something different: the “Performance Sail” algorithm, which uses an internal mathematical boat model with fixed gain tables you can’t see or adjust. The blend isn’t linear either. Experienced users describe it as more of an exponential curve than a volume dial: the step from Perf 2 to 3 is small, but from 4 to 5 it’s large.

Why this matters: AutoTune mainly calibrates the Perf 1 algorithm. If you run Perf 3 (which handles most conditions on a boat our size), your Rudder Gain setting has diminishing effect. If the pilot steers poorly at Perf 3 and you start tweaking Rudder Gain, you’re adjusting a parameter the system is mostly ignoring. The problem is almost certainly somewhere else.

On top of the steering algorithms sit the Expert Systems, and understanding what they do is just as important. Gust Response and TWS Response don’t change how the rudder is controlled. They change the target TWA that the pilot is aiming at. When a gust hits, Gust Response shifts the target a few degrees lower (bear-away). When the gust passes, it slowly returns. TWS Response does the same thing but for sustained wind increases, and it stays low as long as the wind stays up.

The exception is Heel Compensation, which acts directly on the rudder as a fast-learning weather helm compensator. It tracks average heel and reacts to rapid increases. Think of it as a much faster version of AutoTrim that responds within a gust cycle rather than over 60 seconds.

For troubleshooting: if the boat bears away too much in gusts, look at Gust Response gain before Rudder Gain. If the boat rounds up in gusts, look at Heel Compensation before the Perf level. Wrong layer, wrong fix.

This is also the principle behind products like Pixel Sur Mer’s Exocet system, which takes external target management much further. Their overlays can steer to heel angle, target speed, or polar-optimised wind angles, using data the H5000’s internal systems don’t have access to. Pixel Sur Mer have been B&G’s development partner for many years and the collaboration has improved the H5000’s own algorithms. If you’re racing at a level where the built-in Expert Systems aren’t enough, their Exocet Essential is worth looking at.

2. Don’t turn everything on

It’s tempting to enable every feature; Gust Response, TWS Response, Heel Compensation, Auto Response, Recovery Mode, because they all sound useful. Don’t.

Start with the basics: Perf 3, Wind Mode Auto, everything else off. Sail the boat. Understand how it steers with just the core algorithm. Is it holding course? Tracking through tacks? Handling waves?

Then add features one at a time. Enable TWS Response first (generally more useful than Gust Response for sustained conditions). See if it helps. Then try Heel Compensation. Then Gust Response. Each time you add a feature, sail enough to understand what it does to your boat in your conditions.

The reason this matters: the Expert Systems interact. Gust Response and TWS Response can fight each other if their gains overlap. Heel Compensation at too high a gain looks exactly like oscillation from too much Rudder Gain. If you turn everything on at once and something goes wrong, you have no idea which feature is causing the problem.

B&G’s own Level 300+ training material says the same thing: only activate advanced features one at a time.

3. Turn some things off

Two settings should be off by default for racing, and one commissioning step is easy to overlook.

Turn off Adapt — especially after the 2.0.0.2 software update. This is the single most important setting change you can make, and the one most people miss. Adapt is a background learning function that continuously refines the pilot’s understanding of how your rudder translates to turn rate. The 2.0.0.2 release came with a Technical Bulletin specifically calling out that Adapt should be disabled after the update, but in practice most owners haven’t picked up on it, and they don’t see the performance improvement the new software is supposed to deliver. If you’ve updated and feel like nothing changed, this is almost certainly why. Go into the Pilot menu and turn Adapt off. Re-run AutoTune any time your boat’s characteristics change significantly (different keel position, rig tune, bottom state).

Turn off leeway-corrected wind for the pilot. The H5000 can include leeway in its TWA calculation, which is great for navigation displays and tactical decisions. But for the pilot it can create a feedback loop: leeway changes with heel, heel changes with gusts, so if the pilot steers to a leeway-corrected TWA, every gust shifts the target angle. The pilot chases this moving target, producing constant rudder activity and unstable steering. Keep leeway correction on for your displays and Expedition, but verify it’s off for the pilot input via the H5000 web interface.

Save Rudder Drive settings before commissioning. This step is documented in B&G’s Level 300+ training material but easy to miss. Before starting Dockside Commissioning or the Rudder Test, you have to visit the Rudder Drive settings page and press Save, even if all settings are already correct. Otherwise your settings aren’t stored, and downstream steps (Rudder Test, AutoTune) may not behave as expected. I suspect this explains some of the “I can’t get my pilot to work” reports on forums where people have done everything right but the system refuses to cooperate.

4. Cruising Speed is not what it sounds like

The name is misleading. Cruising Speed is the reference point around which the PID controller scales its response. Below this speed, the pilot applies more rudder (the boat is sluggish, needs more authority). Above it, less rudder (the boat is responsive, needs less). It functions as a gain reference more than a speed setting.

B&G’s guidelines: 5-6 kn for 10m, 7 kn for 12m, 10 kn for 20m. In practice, many racing setups run higher; a SunFast 3300 (10m) uses 8-12 kn. Our J/111 ran 6-10 kn. Experiment, but know what you’re actually changing. It also serves as the failsafe speed if both BSP and SOG fail, so don’t set it to something absurd.

5. The pilot watches you steer for 30 seconds before it takes over

When you press AUTO, the pilot doesn’t start from zero. It samples your average rudder angle over the preceding 30 seconds or so and uses that as its starting weather helm offset. That becomes the initial value for AutoTrim.

Practical consequence: steer straight and steady for at least 30 seconds before engaging. If you hand over during a luff, a course change, or while fighting a wave, the pilot starts with a wrong baseline. It will steer poorly for up to 60 seconds while AutoTrim re-learns the correct offset. On a doublehanded boat where you engage the pilot to go do something on the foredeck, those 60 seconds matter.

6. AWA isn’t always best upwind

The conventional wisdom (and B&G’s default Auto mode) is AWA upwind, TWA downwind, switching at 70° TWA. Works well in moderate conditions. But there are two situations where TWA can be better upwind.

Light air (under 8-10 kts): AWA fluctuates constantly with every puff and lull. The pilot chases these fluctuations, producing constant rudder activity that costs speed. TWA is more stable, giving smoother steering. Some experienced pilots use TWA upwind most of the time for this reason.

Heavy air (18+ kts, TWA 55-62°): when sailing wider angles for safety and speed, TWA provides more stable target tracking and prevents the pilot from pinching in lulls.

Our default remains Auto mode, which works well for the J/99 in 10-18 kts. But it’s worth experimenting at the extremes.

7. Check your hard-over time

One setting that doesn’t get much attention: how long it takes the rudder to swing from full port to full starboard. B&G’s spec is 12-15 seconds for 30-90 foot yachts, never above 25 seconds. Performance boats are usually toward the lower end.

Several owners on Sailing Anarchy have reported that their pilot steered noticeably better after slowing the hard-over time to around 12 seconds, especially when it had been set faster. The mechanism isn’t obvious; the drive speed is something the PID values are implicitly calibrated against, so changing one without retuning the other might be part of it. Either way, worth knowing your number.

To check yours, put the pilot in NFU (Non Follow Up) mode, drive the rudder fully one way, then start a stopwatch as you drive it the other way. If you change the setting, re-run AutoTune afterwards.

Thanks to the UK Double Handed Offshore Series and Matt Eeles at B&G for letting me tag along on their H5000 training session in Southampton.

8. Troubleshoot in layers, not parameters

Think of the pilot as a stack. Hardware at the bottom, target intelligence at the top:

• Layer 5: Expert Systems (Gust/TWS Response, Heel Compensation)
• Layer 4: Steering algorithm (PID at Perf 1, PS algorithm at Perf 2-5)
• Layer 3: Mode and target (compass, AWA, TWA)
• Layer 2: Sensors (BSP, heading, wind)
• Layer 1: Drive (motor, ram, rudder feedback)

When the pilot misbehaves, resist the urge to jump straight to Rudder Gain. Work from the inputs upward instead.

Start with the sensors. Is BSP correct? Compare to SOG in zero current. Is heading calibrated? Is wind data trustworthy? A fouled paddlewheel produces the same symptoms as bad PID tuning.

If sensors check out, look at mode and target. Are you in the right mode? Is TWA calibrated for the current conditions?

Then look at the Expert Systems. Are Gust Response and TWS Response interacting badly? Is Heel Compensation gain too high? That looks exactly like oscillation.

Only then consider the steering algorithm itself: Rudder Gain and Counter Rudder. And only if you’re running Perf 1-2. At Perf 3+, these have limited effect.

Finally, check the drive. Is Motor Output sufficient? Is the ram hitting its end stops? Is battery voltage holding up under load?

The most common mistake I’ve made is adjusting the steering algorithm when the real problem was a sensor. Fix the inputs first.

If you have experience with the H5000 pilot, especially on 30-40 ft performance boats raced doublehanded, I’d love to hear what works for you. What Perf level do you run? Do you use Gust Response or TWS Response? What was the single most impactful adjustment you made? Comment below.

Thanks to Matt Eeles at B&G for input and sharing the field experience that shaped much of what’s here.