LOHAN test flight: Results in from Oz jury
Andrew Tridgell picks over Vulture 2 avionics logs
Low Orbit Helium Assisted Navigator (LOHAN) brain surgeons Linus Penzlien and Andrew Tridgell* have scrutinised the log flies from the recent test flight which saw a Pixhawk autopilot, batteries and servos (pictured below) sent aloft to 27,700m to determine in real-world conditions how the Vulture 2 spaceplane's bulging electronics package would handle the cold.
We'll get to the results in a moment, but first up is another tip of the hat to our US allies at Edge Research Lab, who carried the "LOHAN Experiment: Stratospheric Test of Energy Reserves" (LESTER)** payload on their EDGE17 flight.
The mission's principal objective was to test the "BEACON e-field sensor" as part of the "Balloon Enabled Atmospheric Conditions Observation Network" project. Ultimately, Edge will release a sensor cluster into a thunderstorm "in an effort to more accurately profile the electrical characteristics of convective weather", and there's more on their electrifying work right here.
As you can see from the entertaining video above, EDGE17 was a textbook operation, and we got our avionics back in good shape, with the servos still operating in the custom APM AUTO mode command MAV_CMD_NAV_ASCEND_WAIT, which wiggles the servos every 15 seconds to prevent them freezing on the ascent to Vulture 2 launch altitude. There's more info on this and other custom LOHAN parameters/commands here.
Pixhawk peripherals along for the ride were a GPS/magnetic compass unit and a digital airspeed sensor identical to the one already installed in the Vulture 2's very pointy beak. The autopilot was powered by four Energizer Ultimate Lithium AAs, and the servos by eight of the same batteries.
So, what's the verdict? First up, while the external temperature dropped to a nippy -50°C, inside the enclosure was a positively balmy 0°C, pretty well in line with our experience on previous test flights, and what we expect to happen inside the spaceplane when the big day arrives.
We'll now hand you over to Tridge for a breakdown of the other results:
First the easy bits. The two 3-axis accelerometers worked perfectly, as did the gyros. The way we can tell they worked well is they matched. That is the advantage of having redundent sensors of different types. When they match you can be pretty confident both are right.
The internal compass also seems to have worked well, at least to the degree we can tell from the log. We can only really look at whether it gave plausible readings, not whether it was actually correct, as we have no way to validate it in a balloon. Once we are in fixed wing flight we can properly validate it, but in a balloon you can move in any direction, so compass can't be checked against other sensors.
The airspeed is interesting....
The blue line is the actual airspeed reading, which is apparent airspeed. It was very noisy over a small range. The green line is the true airspeed calculated by TECS during the flight by using the EAS2TAS ratio. The red line is the GPS vertical velocity, which ideally should match the true airspeed.
The GPS vertical speed and true airspeed do follow the same curve, but are offset by a factor of around 1.5. It would be nice to work out why that is.
I think there are two likely causes:
1) Looking at the logs, I see that it had quite a large ARSPD_OFFSET, probably because the electronics wasn't warmed up enough on the ground before airspeed calibration. That is something we can fix for the next flight. We need to cover the sensor loosely for several minutes, then get the GCS to do a airspeed calibration (offset zero) before removing the cover.
2) It could be the placement of the airspeed sensor, and that we're not getting clean airflow over it. Looking at the setup here...
I wonder if the edge of the box is interrupting the airflow? I suspect the problem was the offset, but if we do another test flight it would be good to get the airspeed sensor in clean air.
The next thing to look at are the batteries:
There are a few things notable about this graph. First off, the servo voltage is a bit high. Apparently the CC UBEC was set to 5.6V? It should be set to 5.3V or so I think. We want plenty of safety margin below the 5.7V limit for the servo rail being a backup power supply for the FMU.
We also see some voltage spikes on the servo rail. I think we should install a reverse biased zener diode on the servo rail to clip those. That will ensure that if primary FMU power fails the servo rail is a good backup. See here for details.
The next thing that is noticeable is the POWR.Vcc is quite low. That is normally steady at 5V. Seeing the match to the CURR.Volt value, it looks like this is just diode drop in the power brick, combined with voltage drop in the main FMU power supply with the cold temperatures.
The Vcc stayed well above the level the board can handle, so it is OK, but it sure would be nice to fix the servo rail as a backup supply in case it drops any more. That battery pack dropped to 5.1V (see below), which is quite a drop.
In summary, then, we've got a few issues, but nothing insurmountable. Just when we hoped we'd done the final test flight, it looks like a further jaunt is on the cards, with the following adjustments:
- Enhanced battery monitoring for both packs
- Move pitot to be in clear air
- Zero airspeed offset before launch
- Change CC UBEC voltage down to 5.3V
- Add zener clipping diode
And while we're at it, Tridge has suggested we give the Vulture 2's RFD 900u Radio Modem a run for its money.
We recently installed this long-range kit, which replaces the standard 3D Robotics radio. With its big brother the RFD 900 (at top, above) on the ground, we'll be able to monitor the aircraft live via the APM Mission Planner.
Of course, that opens the possibility of Tridge ultimately monitoring and controlling the spaceplane via VPN from the comfort of his PC in Australia, which is a very provocative possibility indeed.
We'll bring you details of quite how that hook-up works in practice when Tridge and Edge have mulled the practicalities. ®
The simple premise of the challenge is to detect a dummy dubbed "Outback Joe" and drop a water bottle as close to him as possible. As you might imagine, there's some serious work involved in making that happen, so we invite readers to raise the traditional glass to CanberraUAV for a truly impressive feat.
**Not vanity on my part, I assure you, but rather a piece of unprompted backronym tomfoolery by Edge, very much in the spirit of LOHAN.
More from the lovely LOHAN:
- You can find full LOHAN coverage right here.
- If you're new to LOHAN, seek out our mission summary for enlightenment.
- There are photos our our magnificent Vulture 2 spaceplane here, and detailed structural plans here.
- For your further viewing pleasure, we have all our photographic material stored on Flickr.
- Our LOHAN and Paper Aircraft Released Into Space (PARIS) videos live on YouTube.
- We sometimes indulge in light consensual tweeting, as you can see here.