Attitude Movies


Description of frames in each movie

Sample orbit plot

The above image is a single frame from one of the following movies (orbit 580); there are about 968 frames, one per six seconds, in each of these movies. The links below are to pages describing the attitude behavior for the given orbit, from which the movies themselves (about 3.4 MB each!) are accessible.

Each frame contains the same elements. The main plot at the left shows the SAMPEX orbit and the Earth, to scale, with a dot showing the position of the satellite at the time of the frame. The view is from the sunward direction, so that you are looking at a "full Earth"; the vertical axis of the plot is along the projection of the Earth's rotational axis into the plane normal to the Earth-sun line (north is upward). All movies depict one full orbit, starting at the ascending node (northbound crossing of the geographic equator); if a portion of the orbit is drawn as a dotted line, as in the example above, then SAMPEX is in eclipse when the dot giving its position is on that portion of the orbit.

The moving sketch of the SAMPEX spacecraft that follows the dot around the orbit shows the variations of the spacecraft's attitude projected normal to the sunline; these variations consist almost entirely of rotations about the sunline. The position of the spacecraft in the plot is not significant; it is merely intended to keep the sketch of the spacecraft near the dot marking its location in orbit. The attitude of the spacecraft is calculated either from the APID-11 quaternions directly or, in coast mode, according to the method I developed early in the mission based on the guidance of the late Tom Flatley. The color of the spacecraft sketch will be either green, yellow, or red: green indicates normal attitude readouts; yellow indicates that there has been no quaternion update for 40 seconds, which I was originally told was a sign that the spacecraft had entered coast mode (but see the lessons learned below); and red indicates that the spacecraft has received no attitude update in 40 seconds and the angle between the model magnetic field and the sunline is 5° or less.

Several lines are plotted on top of the spacecraft sketch, as unit vectors projected into the sun-normal picture plane. Lines along the instrument look direction and the solar-panel normal are plotted in the same color (green, yellow, or red) as the spacecraft, though the latter is nominally perpendicular to the plane of the picture and thus of zero length. The magnetic field direction is plotted in blue; the blue will be dark or light depending on whether the local model field magnitude is above or below 0.3 Gauss (the color of the dot marking the spacecraft's position in its orbit changes color in the same way). Finally, the direction of travel (ram direction) is plotted in white.

On the right of each frame are several diagnostics to help the viewer interpret the spacecraft's "attitude adjustments". At top are the day and time, in YYDDD format and seconds UT respectively. Below the text are two back-to-back angle gauges and two color bar gauges. The left semicircular "gauge", with orange pointer, is the angle between the sunline and the spacecraft solar-panel normal, centered on 0° (this gauge doesn't move at all, most of the time). The right half of the circle has a blue pointer that indicates the pitch angle of particles coming straight into the instruments, with 0° straight up and 180° straight down (an angle of about 40° is shown in the example above); the pointer changes from dark to light blue as described above depending on the magnitude of the local model field.

The blue bar gauge to the right of the circle is the field magnitude, from 0.1 to 0.5 Gauss; the color of the bar changes when it crosses 0.3 Gauss, at the position on the gauge indicated by the change in the color of the outline. The rightmost gauge is the angle between the sunline and the local model magnetic field from 0° to 90°, with 180° to 90° rectified to 0° to 90°; this is the inverse sine of the absolute value of the "inertial dot product". Onboard calculation of this angle is what triggers coast mode; if the spacecraft is in shadow, then the spacecraft enters coast mode when this angle drops below 40°, while if it drops below 5° then coast mode is entered even if the spacecraft is in full sun. The color of this bar turns from green to yellow to red as these thresholds are crossed, at the points indicated by the color changes on the outline of the gauge.

Below the gauges are three static views of the spacecraft along its principal axes; in each view are projected the same unit vectors as on the moving sketch, namely the magnetic field in light or dark blue, the ram direction in white, and the instrument look direction and solar-panel normal (these two, of course, are fixed in these views). The spacecraft and the latter two vectors change color from green to yellow to red at the same time as the moving sketch. In addition, the sunline is plotted in orange (this would always be projected to zero length in the moving sketch).

List of available movies

I generated movies for several different cases, as outlined below. There are three attitude-control modes of interest: the original "orbit-rate rotation" pointing strategy; the "j-perp" mode, which attempted to orient the instruments perpendicular to the magnetic field when the model field magnitude was below 0.3 Gauss and upward along it when above; and the current mode, which is a "j-perp" strategy with the original ram-avoidance constraint (which prevented the delicate front foils of the HILT from looking too close to the direction of travel, while it still had gas pressure) disabled.

For each orbit, I also list below the magnetic local time and geographic longitude at the ascending node. Orbits with MLT of the ascending node near 0600 or 1800 are dawn-dusk, full-sun orbits; orbits near 0000 or 1200 are noon-midnight orbits with some period in eclipse. Longitude of ascending node was selected to place the orbit in or away from the South Atlantic anomaly, in sun or shadow, as described on the linked page below for each orbit (except for the first two, where no magnetic-field constraint was in effect).

Orbit number

Attitude control strategy

MLT at ascending node

Longitude of asc. node


orbit-rate rotation




orbit-rate rotation




j-perp, ram avoidance




j-perp, ram avoidance




j-perp, ram avoidance




j-perp, ram avoidance




j-perp, ram avoidance




j-perp, no ram avoidance




j-perp, no ram avoidance




j-perp, no ram avoidance




j-perp, no ram avoidance



Lessons learned

The most obvious lesson for me was that my original criterion for identifying periods of coast mode was far too conservative. As can be seen in every one of the noon-midnight orbits linked above, there are often extended periods when the spacecraft is not in coast mode (according to the inertial dot product angle gauge) and yet the quaternion is not being updated. The reason, which I should have figured out long ago, is that the quaternion is only updated when the spacecraft attitude changes by at least 1°; if the spacecraft is not in coast mode but is also not changing its attitude in inertial space for at least 40 seconds, then my algorithm identifies this as a period in coast mode (and the spacecraft turns yellow in the plots).

This explains the variation that I saw of the apparent time per day in coast mode, as a function of pointing strategy: in the latest plot I generated, I found that time in "coast mode" (as identified by my algorithm) under the original pointing strategy varied from 0% in full-sun orbits to about 40% in deep-eclipse noon-midnight orbits; in the j-perp mode with ram avoidance, maximum "coast mode" was still about 40%, but even in full-sun orbits there was a "floor" of about 4%; in spin mode the fraction varied from 0% to only 25%; and in the current j-perp mode without ram avoidance, the variation is greater than ever, from 0% to 55%! As can be seen by examining the behavior under different conditions in the orbits above, the true fraction of coast mode is, as I should have been able to figure out already, a function only of the orbit and not of pointing mode, and is most accurately reflected in the spin-mode figures of 0% to 25% above (since the spacecraft, spinning at 1 RPM, is putting out quaternions as fast as it can, it never experiences "pseudo-coast-mode" periods when the quaternion is not updated for 40 seconds!).

The practical upshot of this is that the apparent increase in the amount of time spent in coast mode under the current pointing strategy is not a cause for concern, but simply reflects the fact that the spacecraft can spend more time at a given attitude without having to drift sideways because of the ram-avoidance constraint, thus generating more periods of "pseudo-coast-mode". In light of what we see in these movies, the spacecraft appears to have been doing exactly what we had told it to do under each of the pointing strategies. I need to revise my algorithm that "flags" periods of coast mode, but the attitude correction that I have been doing all along does in fact appear to be valid, as reflected in the smoothness of transitions as the sketch of the spacecraft changes color (and also as reflected in the fact that data organized according to the corrected attitude behave reasonably!)

With regard to improvements that might be made to the attitude control strategy itself, well, the most common cause of large, rapid swings in attitude is the attempt to maintain the instrument look direction parallel or antiparallel to the projection of the magnetic field into the sun-normal plane, in the j-perp mode with local field magnitude above 0.3 Gauss. This does not add any value, and indeed, I suspect we would be better looking toward the zenith than along the field projection when we get to lower latitudes and the field rotates well away from vertical and well out of the sun-normal plane. Changing the high-field pointing criterion to be zenithward rather than the current approximate field-alignment (or anti-alignment, in the northern hemisphere) would eliminate some these "attitude adjustments". I'm still not quite sure what we could do to get rid of the 180° flips when the spacecraft crosses the equator in a low-field region.

Back to top of Data PagesBack to top of Data Pages

new 9 March 1999

Valid HTML 3.2!