This flight belongs to the test series that includes NSL-74 and NSL-77.   The same small box, line, and parachute now held:

Paul planned on some sort of test flight for the weekend, but schedule conflicts pushed it to early Saturday morning.  This happily coincided with a SpaceX launch from FL, so maybe this flight could capture an image of the launch.  

At 3:30am (yawn), with the SpaceX launch still "go", a 600g cell was filled with about 1300g of neck lift.  The balloon was filled prior to payload activation to save as much battery power as possible.  It will be a cold flight without solar heating and this will degrade LiPo battery performance.

NSL-82 launch occurred at 4:03am.  This should put the flight around 25km (82kft) when the Starlink flight launched.  Like NSL-81, the payload should land about 30 minutes later in the Bunn, NC area.

The launch proceeded normally, following the projected flight path almost exactly.  APRS and LoRa transmissions worked well, but the battery drain the in the early morning cold was very noticeable -- worse than expected.  Will the batteries hold out?

Unaware of the Mobius issue, Paul waited at the predicted landing area near Bunn with his much needed coffee.

The T-Beam announced burst with a special packet:

This noted that the tracker had detected burst and high-speed descent from a maximum altitude of 31,398m (103,012 feet).   Also the onboard watchdog had not noted any system crashes and that 41 data points had been logged into the T-Beam's EEPROM for later recovery.

Meanwhile, the AP510 recorded a maximum altitude of 31,517.6m (103,404 feet) on its SD card.

As the payload descended the reported battery voltages continued to drop along with the internal temperature.  The external temperature was already pegged at the minimum sensor reading of -46.5C.  At 15km, the frozen AP510 shutdown due to low voltage.  Only the T-Beam remained and its voltage hovered around a low 3.3V.   Ten minutes later, it began reporting warmer temperatures and increasing voltages!   The predicted landing track continued to be calculated and relayed to the chase car.

The track would carry the payload under chute over the chase car towards the east, and then turn around and pass over it again going west.  It was to land about 2km away.  If the T-Beam ran out of battery, this information may be the only way to spot the parachuting payload.  Paul had two choices on Google Maps to get to the landing area.  The northern path was a looping set of country roads.  The southern path was a straight line.  He chose the southern path...  and found that it was a dirt road with a locked gate.  He sped back around to the northern route, catching sight of the orange parachute as it went overhead.  It landed on a fallow field beside the road and was an easy recovery.

The on-board landing prediction software worked rather well.  It's guesses during descent were within a couple kilometers of the actual landing site.  The map above shows the path of the guesses as the flight descended.  Noted are the approx altitudes where the guesses were made and their associated number of minutes before touchdown.   The prediction used data collected during ascent over two hours prior.  It also landed in an area that was lower than the launch site.

The larger GPS antenna worked great.  It recorded 9-12 satellites and not sign of RFI.

The longest receive distance on the 100mW LoRa was 27km.