Went for our long distance flight attempt today, flew a test flight with a 6S 2600mAh LiPo and then put in the 18650 batteries in, the log unfortunately shows both flights as I forgot to reconnect so I have calculated the time and total distance from the launch of the second flight to the landing.
Basically we did a 300m diameter Loiter for most of the flight, biggest problem was the phugoid motion of the plane losing height on every downwind leg and having to regain the lost height on the upwind leg using precious amps to do so. I tried just about every parameter to tune the plane better throughout the first hour but could never get it to behave.
Special thanks to the devs (especially Tridge and Paul Riseborough), to Steve B. for assisting and UAVDS for making it possible
Plane: Super Sky Surfer
AUW: 3078g
Motor: Maytech MTO3548 790Kv
Prop: APC-E 9x6
Batteries: 6S4P NCR18650b 25.2V, 13600mAh
Battery weight: 1172g
AutoPilot: APM2.5, APM:Plane v2.78.
Weather: Calm (0-4km/h), clear blue sky, 26°C. Altitude: 1668m ASL
Time flown: 126m 53s
Total Distance flown: 116.475km
Calculated consumption on nominal capacity: 115mAh per km (will update after recharge).
Estimated current draw: 5.2A average
Lowest voltage on pack 18.0V
Average speed: 55km/h
Droneshare: http://www.droneshare.com/view/bfgodhs
Tlog: https://db.tt/UfVPiyBE (12MB)
KMZ: https://db.tt/1Tmzim1W (3MB)
Comments
and this is the case when using airspeed sensor or without it ? generally seems to be some sort of PIO or in this case aPIO ;)
Is the attitude estimation on APM really that bad during sustained turns? Is for instance autonomous thermalling even possible? (and by this i mean only climbing in a thermal not finding it which is a whole different can of worms ;) )
Paul, thanks for the insight, very interesting, I'll try your suggestions on our next flight. I wonder why we keep getting this problem to a greater or lesser degree on every single one of our planes, the X8 was the worst, so much so that we decided not to use them any more.
FYI, I changed the parameters quite a bit during this flight but didn't refresh them at the end so what's in the log are what we started with not what we finished with.
@Doogie, 2C is correct but we used 6 series 4 parallel so have theoretically 6.8Ax4=27.2A max. I tried to keep the launch current low by not climbing too steeply and max current during the flight if I recall didn't go over roughly 14A.
The true measure of batteries for our purposes, the watt hour/kg, is very high with these cells.
Average LiPo's are around 140-160Wh/kg, the 18650b's are about 250Wh/kg (3.100Ah x 3.6V / 0.045kg).
Here's some good info: http://lygte-info.dk/review/batteries2012/Panasonic%20NCR18650B%203...
I have many questions and want to learn more about Li-ions in our UAVs. Mainly, I understand that the max discharge of the NCR18650b is 2C or 6.8A. How is that affecting your flight when compared to the 20-40C rating of most popular Li-pos? Can you point me in the direction of where I may find more literature on Li-ion for RC? Besides price, what are the disadvantages. It seems to me you're getting the same energy from half the weight. Is that right?
Further to my point.. all simple autopilots used on aircraft have absolutely no control over power or throttle therefore if you command an increase in altitude you must add power manually so the aircraft is able to maintain AIRSPEED otherwise the autopilot will just pitch nose up and you will stsrt to climb but aispeed will decrease proportionately to the pitch attitude. This is fine however if the rate of climb selected is relatively small.
Advanced auto pilots like on citation x and airbus / boeing will have auto throttle. You can select a constant airspeed climb or just select a rate of climb. The pitch is first set, then the auto pilot detects a decrease in airspeed (after a short delay due to inertia) and adds power to maintain speed.
Primary effects of PITCH= altitude inc or dec
Secondary effects= inc or dec in speed.
Primary effects of POWER= increase or decrease in speed followed by an increase or decrease in altitude.
Well, that is exactly what I've advocated all along but I really don't understand anything about the TECS algorithms used so am just going on what I observe and understand as a pilot.
I think the requirements for a full size autopilot are vastly different from those of a non manned model plane and so our planes currently are not behaving correctly.
I think the problem lies in the programming assumption that throttle controls altitude. An age old argument.
The apm needs to use pitch as the primary altitude control and throttle (power) for airspeed control. This is obviously not the case if your gliding. You would need to set the airspeed and max pitch limits otherwise the apm might try to climb at an unsustainable attitude.
Pitch + POWER= PERFORMANCE.
If you fly a basic trainer like a cessna 172 the you might get away with power for altitude but if you fly an aircraft with the thrust line above the CG like most fpv flyers then power for altitude fails to work.
The link you sent confirms the issue of trying to use power for altitude control. Maybe ask the developer for some feedback on this.
Hi Ben, we flew at 13m/s with a plane which has a stall of around 8m/s, airspeed was fairly constant throughout the flight, much more constant than the altitude. As the wind speed was very low peaking at about 1m/s, one would expect the autopilot to at the very least, be able to compensate for this.
I have had this problem with all planes and all versions of APM:Plane since the introduction of TECS with v2.74 (see http://ardupilot.com/forum/viewtopic.php?f=77&t=5162) and have spent hours & hours flying & tuning and have also had a lot of communication with the devs regarding this (who's logs also show the problem).
As I have said in the past, from a practical pilot's POV, not a mathematician's, it seems there too little "yank" in the "bank".
You can see how dramatically the throttle varies with the height below. Maybe I'm being pedantic but efficiency of this flight would've been greatly improved with better handling of pitch and airspeed.
Great flight.
Just a note on the problem you mentioned regarding the phugoid motion...
It look like you are flying a constant radius with reference to the ground? The problem there is the need to speed up and slow down. The downwind leg could require you to fly too slowly resulting in a loss of lift not to mention higher drag. Remember that the aircraft should be flown at a constant airspeed not ground speed. Changing ground speeds should be expected in windy conditions and its not practical to compensate.
Another way to fly a circle amd without changing the speed is to change altitude slightly. What your
Looking for is the pivotal altitude. This changes with ground speed. Calculation is: groundspeed squared / 15= pivot altitude. (For mph.)
Otherwise stick to one efficient airspeed and accept a wobbly looking circuit.
Congratulations, Graham! I step away for a few days and the 100+km club keeps growing! I am anxious to see these capabilities put to useful applications.
@Criro, regarding the total distance traveled, I relied on reading the KML in Google Earth afterwards as Graham described. I could not find the total distance travelled in the log files, but I did monitor total distance in MP during the flight.
-
1
-
2
-
3
of 3 Next