NavList:

Gary LaPook asked [993]-|

| My car has an electronic compass installed and the directions for
| correcting it for deviation (of course the car manual didn't use
these
| technical terms) has you pushing a calibration button and then
driving
| the car in a slow circle. How does the compass figure out the
deviation
| from just the data it can capture while the car is driven in a
circle?

and Red responded [994]-

"I am guessing that the compass reads deviation versus time (the rate
of turn
must be constant) and then it "remembers" the min/max and when the
pattern
repeates itself. Something in the pattern must be perceived as "most
north" and
from there, you just need to note the other readings versus time, i.e.
halfway
between "most north" readings would have been south, etc. and you
divvy those up
into a lookup table.

Just guessing. Maybe they've got something simpler and more elegant."

=====================

Yes, they've got something simpler and more elegant than Red's guess.

I have long been sceptical about such claims, that one could calibrate
a fluxgate or other electronic compass against deviation caused by its
surroundings within the vessel, simply by travelling in a circle (or
perhaps two). I had thought, as Red did, that the presumption was made
that the true bearing would be changing steadily with time, and I
didn't see how, in any sort of vehicle or vessel, that could be made
to hold good to nearly sufficient accuracy. Now I know better; because
the basis of the method is in fact quite different.

Bill mentioned Richard Langley's article, referred to in [894] as-

""Getting Your Bearings: The Magnetic Compass and GPS":
<http://gauss.gge.unb.ca/papers.pdf/gpsworld.september03.pdf>
Mostly about electronic compasses. Nothing in it that the cognescenti
on the
list don't already know but it might be of interest to some."

and for me, that paper has been the key to understanding what was
going on. With Fred Hebard, I recommend it strongly to anyone keen to
learn about the state of the art, in modern-day compasses. Together
with one of the references therein, "Applications of Magnetic Sensors
for Low Cost Compass Systems", by M J Caruso
http://www.ssec.honeywell.com/magnetic/datasheets/lowcost.pdf

Here's how I see it-

It depends on the fact that a such a magnetic sensor, to measure field
direction in the horizontal plane, is actually a combination of two
such sensors at right angles.  One measures the horizontal component
of the magnetic field lengthwise, parallel to the centreline of the
vessel, Hl, the other measures the same thing, but crosswise, athwart
the vessel, Hc.  You can think of these two sensors being mounted on a
gimballed platform to keep them horizontal (though there are
alternatives to that, as a "strap-down" configuration, which those
articles deal with, but I won't, here). To keep things even simpler,
think about a situation at the magnetic equator, where the field is
horizontal only, with no dip to worry about. Hl and Hc can be positive
or negative, depending on the way the vessel points with respect to
the horizontal field H.

From those two components, Hl and Hc, we can derive a magnetic course
C, from arctan Hc / Hl. That course is the only information, that an
ordinary needle-compass would provide, but now we have extra
information, because we know those two components of the field
strength. A needle-compass tells us nothing about field strength, just
the direction.

First, let's assume that there are no sources of deviation. No iron
anywhere around, so the compass tells the truth  about the strength
and direction of the Earth's horizontal field. Then Hl will vary as H
sinC , and Hc, as H cosC. If we plot Hl against Hc, the result is a
true circle, centred on the origin, radius H.

Next, assume there's some permanent magnetism somewhere aboard (what
is often, somewhat incorrectly, attributed to "hard" iron). That
implies that at the position of the sensors there's a constant
perturbing field, with horizontal components which we can call  Hl'
and Hc', which do not change as the vessel's heading alters, and which
would be the same even if there was no field from the Earth.. Now the
two sensors indicate a total field (H sin C + Hl'), and (H cos C +
Hc'). If we plot these two values against each other now, we still see
a circle, but it's no longer centred on the origin; the centre is
displaced, to a position Hl', Hc'.

The compass has to have associated with it a simple computer or
microprocessor, which can record the pattern of outputs from the two
sensors as the magnetic course changes through at least 360 degrees in
a calibration run. These are, effectively, "plotted" internally,
against a nominal course angle derived from the arc tan of their
ratio. The whole range of angles should be passed through, but those
angles are NOT related to the time within the test run, which does not
rely at all on the turn rate being held constant. It's a simple
business for the computer to analyse that information, to determine
the offsets of the circle, Hl' and Hc', and from then on  to subtract
those offsets from the transducer outputs, and compute a true heading
direction based on those corrected values.

There's another source of compass deviation, relating to induced
magnetism (often known as "soft" iron), which is actually caused by
the effect of the Earth's magnetic field on the iron or steel, and
would vanish if that field were removed. That effect does not displace
the centre of the resulting (Hl. Hc) circle, but lengthens it, in one
direction, into an ellipse. The computer should have no difficulty in
assessing, separately, the results of permanent and induced magnetism,
and allowing for them appropriately.

I can imagine one problem that might arise when making a calibration
run in that way. It's the horizontal components that need to be
measured, and so (except near the equator) it's important that the
measuring platform is kept truly horizontal by gimballing (or else by
some form of tilt-compensation). If, in a calibration run, a vessel
was taken through its range of heading angles by steering in a circle,
and not by stationary swinging (for example, by ropes between
pontoons), then the "horizontal" level of the gimballing would be
affected by the centrifugal affects of the turn. Vessels really ought
to be instructed to travel sufficiently slowly while making the turn
that such accelerations are quite negligible. AA tight, very
slow-travelling, turn is better than a big circle travelling fast,
over the same time period.  A good procedure might be to to attach to
a mooring buoy, then push the stern slowly around in a circle, with
the dinghy. This could be a particular difficulty when calibrating an
aircraft compass, which presumably must be done on the ground, rather
than in the air, for that reason.

George.

contact George Huxtable at george@huxtable.u-net.com
or at +44 1865 820222 (from UK, 01865 820222)
or at 1 Sandy Lane, Southmoor, Abingdon, Oxon OX13 5HX, UK.



--~--~---------~--~----~------------~-------~--~----~
To post to this group, send email to NavList@fer3.com
To , send email to NavList-@fer3.com
-~----------~----~----~----~------~----~------~--~---