Saturday, 26 November 2016

The TT7F's First Launch

This is a throwback to the August launch I wrote about in The TT7F Payload but haven't finished the story yet. Now, three months later, it is probably time to conclude it.
Early in the morning with the 8:00-9:00 launch in mind I set up all the equipment in the garden. The latest predictions remained favourable suggesting 8:30 launch might have been ideal.
I filled the balloon while estimating the proper lift with the standard 'water in a bottle' technique. Not having a regulator on the hydrogen cylinder, the balloon was ready pretty quickly. I also filled a small party balloon to have it fly first and ensure there wasn't any strong wind higher up that could blow the balloon to one of the plenty of trees that were around.
Powering up the tracker, tuning in with the receiver (an rtl-sdr dongle with a helix antenna), verifying that the balloon has appeared on the map, turning on the camera - all checked, I moved to the front yard and let go of the balloon.
The unit cleared the surrounding obstacles without any problems and slowly went on its way.
The receiving station was somewhat unorthodox, nevertheless, a significant improvement on the previous lousy notebook with a mobile Internet. The snow shovel supported helix antenna performed quite nicely throughout the majority of the flight. Like this, the tracking experience was much more enjoyable than on my previous flights.
The received signal was sufficient even though a little 'shaky' (perhaps the thin guitar string antenna was responsible for that) and decoded without problems. The wind took the balloon predominantly towards west
and in a similar direction from where I was which meant that I didn't have to redirect the antenna too often.
Just a few hundred meters in ascended altitude and receivers from Poland (many thanks) started to upload TT7F's telemetry as well. The furthest reception was by SP1TMN at about 484km (based on his position on the map). The final proof that the tracker could operate at large distances not just in the somewhat closer proximities as used in testing.
Initially I was a little worried about the ascent rate. The balloon started cautiously with about 2m/s (with all the calculations done for 4m/s). Especially the moment where the speed fell from 3m/s to just about 1.5m/s quickly summoned flashbacks of last year. Luckily the speed eventually grew and brought the final average up to 3.69m/s.
The balloon rose and rose and since about 15km in altitude, thanks to the clear skies, I was able to track it with a pair of binoculars all the way to its burst. That came after 3 hours and 8 minutes at 42337m which was just a few tens of meters shy of the the predicted 42383m.
This is a graph of TT7F's horizontal speed I calculated from the GPS data. Particularly interesting to me were the numbers towards the end that show what winds the balloon encountered at around the 40km mark. It seemed to me while watching the balloon live on the map that it quite sped up at those altitudes so I was intrigued to verify that in data. Generally the everyday soundings by the met offices don't go that high so the wind speeds up there were an unanswered question to me.
The tracker ran the old software that didn't have the ADC measurements properly handled yet so the readings jump around quite wildly. Despite that a little averaging shows the expected temperature profile as measured by the mcu's on-chip temperature sensor.
The battery voltage seemed to react in accordance with the temperature. After all the two AAA Energizers were exposed to the outside atmosphere without any insulation.
With the descent rate graph above it's probably time to address the elephant in the room and the main reason for why I am writing this summary three months after the launch. Put simply, I didn't find the payload nor have since. As I said, I was watching the balloon with binoculars and actually managed to see the burst itself. When I returned to the PC, I found out the signal got a bit wild and the frequency had drifted out of the dl-fldigi's decode bands. Also none of the Polish receivers seemed to be able to decode the signal anymore. Upon restoring decoding, I saw the descent speed substantially higher than expected. I remember seeing the balloon in binoculars even after it burst suggesting most of it remained attached. That would explain the plummeting speed. The main issue, however, occurred at around 12km in altitude where the signal simply disappeared. Habitat's last upload has the payload at 14916m while a partial decode pulled out of dl-fldigi puts the last position at 49.4978N 17.5210E and 12471m. As to what might have happened. One possibility is that the tracker, light as it was, hit a layer of strong wind that might have slowed it temporarily causing it to collide with the rest of the balloon breaking of the guitar string antenna (honestly, not build for this). Another possibility is that the exposed batteries simply froze (at least temporarily, the region coincides with the lowest temperatures this time magnified by the increased speed) despite being Energizer Ultimate Lithiums.
I packed a notebook, the sdr dongle, a yagi, a pair of binoculars and after roughly estimating the possible landing location (simply projection the fastest descend prediction I had from the actual burst coordinates) I drove off to the site. Comparison between a prediction some 24 hours to launch and the actual trajectory can be seen above.
Upon arrival I couldn't find the signal from any of a few locations overlooking the potential area of impact. If the thin antenna survived the rough descent, it quite possibly didn't the consequent hard landing. Wandering the endless fields with a pair of binoculars (after all a 4m² white sheet of latex can't just disappear without a trace) for the following three hours led nowhere, so I returned home with nothing but a tank top tan lines.
That started off a series of potential descent trajectory modellings, consecutive visits and lengthy however unfruitful searches. On one visit I spent the whole time in a corn field. In a 4 hour purgatory I manged to cover just a fraction of it gaining nothing but eternal hate towards corn.
On another occasion I closely inspected the grounds of an electrical substation that unfortunately happened to be in the middle of the suspected area. But the payload didn't seem to be there nor on any of the many high-voltage power lines.
The last visit came with the beginning of November. The main reason was that the corn fields should have been harvested by then and I wanted to walk through them to see if there weren't any remnants. No luck on that occasion either. Above is a photo overlooking the area to give a little perspective on the balloon's whereabouts while it was still transmitting. The view is oriented south-east.
The map shows the overall searched area. Most of it are fields out of which many were cropless thus easily explorable with binoculars. By the November visit most of the fields seemed ploughed so it seems unlikely the payload may still be lying in one of them without anyone having noticed it yet. It may have been destroyed during the harvest or by one of the ploughs. Somebody may have found it and kept it. Or it may have landed in a bushy ditch or a row of trees between the fields. I may have walked within 10 meters of it unable to see it. It's really hard to say.
This is an example set of potential descent trajectories I made. They differ in origin to see to what extent they follow the part of the actual descent I had received (42337m to 12471m). Another variable is the air speed and direction data used. From the start I was using data from the closest met office's sounding (Prostejov) published at 12:00AM GMT with TT7F bursting at 9:37:30. Only during writing this post it occurred to me to use TT7F's own ascent data as input (possibly more accurate) as well. The result can be seen in the blue to purple trajectories. Since not even these match the real part of the trajectory perfectly, I reckon the conditions have evolved a little in the three hours of flight (and possibly in the 79.8km the balloon covered) so the truth will probably be somewhere in between. The modelling was done second by second in Excel. Another variable I tinkered with was descent speed. I dived into the depths of Habitat's archives to dig out a couple of 'rushed' descent telemetries to match TT7F's own. I also modelled a few slow descents for comparison (could the balloon have formed a sail and glided away at lower altitudes?). The variables are plenty and actual data scarce.

Above is a more interactive view of the probable landing area with all the search attempts and photos in a little more detail. It was made in Google's My Maps.

Given the trend in my recovery attempts: TT7 several hours, TT7-40 two months, TT7F three months and counting. It's probably not for the worst that I am slowly shifting my efforts towards floaters. In the end I am glad that I have finally manged to get above 40k in altitude (officially, who knows how high TT7-40 had gotten). The 42337m for a 150g payload seem quite nice. It is a shame I didn't get the video recording back. Given the altitude it reached and the flight time, the SD card quite possibly holds one of the highest full flight recordings. May it rests in peace... or a ditch probably.

No comments:

Post a Comment