- To educate and inspire the scouts and the leaders
- Promotional for Crystalmark Aerial Photography
- But most of all because it’s thrilling, interesting, geeky and blooming good fun!
We started with a balloon kit from Sent Into Space. The balloon, foam box, GPS trackers and rope are from the kit but we’ve added the cameras and flight computer.
Locating and chasing
The balloon goes up, pops (technically it shatters) and the payload drops. A parachute slows the decent hopefully enough to stop the payload being damaged or damaging things on the ground as it lands.
We know where it takes off from but how do we know where it’s going to land or indeed, where it’s actually been?
SMS GPS Tracker
A very light backup tracker. It communicates back to us by sending an SMS with lat and long when we call the mobile telephone number. Since it only works below 1000m metres, it’s not much use apart from a backup. This device will be out of range for 90% of the flight.
A little more expensive and requires a subscription but this device works to about 7,000m so we should get location information for about 25% of the flight and have a much better idea which way to drive. It uses satellite communications to send location back so it should work even where the SMS tracker doesn’t. If the payload lands in the middle of moorland and there’s no mobile signal, we should still be able to receive from the SPOT Tracker.
A SPOT Tracker bonus is the internet API, which I’ve used to update the live location page.
Flight Computer GPS
See below for more details. The Flight Computer is a USB GPS attached to a raspberry pi zero. It will record raw GPS every 10 seconds of the flight.
Why another GPS? SPOT Tracker and the SMS Tracker will only record upto 10,000m. We’re hoping to go much higher than that. The other trackers don’t record height and we really really really want to know how high the balloon goes.
The COCOM limits will mean that most consumer GPS receivers will stop working at 18,000m. The main purpose of the flight computer is to record how high the balloon goes and the predication is 37,000m. We have two USB GPS receivers: G-Star IV (SiRF Star IV chipset) and a uBlox dongle. According to the specs, the G-Star is limited to 18,000m but the uBlox VK 172 (which probably uses the G7020-KT chipset) is rated to 50,000m, yay!
After searching around various forums it seems that recording the raw NMEA data from the receiver before it gets processed into latitude, longitude and altitude could bypass the COCOM limits. I’ve tried writing a script to record the raw data from the serial port on the Pi but I think that it’s only a python library converting the raw data anyway, so there’s not much point. If the readings stop at 18,000m, we’ll just have to fly again. 🙂
Chase Car Computer
I’ve cobbled together a raspberry pi with a USB GPS receiver, based on the Flight Computer (see below). This updates the webserver via ssh so that the current location of the chase car can be seen. The code/design/scripts are available on github.
Balloon size vs payload weight vs cost
The standard balloons come in many sizes, here’s a general guide to size vs payload weight vs cost:
|Size||Balloon weight||Recommended payload weight||Cost||Estimated burst height|
|ultra large||3000g||1500g||~£280||40km (maybe!)|
We want to send a very heavy camera as high as possible so we have shelled-out for the ultra large balloon. The volume of helium needed to fill this balloon is huge! We need to get > 6.2m3 of helium. That’s about £100 plus daily rental of the bottles.
For the 3000g balloon, the recommended payload weight is 1.5kg but with the OSMO, extra battery and all the other stuff we are cramming in there, it currently hangs at 2.5kg. This is still within limits but adds risk and costs more in helium to counter the higher weight. I initially wanted to use the DJI Phantom 2 battery but it’s very heavy (500g) so I pinched a 1800mAh 3C LiPo battery from a racing drone which powers the OSMO X3 for 2 hours and is only 130 grams.
More gas in the balloon at take-off means that it will expand to maximum at a lower altitude so it’s a balance between heavier payload and higher flight. According to the balloon burst calculator, the estimated burst height is 37,000m instead of 38,200m. That’s about 1% difference in height for an extra kg of payload, not too bad.
An extra kg of payload means more helium. To get the magic 5m/s assent rate, the balloon needs to be filled to 7m3. That’s a lot of gas. A lift of 4.2kg (as measured using a fish scale as the balloon is filled) is needed to achieve the desired assent rate and therefore height. Estimated time to burst is 130 minutes with about 40 minutes to descend on the parachute.
This is a gimballed camera for smoother shots. I’d like to prove that sending a gimballed camera into near space is possible and I’m hoping the results will be spectacular. It weighs much more than the other cameras but, as a proof of concept, it should produce some nice film and photos.
Hobbled – The OSMO is only able to record up to 30 minutes of video without manually pressing the record button again. This means there’s no hope of recording 2.5 hours of 4K film. I’ve thought about extending the function of the flight recorder with a solenoid and light detector so if the recording light has not flashed for over 1 second, the record button is pressed. I’m still working on this and if it works, I’ll be very happy but it is probably a bit advanced for the first flight, maybe next time…
The usual GoPro or other action cameras give great results but the video is shaky and so only really useful for fun, not professional footage.
A very light (49g with battery) camera with a very wide angle lens. This camera will be pointed at the ground so we must remember to wave as the balloon takes off.
Pointing up and attached to the Flight Computer, this is taking low resolution video at 90 frames per second. Hopefully it will capture the balloon shattering. The flight computer is mounted on the underside of the foam box and the ribbon cable fed though a hole to the outside. This means that some of the cable and the whole camera will be exposed to the very low temperatures. It will be interesting to learn how it copes. There is the risk of the cable breaking and causing a failure in the flight computer, loosing the GPS and temperature logs too.
Home made flight recorder – camera, temperature (inside and outside of the payload box) and GPS.
Powering the flight recorder – 1000mAh battery powers the Raspberry Pi Zero + GPS +PiCam + temperature sensors for about 4 hours. The battery weighs 120 grams. If we’ve planned well, the payload should be in the air for between 2.5 and 4 hours so the battery must be able to power the flight computer for this time.
Why extra recorder? The GPS locators don’t work over 1000m for SMS and 7000m for the SPOT Tracker and we would like to know how high it went. The addition of temperature gauges is easy and they only add an extra few grams, not necessary but they may help plan a better mission next time with respect to battery life. The G-STAR GPS dongle is rated up to 40,000m AMSL and if we go above this height, my planning was very wrong!
Setting the time on the flight computer. Accurate time on the logs is important to synchronize logs with camera footage so the Pi is setup to use NTP from GPS. This is working well. Also the GPS starts to log as soon as the Pi boots so no fiddling around on the launch day, just plug and play.
I’ve shortened the cables on the GPS receiver, power supply and temperature sensors in an attempt to save a few grams.
- Camera on a stick – running camera tests with the external battery to see how long the OSMO and GoPro will film for. The GoPro 3 Silver filmed for ??XXX?? minutes and the OSMO took time lapse photos (2 second interval) for 130 minutes.
- Washing line – when the payload is stuffed full of cameras, batteries, locators and flight computer, it was hung from a washing line and everything switched on. We expect the flight to be 3 hours so it was left untouched for 4 hours except a few harsh shakes of the line to simulate turbulence.
- Driving around with the payload in the roofbox to test the flight computer GPS recording and temperature changes.
There will be a live location tracker page so people (scouts) can see the balloon’s progress and the chase car’s attempt to follow it.
Soon. We’re desperate to launch but also desperate to get it right. Maybe late October. We have to wait until we receive permission from the CAA before launch.
Exemption from the Air Navigation Order is needed to release a big balloon. We applied to the CAA on the 18th of September and received the certificate on the 18th of October. The CAA will issue a NOTAM, which will appear on here, when we call them 72 hours before the expected launch. See UKHAS FAQ for some really useful information.
Outside the scout hut in Addingham. There is a football field outside the hut which is easily big enough to launch from but there are a few considerations:
- It’s close to the landing circuit for Leeds Bradford Airport so ATC may not give permission.
- There are tall trees surrounding the park
- Footballers playing?
- Public land and regular dog walkers and kids playing
Currently we are ready to go for Saturday 22nd of October. The weather is looking very good but the forecast is for an Easterly wind. Not good. There is a high pressure due to be sitting over the whole coutry and if it moves South a little, we will have Westerly winds and we’re a go!
Launch site set-up
TODO: Balloon neck, tying the balloon off, scales, filling with helium
As part of every aerial filming launch, whether it be a TV drama or a forestry survey, we always used checklists for pre-sight, pre-flight and post-flight to ensure we don’t forget anything. This is the same for the balloon launch. Pre-launch checklist included:
- Charging batteries
- Formatting/clearing memory cards
- Making sure all batteries are plugged in
- Start cameras
- Lens protectors removed
- Glass dome cleaned
It worked. Final estimated height is 35,911 metres – 117,818 feet!
The DJI OSMO photos are great
The GoPro camera filmed the whole journey
The Runcam shots were clear
The HabHub predictions and calculations worked almost perfectly – landing and gas amount for lift
The Spot Tracker worked really well
Help from friends was great
Education and energy from the Scouts was great
To view some of the photos, see here.
What didn’t work?
- Runcam overheated
- Runcam down view was too spinny
- PiCam did not start
- No status lights
- Took an hour to fill the balloon
- The GPS Location Tracker did not charge and turned off before landing. The Spot Tracker worked really well though.
The photos stop before the external battery ran out on the for the DJI OSMO. The internal temperature was about 20C so I doubt that was a factor. From the photos, it looks like the balloon hit a lot of turbulence at 13,000m/45 minutes and the gimbal shifted on it’s mount. The gimbal safety probably shut down the motors and the camera. It’s a pity, there would have been a few even better photos further up.
What would we do better next time?
- Protect the glass dome better so there’s less scratches to spoil the photos
- Maybe use a gimballed camera to face straight down
The down camera shots were good but with all the spinning and rocking, it’s quite difficult to watch and impossible to speed the film up without it looking like an earthquake
- Black out the white foam with paint or tape so there’s less reflections on the inside of the dome
- Telemetry recorded at same fixed intervals for GPS and temperature so it’s easier to match up readings
- Motion/g-force recorder to get an accurate timing for take off, burst and landing
- All telemetry files have leading zeros or timestamp so they can be easily sorted
- Find a way to record GPS over 12,000m
- Fasten down the DJI OSMO camera better so that the gimbal is not disturbed during turbulence and the camera can take photos for the whole two hours of battery.
- A last test of the flight computer and pi camera on the night before – the status lights and camera did not work but there was not enough time on take-off to debug.