Emergency tools and strategies for cloud flying without gyro instruments in "conventional" aircraft and hang gliders
September 27 2005 edition
steve at aeroexperiments.org
This article is aimed at shedding some light on some interesting physical relationships, and providing some emergency strategies. This article is not aimed at encouraging pilots to intentionally enter clouds without gyro instruments.
Because banking creates an immediate turn in any aircraft, regardless of whether or not the pilot applies back pressure on the control stick or yoke to increase the G-loading, a heading indicator of some sort is very useful for helping a pilot to keep the wings level in clouds. If the bank angle is adjusted as needed to prevent rapid changes in heading, the wings will stay nearly level, and if the bank angle is adjusted as needed to keep the aircraft on or nearly on a pre-chosen heading, the wings will stay nearly level and the aircraft will travel in the desired direction. Normally this is done with the aid of one or more gyro instruments. An artificial horizon is nice, but a turn rate indicator or a heading gyro are adequate for keeping the wings nearly level if the air is not too rough. If the wings are kept nearly level, an airspeed indicator provides adequate guidance as to the aircraft's pitch attitude.
How might we obtain heading guidance without the aid gyro instruments?
First let's take at look at this Bohli compass link. (Manual for Bohli compass (Adobe document)) This remarkable non-gyro non-electronic compass is free from the turning errors and linear acceleration errors that affect a standard magnetic compass. By accurately sensing the earth's magnetic force vector in 3 dimensions, in a way that is completely independent of the direction and magnitude of the G-force acting on the compass, the Bohli compass not only avoids the errors that plague a standard compass, but also provides information about an aircraft's attitude in all three axes. (Bohli company home page)
Please be certain that you carry a parachute if you actually try to use a device like this in clouds, without any supporting gyro instruments!
Food for thought: it would be possible to make an equivalent device using electronic sensors to detect the three-dimensional orientation of the earth's magnetic field, especially if the aircraft's "magnetic latitude" is known. For this project, it would be critical to avoid the use of any kind of an inertial "level" sensor. The fact that conventional compass needles or cards "hang" from a pivot or "float" in liquid, and thus orient themselves "upright" in relation to the total G-load acting on them, rather than "upright" in relation to the external world, is what renders them largely useless in turning flight. We'll examine this in more detail below.
Some inexpensive, non-aviation GPS units are advertised as having "3-dimensional" or "2-axis" electronic compasses, but the one that I've actually had a chance to examine (Magellan Meridian Platinum) clearly incorporated some sort of inertial "level" sensor, and exhibited exactly the same "northerly turning error" in flight as does a standard magnetic compass. This was true even though it also clearly had some sort of algorithm to factor the magnitude latitude into some of its calculations.
Other specialized non-electronic, non-gyro cloud-flying compasses: the Shanz compass is a gimbaled compass that has a more conventional type of display card, and the KT1 compass is another gimbaled compass. I don't know much about either of these compasses. The Cook compass is somewhat like a conventional aviation compass, but the card rotates around a fixed axle and cannot rock or tilt to orient itself in an "upright" manner in the aircraft's own reference frame, or in relation to the direction of the G-loading that the compass is "feeling" at any given moment. The whole unit can be manually tilted in the roll axis to compensate for a given target bank angle. These two features together keep the compass card level with respect the earth, avoiding turning errors--but only so long as the pilot maintains the target bank angle to which he has set the compass. The rigid axle also avoids compass errors due to linear accelerations ("accelerate north decelerate south"), but introduces a new class of errors--even in 1-G non-accelerated flight, if the aircraft is a nose-high or nose-low pitch attitude, this will create an error in the compass reading, unless the aircraft is flying due magnetic north or south. Due to these limitations, the Cook compass would probably not serve any better than a standard aviation compass as an aid to cloud flying without gyro instruments. (Note that the Bohli compass housing can also be manually tilted in the roll axis to match a target bank angle but with the Bohli compass the purpose of this adjustment is to make the display easier to interpret, not to avoid compass errors. The compass errors are avoided by the internal design of the compass.)>
Here's a link to a good explanation of the northerly turning error by Ed Williams -- this is why it is generally very unwise to try to fly in clouds depending on any conventional type of compass for heading guidance. The explanation given in this link is for the northern hemisphere. On a northerly magnetic heading, when a vehicle starts to turn to the west (for example), a standard compass actually indicates a turn toward the east! On a southerly magnetic heading the opposite is true: when a vehicle begins a small turn toward the west (for example), the compass indicates a much larger turn toward the west, and then has to spin back the other way to "catch up". The crux of the problem is that the earth's magnetic field contains a vertical magnetic component, so if the compass is tilted for some reason, it can better align itself with the earth's magnetic field (in the northern hemisphere) by rotating in such a way as to dip the north-pointing end to point more downward, which creates a false change in heading. Only a heading of due magnetic east or west avoids this problem. Most compass needles or cards--the needle of the Bohli compass being a notable exception--"hang" from a pivot or "float" in liquid, and thus orient themselves "upright" in relation to the total G-load acting on them, rather than "upright" in relation to the external world. In a 1-G unaccelerated situation, they are not affected by the pitching and rolling motions of the vehicle, but in a turn, they tip to the side in relation to the external world, and this is why they suffer from the "northerly turning error" described in this link. This is true of compasses in boats and cars as well as in aircraft: in a turn the G-forces are not vertical with respect to the earth, regardless of whether the G-forces act in the "upright" direction in the vehicle's own reference frame, as is the case in a "coordinated" turn in an aircraft, or include a "sideways" component in the vehicles own reference frame, as is the case with a skidding turn in an aircraft or a normal turn in a car or a boat. If we use a simple 1-axis electronic flux gate sensor and mount it rigidly to the vehicle, we solve the problem in the case of a vehicle that always stays rigidly upright, but not in the case of a vehicle that banks when it turns. As was suggested in our discussion of the Cooke compass, if the vehicle pitches up and down or rolls from side to side, a rigidly-mounted electronic magnetic compass with a simple 1-axis sensor will experience compass errors both due to the rolling motion and due to the pitching motion. A 3-dimensional (2-axis) electronic flux-gate compass could avoid these tilt-related problems, but to also be free of linear acceleration errors ("accelerate north, decelerated south" ), and the northerly turning error described in this link, it would need to be a very sophisticated design that did not use an inertial "level" sensor and emulated some of the principles employed in the Bohli compass.
We've mentioned linear acceleration errors ("accelerate north, decelerate south") several times now. Here's a link to a good explanation of linear acceleration errors by Ed Williams. Because of the way that the direction of the G-loading (net aerodynamic force) vector tilts away from the vertical when an aircraft is decelerating or decelerating, the compass card tilts with respect to the earth. Once this happens, the compass card can better align itself with the earth's magnetic field (in the northern hemisphere) by rotating in such a way as to dip the north-pointing end to point more downward, which creates a false change in heading. If the aircraft is accelerating the compass card will swing to indicate a more northerly heading than is actually present, and if the aircraft is decelerating the compass card will swing to indicate a more southerly heading than is actually present. If the aircraft is flying due magnetic north or south, linear accelerations will not create errors. In actual practice, the "northerly turning error" is a much more serious problem than the linear acceleration error.
I have found that a standard aviation compass does work surprisingly well as an emergency blind flying instrument in a light airplane, substituting for a turn rate gyro and heading gyro and allowing the aircraft to be kept under control without the aid of gyro instruments, provided that several critical conditions are met.
The first critical condition is that the air is relatively smooth.
The second critical condition is that the aircraft can be trimmed to a relatively pitch-stable, airspeed-stable configuration. In the case of a powered airplane, if the goal is a steep descent through a cloud layer, low power (but enough for carb heat) and full flaps and the airspeed in the middle of the white arc will likely work well, and should provide a good margin against stall or overspeed. Controlling the airspeed is very important, because the airspeed indicator is the pilot's only source of pitch attitude information. If something disturbs the aircraft's equilibrium, the pilot should make pitch inputs as needed to stabilize the airspeed and should not complicate things by trying to closely control the altitude or descent rate. The throttle should be left alone because any change in power will make a strong change in the aircraft's pitch characteristics, for a given trim setting.
The third critical condition is that the pilot does his utmost to minimize the rate of change of heading, which minimizes the aircraft's bank angle. If the bank angle gets too large, compass errors will cause the pilot to lose control of the situation. Bank angles over 10 or 15 degrees are likely to be too large.
Of course the pilot's technique and the aerodynamic characteristics of the aircraft strongly affect the odds for success when a standard aviation compass is used as a primitive substitute for gyro instruments. The successful tests described in this article were carried out in a Cessna 152. In this aircraft, using very gentle rudder inputs for steering works much better in the situation we're discussing here than using the ailerons and rudder together. There are several reasons for this. When a rudder input is only held for a short time, it turns the aircraft mainly in a "flat" manner (by yawing the fuselage slightly sideways in relation to the airflow, which generates an aerodynamic sideforce.) As far as a standard aviation compass is concerned, a flat skidding turn will create the same errors as a banked turn will. But the advantage of a flat skidding turn is that the turn will stop immediately when the pilot relaxes the rudder input. This is extremely helpful in the situation that we're addressing here, where it is critical that the pilot not over-control and turn too far (more on this below) or accidentally put the aircraft into a steep bank. When a rudder input is held for a longer interval of time, it creates a significant rolling effect as well as a yawing effect, but because the rudder pedal forces are heavy and the aerodynamic yaw-roll coupling is mild, the pilot is still much less likely to over-control the aircraft if he is using the rudder alone than if he were using the ailerons and rudder together. Another reason the rudder alone works better than ailerons and rudder together in this situation is that if the pilot is making aileron inputs, it is very easy for him to make unconscious roll inputs that reflect his own, often erroneous, sense of which direction is "up". A pilot in clouds without a full 2-axis artificial gyro horizon absolutely must expect to be very disoriented and must be prepared to ignore his impressions of which way is "up" and which way the aircraft is turning--in fact we should say that this knowledge and discipline is the fourth critical condition that must be met if the pilot is to have any chance of controlling the aircraft in cloud by reference to the compass alone.
A good way for a pilot to avoid making unconscious aileron inputs is to keep his hands off the control yoke unless a corrective pitch input is needed to stabilize the airspeed, in which case he can touch the yoke or stick with an "open hand" in a way that cannot exert a twisting or sideways force on the control yoke or stick.
A fifth critical condition must also be met in order to allow a standard aviation compass to be successfully used as crude substitute for gyro instruments. This condition is the least well-known and is perhaps the most important of all: in the northern hemisphere, the pilot must do his utmost to keep the aircraft on a heading that does not contain a northerly magnetic component. Therefore the safest heading affording the greatest margin of error is due magnetic south. Headings up to 30 or 45 degrees away from this might be alright, but the closer the aircraft's heading is to due magnetic south, the larger the margin of safety (survival) the pilot has in the case of an accidental turn. The oversensitivity of the compass (the flip side of the "northerly turning error") is greatest on a due south magnetic heading, but this can be tolerated, at least at light-aircraft airspeeds. Headings that are closer to due magnetic east or west may be theoretically better than a due south magnetic heading, but the problem is that if the aircraft accidentally wanders far enough off course to gain even the slightest northerly magnetic heading component, all control will be lost. The aircraft will end up in a spiral turn of unknown direction with the compass locked up or reversing directions at random. We can't emphasize this point strongly enough, because it will swiftly lead to catastrophic structural failure.
There's another factor that makes this all very complicated. Due to the way that a standard compass card is laid out, when the aircraft starts to wander off course, the pilot must realize that he must apply rudder (assuming for the moment that he is using the rudder for roll control as described above) on the side opposite the symbol representing his chosen heading. For example, when the aircraft is pointing south, the "west" heading icon is on the part of the card that is to the pilot's left, and the "east" heading icon is on the part of the card that is to the pilot's right. If the nose swings a few degrees to the west, the "south" icon, representing the pilot's chosen course, will actually move to the pilot's right, as the "west" heading moves closer to the compass index line. To bring the aircraft back toward the south, the pilot must apply rudder on the opposite side from the south icon--i.e. the pilot must apply left rudder. So the pilot must always "step away" from the icon representing his chosen course. This is extremely confusing when the outside world is no longer visible. (In a hang glider, a trick is to move the control bar toward--and the pilot's body away from--the icon representing the desired course.) Note that with a heading gyro, because of the way the card is laid out when the instrument has a vertical face, this problem is avoided.
In a real-world situation where a pilot was really contemplating using a compass to enter cloud in a light airplane, he would be wise to give himself a chance to practice first. For example if he had to descend through a cloud layer, if possible he should climb at least a thousand feet above the cloud layer before setting up a stabilized, constant-heading descent.
For minimizing compass errors, it turns out that a lower airspeed is better than a high airspeed, because a given bank angle is associated with a larger rate of change of the aircraft's actual heading, so the compass errors become less important in comparison to the real changes in heading. This is not a panacea--the "northerly turning error" is a very serious problem even at hang-gliding airspeeds, and I've even observed the "northerly turning error" while driving a car at speeds below 10 mph. And in reality the characteristics of the aircraft will dictate the choice of airspeed, as described above.
At this juncture we need to pause to note that the successful tests described above were carried out in Oregon (USA) at a latitude of roughly 45 degrees. Of course, the relevant factor is actually the "magnetic latitude", or more accurately, the angle of magnetic dip (also called "inclination"). The angle of magnetic dip at the test location was 67 degrees. Charts of lines of equal magnetic dip (inclination) are given at the end of the Bohli compass manual (see the link provided earlier in this article), as well as on these links (#1, #2, #3). To give some context, we'll note that much of the central and eastern part of the US is closer to the north magnetic pole and has a higher magnetic dip angle than the test location, and that the southwestern tip of the Alaskan mainland, the northern tip of Sakhalin island, the northern tip of the Caspian sea, and the southern tip of Ireland all have about at the same magnetic dip angle as does the test location. Observers closer to the line of zero dip (i.e. in regions with lower magnetic dip angles) will find that the northerly turning error is much less problematic than implied this article. Observers further from the line of zero dip (i.e. in regions with higher magnetic dip angles) will find that the northerly turning error is much more problematic than implied this article. The specific guidelines mentioned above, as well as the suggestion that these techniques have any chance of working at all, are really only appropriate for locations with equal or less magnetic dip than was present at the test location. When the aircraft is too close to the north magnetic pole and the angle of magnetic dip is too high, the techniques described in this article are certain to fail. So we have a sixth caveat for successfully using a standard aviation compass as a primitive substitute for gyro instruments--the aircraft must not be at too high a "magnetic latitude", i.e. the aircraft must not be in a region where the angle of magnetic dip is too high.
Of course, observers in the southern magnetic hemisphere will encounter a southerly turning error rather than a northerly turning error, and will find that a course of due magnetic north is optimal when a standard aircraft compass is employed as a primitive aid to cloud flying. Unless stated otherwise, in this article we're assuming that we're in the northern hemisphere. Note that almost all of the large landmasses in the southern hemisphere (except for Antarctica, the far southeastern tip of Australia, Tasmania, and the southern half of New Zealand) have less magnetic dip than does the test location for the experiments described here. Large parts of South America, Africa, southeast Asia, and the islands of the southwest Pacific are close to the region of zero magnetic dip, where compass errors should be minimal.
In a light airplane (again in Oregon USA), I've also experimented with using similar piloting techniques as described above--except that any arbitrary heading could be used--in combination with an inexpensive, non-aviation GPS unit with a 1-second update rate, set to the analog, compass-like (but GPS-driven), "heading" display screen. This yielded a less controlled flight path than did using a standard magnetic compass on a southerly magnetic heading. That's right: providing that I had the freedom to choose a southerly magnetic heading, the magnetic compass actually outperformed the GPS unit as a blind-flying aid! The key limiting factor with the GPS unit was the low update rate of the display. As a result, the flight path ended up wandering up to 30 degrees or more to either side of the intended heading. However, in smooth air, using extreme caution as described above, the risk of entering a spiral dive still seemed low. In fact the overall safety level was undoubtedly higher with the GPS unit than with the magnetic compass, because if the heading had accidentally deviated 90 degrees or more from the intended heading, the situation still could have been salvaged, as long as the bank angle was still shallow. However the slow (1-second) update rate was a very serious problem and made the GPS display behave in a very "twitchy" manner. In a steep spiral dive with a high turn rate, the low update rate would have made the GPS unit unusable.
With a GPS unit of the type described above, it turns out (for a given, preferably shallow, bank angle) that a high airspeed actually provides a more stable heading indication than does a low airspeed, because the turn rate is minimized for a given bank angle, so the low update rate is less of a problem.
If a magnetic compass and a GPS unit of the type described above were both available, and if no gyro instruments were present and entry into clouds could not be avoided, the best plan might be to rely primarily on the compass, but be ready to swiftly switch to the GPS if the heading deviated far enough to gain a northerly magnetic component, or if there was any other evidence that the situation was starting to "get away" from the pilot and the compass readings were becoming unreliable. Of course, as noted above, if the aircraft is truly entering a spiral dive this type of GPS unit will not be of much use either, due to the low update rate.
I don't want to imply that any of the techniques described above afford any kind of reasonable margin of safety. They absolutely do not. They are only strategies of last resort.
Loss of control in clouds in any type of "conventional" aircraft almost always leads very swiftly to structural disintegration due to aeroelastic flutter from excess airspeed, or due to excess G-loads from abrupt pilot control inputs and/or excess airspeed. This is doubly true in a sleek aircraft like a sailplane or retractable-gear airplane. If you want to experiment with these techniques, do so in a 2-seat aircraft with the appropriate IFR clearances and with a qualified IFR pilot in the other seat watching a full set of gyro instruments to keep you out of trouble. That will be an enjoyable experience. Finding yourself alone with the wings departing the aircraft will not.
Hang glider pilots have all sorts of ideas about using compasses in clouds. It's worth emphasizing that compass errors are strongly related to magnetic dip and a pilot who has been successful in using a compass to fly in clouds in Hawaii (for example) might not meet with the same success at more northerly latitudes.
While flying a flex-wing hang glider in Oregon (USA), using the techniques described above (where applicable), I've not been consistently successful in using either a magnetic compass or an inexpensive, non-aviation GPS unit with a 1-second update rate to maintain orientation in clouds, even in smooth air. This is largely due to the flight characteristics of flex-wing hang gliders. Flex-wing hang gliders exhibit enough adverse yaw to be a real problem when a pilot is trying to fly by reference to any kind of heading instrument (though adverse yaw ought not have much effect on a GPS display, because the GPS receiver measures the actual direction of the flight path, not the glider's heading). Also, the significant time lag between a flex-wing hang glider pilot's roll input and the development of a significant change in bank angle creates problems. And in a case where the pilot wishes to fly fast, the tendency toward yaw-roll oscillations that many hang gliders display at low angles-of-attack causes further problems. Installing a fin to help damp out adverse yaw and yaw-roll oscillations would definitely help with some of these problems. But fin or no fin, a flex-wing hang glider just doesn't have the same "solid", yet responsive, "feel" that a typical general aviation light aircraft being flown primarily with rudder inputs does.
While hang gliding, in an emergency cloud-flying situation where I had both a compass and a GPS receiver of the type described above, I would use the compass as my primary reference, so long the situation allowed me to aim for a southerly magnetic heading, and I would keep an eye on the GPS receiver in case the heading accidentally strayed to include a northerly magnetic component. With the low airspeeds typical of hang gliders the useable portion of the magnetic compass card may extend 20 degrees or more north of the due east and due west magnetic headings, but if the heading contains too much of a northerly component, the compass will still definitely become utterly unusable. Keeping the glider's heading between the magnetic southeast and the magnetic southwest compass points would be a good policy, if possible.
With a weight-shift control system, it is easy to make an unconscious roll input (especially if the pilot is also pulling in the bar for airspeed rather than letting the glider fly at trim), so the pilot must be very vigilant about remaining neutral on the bar in the roll dimension unless he is making a deliberate correction. Small "bump" corrections may work best. As we've already noted, any pilot flying in cloud without a full 2-axis artificial gyro horizon absolutely must expect to be very disoriented and absolutely must be prepared to ignore his impressions of which way is "up" and which way the aircraft is turning.
For hang gliding, if a pilot feels it is worth carrying a compass despite its many limitations, a small marine-type compass is best. Aerodynamic sideforces are minimal in hang gliding so the compass card won't need to tip very far to the left or the right of the "upright" direction in the glider's own reference frame, but the compass card does need to be free to tip forward and aft a bit so that the angle of mounting (in the pitch sense) does not become too critical. See the images in the "Experiments" photo gallery on this website; the particular unit shown here came mounted on a "ratcheting" base so the whole unit can be adjusted up or down at will.
It's worth noting that hang gliders often fly in enough wind that the groundspeed can potentially become very low or even negative. A GPS unit will give extremely confusing indications in this situation. In fact, any time the wind speed is in the neighborhood of half or two-thirds of the airspeed or higher, a GPS unit will start to become extremely sensitive to small heading changes when the glider is flying into the wind, and will become extremely insensitive to small heading changes when the glider is flying downwind. With a low (1-second) update rate this can render the GPS unit almost unusable. This is one reason why a hang glider pilot might want to consider flying with a magnetic compass. This is also a reason why a hang glider pilot might want to fly fast during an inadvertent encounter with a cloud. Another obvious reason to fly fast would be to penetrate forward away from high terrain, and to increase the sink rate to descend from the cloud. A hang glider pilot who is counting on a GPS to provide some assistance in the case of an inadvertent encounter with a cloud should consider mounting the GPS unit where he can read it even when he has pulled the bar all the way in.
No pilot should intentionally enter cloud without gyro instruments under any circumstances, but a hang glider pilot should be aware that the risks of entering cloud are multiplied 10-fold if there is not a great deal of vertical clearance between the bottom of the cloud and the nearest terrain, and plenty of landable terrain in all directions, especially downwind. In other words, don't go try to fly in the cap cloud that forms low over your local ridge-soaring hill. It's no fun to be in the "white room" when the clouds are likely to have rocks in them! And note the catch-22: in general, the only time a hang glider is likely to be near a cloud high above the ground is when strong, turbulent updrafts are present, making controlled blind flight a dubious prospect.
I have noticed that in a steep turn in a hang glider, a yaw string blows slightly toward the outside of the turn. This knowledge could be useful if a disoriented hang glider pilot ends up in a steep spiral of unknown direction, in which case a compass and a GPS unit of the type described above would both be worthless. However a yaw string is also responds strongly to a pilot's roll inputs, as we'll describe in detail elsewhere on this website, so its value as a blind flying aid is extremely limited. Whenever the pilot is making a roll input, the associated adverse yaw will shift the yaw string strongly to one side, and the pilot will not be able to use the yaw string for bank angle information.
In a flex-wing hang glider, releasing the VG and pulling the control bar most of the way in and shifting a significant distance to one side of the control bar (in the same direction as the glider is already turning, if this is known) may be the safest choice when orientation is truly lost or is about to be lost. The glider should be fairly stable in this configuration--moderately steep, fast turns with the VG loose generally need some low-siding to prevent the bank angle from decreasing. The most severe upward and downward curvatures in the flight path, with severe changes in airspeed and pitch attitude and a risk of a tumble, generally occur when the glider is randomly fluctuating between a wings-level attitude and a steep bank.
Of course the ultimate solution to the possibility of being "whited out" in a hang glider would be some sort of lightweight 1-axis turn rate (yaw rotation rate) indicator, to supplement the information available from an inexpensive GPS unit. These have been available in the past. Today's piezoelectric "gyros" as used by RC helicopter and airplane pilots could undoubtedly be incorporated into such an item.
With a rigid-wing hang glider, which doesn't share the slightly forgiving load-shedding structural characteristics of a flex-wing hang glider, I would not dream of intentionally entering into clouds in any circumstances whatsoever, and minimizing the risks of accidental entry into a cloud would be higher priority than it sometimes is with a flex-wing hang glider, even though the roll response and roll stability characteristics of a spoileron-controlled rigid-wing hang glider are better than those of a flex-wing hang glider.
As extra insurance while hang gliding, I have sometimes carried a compass, GPS unit, and yaw string, all mounted on a long probe that inserted into a socket on the centerline of the base bar so that I could read them even with the bar nearly "stuffed". (Photos are included in this section of the "Experiments" photo gallery of this website). But all pilots in all aircraft should play it safe and observe the legal cloud clearance requirements. Realistically, in most cases there's no good reason for a VFR pilot or sport pilot to carry some sort of rudimentary blind-flying instrumentation. Don't let this theoretical discussion give you a false sense of complacency or an over-active curiosity. Even apart from the legal issues and the danger to other air traffic, the hazards of flying in clouds without gyro instruments cannot be emphasized strongly enough. Even flex-wing hang gliders have broken up in clouds, especially after a tumble. By no means do I want to give the impression that hang glider pilots consider it a routine manner to fly in or near clouds, or that they are routinely unable to prevent themselves from being "whited out". All pilots place themselves in great danger by entering clouds without the appropriate gyro instrumentation and training. Everything may work well for long minutes, but once things start to go wrong, they can go very wrong very quickly.
I'll close by noting that my experience with flying with inexpensive, non-aviation GPS units that included magnetic sensors has not been good in any type of aircraft, even in normal VFR conditions. I've found that due to the low (typically once per second) update rate, the electronic magnetic compass display is much harder to use than a standard magnetic compass, even though there are no longer errors related to the inertia of a rotating needle or card. In a "conventional" aircraft the standard aircraft compass always outperforms the magnetic compass function on an inexpensive non-aviation GPS unit, and in a hang glider, a small marine-type compass always outperforms the magnetic compass function on an inexpensive non-aviation GPS unit. As described above, unless an extremely sophisticated design is used, these electronic magnetic compasses will suffer from the same "northerly turning error" as does a standard magnetic compass, even if the electronic magnetic compass is designed to be held a wide range of pitch and roll attitudes in a 1-G, unaccelerated environment. With a simpler electronic magnetic compass incorporating only a 1-axis sensor, we'll see the same effects that we saw in the "Cook compass" which used a rigid axle: linear acceleration errors ("accelerate north, decelerate south") will be avoided, and a flat skidding turn will create no error, but errors will occur whenever the compass is in a nose-up or nose-down pitch attitude (unless the aircraft is flying due magnetic north or south) and also whenever the compass is banked in relation to the horizon (unless the aircraft is flying due magnetic east or west). As a more general extension of this idea, the fact that a simple 1-axis magnetic flux sensor is insensitive to the direction of the G-loading, but is sensitive to direction of tilt of the platform with respect to the horizon, seems to end up making it even more unstable and difficult to use in flight (and in virtually any other application) than a conventional magnetic compass, which is sensitive to the direction of the G-loading but not to the direction of tilt of the platform (within limits, until the movement of the needle or card is physically impaired). And of course any type of electronic magnetic compass will suffer from deviation effects unless the calibration routine is performed after the compass unit is mounted in a fixed position on the aircraft. This may not always be practical, especially with the so-called "3-dimensional" or "2-axis" sensors that are designed to be held in a wide range of pitch and roll attitudes in an unaccelerated 1-G environment--these units require calibration over a wide range of pitch attitudes and roll attitutudes as well as over the full 360 degrees of the compass rose. When a compass has not been re-calibrated after it has been mounted in a fixed position on an aircraft, the resulting deviation errors may be quite severe, if the unit is being used on a "conventional" airplane with lots of ferrous metal parts and electronics. As a result of all these problems, when an inexpensive, non-aviation GPS unit includes an actual magnetic sensor, this feature should be turned off before flight so that the compass-like display will reflect the satellite-derived direction of the ground track.