You’d never know it from the high of 85deg today, but the leaves have finally started to change for the season here in North Alabama, so I took my Phantom 4 up this afternoon to get a few shots of the Fall colors.
I’ve been getting more comfortable flying the Phantom, so after I got a few shots I decided to go a little higher than I’ve ever gone before. The legal limit is 400ft AGL, but I’ve never been much higher than 200. Starting from about 120ft, I climbed gently for a few seconds, then gave it max throttle to see what she could do. The drone shot straight up into the air, crossing 200ft in seconds. After a few seconds, my flight display showed a warning “Max Motor Speed Reached”, so I let off the throttle and the warning went away quickly…but immediately my heart started to sink, as the video feed from the drone showed that it was falling from the sky.
I could still see it above me (this was a more or less straight up and down flight), and I could hear the motors changing speeds up and down rapidly. I experimentally gave it some throttle and could hear the motors speed up and see the fall slow down on the video feed, so I could tell I still had some control. Over the next several seconds, I continued to feather the throttle, giving it just enough to keep the sink rate slow but not daring to hold constant throttle against the fall. Somewhere above 100ft, it stopped losing altitude and hovered in place. I landed gently and found no damage upon inspection; neither the motors nor the battery were even particularly hot.
A propellor is just a wing that gets its lift from moving fast in a circle instead of fast in a straight line. Just like any other wing, a propellor can stall. When a wing stalls, there isn’t enough lift being developed to keep it in the air anymore. I am not an aeronautical engineer, so my analysis could be wrong here, but I assume a quadcopter can stall because I do know that helicopter blades can stall: see https://www.aopa.org/news-and-media/all-news/2014/may/08/rotorocraft-rookie-helicopter-stalls and https://en.wikipedia.org/wiki/Retreating_blade_stall.
Just as an intellectual exercise, I decided to perform a little investigation. According to Wikipedia, a helicopter blade stall can be caused by any of the following factors:
- High gross weight
- High airspeed
- Low rotor RPM
- High density altitude
- Steep or abrupt turns
- Turbulent ambient air
We can rule out some causes easily. High gross weight: I haven’t added anything to the craft, it is as it came from the factory. Low rotor RPM: according to the warning message, I was in the opposite case. Steep or abrupt turns: I was climbing straight up.
Density altitude had actually been my first thought when I realized the propeller may have stalled. Today was a pretty hot day after all, I figured. It’s not actually the most likely cause, but let’s work it out anyway. Because the air gets thinner as you climb, a wing has to go faster to develop the same lift at higher altitudes than at lower ones. Temperature and pressure affect this too (as in high and low pressure systems, from the weather report), since air expands (gets less dense) as it gets hotter, and a low pressure system literally means less dense air. Density altitude is a formula that tells you how high the aircraft “feels” that it is, taking into account temperature and pressure.
I live pretty close to an airport and have an app that gives weather and altimeter for other nearby airports as well, so I gathered or interpolated the following information, based on the available information for the closest airports:
Altitude: 692ft MSL
Altimeter: 30.14 inHg
Temp: 81deg F
Dewpoint: 50deg F
MSL means “mean sea level”, so the measurement above means that ground level where I am is 692ft above sea level. AGL means “above ground level”, meaning the actual distance above the ground. The numbers above are all approximate, b/c 1. I’m not the NTSB, and 2. the stakes are pretty low here. For instance, I used the reported altitude of the nearby airport even though I was actually launching from higher ground, maybe 20 feet higher.
According to the flight log kept by the DJI app, the highest altitude I reached was 282ft AGL, so that puts me at 692ft + 282ft = 974ft MSL. Plug all those numbers into a density altitude formula and I get 2515ft, so at the point where the drone started falling, it’s as if I was flying at 2515 feet above sea level, though I was actually only 280 feet above the ground.
(I want to point out here, for those who may have missed it, that density altitude is a measure of performance of an aircraft, not an actual height above the ground. My max height above the ground on this flight was 280 feet. Flying a remote aircraft at 2515 feet anywhere, much less near an airport, would have been both dangerous and illegal in my circumstances.)
I couldn’t find any official numbers on this, but I found one reference online that the service ceiling of a Phantom 4 is 19,685 ft. More importantly, I found many people in online forums reporting that they use their Phantom 4 in places like Colorado, at density altitudes around 8000 to 12000 feet. So it should be safe to say that I didn’t exceed the capabilities of the drone flying at a density altitude of 2515.
Wind this afternoon was calm. I was going straight up and down so I should have been in the same air column the whole time. That doesn’t necessarily rule out turbulence, but I think it makes it less probable.
That leaves high airspeed, and that is my best guess: I accelerated so quickly that I reached a vertical climb speed that stalled the propellors. The motors tried to compensate by running all the way up to full speed. Stalled, the craft lost altitude until the combination of the craft’s flight controller and my own feathered inputs got it back under control. That’s my theory at least.
Edit: After a good night’s sleep, one other possible cause occurred to me: altimeter error. Every time you take off, the Phantom 4 counts your current position as altitude zero and displays your altitude then as an offset from that. One thing I failed to mention above is that the log for this flight showed an altimeter reading of -26ft at landing, though it started at zero and I launched and landed from the same spot. I originally chalked that up to the device failing to track altitude change during the time that it was falling, but I didn’t look into a mechanism for how that could happen. I researched it this morning and it turns out that the Phantom 4 uses a barometric pressure altimeter. I did do something just before the flight that may have affected the accuracy of an altimeter: I took the Phantom from inside (about 72deg F) to outside (about 81deg F). Air density decreases as temperature increases, so if the device was just hovering in mid-air, not actually moving, but warmed up to ambient outside air temperature, it may have “thought” it was gaining altitude and tried to compensate.
I’m not sure if that really tells me anything, though. Presumably the unit had done some warming to ambient temperature before it self-calibrated upon being turned on, so I could cherry-pick any intermediate temperature that worked and say that accounts for a 26ft drop. Further, though I was paying more attention to keeping the drone out of free fall than I was the altimeter, I am fairly certain that the actual drop was greater than 26 ft. Regardless, reading the DJI forums, it sounds like the general consensus is that the barometer in a Phantom 4 does experience temperature-driven inaccuracy in its height readout, but the effect is small, and it’s common for the height to drift by several feet during a flight. I still have a feeling that the -26ft landing height is telling me something about what happened to the drone at the apogee of its flight, but I don’t know what.