Fluxgate compass wrong even after recalibration

Any fluxgate compass gurus out there? I have a B&G setup. The main compass is a Precision 9. Back in late November I noticed when coming into Batemans Bay that the compass heading did not chime with visual ref points by about 20 degrees. I'd updated the chartplotter software as we were coming in once we got within phone range, as I was sick of the nag screen, which frequently reminded me there were updates available. When I contacted B&G support, they said the problem was a bug between the updated chartplotter software and the autopilot computer software, which needed to be updated also, requiring a microSD card to load it after download on a laptop. Went through all that, and a recalibration of the compass - slow 390 degree turn - and a recalibration of the autopilot, first dockside then on the water. Problems seemed to be fixed after one more compass recalibration. Heading agreed with the Square Head lead light bearing.

Next day, coming into Eden, compass heading was off again by about twenty degrees. When heading by eye into Twofold, the compass heading thought the boat was pointing south of Boyd tower at one point.

So an intermittent error of about 20 degrees.

Currently down in Tassie, and have been dealing with local certified B&G chap. Of course the first suspicion was stray metal near the compass. So shifted every possible culprit at least half a metre away from compass. Did multiple recalibrations, but compass heading would drift off not long after. So that wasn't the problem.

Next, added an extension lead and repositioned the compass away from any possible electrical interference, re-did calibrations, but problem remained.

In the last week or so it has got much worse. To the point where I have apparently mastered the art of sailing almost backwards. On a short trip under motor, compass insisted boat was headed 150 degrees off the actual heading. Needless to say this mucks up all the heading dependent information. So wind direction display is nonsense, current is nonsense, and of course the autopilot is unusable.

Next step is to replace the compass. Currently awaiting arrival of replacement. My concern is that I don't know why the compass has developed an error. And if there is an external cause, will I have a repeat problem with new compass (they aren't cheap, and they are central for the navigation system). Two possibilities I can think of is a broken winding in one or more of the nine coils used to measure the magnetic field. Another is that the EEPROM, which I assume is how the calibration is stored, has stopped working.

If so, new compass should fix the problem. But leaves the question as to whether it was just an unlucky dud compass (it's only ~2 years since installed by PO) or something has fried its innards. I have fired up the inverter, the HF and VHF radios and used a gauss meter app on an iPhone to see if there are any stray fields coming from wiring and aerial cables that are nearest the compass but have not detected anything out of the ordinary.

Is the 30 degrees deviation constant ?

I dont suppose you have a metal boat?

Turns out it wasn't the compass. It was the 7 amp current to the electric fridge compressor running through wires in a conduit pretty much next to the compass. Compressor on, heading deviates, compressor off, heading goes back to being correct. Of course if you calibrate the compass while the compressor happens to be running, the calibration is thrown out.

Solution will be to relocate compass away from wiring, while also staying away from engine, etc.

Whenever I raise an issue I try to come back with the final resolution, even if it is sometime later, on the assumption it may one day be useful information for someone having a similar problem in future.

The field strength from the fridge compressor wiring is more than I expected. It is running about 7 amps at 12 volts - so 84 watts. The compass was no more than 2 to 3 inches away from the conduit (built in and running vertically down the inside of the hull inside a hanging closet on the starboard side). I presume the compass was put so close to the hull (and quite high up) to meet the installation spec of "at least 2 metres from the engine"). The cupboard is abreast of the motor, and my boat's beam is 3.71m. So getting 2m from engine means going for'ard or aft or well up and/or close to outboard. Whoever installed it went for close to outboard and up a bit.

I have now shifted it to be 0.5m inboard from it's old position. So violating the 2m from engine requirement, but further away from the conduit. As I understand it, magnetic force is s'posed to decrease according to an inverse cube law. So double the distance means 1/8th of the force. Yet even at the new location (a good deal more than double the distance away from the conduit), the compressor running still deviates the heading by ~3 degrees. I can likely live with that, as any placement I can think of will be a compromise, and an intermittent 3 degree deflection is a lot more tolerable than 17 degrees. By the way, that 17 degrees is while the boat is heading 31 degrees magnetic while tied up at a marina. I haven't measured the effect in all quadrants, and presumably it will vary.

I didn't get the the chance to go out and calibrate the compass in this new position before I came away (had other fixing also on the go). When calibrated it records a value known as local field per cent, which I understand is essentially a summary of the degree to which it has to make allowance for external magnetic interference when calibrating through 360 degrees. For the purposes of my autopilot computer this value is meant to be less than 12%. As I recall, when the fridge was not in use (switched off at the switch board), the local field percent in the old location was 7%. But obviously a lot more (but I don't know how much) if calibrated when the fridge compressor was running during all or some of the calibration circle (which it likely was during my attempts to resolve the issue before the cause was identified).

I'll report back on the local field per cent when I calibrate the compass in its new position. If less than 12 per cent, then there it will stay. If it is more than 12% I'll go hunting for a better location.

Fascinating to me, I have a friend with a steel boat that has the exact same compass mounted under the companion way stairs and which he reports performs faultlessly, and has done so for some years, although surrounded by ferrous metal. (I note in passing he has a eutectic fridge that runs off a mechanical compressor driven by the engine.)

I'm told shielding to prevent magnetic field is quite difficult. Apparently the best option is not shielding but rather using twisted pair wiring. Latter is not perfect but does help reduce magnetic effect by having the field from one wire offset by the field from the other. Requires the current to run in both directions, so can't use common ground. If there's an electrical engineer reading this perhaps they can clarify.