Hubble Space Telescope was launched by the shuttle Discovery (STS-31) on 24 April 1990 at 12:34 UTC. For those of us who worked on the project, the inside joke that the “Hubble Constant is 2 years until launch” had been broken. No longer was this a mission that we were working towards but, rather, a mission that was about to become reality. The question in all our minds was, “Will it work? Will all the years of hard work and planning pay off?”
I first joined the Hubble project in 1982. It hadn't even been named for American
astronomer Edwin Hubble yet. That was to
come a year later in 1983. When I
started, it was simply ST, Space Telescope.
I was a comparative latecomer to the project. For those who had been there at the beginning
in the 1970s, it had been the Large Space Telescope, the Large being dropped as budgets and the realities of operating a
telescope in space began to settle in.
Still, it was to be Big with a
capital B, a 2.4-meter optical
telescope that would operate above the distorting layers of the Earth's
atmosphere. It would be controlled
remotely from a control center at Goddard Space Flight Center (GSFC) in
Maryland with all science planning done at the newly established Space
Telescope Science Institute on the campus of Johns Hopkins University in
Baltimore. To those of us who worked
there, ST ScI would become known simply as the 'tute, truly an
internationally-run observatory whose telescope just happened to be in orbit.
My own role on Hubble was a modest one. I had an MS degree in astronomy with
specializations in celestial mechanics and in astrometry, the science of
positional astronomy that compiles positions of stars and other celestial
objects. Thus it was no surprise that my
first assignment was to work on the attitude determination system that would
use data from spacecraft sensors to determine Hubble's pointing to an accuracy
of better than an arcsecond. The name of the project was PASS, an acronym for POCC Applications Software Support, with POCC itself being an acronymn for Payload Operations Control Center. Our often repeated inside joke was that we had to be an acronym within an acronym in order not to be a somewhat impolite-sounding POCCASS.
Hubble was to be controlled to an accuracy of 3 milli-arcseconds, finer than
any pointing control that had been attempted until that time. It was to be done using data from gyroscopes,
sun sensors, and star trackers. In 1978,
on an earlier mission, I had already made my acquaintance with the Fixed Head Star
Tracker (FHST). Hubble was equipped with
three of them, and they would be used to update gyro-based attitudes after
every spacecraft slew to a new target.
An FHST had a field-of-view (FOV) of 8-deg by 8-deg and could measure a
star's position to 20 arcseconds, good enough to go to the next step of
determining what stars were in the field-of-view of the telescope's main optics
and use them to determine Hubble's pointing to the sub-arcsecond level. Little did I know when I joined the project
that I was to become the most knowledgeable person on FHSTs, eventually
becoming known to many as Ms. FHST.
Cutaway Diagram of a
Fixed Head Star Tracker
Hubble was scheduled to be launched by the Space Shuttle in
October 1986, and we were all under pressure to complete the ground control
systems on time. The pace was frenetic,
and from one system audit review to the next, it was becoming clearer that we
would not be ready. But a shuttle launch
could not be changed without upsetting all of NASA's mission schedules. Senior managers began to think that we would
launch Hubble and let it sit in safe mode in orbit, a sort of minimum energy cocoon mode, until the ground systems could be finished and
tested.
That went out the window on January 28, 1986, when the Space
Shuttle Challenger exploded 73-seconds after launch, killing all on board in
the most tragic space accident experienced by the US until that time. After the tragedy of the loss of all the
astronauts on board Challenger had sunk in, we began to realize that our own
problem now was not whether we would be ready for a launch in October 1986 but,
rather, whether Hubble would be launched at all. Would the Shuttle ever fly again? After a few months we were assured that
Hubble's launch would take place in 1988.
That launch date soon began to slip, however, leading to our inside joke
that we knew the true Hubble Constant[1] to be “two
years until launch.”
For me the morning of 24 April 1990 was one of sitting in
front of the television and watching the launch and feeling the same thrill I
had felt at every launch since the early days of the space program. This time, however, the thrill was even
greater, for Shuttle Discovery was carrying a mission that I had a direct role
in.
My own launch excitement in the sense of work, however,
began two days later on the morning of April 26. That afternoon the Canadian-built manipulator
arm was to remove Hubble from the shuttle bay and release it into space. Hubble's systems were being turned on
one-by-one and tested before the release.
I had just arrived at my office a short distance from GSFC when a PASS friend and colleague called. I don't remember his precise words, but they were something like, “Robyn, get out here. We can't identify what stars the FHSTs are
seeing.” A chill went down my
spine. If Hubble were to be released
without the FHSTs being able to identify star patterns, Hubble would be
literally lost in space, locked into
its cocoon-like safe mode until engineers like me could figure out what
had gone wrong.
An hour later I was sitting in front of a terminal in the
Space Telescope Operations Control Center (STOCC) at GSFC. My colleague and friend explained, “We've
been trying ever since the FHSTs were turned on, but no matter what we try, the
algorithms can't identify the star patterns.”
As calmly as I could, I asked, “Can you get me all the FHST telemetry since the
trackers were turned on? Let's start
reprocessing from scratch, taking it step-by-step and paying close attention to
detail.”
 |
| The STOCC at GSFC |
From my experience on an earlier mission, I already knew
just how temperamental FHSTs could be.
These were instruments from before the days of charged couple devices (CCDs). They used simple optics and an image
dissector tube, and they could observe only one star at a time. A controllable magnetic field was used to
cause the dissector tube's photomultiplier and photocathode to scan the FOV in
a serpentine pattern and lock onto any object brighter than a threshold
magnitude for 20 seconds before breaking track and continuing the scan. FHSTs had been known to track not just stars
but the Moon, planets, nebulae, other satellites, space debris, and even bright
cities on the Earth's limb. The trick
was to edit out all the junk so that only star tracks remained and then
massage those tracks into point images using gyroscope rate data that measured
moment-to-moment spacecraft motion.
Finally, these FHST-measured star positions would be passed into a
pattern match algorithm that would take the measured positions and compare them
with positions in a star catalog. That
pattern match algorithm required fine tuning in order to work reliably. All-in-all we had just a few hours to get it
right before Hubble would be released into orbit on its own.
Slowly, as calmly as we could, we began reprocessing
telemetry from the start. We edited out
spurious objects. We adjusted the
editing parameters to get star images with the smallest possible clump size. As we worked, I became dimly aware of the big
screen that hung at the front of the STOCC.
There was Hubble, perched on the manipulator arm, as the solar arrays
began to unfurl, unrolling from their containers and glistening like
ever-lengthening, golden sails in the bright sun. Just as the second solar array finished
unfurling, we did it. We identified the
stars that were being seen by the FHSTs.
We did some hand calculation sanity checks to make sure we had
identified the right stars. We had. “Now let's do it again with another data
set,” I said.
 |
| Hubble on its Own, Released from the Maniupator Arm |
One data set after another, we repeated the process, making
further adjustments until we could identify stars correctly without further
intervention from us. The algorithms we
had designed were working. A higher
level mission manager approached and asked, “Are we GO with the FHSTs?” We nodded yes. Shortly after we watched in real time as
Hubble drifted away from the arm and from the shuttle. We had done our part. Hubble would not be lost in space.
That was my role 25-years ago. My day in the STOCC as the solar arrays
unfurled is one of those images frozen in my long-term memory. Hubble didn't have an easy start. Soon the newspapers were joking about Hubble
Trouble when it turned out that the telescope's main mirror had been ground
to the wrong figure and suffered from spherical aberration that was
giving blurry images. My FHSTs were not
out the woods yet either. Another part
of PASS, the Mission Scheduling System, was attempting to
use the FHSTs in a way they had never been used before by commanding them to
lock on to preplanned reference stars after each telescope slew to a new
target. The FHSTs were failing to find
the right stars one time out of three, each failure resulting in the loss of
science observations for a good part of an orbit. It was the second largest problem in Hubble's
early operations right behind the flawed mirror.
As they say, however, the rest is history. Once the mirror's spherical aberration was
understood, it was possible to grind corrective lenses that were installed by
astronauts on the first servicing mission to Hubble in December 1993. Those corrective lenses were known by the
name of COSTAR, Corrective Optics Space Telescope Axial Replacement, and they
silenced the cries of Hubble Trouble, enabling Hubble to give the crisp images
that have become part of both our scientific and cultural lives.
For my part, I was brought onto a team whose mandate was to reengineer the Mission Scheduling System. We
were known as MSRE, the Mission Scheduler Re-engineering team. We pronounced MSRE like ms'ry, and thus our inside gallows humor was “MSRE loves company.” My part of the mandate was the Pointing
Control Subsystem. Over the next several
years, working as a team, we improved the FHST reference star success rate to better than 99%.
The last effort I had a small hand in before leaving HST and PASS in 2005 was the design of what became known as the Two-Gyro Science Mode
that would radically change the pointing control algorithms in a way that had
never been attempted before. A gyro
gives information in one dimension, and thus three gyros are needed to know a
spacecraft's orientation in three dimensions.
Six gyros were installed on Hubble for redundancy and in the knowledge
that gyros are mechanical devices that eventually wear out and fail. Hubble's gyros began to fail within a few
years after launch, but they were replaced during servicing missions. After the Space Shuttle Columbia disaster in
2003, however, all future servicing missions to Hubble were canceled. Of the six gyros on Hubble, three had already
failed. It was only a matter of time
before yet another would fail and force Hubble into permanent safe mode, ending
its mission of scientific discovery.
The idea behind this last effort on Hubble was to take me
back to my FHSTs. Gyroscopes give rate
information, whereas FHSTs give position information. But could we watch stars as they
moved in an FHST FOV? Could those position measurements be used to compute a rate,
effectively allowing the FHSTs to take the place of one of the gyros? The answer was yes, they could. The newly designed control algorithms were so
successful that NASA shut down the third of the three remaining operational
gyros in August 2005, keeping it in reserve and thereby extending Hubble's operational
life. Even after a final servicing
mission to Hubble was reinstated and six new gyros were installed in 2009, Two-Gyro
Science Mode has remained the primary control algorithm for Hubble.
How long will Hubble continue to provide us with the
beautiful photos and ground-breaking science for which it has no equal? Current estimates are that Hubble will
continue to operate at least until 2018, when the next generation James Webb
Space Telescope is scheduled for launch.
It may continue in operation well beyond that as long as budgets allow and
spacecraft systems continue to function.
Not bad for a telescope that was designed and built with 1970s and 80s
technology and that many thought would not last for its original projected
lifetime of 15 years.
If you're wondering by now how it was that this engineer
left the Hubble project to start a diplomatic career with the U.S. State
Department, the answer is that even in those days, I had something of a double
life. Outside of my day job on
the Hubble project, I was known as a historian of Soviet science. In the summer after Hubble's launch, I
published perhaps my most important history work on Soviet astronomy in 1936-37
during the height of Stalin's Great Purges.
When I left the Hubble project in 2005, in a sense I exchanged my hobby
for my career, my career for my hobby.
But on this April 24th, on the 25th anniversary of Hubble's
launch, my mind will be back there, reliving the moments of frustration and
exhilaration and recalling the faces and names of so many colleagues and
friends from the PASS project who were there at the beginning.
And Ms. FHST will smile and feel an inner warmth to know that her
children-in-engineering, those three Fixed Head Star Trackers on Hubble, have
not missed a beat and continue to guide Hubble on to discoveries that take us
back ever further towards the dawn of our Universe.
 |
| 2014 Reunion Picnic with PASS Friends and Families |
[1] The actual Hubble
Constant is a measure describing the expansion of the Universe. The current best estimates are in the
vicinity of 71 km/s/Mpc, where Mpc is a megaparsec, a distance of approximately
3.3 million light years.
No comments:
Post a Comment