Lessons from an Autopilot failure

Jungle of hydraulic hoses surrounding the AP drive unit; note the confined space!

​Where did it all begin?
Well, that is a very good question. I recall writing in one of my first blog articles, going back to our Biscay crossing, that the AP sounded like a sick hydraulic wildebeest calling its young. It was loud! It was constantly working back and forth and made it difficult to sleep in our starboard bunk (the autopilot motor is just behind our bed head). You could even hear it over the top of the starboard engine and creaky window when lying in bed and it is a sound you really fixate on. We assumed this was normal though, and maybe it was. It’s a brand new boat so there couldn’t be anything wrong with it, right?! Also, we had nothing to compare it to. But looking back now, we wonder if everything really was perfect.
 
After not missing a beat since we started cruising in August 2018, the first sign of a glitch in our autopilot (AP) was on the way from Gibraltar to the Canaries. In relatively benign conditions, It just quietly, and without any warning or error messages, still displaying “Heading Hold” on the GHC20, the AP stopped holding a course and started wandering off in an alternative direction. This rather surprised the person on watch who quickly hit “Standby” and hand steered back onto course before re-engaging heading hold. Hmmm...

This happened again a couple of times on our trip down from The Canary Islands to Cape Verde – the AP was having to work very hard in the rough conditions, with the confused sea and 3 different wave directions constantly knocking us back and forth, smacking us on one hull and then the other. We blamed the conditions for the issue as we have never been in such bad sea state before. A couple of times we even restarted the instruments (the old “turn it off and turn it on again” trick), which seemed to help but in hindsight was probably coincidence. Then on our Atlantic crossing, about halfway across, it started happening with increasing frequency. Several times during every 3 hour watch was becoming the norm, which did give us all some practice at hand steering. Getting the AP to re-engage was not always straightforward either, hand steering in rough conditions, manually holding a course sometimes without a visual datum and often trying 2-3 times before it would engage in heading hold. Finally on our last day, after one last gasp mid-morning, the autopilot gave up completely and we had to hand steer the final 20 hours to Barbados. Gruesome! But at least it happened on the last day and not earlier!

Needless to say the manual did not help...

Troubleshooting
Once we got to Barbados we spent a lot of time on the phone to Garmin US. They were really good, going through all sorts of trouble shooting, recalibration and eventually agreeing to replace our heading sensor (Reactor 40) under extended warranty (thanks Covid!). Getting it to us would be the difficult part, but as luck would have it Keith’s niece was coming over to St Lucia from the UK and, long story short, we were able to get her to bring a replacement unit with her (thanks to Garmin UK for stepping in). This meant hand steering from Barbados to St Lucia, however this was not too bad in the end as conditions were much smoother and we were prepared for it. Sadly though, replacing the part did not immediately solve our problems. We fitted it and promptly performed all of the calibration wizards but alas, still not holding a course. The course computer was simply not engaging with the AP motor which drives the hydraulics and moves the rudder. The only good news that came out of this exercise is that we finally got an error message – Drive Unit Overload – this (eventually) turned out to be GOLD!
 
More time on the phone to Garmin and they had agreed to replace the ECU, the computer that drives the hydraulics. Having failed to solve our problem so far though, we were beginning to wonder though whether electronics was indeed the issue, and again we have the hassle of getting a part to us with no fixed address. Garmin US did not even have an agent listed in the Caribbean at all. We decided to head to Le Marin in Martinique where there are a huge range of yachting services available, including a company called DigiNav, that had great reviews on NoForeignLand.

---

Happy to recommend Jacques at Diginav in Le Marin

An answer at last!
We made contact with Jacques at DigiNav shortly after arrival and as soon as he saw our error message and heard the background he immediately identified the problem. The drive motor was overworked and had worn out. He told us (well Keith actually) how to remove the drive motor for the autopilot and lent us a couple of tools that we would need to do it. He explained what he expected to find and how he would fix it for us when we brought it in to him. He inspired confidence from the start, although we still had some “homework” to do ourselves.  This would probably be a good learning experience.
 
Keith set to the task of removing the drive motor, which was no easy feat, but he was up for the challenge despite having never touched hydraulics before. In the first place, getting to the motor is very difficult. It is in the starboard engine bay, on a shelf forward of the engine with everything you need to access facing forward (ie away from you) and its surrounded by stiff hydraulic hoses. Oh and it’s up against the bulkhead at the top so you really have limited space to work with. We are wondering if they built the entire boat around this particular part!

The old and new. Note elongated hole in the old part and gouged pin!

The steps for extraction went something like this

1.Disconnect the tiller cross bar, to give you access to the electric motor behind the hydraulic pump, leaving it to swing back and forth as the boat moved at anchor, while you continue working around it.

2.Unbolt the whole drive unit (motor & pump) from the shelf in the engine bay to manipulate it into a position that you can access the 4 bolts holding it the two components together.

3.Unbolt the motor from the hydraulic pump. To do this you need a special ring spanner ground down on one side to fit the narrow space available to access the bolts (Jacques loaned us his). Oh and the hydraulic pump has 4 very stiff hydraulic hoses coming into it restricting access to the bolts.

4.When removing the electric motor, which has the connection to the hydraulic pump’s impeller, you need a plug to stop the all the hydraulic oil draining out of the hydraulic system where you have just removed the motor from. Jacques said to expect a small amount to come out, but it felt like we lost a couple of litres at least! The engine bay resembled an episode of MASH, with the blood soaked surgeon calling for more paper towel!

5.The final step in removing the electric drive motor was to disconnect the two power supply wires. Usually these need to be cut to as they should have been crimped firmly in place, however one of them (the positive) slid cleanly out of its crimping sleeve. It had never been crimped properly. It was a smaller diameter to the wire it was connecting to, and the crimping sleeve was for the larger size wire. By contrast, the negative wire needed to be cut to remove it. Could this “loose” wire have been the source of or at least a contributor to our problems? Intermittent loss of power while it worked its way loose would be consistent with the pattern of behaviour we experienced as it failed.

​6. Once the motor had been disconnected it was obvious that there had been a lot of strain on the drive arm as the sleeve that has a cotter pin holding it in was very badly elongated and the pin itself was heavily gouged and nearly sheared through. This would have made the autopilot work even harder, as there would have been a lot of slack between each movement.

Drive unit extracted, and loose negative wire

Feeling as if he had just extracted a living organ from a reluctant donor, Keith proudly presented the drive motor to Jacques for him to perform further surgery, or some kind of miracle. As soon as Jacques saw the worn sleeve and cotter pin and heard about the lose wire, his suspicions about our problems were confirmed. Of course the brushes on the motor were also shot from working too hard and would also need replacing. With the vital organs now in the workshop we are in Jacques capable hands! He would rebuild our drive motor!
 
It was a nervous wait over the weekend and although our anchor was well dug in after a week in a very protected and calm anchorage (actually a hurricane hole) it’s unsettling to know that you have no steering and if you dragged anchor, resetting it would be interesting. It would also be difficult to take evasive action should one of the thousands of other boats anchored in the bay, to windward of us, dragged towards us.

Tiller arm and hydraulic ram

Re-implantation
A few days later Keith dropped in to pick up our AP motor, duly rebuilt and ready to be re-installed. The photo above shows the old and the new parts for comparison.
 
To re-install the drive motor, repeating the steps above, but in reverse order, with a slightly different combination of swear words, and haemorrhaging of another couple of litres of hydraulic oil. We refilled the hydraulic reservoir and tried to move the oil through the system but had missed the vital step of opening the bypass lever, so the rudders were stuck on one side. Fortunately Jacques was coming to the boat to rescue us the next day! He showed us the bypass, quickly got the fluid moving and most of the bubbles out, but there was still some air left in that was causing a problem. He explained what we needed to do and later provided the poly tubing we needed after we dropped him back to shore.
 
We started early the following morning with the tubing leading from either end of the hydraulic pump into a jar half full with hydraulic oil. Keith held this as I moved the steering wheel back and forth so the oil (and air bubbles) flowed out into the jar. We could then refill the reservoir at the helm station with it. We repeated this process 3 times until we were happy that all the air was out.

Hydraulic oil reservoir behind the helm station

And now finally the moment of truth… Oh wait no, one more thing
Jacques had noticed a bit too much lateral movement in the ram of the autopilot, 2-3mm is tolerable but we have over 1cm! He suggested that the nylon bearings on the universal joint are probably worn and that these needed replacing, this along with residual air in the hydraulics would be causing the excess movement. He provided the part and instructions. He has a lot of faith in Keith’s mechanical prowess by now!
 
Needless to say replacing this bearing was not a straightforward task either. Again, access to the part was hampered by the jungle of stiff hydraulic hoses surrounding it. The entire unit had to be removed so that the top half of the flange could be accessed from under the hydraulic hose fitting. The bolts holding the flange were way too long and had to be ratcheted all the way out to undo them (ie could not be hand turned once loosened). Then of course this needed to be reversed to put everything back together. Unfortunately, Keith cross threaded one of the bolts in the aluminium flange in the reinstallation process and had to pause the process, go ashore and drop it into a workshop to get it re-machined before continuing on. In re-fitting it he decided to reverse the direction of the bolts to make them easier to get out if we ever had to do this again (heaven forbid). Now that all of this is done, however, there is absolutely no movement in the arm! Jacques is impressed.

Putting it all together
Finally we take the plunge and lift the anchor, heading out of Le Marin with our refurbished drive unit and a fresh carpet of green growth decorating the hull and the anchor bridle. We go through the now familiar Garmin calibration exercises which involves setting the compass by turning in circles, setting the rudder gain and counter gain by doing a series of zig-zags. Once that is complete we check the AP responsiveness with some course changes to ensure she goes straight to the heading and stays there. A few days latter we reset the rudder sensitivity to lower it, so the rudder is not working too hard.

Root Cause Analysis
So in true “Air crash investigation” style we are keen to know what caused the problem? Was there a single issue/error/mistake that had a series of knock on effects? Or, as is often the case, were multiple factors at play?
 
A few options to consider
 
1.The loose wire to the AP motor has to be a prime consideration. Losing power intermittently would have explained the loss of heading hold without error message. It would also lead to the motor working too hard.

2.Air in the hydraulic system – many people have this issue and we may have as well.

3.Rudder settings – we had never adjusted the gain and counter gain or rudder sensitivity/responsiveness from the factory settings. Once we have completed the auto-calibration we set the rudder responsiveness to the lowest level. The difference is noticeable, the autopilot is working a lot less now but still functioning quite well in terms of holding us on course. Was the higher sensitivity setting working the hydraulics too hard? Particularly in the rough, cross current of the ocean crossing.

4. Rough sea conditions on the crossing from Canaries to Cape Verde and on to Barbados making the AP work too hard?

5. Other (intermittent) electronics issues or glitches with the ECU and/or CCU?

What do you think? Leave a comment if you think of any other possible causes.