Whither Pointest Thou?

Science Nugget: October 27, 2000


It's been nigh on 10 years since the launch of Yohkoh, and by all appearances the instrument continues to work nearly flawlessly -- or so we would have you believe, as we continue to think up new bits of software to better correct the occasional blemishes we see in SXT data. Stray light, CCD dark current and the Point Spread Function are all recurring topics in the Yohkoh Operations Room at ISAS.

There are other aspects of Yohkoh operations that generally escape the eye of the casual armchair observer. One aspect of operations that might not escape the casual observer is illustrated in the following movie:

Yohkoh Roll Demo

This is an example of what happens with the Star Tracker on Yohkoh loses Canopus, our trusty Guide Star. (The Star Tracker is explained below.) In this case, it's apparent that we lost that aspect of the pointing system which kept track of the the roll angle. The vertical bouncing later in the movie arose when the tohbans (spacecraft operators) sent commands to bring Yohkoh back into nominal pointing, with North pointing Up. Some of the images in the above movie show improper subtraction of stray light in the telescope arising from these non-normal pointings. (A tutorial on "stray light" removal appeared here last year).

A Satellite Pointing mini-tutorial

Sun-pointing satellites such as Yohkoh, TRACE, and SOHO all require some method of keeping pointed at the sun to pretty tight tolerances. Various methods are employed, ranging from all-sky star trackers ("Any bit of the sky in, exact pointing coordinates out") to primitive photosensitive diodes that are only really able to tell that the sun is shining on them, so hence it must be somewhere In front of us. (Think of walking outside, closing your eyes, and figuring out where the sun is, based on how bright things look shining through your eyelids. That's about the level of precision we get here.)

Yohkoh uses two stars and three instruments to define its orientation in space. The Sun is the first obvious guide star. The second star may not be so obvious: Canopus. With these two stars, we are able to (a) able to keep the Sun centered in Yohkoh's sights, and (b) able to keep the Sun from spinning around on us, leading to the (possibly?) mistaken impression that the heart of our solar system has decided for a change to revolve about its axis for a while.

The Details

While we've already mostly summed up how Yohkoh pointing works, this is a good time to point out that - as in many things - "the Devil's in the details." Specifically, just what are these telescopic sights that keep the Sun and Canopus centered, and Yohkoh on track? As I mentioned, there are three. Much of what follows was stolen wholesale from a pivotal paper by Wuelser et al.

The HXT Aspect sensor system (HXA)

The HXA aspect sensor system is part of the HXT instrument, and is composed of two linear CCD detectors, arranged at 90 degrees to each other and at a 45 degree angle to the HXT coordinate system. To put this in layman's terms, the HXA sensors consist of two rows of pixels criss-crossing the Sun. By looking at the signals on these pixels, one can then infer where the solar limb is, and make a solid determination of where we're pointing. In the normal usage of the HXA sensor, these four limb points sent down from Yohkoh in telemetry and then processed to determine solar pointing.

This system used to work well. But the HXA sensors (the linear CCDs) have degraded over time: the sensitivity has dropped, and the signal-to-noise level has worsened. (This is a consequence of putting a CCD detector in space. Radiation damage has probably taken its toll on both of these sensors after 10 years on-orbit.) The noise level was such that it became difficult to determine the location of the limbs, and often the returned pointing was incorrect.

The solution, again implemented by our Pointing Expert Jean-Pierre Wuelser, was to collect the entire row of CCD data for each of the HXA limb scans - one-dimensional images of the Sun - and fit a curve to the result. This method overcomes the worsening signal-to-noise problem in the sensors and gets the HXA sensors, and Yohkoh pointing, back on track.

Back on track, that is, except for solar eclipses which occur on regular schedules but sometimes surprise us anyway. Solar eclipses as viewed from a satellite whizzing around in low Earth orbit are kind of different... [1], [2], or search on "eclipse" in the Nuggets home page. During an eclipse of course the Sun's apparent shape changes erratically, so routine pointing procedures may (and sometimes do) go awry temporarily.

The Inertial Reference Unit

The Inertial Reference Unit, or IRU, is a bunch of fancy words used to label a set of four gyroscopes. The gyroscopes observe changes in the orientation of the spacecraft and feed an error signal back to the spacecraft computer. (Three are used at any one time, with one as a backup unit.) All gyroscopes, including those on Yohkoh, exhibit a slow drifts and can't be used to provide absolute pointing information on their own.

To compensate for this, a time-averaged signal from the HXA sensor is used as a baseline and the error signal from the IRU sensors is used to calculate the drift and jitter since the time of the HXA measurement. The HXA unit provides pointing information with about 2 arc second resolution on a time scale of 1-8 seconds, dependent on operating mode. Meanwhile, the IRU unit provides 0.08 arc second resolution on a time scale of 0.25-2 seconds. By combining these two datasets in ways beyond the scope of this nugget, we get fine-scale pointing information for Yohkoh, at a level generally considered sufficient for a detector with 2.5 arc second resolution.

Even this level of pointing isn't always good enough. The current algorithm doesn't use all of the IRU datapoints, so it's possible for the pointing between sequential images to be off by maybe 0.20 arc seconds, substantially less than an SXT full-resolution pixel. Usually this isn't a big deal. But it becomes a lot more important when someone decides to do temperature analysis with SXT images! Temperature maps of soft X-ray images involve differences and ratios of actual data, always a tricky business prone to systematic errors. Further information (for experts only) may be found in the literature: (Acta Astronomica, v.46, pp.15-28, 1996).

The Canopus Star Tracker

The Yohkoh star tracker is a special-purpose CCD camera that observes the bright southern star Canopus. The theory is that if Yohkoh pointing is centered up on the Sun, and the roll angle with respect to heliographic north doesn't change, then Canopus will follow a well-defined path in the star tracker's field of view. (This is all paraphrased from the Wuelser paper mentioned earlier.) The star tracker is somewhat tricky, as it has to be told when Canopus is visible and when it will be obscured by the Earth. It's also true that the star tracker CCD detector suffers from the same degradation that affects the rest of the instrument, and the signal-to-noise ratio gets worse with time. Finally, Canopus is a pretty dim star, especially when compared to, say, the lights of (say) a squid fishing boat out at sea. Yohkoh has been lured away before by such phenomena.


The obvious conclusion is that the Yohkoh pointing system has proven amazingly resilient to almost 10 years of space environment. The spacecraft design is excellent (or we've been lucky) in that there have been ways to fine-tune and improve the pointing system. In fact we haven't explored all of these pathways yet. As researchers push the analysis envelope harder, and as things on the spacecraft inevitably wear out, these methods of fine-tuning the pointing coordinates will continue to be developmental items if not hot research topics.

October 27, 2000

Brian Handy <handy@isass0.solar.isas.ac.jp>