Launching & Activating Swift

By Frank Marshall

Swift is a medium-sized Explorer (MIDEX) to study gamma-ray bursts (GRBs). It was developed as a collaboration between NASA; the UK; and the Italian Space Agency (ASI), which provides the primary ground station (the Broglio Space Center in Malindi, Kenya). Swift is managed by GSFC. After numerous delays, Swift was finally scheduled to be launched on November 17, 2004, at 12:09 pm (unless stated otherwise, all times are EST). The Pennsylvania State University (PSU) operates the control center (MOC) at State College, PA. As the lead of the Ground NEtwork for SwifT (GNEST), I was at the MOC to help prepare for the launch. This page relates my personal experiences leading up to the launch, the launch itself, and the first days of the mission. It is not intended as any sort of official report. I use initials or console position IDs instead of names, but I am sure most of the Swift team can easily identify the people.

Sunday, Nov. 14, 2004

One of my responsibilities was to report the readiness of the ground system to support the Swift mission to the launch team at Cape Canaveral. This is part of a carefully scripted launch countdown procedure that starts 7 hours before the scheduled launch. The GNEST runs data flow tests with its various elements to verify readiness and also does status polls with the elements. The final rehearsal of this procedure was on Monday, Nov. 15^th. I traveled to PSU the day before to review the final preparations and to be available early Monday morning.

It is about 200 miles from my home in Columbia, MD, to State College, and the roads most of the way are limited access with at least 4 lanes.  The trip takes about 3½ hours if there is little traffic. There are several points of interest along the way. I-83 runs from the Baltimore Beltway to Harrisburg, PA. In Maryland, it parallels the Northern Central Railroad Park, one of the best bike paths in central Maryland. In all my trips to PSU, I only managed to use the path once because the launch was delayed until late in the year. The path uses the abandoned bed of the old Northern Central Railroad, which provides a relatively flat path. Except on weekends, the path is not crowded and provides a lovely tour of the rural countryside. At the Mason-Dixon line, the bike path becomes the York County Heritage trail, which continues north for about 20 miles to York, PA. York is the home for the York Barbell Co. They take advantage of their location next to I-83 to advertise with a rotating sign to the many passing motorists. A few miles south of Harrisburg on the Susquehanna River is the site of the Three Mile Island nuclear power station where Unit 2 had a partial meltdown in 1979. It is just visible from I-83 as you cross the Susquehanna. North of Harrisburg is a 1/17th scale model of the Statue of Liberty that was placed by helicopter on a pier of an old railroad bridge in the middle of the Susquehanna. The bridge itself is no longer there.

Monday, Nov. 15^th

The final rehearsal starts at 9 am Monday and runs until a simulated launch at 12:09 pm. At the MOC are the Flight Operations Team (FOT) who will be sitting at the consoles communicating with the observatory, engineers from Spectrum Astro (now General Dynamics) who are responsible for activating the spacecraft they built, and several people from the GSFC Project Office. After launch, representatives from the three instrument teams will arrive to oversee activation of the instruments. Officially there are separate countdown procedures for the MOC and the launch team at KSC, but at the MOC we follow a merged script I produced using some Visual Basic code in Excel. We skip several of the least interesting stretches of the procedure to reduce the length of the rehearsal. The test team deliberately creates problems for us to make things more difficult, and we manage to create some problems without their help. Overall our team does well, some slight modifications are made to the procedure, and we feel that we are ready for launch.

At the end of the day, we all head for Prospector's Allegheny Rib Co., a nearby restaurant, for a team-building exercise. The exercise has an inauspicious start as the Spectrum Astro people arrive late and have to sit at their own table. After a round of drinks, team building starts in earnest. Someone on the Spectrum team decides that we should nominate people for faux awards. The menus are a convenient source of writing paper, and since they are in the shape of a pig, the awards become known as the "Pig Ass Awards". While started by Spectrum Astro, soon everyone joins in. That evening people are only nominated for awards, and there are often multiple candidates. I am honored to be nominated for the best dressed at a formal review award, which is obviously a reference to my casual dress code. The next day the menus are taped to a wall (left photo) in a hallway in the MOC. Somehow the awards do not seem quite as humorous to some who were not involved in producing them, and they disappear after Thanksgiving.

Tuesday, Nov. 16^th

Tuesday was pretty slow at the MOC. We produce a new integrated countdown procedure to incorporate changes from the rehearsal. I went to bed at about 8 pm after asking for a wake-up call at 1 am Wednesday morning. I am preparing to work a  midnight-to-noon shift.

Wednesday, Nov. 17^th

There were many simulations of operations and the countdown, but we are unprepared for what really happens. Very early in the morning, the launch attempt is scrubbed, and the Swift team scrambles to notify everyone quickly. Almost everyone gets contacted, but there are a few notable exceptions. Somehow the MOC system administrator is not informed, and he does not look happy when I arrive in the MOC. They call my personal cell phone, which I left home, instead of the Blackberry I have with me. They call the hotel that I checked out of the previous day and are put through to my (old?) room. Fortunately no one answers. I was sent an e-mail at 1:35 am, and I read it about 2 am in my hotel room well before I need to be in the MOC. The Deputy Project Manager arrives in State College in the morning unaware that the attempt has been scrubbed.

We later learn that there was an intermittent problem in the command destruct system, which is used to destroy the rocket if it veers off course. By that evening it is decided that there will be no attempt Thursday, and Saturday appears to be the most likely day.

Thursday, Nov. 18^th

I wake up as planned at 1 am, read the latest e-mails, and decide to go home. I check out at 2 am. It is warm, and a light rain is falling. There is almost no traffic except for large trucks until just outside of Harrisburg.

Friday, Nov. 19^th

I was not sure whether I should drive up to the MOC today or wait until early Saturday if we attempt to launch. Eventually I decide the slight risk of not making it Saturday morning is not worth the extra day away from home. (Actually I just couldn't seem to get moving.) At a 5 pm telecon with most of the Swift team it is announced that the problems with the command destruct system have been solved and we are go for launch. Everyone sounds very happy and is having a good time. Almost giddy.

Saturday, Nov. 20^th

I awake at midnight, quickly eat breakfast, load the van, and hit the road about 12:30 am. Unlike the drive back from PSU Thursday morning, there actually are a few cars on the road in Columbia. Once I get past Harrisburg, though, there is very little traffic. I only encountered 3 cars on my side of the road until I reach State College. The MOC (upper left photo) is almost empty when I arrive at about 3:45 am. I head to my "office" (lower left photo) to change into the formal shirt that I have not worn since the launch of BBXRT in 1990. It takes several attempts to make an acceptable knot in the bow tie. Add my curiously smaller cumber bun, and I am ready for launch. Gradually people arrive at the MOC. My shirt is a big hit. DS takes a picture (right photo). Get a big smile from LP when she arrives. We have a history of being over/under-dressed for reviews (mostly under). Countdown starts at L-430 minutes (5 am EST).

Things don't go well right from the start. Two positions for the ground (FDF and USN) are not on console. We managed to contact USN by cell phone. Since both positions are staffed in time their activities, in the end it did not really matter.

Shortly after the start, Spectrum Astro comes into the control center and points out that their method for calculating the start of one of the early contacts (with USN, I think) produces a different answer than the MOC's official product by about 19 seconds. Further the FDF product is also different. SS and SA investigate the problem and eventually discover that 3 different positions are being used for the station. In addition, SA was using a slightly different separation time. FDF may also have used a different viewing constraint (0° above the horizon instead of 2°). Eventually we conclude that the official MOC product is accurate to about 1 second and that this is a negligible difference. SS then checks the Malindi predictions and verifies that these are also accurate to about 1 second.

Malindi has trouble dialing into TechCmd (technical commentary via a commercial phone line) for reasons that never become clear. Data flow from Malindi goes well (starts at L-391). USN data flow starts at L-350 and also goes well. We cannot confirm the byte count at the call to FSM at L-325, but it confirms a short time later. The data flow with SN starts at L-320 (about 11:49 UTC) and becomes a real circus. We test uplink and downlink with both ground terminals (WSGT and STGT). There are also two data paths to each. The NOM suggests testing both paths by having the prime link fail even though it is not in the script. The result is that we cannot confirm that the byte counts are correct (we apparently did not get all the data). The cause of the discrepancy is not immediately clear. I have to report at L-275 to FSM that byte counts don't match, and we have a RED condition (not ready for launch). This got everybody's attention. FSM calls back and wants to know the number of the criterion that was not being met. This confused me because my version of the document did not have the criteria numbered (it was in an Excel spreadsheet). After looking at the document and the accompanying spreadsheet, I gave up on the number and simply read the description. FSM agrees that we are red. Eventually the FOT figured out that the MOC only captured the data from the primary data path and our byte count only included these data. We were surprised that none of the data using the backup path were captured, but a phone call to DA confirmed that this was the design. I talked to SMD and FSM on the SN COORD voice loop, explained what had happened, and reported that we were GREEN (ready to launch). We then ran another data flow with the SN and practiced the plan for moving to the back up data path. It went smoothly (but on-orbit the FOT will first have to notice that we are not getting data).

The initial attempt (at L-275?) to ping the Swift Contingency MOC (SCM) at GSFC failed. Since there is normally nobody there, we were not sure how to troubleshoot the problem. We called the SCM (it is shared with another mission), but no one was there. We were about to talk to the NOM when another attempt to ping worked. It was never clear why it did not work the first time, but it was not required for launch anyway.

At L-205 FDF reported that they had not been able to confirm that Kwajalein and Kaena Pt. were ready to do the early (C-band) tracking. This was not a mandatory requirement though. At this point everything on my worksheet was green except these two tracking sites. I had "improved" the worksheet created by LP, and had to correct a couple of my mistakes to get everything green.

At some point early in the countdown, it was noticed that one of the cables from the tower to the rocket was hanging down lower than expected. There was concern that it might get tangled on one of the solids after launch. A team went out to the pad to adjust the cable. This delayed loading the fuel and may have contributed to the 6-minute launch delay.

Eventually the countdown clock reaches zero (L=0), and Swift launches! We had invited everyone into the control room to watch NASA TV on the monitor. As the audio announced ignition and liftoff, nothing happened for several seconds on the screen in the MOC. This weird situation was caused by the delay in getting the streaming video. In any event, the launch went flawlessly. We watched (left photos) as the Delta II cleared the tower and streaked into the clear Florida sky. The three solid rocket motors burned out after about a minute and dropped off. Ten minutes later Swift was effectively in orbit although there were subsequent adjustments to the altitude and inclination of the orbit. Team members took lots of photographs in the control room. It was like getting married (a little). Images taken by KSC of the launch are available at the Swift Science Center web site. There was also a PSU video camera, which can be seen in the near left photo, that was running during most of the countdown. The audio was enabled at about L-20. The local press and MC's husband came in to take a few photos.

Swifties at both the MOC and the Cape were jubilant. So many years of work by so many people (left photo) were finally about to be put to use. A lot of congratulations on the voice loops. The FSM wished the MOC "good luck". We may need it. I called the FSM and congratulated and thanked him. I also called LS at the Malindi Ground Station, and we exchanged congratulations. Upstairs people were enjoying themselves (right photo).

Because the launch was 6 minutes late, we needed a new ephemeris (which contains the information needed to predict the satellite's position) for Malindi and the SN. FDF had the responsibility, and we had reminded them when it was announced that there was going to be a slip. The first two-line element (TLE) file (a specific format for ephemeris data) produced by FDF indicated that the first contact with Malindi was going to be delayed by about 30 minutes. This was obviously wrong, but FDF maintained that the TLE was correct. (We later learned that FDF was merely forwarding a TLE provided to them.) After a bit, they agreed to go back and try again. Meanwhile separation of the spacecraft from the launch vehicle was rapidly approaching (L+80), and we would immediately try to contact TDRSS. We decided to have SS compute our own TLE with STK by simply adjusting the epoch of the pre-launch TLE by 6 minutes. He then sent this to Malindi. Before separation, FDF had their new TLE and stated on the voice loop that this version computed a 12-minute delay for the first contact. SS went off and compared the predictions of the new FDF TLE and his adjusted TLE, and they actually agreed very well. Apparently the "12-minute" call was simply a misstatement. They were so close that we decided not to update the TLE that we had already provided to Malindi. USN did their own adjustment without any "help" from us.

There was a camera on the launch vehicle to show separation in real time, and we all watched it live to much applause. (I later heard that other people had seen what appeared to be another satellite off in the distance that stayed in the image for many minutes. If true, this would not be good news because there would be a small chance of a collision as the orbits evolve. There are actually short stay-out intervals in our launch window to avoid other satellites.) We now waited anxiously for the TDRSS contact that was to start almost immediately. We were not even sure which version of the ephemeris the SN was using. Great relief when we started getting data. Videos of the launch and separation are available from KSC. The first Malindi contact was at L+133. Before launch DA said that he thought the contact would not work. I offered him a $10 bet that it would work, but he turned me down. After the confusion with the TLEs, he decided to take the bet. If we got any data from Malindi, I would win the $10. Everyone was worried about the contact, but it went perfectly. We sent a no-op command to verify the uplink, sent a few individual commands, and then uploaded the ephemeris that SS managed to compute just in time. Wow! At the end of the pass, DA handed me the money (left photo) and said that this was one bet he was happy to pay off. We then all applauded SS's intense efforts. People were happy  in Malindi  as well (right photo).

I later learned that the next contact (the first one with USN) failed (apparently because of a configuration problem at the ground station). This was rather ironic because our testing with USN had been amazingly reliable while testing with Malindi had lots of minor problems. Subsequent contacts with USN later in the day were all successful.

Oddly there was a home PSU football game with Michigan State with kickoff also scheduled for 12:10 pm. It was generally expected that PSU would lose badly, but they actually won 37 to 13. A happy day for Joe Paterno, enshrined in a bronze statue outside the stadium (left photo), and the Nittany Lions. Because of the football game, it was difficult to find a reasonably priced hotel room in State College on Saturday. I left for my hotel in nearby Altoona about 3 pm. I take back roads most of the way, but on the final leg I drive I-99, whose out-of-sequence number is due to Congressman Bud Shuster. Getting briefly lost in Altoona gave me the opportunity to see the Jaffa Shrine. It would fit right in at Malindi.

Sunday, Nov. 21^st

The launch is front-page news (but below the fold) in the local State College newspaper the Centre Daily Times. Erin Nissley's article, (or local copy) with the headline of "PSU telescopes rocket into orbit", had a decidedly local slant and a nice picture of TT.  The launch was also reported by CNN, ABC, and the BBC.

There was a lot of good news when I returned early Sunday morning. The solar arrays deployed exactly as planned, the solar cells produced the expected amount of power, and the battery performance was nominal. The solid state recorder was working, and the attitude control system was working using the Sun and the magnetic field as references. An ephemeris for the spacecraft had been loaded, and the spacecraft was autonomously selecting the correct hemispherical antenna.

On Sunday we began what was expected to be a two-week process of checking out and calibrating the attitude control system (ACS). Late Sunday we attempt to transition to inertial pointing (the mode that we needed for the mission), but immediately had a problem, and went back into Sun-pointing mode. This pattern of try something, learn something, make adjustments, and then test something else continued for the rest of the week.

During the early part of the mission most people are working 12-hour shifts with another hour or more at the MOC during handovers. Since I had worked early in the morning during the countdown, it was easier for me to take the night shift and let LP take the day shift. Things were calmer on the night shift. Fewer activities were planned in part because the Malindi gap (a span of several orbits in which no contacts with the spacecraft could be scheduled) was at this time during the night shift, and fewer people were around.  In addition, KM, the Spectrum Astro lead on the night shift, kept things on an even keel. The long hours gradually wear everyone down, but the stress must have been worst on the small Spectrum Astro team who were responsible for the safety of the observatory.

Monday, Nov. 22^nd

This morning someone decides to BBQ hot dogs in the kitchen with their propane grill. The dogs were tasty, but the cook was encouraged to move the grill outside.

Most of NASA's communications network "freezes" their configuration shortly before and after launch to avoid any unexpected problems during this critical time.  PSU also temporarily  froze their communication links. We knew the freeze was effectively over on Monday when construction began on a new building across the street from the MOC (right photo). Fortunately they managed not to dig up any utility lines.

Tuesday, Nov. 23^rd

The rumor that Swift was in an unexpectedly low orbit finally reached the MOC. Apparently it had been bouncing around NASA HQ and the Explorers' Project Office ever since launch. The rumor was completely bogus as anyone could tell by checking the publicly available orbit information. It was not clear how the rumor started.

Today the electronics for the X-Ray Telescope (XRT) were turned on. Everything looked normal. This was the start of a several week process of activating the 3 instruments.

Thursday & Friday, Nov. 25^th & 26^th

Thanksgiving. We have a lot to be thankful for, but the MOC is not where most of us want to spend the holiday. PSU graciously provides a large dinner for each of the two work shifts. I skip town, though, and meet my family in Harrisburg. We have an enormous buffet-style Thanksgiving dinner in a local restaurant, and then top it off with another all-you-can-eat breakfast (left photo) at the hotel. After dinner we drove about 10 miles to see the Three Mile Island nuclear power station (right photo). It is a bit unnerving to find that one reactor is still in use. We visit Hershey mid-day Friday and learn how to make chocolate. Friday afternoon I return to State College.

Saturday, Nov. 27^th

I decide to arrive at the MOC after the Malindi gap, which means showing up about 3 am instead of 10 pm. I end up staying later, but I still have an extra couple of hours off-duty every day.  I finally get to use the hot tub and pool at my hotel.  This morning BAT starts the long process of activating their large detector array.

Monday, Nov. 29^th

BAT turns on its second group of 256 pixels (out of a total of 32,768 pixels) shortly after the Malindi gap. Again everything looks normal. Things are going well, and I am comfortable with returning to GSFC. I leave about 11:45 am. Traffic is light, and I'm home shortly after 3 pm. There was an unusual number of pick-up trucks parked along the road between State College and Harrisburg. Presumably these were left by hunters enjoying the opening day of deer season.

Initial Summary

It is risky writing a summary only 10 days into the mission, but overall things have gone well. We are very close to the so-called Sunny Day Timeline, which no one thought we would keep to. A few mistakes have been made, but they were quickly overcome. Despite extensive testing of procs, one proc was run with a mistake that prevented it from running to completion. There is some truth to the description of the operations team as a collection of co-located tribes, but at least all the tribes are intensely interested in making Swift a success. The FOT and the Spectrum Astro team are working together well. The minor problems with the MOC hardware and software and support from the Space Network and ground stations have not impacted the activation process. Hopefully we are well on the way to a very successful mission.

End of the Beginning

The contract with Spectrum Astro called for the official delivery of the spacecraft after a 30-day on-orbit checkout period, and the review provides an opportunity for Spectrum to show that the spacecraft meets its requirements. The review goes very well. Toward the end of the SAR, the PI announced that earlier that day BAT had imaged its first gamma-ray burst, and the team would announce the discovery to the world later that day. There are a few things (mostly documentation) that need to be finished, but at the end of the review the Project Manager accepts delivery of the spacecraft for NASA. A brief celebration with champagne follows (photo at left). We trade stories about our adventures and misadventures during the past several years. Smiles all around.