The United Kingdom High Altitude Society’s 2nd Annual Conference, Sept 22, 2012, in London, UK, will feature high altitude balloon talks from several balloonists, including myself, this year in person!  I gave a talk last year on a few challenges facing small superpressure balloon development.  Since then I have discovered much an incredible amount of historical information on the subject.  This year I will be speaking about the fascinating engineering history of hundreds of small superpressure balloon test flights that have taken place around the world since 1950.

Watch live at the UKHAS conference page.  After the conference, expect video, slides and references on that site, as well as my ballooning advice website, High Altitude Balloon Consulting.

More details will be posted soon, in the mean time check out http://track.whitestarballoon.com for the flight data.

5/12/2012 1:22 am: Sensor controller malfunctioning.  Removing camera and humidity sensor.  Going to fly without them.  If all we have is a bag of gas were still flying!

Follow the live twitter stream: https://twitter.com/#!/Lallynaut

Track Speedball-1′s flight: http://track.whitestarballoon.com

5/12/2012 10:36PM: Starting final pre-flight checklist.  (Here we GO!!!)

5/12/2012 9:45PM: We’re just getting rolling here at mission control.  Launch crews are on their way to the launch site now and will be joining the live stream momentarily.  Click play on the video below to watch the live stream of Speedballs first flight!

Link for live video stream and chat will appear here.

Tracking page: http://track.whitestarballoon.com  (note this may show (and be marked as)simulated testing data up until about 5PM ET Saturday.

Launch time: 7:45PM ET Saturday (2345 UTC 12 May) 10:00PM ET (02:00 UTC) May 12th.

Launch Location: Spaceport Indiana (KBAK Airport)

Cruise Altitude: 70,000 ft  (We’re movin’ on up!)

Flight Duration Expected: 24-72 hours

Estimated Ground Track:  Very slow wandering around the middle USA for several days.

Balloon System: A hybrid floater, with a latex balloon doing the heavy lifting to 21km altitude, far above the ZP’s original design altitude of 10km.  The ZP will still provide the function of venting helium to arrest the climb and establish a float.

•  Zero Pressure Polyethylene envelope, 660 ft^3, custom made Global Western design, with White-star designed one-way plate valve, featuring a silicone seat, and PVC pipe port at the top for He temp sensor.
• Latex envelope, 1600g, Hwoyee latex tow balloon balloon courtesy of WB8ELK, Bill Brown.

Payloads of note: 

• Remotely triggerable Go Pro Camera (Simultaneous wide angle view of both balloon above and horizon/ground in front)
• Supervising AVR Flight Computer communicating with 3 Arduino modules via I2C, including SatcomController (Mega 2560), SensorController(FIO), and BallastController(Boarduino)
• Iridium-Arduino Satellite Communication Shield by White Star with command uplinking and adjustable telemetry intervals
• Nichrome deadman cutdown by Carl Lyster, WA4ADG
• ZP balloon lift gas anti-dilution valve by Gary Flispart
• 54 Energizer Lithium AA Batteries
• 5.5lbs of C2H6O ballast in a SNOX-style electric solenoid drain ballast bottle
• Humidity Sensor
• Cloud particle sensor
• Temperature sensors
• Helium internal temperature sensor
• Raw battery pack voltage sensor

Lift Helium proudly provided by SpacePort Indiana and Praxair Gases.

Flying across the ocean is no small feat.  It takes the concerted efforts of dozens of people, working hard at lots of difficult problems, from modeling balloon volume and flight dynamics, to planning interactions with air traffic control.  The diagram above gives a little bit of an idea of the effort involved in getting across the ocean.  Any single block represents tens to many hundreds of man-hours worth of effort.

Components in Purple represent things which will actually be flying across the ocean.  This hardware and software must perform flawlessly at all times.  Components with a red heptagon represent significant software efforts.  The red square shows the components which lie on amazon EC2, spread across three instances, with a total cost of $100 a month (during flight season) to maintain.  Pink commands are sent using PubNub, a service without whose generosity our public page would not be possible.

All of these systems are in the critical path, and a failure of any single flight system will compromise science data.  Fortunately, we always have positive control of our craft, thanks to a dead-man cutdown, which operates entirely autonomously, and a 9602 modem which will respond with rough location coordinates even if all other flight systems have failed.  Our ground systems all have hot-backups, and can all be operated from anywhere on the Internet, so these systems are as redundant as they can be.

Stay tuned for more updates at http://whitestarballoon.org , we are on track to launch in the next month or two!

Dan Bowen
@SteamFire
White Star Team Lead

Hi All, the flight of the SuperPressure Initiated Termination Balloon Flight 1 was successful as far as we can tell so far.  We don’t know for sure because we haven’t gotten the SD cards that hold the video and science data on them.  Technical results as they come in over the next week will be posted on our wiki site, here.

With just skeleton assembly, launch and mission control crews preparations for this flight had to be less organized than we’ve done in the past for SpeedBall flight attempts.  We allowed for that by targeting an early launch time, with plenty of daylight left for schedule slip.

The massive zero pressure balloon was launched from SpacePort Indiana around 3:30 PM EST, by having the payload support crew run with the wind as soon as they released the balloon, allowing the balloon to rise straight up as it lifted each payload.

The balloon landed about 1.5 hours later on a family’s home in Miamisburg OH and was recovered shortly after.  The family made a youtube video of the balloon on the house!

Thanks to all who helped!   Some science observations follow.  The three pics above were emailed to us by the recovery crew in Ohio.   More pics available on Flickr here.

Science objectives and observations thus far (not final yet):

Rotation rate of balloon envelope: Observed a rate of ~10 revolutions per minute in the first 30 seconds of flight before cloud entry.  This may not be solely due to the asymmetric balloon shape, as the inversion rope was seen to have been bearing a part of the load in the flight videos.  It should have been loose and not bearing load.  Bearing the load would cause it to tip the balloon over slightly, and may exaggerate the aerodynamic asymmetry.  This unplanned rope tension may have invalidated the experiment.  Likely caused by simple miscalculation of required length.

Cutdown inversion method performance: The cutdown was set to a 61 minute timer.  All that is really known about the performance is that the cutdown did fire as designed, as seen in the recovery video.  It also did not simply rip the top off the balloon, which was one theorized possibility.  The action of the balloon inversion is unknown until we get the onboard video.

Vent valve performance: The performance of the vent can’t be determined until we see how the balloon inverted from the video.  The helium did not completely empty from the balloon, as seen in the recovery videro but that may not be because of vent failure.

Burst pressure measurement: Unknown until we get the data cards back.  Burst appeared to occur just a few minutes before cutdown, based on radio telemetry of altitude, but that was not clear.

Altitude vs. Time Chart Small with Climb Rate Color

Update: DATE CHANGED TO SUNDAY November 20 2011

See live streaming and live web tracking!

Full details available on the White Star Wiki page, SPITBall-1

Slides from presentation and Source Paper Citations in text format

Now, background info to get you up to speed on the state of amateur superpressure:

This type of balloon has the potential to stay in the air for extremely long times, much longer than the 3 days flight estimated for the White Star’s trans-atlantic ‘zero-pressure’ balloons.

Superpressure is merely the condition where the gas inside the balloon is pressurized higher than the air pressure outside of the balloon. This can only happen with a balloon that doesn’t swell up like rubber nor can it be allowed to burst. Practically, you need a super strong plastic bubble.

Mylar is very strong, so Mylar foil (polyester) balloons from parties have been tried many times. Much math has gone into trying to calculate just the right lift and weight are needed to make the balloons pressurize and achieve level flight. Unlike the success in making and using ‘zero-pressure’ balloons, amateurs have had little, if any success making and using superpressures. The goal is a level float that lasts day and night, without needing to drop ballast.

Most flights seem to rise and simply burst, or more rarely, achieve float, but sink to the ground at sunset. I’ve taken the time to begin researching publicly available science research papers from the wonderfully open, various US meteorological journals. What follows in the presentation, is a summary of the things I’ve found that might explain why it isn’t working, and how we can make it work.

Here’s a link to the rest of the presentations on video.

We kicked off the first meeting this past Thursday (Sept. 8th) with a presentation by Dan Bowen. Topics discussed were the design & mathematics behind super pressure balloons, and the valve design used in Japan’s old incendiary, zero-pressure balloons.

This week’s meeting (Sept. 15th) will focus on defining tasks to accomplish this season in order to launch Speedball-1 and develop future super pressure balloons.

Altogether, expect to see lots of super pressure balloon experiments, a revision to Speedball-1′s envelope, and the launch of Speedball-1 this season!

Want to get in on the action? You still can! Meetings are open to everyone. Join us weekly on Thursdays, 8PM EST at LVL1 Hackerspace!

For everyone we met – remember to put your email address in the field on the right, we’ll send you an email the day before we’re about to launch. –>  We have to wait for winter, and then wait for the Jet Stream winds to be just right before we can choose a launch date.

After months of prep work, we had a flawless demo at MakerFaire Detroit. The electronics from SpeedBall-2 flew Gary’s exhibition balloon up and down and up and down for 9 hours straight, two days in a row, with no intervention needed from the crew.  Before we unveiled SpeedBall-2′s indoor flight, we spent days flying it at our LVL1 Hackerspace in Louisville, KY. The ballast algorithm needed much adjustment, which gave us some invaluable experience in understanding how the balloon reacts to the real world.

We were interviewed by Make Magazine, The Henry Ford museum, and talked to hundreds of visitors over the course of two days.  Thanks to everyone that stopped by!  If you have any ongoing questions, feel free to ask us by email, comment, tweet, or carrier pigeon.

The device works by inserting a dowel rod with a specific pattern of holes perpendicular through the dispensing pipe. A servo with a custom made dowel-to-servo coupler (made with LVL1’s new laser cutter!) spins the dowel rod back and forth at a programmed rate determined by code on a Teensy 2.0, kindly donated by PJRC (http://www.pjrc.com/). As the dowel rod spins, BBs drop into the holes from the top and eventually fall out the bottom as the dowel rod rotates around.

As one could imagine, a major problem with any dry ballast design is jamming. We made two design choices to prevent this. First, we chose to spin the dowel rod back and forth, which prevents pinch jams between the holes and the pipe wall from permanently jamming the device. Second, to prevent bottle neck jams in the pipe, we chose to use a Sprite Zero bottle circa 2011 for its convex bottle neck and to taper the PVC pipe where it meets the bottle in order to create a smooth, fluid funnel.

After several days of testing, the dry ballast system has been successful and is ready to strut its stuff at Maker Faire. Come by the White Star Balloon booth at Maker Faire Detroit 2011 and check out Speedball-2 in action! You’ll be glad you did!

SpeedBall-2 is slated to make it’s public flight debut at the MakerFaire, (sorry SB-1!) as a lean-mean indoor flying machine.  It’ll be stripped down to the bare necessities required to control it’s altitude – arguably the most important innovation of the whole program.  Consisting of the classic SNOX Ballast bottle, the Flight Computer, the Comm Controller, and the Ballast Controller, the load frame and a small 12v pack of AA batteries, SB-2 will hang from a small custom plastic balloon.

Guided by a vertical cable to keep it from wandering around the museum, SB-2’s balloon it will have a calibrated leak, causing it to slowly sink down, a high-speed simulation of helium loss over a 3-day air voyage.  This will allow you to see the prowess of the ballast system and control algorithm, as it manages ascents, descents, and altitude holds, fully autonomously – look ma, no wires!

We’re working now on putting it all together and testing the demo setup at the LVL1 Hackerspace.  Follow our Twitter @LVL1WhiteStar for up-to-the-minute progress updates.

After 4 tries, and some expensive test equipment, the end result was a Quadrifilar Helix antenna weighing only 80 grams!

Our ground test antenna was a 5/8ths wave whip antenna, which works well, but unfortunately needs a ground plane. Tests with both a quarter-wave dipole and a J-pole antenna were lackluster. Documentation from our satellite service provider implied that a quadrifilar helix antenna would provide the best coverage at all. While these antennas are pretty, their design and construction was voodoo magic at first.

Thanks to some design documentation here: http://jcoppens.com/ant/qfh/index.en.php and some help from the balloon community, we had some baselines for creating such an antenna. We still went through *quite* a few revisions.  We went through 3 revisions that didn’t work, and one which works pretty darn well!

Here are the antennas which didn’t work:

And Finally, this is the antenna that DID work.  This antenna worked after we finally broke down and purchased an MFJ Antenna Analyzer.  The antenna analyzer allowed us to very precisely tune the antenna.  A Quad Helix antenna is two full-wave loop antennas, wrapped in a helix, and shorted at the feed-point.  One loop is tuned slightly above the target frequency, and the other loop is tuned slightly below the target frequency.  This affects the circular polarization which makes this sort of antenna exceptionally suited for satellite work.

The antenna is constructed of fiberglass, wood, and music wire.  The central support is 5/16ths inch fiberglass rod.  This is sold in 3 foot lengths at hobby stores as a push rod kit.  Look near the model airplane parts.  The non-conductive cross supports are 1/8th inch dowel rod or bamboo skewers.  The conductive supports are made of 1/16th” music wire, and the helix elements are much, much smaller music wire.  Music wire is spring steel wire, and can also be found in hobby stores, near the model airplanes.

In order to maintain the helical shape of the antenna, it’s VERY important to precisely measure the lateral placement of the supports (where they fall along the length of the rod) and their angle.  It is surprisingly difficult to mark a black fiberglass rod!  The most effective means I found to mark the rod was to take the rod against a 90 degree angled surface, such as a door jamb, and run a file between the edge of the jamb and the rod.  This engraved a straight reference line into the fiberglass rod.  The rod is marked for length, based on the dimensions given in the earlier calculator (with some additional slack for tuning!) divided by 4.  The supports for the long and short loops are offset initially by 1 cm, to give room for the cross-braces to cross paths.

Then, I made a jig to hold the rod and drill holes for the cross-supports.  Using a drill press, I simply made a made a reference hole, then marked the centerline of the hole with a carpenter’s square.  Pilot holes were drill perpendicular to the rod hole by a trial and error method (this is the 3rd iteration of the jig).  Then, angle guides are drawn on the rod hole, to give an angle reference.  This is also done with the carpenter’s square, believe it or not!  It could have been done more precisely with some actual drafting tools, but we work with what we have.

To drill out the rod, the jig is clamped to the drill press, and lined up with the appropriate size drill bit, then the length markings etched into the fiberglass rod earlier are lined up with the appropriate angle marking (they’re lettered, so I don’t make dumb mistakes!)

After all the holes are drilled, the music wire and wood supports are cut to length, and placed in the drilled rod.

At the feed point, the two halves of each loop must be insulated from each other, so the wire is cut in half, and coated with liquid tape before being inserted.  Hot glue provides some structural support here!

In order to solder the helix element wire to the thicker support wire, we had to use silver-bearing plumber’s solder, and plumber’s flux.  You’ll also need one heck of a soldering iron.  At LVL1, we’re lucky enough to have a box full of killer irons. Unfortunately, only the 260 watt iron here worked! The 120 and 140 watt irons didn’t heat sufficiently to create a good joint.  After about an hour of tuning the antenna, we killed the 260 watt iron, and had to switch to a butane soldering torch (which also worked well).

After soldering each of the connections, it was simply a matter of hooking each loop individually to the tuner, and shortening the length of the loop until it tuned at the desired frequency.  (After some potentially heretical advice, we adjusted the one loop to 149.5 MHz, and the other to 148.5 MHz).  This takes a lot of time, but can be summed up quickly:  shorten each loop by 1cm at a time until you’re there.

Wire the feedpoint as an infinite balun, as shown here: http://jcoppens.com/ant/qfh/img/v2/012.jpeg

For no particular reason, we also put 4 turns in the coax near the feedpoint.  It can’t hurt.

With all the coax, but without considering the dowel rod you see in the 4th picture, the antenna weighs a stunning 80 grams.  This is under 3 oz!

After configuring both loops, and hanging the antenna in free space, we saw resonance and SWR dip to 1.2:1 at 149MHz.  PERFECT.

In order to test the antenna with the Digi M10 Dev Kit without staying outside, Dan and I rigged the dev kit with a 2mW Xbee Series 2.5 as a wireless serial link.  Now we can be up to 500 feet away while we test our antenna.  This means no more late nights in the parking lot!

Unfortunately, the receive performance of this antenna is not so great.  The satellite beacon signal is around 137 MHz, and our antenna is tuned for 149 MHz.  The modem won’t transmit unless it can hear the beacon signal.  Whenever it keys up, however, those packets fly out.  We’ve never seen any performance like it.

After 4 tries, and some expensive test equipment, the end result was a Quadrifilar Helix antenna weighing only 80 grams!

Our ground test antenna was a 5/8ths wave whip antenna, which works well, but unfortunately needs a ground plane. Tests with both a quarter-wave dipole and a J-pole antenna were lackluster. Documentation from our satellite service provider implied that a quadrifilar helix antenna would provide the best coverage at all. While these antennas are pretty, their design and construction was voodoo magic at first.

Thanks to some design documentation here: http://jcoppens.com/ant/qfh/index.en.php and some help from the balloon community, we had some baselines for creating such an antenna. We still went through *quite* a few revisions.  We went through 3 revisions that didn’t work, and one which works pretty darn well!

Here are the antennas which didn’t work:

And Finally, this is the antenna that DID work.  This antenna worked after we finally broke down and purchased an MFJ Antenna Analyzer.  The antenna analyzer allowed us to very precisely tune the antenna.  A Quad Helix antenna is two full-wave loop antennas, wrapped in a helix, and shorted at the feed-point.  One loop is tuned slightly above the target frequency, and the other loop is tuned slightly below the target frequency.  This affects the circular polarization which makes this sort of antenna exceptionally suited for satellite work.

The antenna is constructed of fiberglass, wood, and music wire.  The central support is 5/16ths inch fiberglass rod.  This is sold in 3 foot lengths at hobby stores as a push rod kit.  Look near the model airplane parts.  The non-conductive cross supports are 1/8th inch dowel rod or bamboo skewers.  The conductive supports are made of 1/16th” music wire, and the helix elements are much, much smaller music wire.  Music wire is spring steel wire, and can also be found in hobby stores, near the model airplanes.

In order to maintain the helical shape of the antenna, it’s VERY important to precisely measure the lateral placement of the supports (where they fall along the length of the rod) and their angle.  It is surprisingly difficult to mark a black fiberglass rod!  The most effective means I found to mark the rod was to take the rod against a 90 degree angled surface, such as a door jamb, and run a file between the edge of the jamb and the rod.  This engraved a straight reference line into the fiberglass rod.  The rod is marked for length, based on the dimensions given in the earlier calculator (with some additional slack for tuning!) divided by 4.  The supports for the long and short loops are offset initially by 1 cm, to give room for the cross-braces to cross paths.

Then, I made a jig to hold the rod and drill holes for the cross-supports.  Using a drill press, I simply made a made a reference hole, then marked the centerline of the hole with a carpenter’s square.  Pilot holes were drill perpendicular to the rod hole by a trial and error method (this is the 3rd iteration of the jig).  Then, angle guides are drawn on the rod hole, to give an angle reference.  This is also done with the carpenter’s square, believe it or not!  It could have been done more precisely with some actual drafting tools, but we work with what we have.

To drill out the rod, the jig is clamped to the drill press, and lined up with the appropriate size drill bit, then the length markings etched into the fiberglass rod earlier are lined up with the appropriate angle marking (they’re lettered, so I don’t make dumb mistakes!)

After all the holes are drilled, the music wire and wood supports are cut to length, and placed in the drilled rod.

At the feed point, the two halves of each loop must be insulated from each other, so the wire is cut in half, and coated with liquid tape before being inserted.  Hot glue provides some structural support here!

In order to solder the helix element wire to the thicker support wire, we had to use silver-bearing plumber’s solder, and plumber’s flux.  You’ll also need one heck of a soldering iron.  At LVL1, we’re lucky enough to have a box full of killer irons. Unfortunately, only the 260 watt iron here worked! The 120 and 140 watt irons didn’t heat sufficiently to create a good joint.  After about an hour of tuning the antenna, we killed the 260 watt iron, and had to switch to a butane soldering torch (which also worked well).

After soldering each of the connections, it was simply a matter of hooking each loop individually to the tuner, and shortening the length of the loop until it tuned at the desired frequency.  (After some potentially heretical advice, we adjusted the one loop to 149.5 MHz, and the other to 148.5 MHz).  This takes a lot of time, but can be summed up quickly:  shorten each loop by 1cm at a time until you’re there.

Wire the feedpoint as an infinite balun, as shown here: http://jcoppens.com/ant/qfh/img/v2/012.jpeg

For no particular reason, we also put 4 turns in the coax near the feedpoint.  It can’t hurt.

With all the coax, but without considering the dowel rod you see in the 4th picture, the antenna weighs a stunning 80 grams.  This is under 3 oz!

After configuring both loops, and hanging the antenna in free space, we saw resonance and SWR dip to 1.2:1 at 149MHz.  PERFECT.

In order to test the antenna with the Digi M10 Dev Kit without staying outside, Dan and I rigged the dev kit with a 2mW Xbee Series 2.5 as a wireless serial link.  Now we can be up to 500 feet away while we test our antenna.  This means no more late nights in the parking lot!

Unfortunately, the receive performance of this antenna is not so great.  The satellite beacon signal is around 137 MHz, and our antenna is tuned for 149 MHz.  The modem won’t transmit unless it can hear the beacon signal.  Whenever it keys up, however, those packets fly out.  We’ve never seen any performance like it.

We encountered a slew of problems, but through our months of training and planning, we already knew our options and flight impact associated with each hiccup we ran into. The massive communications infrastructure we laid out was revealed to those of you who watched on ustream, and facilitated our preparation and troubleshooting phenomenally well. I’ll detail our cool worldwide voice and video comm systems in another post though.

A ton of things went well, but the exciting bits are always the gory failures, so here they are!

1. Homebrew satellite coverage predictor program didn’t work
The day before launch we run an app that tells us how many gaps in sat service we should have, based on predicting satellite orbits. It said we’d have pretty much solid coverage for launch. In reality, we had very little coverage. That changes from hour to hour, and isn’t a problem if we know about it ahead of time. It severely delayed our departure from LVL1 Mission Control while we waited for a check of up and down telemetry, and caused an emergency rewrite of flight firmware to try to deal with it better.

2. 5volt power bus stopped working when we hooked up the cutdown
Still before departure to launch site, It was discovered that there was no power on the 5v power bus. We determined that there were no flight critical items running 5v, and would just lose a lot of science sensors. Flight was cleared to go without 5v bus, but a last check by switching to the flight batt pack caused it to come back online, and switching back to ground power had it still online as well. We do not know what caused this anomaly and have not seen it since.

3. Gps would not stay locked in transport vehicle
Likely due to the proximity to the metal van roof, the GPS lock was a critical part of preflit setup while in transport. This must still be resolved or the setup steps reorganized to work around.

4. HF telemetry could not be verified in transport vehicle
Portable receiver was left behind, and there wasn’t any provision to verify it’s decoded data or GPS lock anyway. Should be adding an audio cable to allow decoding on the road using one of the laptops in the vehicle. May need to add a Mifi to the vehicle to allow HF telemetry to be sent to the Internet servers after decoding.

5. Mobile ground support battery pack had a cell rupture in transport
A D-Cell was found to be inserted backwards in the 12-cell series string of batteries, and was violently boiling over under the multiple-amp drain of payload setup. The cell was removed, halting payload setup near the end of the procedure. Very little was left to do, and was continued on the launch pad using flight power.

6. Cutdown device fired on launchpad
The rope was cut, but since the payload was sitting on the prep stand, it did not fall. The unexpected cut was due to the long power-off period of the payload. The flight computer’s software was never designed to have power removed for more than a few seconds during the preflight phase. It went bonkers and issued every command it was capable of. We painstakingly restrung a new rope through the cutdown and proceeded.

7. HF Backup tracking radio’s GPS would not pick up any GPS signal
This was recently determined to be caused by the GPS located too close to the HF transmitter circuit board. It is easily movable and works fine now only a few inches farther away.

8. Primary sat telemetry antenna failed to receive strong enough sat signal
No up or downlink telemetry was ever passed to/from the payload on the launch pad, which is a required milestone to fly. We are working on a new antenna design now.

Note: Running into this many problems should have scrubbed the launch sooner for a normal program launch. However, as Project Lead, with much experience in large balloon launches, I recommended pressing on as far as we could get with on-the spot troubleshooting. The reason for taking such a seemingly cavalier stance though was twofold- we needed to press as far into launch as we could to uncover any FURTHER problems, and to try out our launch procedures to see how they worked in a real launch environment. There was also a slight chance that we could overcome all problems found. So we proceeded, and we did uncover very necessary data to improve the payload and procedures.

Hope you enjoyed being along with us from home!

We encountered a slew of problems, but through our months of training and planning, we already knew our options and flight impact associated with each hiccup we ran into. The massive communications infrastructure we laid out was revealed to those of you who watched on ustream, and facilitated our preparation and troubleshooting phenomenally well. I’ll detail our cool worldwide voice and video comm systems in another post though.

A ton of things went well, but the exciting bits are always the gory failures, so here they are!

1. Homebrew satellite coverage predictor program didn’t work
The day before launch we run an app that tells us how many gaps in sat service we should have, based on predicting satellite orbits. It said we’d have pretty much solid coverage for launch. In reality, we had very little coverage. That changes from hour to hour, and isn’t a problem if we know about it ahead of time. It severely delayed our departure from LVL1 Mission Control while we waited for a check of up and down telemetry, and caused an emergency rewrite of flight firmware to try to deal with it better.

2. 5volt power bus stopped working when we hooked up the cutdown
Still before departure to launch site, It was discovered that there was no power on the 5v power bus. We determined that there were no flight critical items running 5v, and would just lose a lot of science sensors. Flight was cleared to go without 5v bus, but a last check by switching to the flight batt pack caused it to come back online, and switching back to ground power had it still online as well. We do not know what caused this anomaly and have not seen it since.

3. Gps would not stay locked in transport vehicle
Likely due to the proximity to the metal van roof, the GPS lock was a critical part of preflit setup while in transport. This must still be resolved or the setup steps reorganized to work around.

4. HF telemetry could not be verified in transport vehicle
Portable receiver was left behind, and there wasn’t any provision to verify it’s decoded data or GPS lock anyway. Should be adding an audio cable to allow decoding on the road using one of the laptops in the vehicle. May need to add a Mifi to the vehicle to allow HF telemetry to be sent to the Internet servers after decoding.

5. Mobile ground support battery pack had a cell rupture in transport
A D-Cell was found to be inserted backwards in the 12-cell series string of batteries, and was violently boiling over under the multiple-amp drain of payload setup. The cell was removed, halting payload setup near the end of the procedure. Very little was left to do, and was continued on the launch pad using flight power.

6. Cutdown device fired on launchpad
The rope was cut, but since the payload was sitting on the prep stand, it did not fall. The unexpected cut was due to the long power-off period of the payload. The flight computer’s software was never designed to have power removed for more than a few seconds during the preflight phase. It went bonkers and issued every command it was capable of. We painstakingly restrung a new rope through the cutdown and proceeded.

7. HF Backup tracking radio’s GPS would not pick up any GPS signal
This was recently determined to be caused by the GPS located too close to the HF transmitter circuit board. It is easily movable and works fine now only a few inches farther away.

8. Primary sat telemetry antenna failed to receive strong enough sat signal
No up or downlink telemetry was ever passed to/from the payload on the launch pad, which is a required milestone to fly. We are working on a new antenna design now.

Note: Running into this many problems should have scrubbed the launch sooner for a normal program launch. However, as Project Lead, with much experience in large balloon launches, I recommended pressing on as far as we could get with on-the spot troubleshooting. The reason for taking such a seemingly cavalier stance though was twofold- we needed to press as far into launch as we could to uncover any FURTHER problems, and to try out our launch procedures to see how they worked in a real launch environment. There was also a slight chance that we could overcome all problems found. So we proceeded, and we did uncover very necessary data to improve the payload and procedures.

Hope you enjoyed being along with us from home!

Time: ~8:00PM Thurs EDT (0000 Fri UTC)
Location: Columbus IN, SpacePort Indiana

Flight Path: A high speed trip to Africa!

Tune in to our Live Stream links up above to see launch and listen to Mission Control throughout the 3 day flight.

With such a complicated system, the only we know that everything works together is to test it, all together.  Gary made a GPS simulator to act like a real gps to our flight computer, but actually replay a predefined balloon flight.  Since the balloon robot makes all it’s decisions based on the GPS and our commands from Mission Control, we can make the balloon perform like it’s really flying.

So for the last two weeks we’ve been doing just that.  It’s been a lot of work, and every night we run a full simulated flight, and every night we discover another problem somewhere.  They’re all being fixed, but we’re running out of time.  Luckily (or unluckily) the jet stream winds are not very good for the next few days, so we’re not missing anything.

SpeedBall-1 could have flown just fine this weekend, however, there’s significant parts of the telemetry pipeline, a long convoluted series of data transfers, that haven’t been tested very much.  This will give us some much needed time to make sure that a silly mistake doesn’t doom our precious robot pilot to a watery grave.

This weekend we’ll be running a GPS simulator built by one of our members, Gary Flispart, that replays the Spirit of Knoxville IV flight from March 2008, the current distance record holding flight.  (with slight deviations from the original track for testing purposes)   You’re welcome to watch along if you like, at http://track.whitestarballoon.org .   We can’t guarantee a time or quality of the entertainment, and it may stop working any time, or restart, as we tinker with it.

SpeedBall-1 is scheduled to be released from testing after Wednesday next week.  After that date, we will fly at the earliest possible less-than-36 hour crossing forecast!

SpeedBall-2 is already under construction!