Wednesday, January 26, 2011

Gyro Compass Calibration


UPDATE : the calibration method described here is valid only if there absolutely no current, a situation which is very difficult to achieve, but can be checked by looking at your GPS traces during the turns. For an alternate calibration method tolerant to currents, see this other post.

Electronic compasses typically provide an auto-calibration feature that requires you to make wide circles on the water during the calibration process. The concept behind this is that your angular velocity will be constant during the turns. Well, in practical conditions with even light winds and almost flat water, the angular velocity is not constant. At some points, the effect of wind and/or small waves will impede the rate of turn, and later on will help to accelerate the rate of turn. The ideal conditions required for this type of auto-calibration are very seldom or never seen, and even if you think they are, you will never be sure of the results.  When these ideal conditions are not met, the auto-calibration routine will see false deviations that do not exist. There ought to be a better way.

After the auto-calibration routine reported success on the Airmar H2183 gyro compass, differences of more than 20 degrees with the regular compass were observed on some headings, while the 2 compasses agreed at dock where the compass offset was adjusted. It is clear that a supplementary deviation correction has to be applied.

Here is how it is done, after making several turns in the water while recording the heading value from the gyro compass and the COG value from the GPS. First, the COG values are converted from true to magnetic heading, using the current local variation value. Then, the diffence between the 2 magnetic values are plotted vs the gyro compass value.

In this way, we obtain directly the deviation correction curve to be applied. The usual way to express this curve is to fit the points to the following equation:

Deviation = A + B sin(θ) + C cos(θ) + D sin(2 θ) + E cos(2 θ)

For the above points, the coefficients are found to be:

               A = 4.103844379
               B = -8.302380526
               C = 15.92628402
               D = 1.519510921
               E = 2.346228572

The fitted curve is drawn in red in the above figure.

The deviation equation is programmed in the microprocessor and the correction is applied to each compass value before being reported.

1 comment:

  1. One has to wonder whether the GH2183 with its onboard GPS does a better job of autocalibration. I don't have one so I don't know. But it should have direct access to the COG as the boat turns its circles during autocalibration.

    Also, an H2183 on an N2K network should have indirect access to COG from a GPS and that should work as well. I do have an H2183 and a GPS so I'll check.