What can I say about my MagpieAP? It was a long and ultimately unsuccessful journey trying to get this plane to do what I wanted. She flew beautifully once she was in her home, up in the sky. It was getting there that was the problem.
Of they myriad problems, some I managed to solve, some half-solved, some remained. If I had had all the time in the world I would probably have been able to get everything working. However, I eventually grew impatient, especially seeing as how my other platforms, the Long-Nose Getter, and later the Cat's Eye 1, have worked out so well. I am quite sure most of these problems were self-inflicted (I have an unfortunate tendency to mess with the design to correct anticipated problems, real or imagined). A lot of people have had good success with this plane, so it must just be me.
The MagpieAP is a commercial airplane kit, put out by Mountain Models. Its history is instructive. The original Magpie was designed by MM as a "trainer," i.e. a plane that inexperienced RC pilots could use to learn to fly. Characteristics that make a good trainer are that it be stable, predictable and survive crashes fairly well.
In general, what makes a good trainer also makes a good AP platform, with a few added criteria. Some enterprising AP enthusiasts bought the kit and modified it for aerial photography, with considerable success. The then-owner of Mountain Models, Doug Binder, being a wise businessman as well as designer of many successful model planes, revamped the design slightly and brought it out as the "MagpieAP", a design specifically for AP. A "bay" was added to the fuselage at the CG (Centre of Gravity) for the camera, with suitable ply reinforcement around the bay, a larger fuselage and wing, and slightly lower wing dihedral. It shares with it's "younger brother" good stability, few bad habits (in the air at least -- more on that below), a simple construction and ruggedness.
I purchased my kit in the spring of 2007 and got it airborne that fall, after spending an inordinate amount of time on the design of the camera mount. It flew very well, once up in the air, with a good glide ratio. It landed "hotter" (i.e. faster) than the Getter, but not particularly fast as model aircraft go, especially considering it was carrying 5oz of camera payload.
The MagpieAP is 36" (91cm) long, with a wingspan of 54" (137cm). Wing chord is 9 3/4" (25cm) at the root, tapered to 8" (20cm) at the tip. Wing area is 486in² (3135cm²). All-up weight of mine (with camera) was 39.9 oz (1130g).
The motor was an AXI 2808/24 turning an 8x4 APC "thin electric" propeller. The battery was a Flight Power 2100 mAH LiPo (Lithium Polymer) battery in 3s configuration (three cells in series). The motor drew about 15 amps at WOT (wide open throttle) on a fully charged battery, and provided about 30oz (850g) of thrust.
The camera was a Nikon Coolpix 5200. This is a 5-megapixel camera with fairly good optics for its price range.
It definitely showed promise as an AP platform. It had rather more "performance" than is really needed or wanted as an AP platform. It flew faster than the Getter and consequently needed more room for landing. It was tossed around by wind a bit more (perhaps simply as a consequence of it being smaller and lighter, perhaps because of the wing's larger dihedral). On the plus side, it was somewhat more maneuverable which I'd hoped would help in tight LZ (landing zone) situations, and it was light enough that it could be hand-launched if a strip of grass or roadway is not readily available as a runway.
As I mentioned, I "maidened" the MagpieAP in the fall of 2007. At the time, I was just used to hand-launching everything, so I hand-launched the MagpieAP. After a few tense moments, where it was "all over the sky," I got control of it and it flew well. At the time, I had some home-made landing gear on her, which didn't work very well. On landing, the gear collapsed completely in a tangle of wire. It was rather amusing really. But that put off further testing until the gear was fixed.
I ditched the home-made gear and bought a good solid carbon fibre landing gear at the Local Hobby Shop. Further testing went well enough for me to chance it with the camera. This included a 180° camera mount, which allowed me to shoot from the horizon left, through straight down, to the horizon right. I filled up a 1GB memory card (just let it shoot for the 14 minute flight) and worried about sorting later. I ended up with almost 800 shots, most of them either duds or the same scene repeated over and over again. Of course, all the shots from the right side of the plane were upside-down, and another unanticipated "problem" was the labour involved in rotating them all right-side up. Irfanview has a bulk processing feature, but you still have to identify which shots to process!
In the spring of 2008, the stability problems I was experiencing just seemed to get worse. It seemed to be an extreme case of what some people on RCGroups were calling the "Magpie Dance". Right after takeoff, it would nose up and lose airspeed, stall, and roll over to one side. I was having to correct both the roll and pitch and try to get the airspeed back up all at once. This was pretty much beyond my abilities as a pilot. On one hand-launch, it ended up behind the flight line and almost beaned a couple of other pilots.
I was still under the impression (mistaken as it turned out) that it was my crummy piloting skills and/or hand-launch technique. So next I tried to take off from the grass field. This fared no better. On one windy day I failed two or three times to get it airborne, having it either roll to one side or pitch up, stall and fall back to earth. I did get it airborne once, the takeoff being one of the worst in recorded history. As always, once in the sky, she flew beautifully.
I tried again on a calmer day. This time I got it airborne fairly quickly, but again it pitched up and I didn't correct quite fast enough (or maybe over-corrected, not sure), caught a wing and cartwheeled in. The only damage was to the motor mount, which was probably the best possible outcome, as it forced me to rebuild it.
At about this time I got my Long-Nose Getter finished and began testing her. This showed me, among other things, that there was nothing wrong with my ROG (rise off ground) takeoff piloting skills, so I finally admitted to myself that there were some severe problems with the plane and little tweaks here and there weren't going to cut it. After discussing the problem with several people in various emails and forums, I compiled this list of suggestions that might help. Thanks guys!
- more downthrust,
- shim the wing for lower AOA,
- reduce the power,
- move CG farther forward,
- reduce elevator throw.
In fact I implemented all but one of these solutions (I didn't try to shim the wing). Here are the mods I made:
1. I redesigned the motor mount to have an adjustable downthrust and ended up setting it at 12º (the downthrust "built in" is only 5º). At this setting it had no tendency to nose up when power was applied. The mount is pivoted at the bottm and has a row of holes along the top. Depending on which hole you put the screw in, you can select 4, 8, 12, 16 and 20 degrees of downthrust.
3. I was running an AXI 2808/24 on 3 cell LiPo's with a 9 x 4.7" APC prop. Under static test, this pulled just over 20A and provided about 40oz of thrust (over 1:1 thrust/weight ratio). This is just crazy for an AP plane! I changed this to an 8 x 4 APC prop. This tested at 15A and about 30oz thrust static. It's still perhaps over-powered, but with the added downthrust this doesn't present as much of a problem.
4. I had originally set the CG at the back of the tape spar. It was suggested moving the CG further forward to provide more stability and reduce elevator sensitivity. On my first flight, I moved this to 2¾" from the LE, or about 28% of chord. This proved to be a bit much, requiring about 20 clicks of up elevator trim. For later flights I moved it to about the centre of the tape spar and it seemed happy there, although there is still a bit of up elevator once trimmed. It appeared from subsequent testing that a CG in the middle of the spar is optimal for the purpose, namely a docile, self-stable plane.
5. I reduced the elevator throw by about
15% and added -40%
exponential. This was to try to prevent over-controlling when it
started to "buck".
...And yet more problems!
On July 22, 2007 I had an in-flight failure of my clever motor mount. There was a buzz, a sharp snap, and several pieces flew off the plane. I later learned that the prop, motor, ESC and external BEC left the plane entirely and landed in the field somewhere. I wasn't watching where they landed, as I was concentrating on trying to land the plane. It took me a good 20 or 30 seconds to realize I had no control!
Luckily, with the motor gone from the front end it just porpoised, making the descent fairly gentle, rather than a lawn dart. It also managed to find the one patch of tall grass on the entire field. The only real damage was to the motor mount.
I next launched my Long-Nose Getter to do some aerial reconnaissance to try and find the missing motor/ESC. The Getter performed flawlessly, and I got about a hundred shots, but after spending about 20 minutes staring at the little screen on my camera, I came up empty handed. Eventually, I enlisted the other pilots to do a ground search and it was found in a few minutes. I've looked at the photos since, and, even knowing where it was, I can't find any sign of it in the photos. I guess it was buried down in among the grass.
I guess the mount was just too flimsy, being out of 1/16" plywood. It should have been plenty strong enough if all it had to do is transmit the thrust from the motor to the airframe. However, I had a bit of an "oops" on an aborted takeoff just prior to that flight, so my theory is the motor mount sustained some damage then, that wasn't immediately visible. After the "incident", the plywood was a mess so I can't really tell what was done during the "incident" and what might have been there beforehand. The plywood broke on the top where the row of holes went through, as well as just above the lower screw mount on one side.
After this incident, I built a new mount using more wood at the top where the line of screw holes is, and leaving a "brace" at the back for added strength. I also used 1/8" plywood instead of 1/16". See photo below.
Also, although there was no evidence of a problem with the screws pulling out, in the back of my mind I had concerns about the screws just being held in by 1/16" plywood on the fuselage. So I decided to be proactive and changed this a bit. On the bottom, I put a 1/8" CF rod through the fuselage, sticking out about 1/4" on each side and the mount pivots on that. On the top, I put a 1/4" dowel through the fuselage, cut flush with the fuse sides, and screwed into the ends of the dowel.
I eventually got the MagpieAP to the point where it was a passible AP plane. Since the above modifications were made I had several successful flights, including several flights with camera, and apart from the problems on takeoff it seemed to be performing well. I had one nose-over on takeoff, but I concluded that was due to "pilot error." I since learned to keep some up elevator until it is airborne, then back off and let it gain airspeed before the climb-out.
If I hadn't had so much success with the Long-Nose Getter and the Cat's Eye 1 I would probably have been able to use the MagpieAP for my AP work. However, as it was, I became discouraged with its shortcomings and eventually removed the electronics and gave the airframe to a friend.
Finally I should mention that I blame all these problems entirely on myself and in no way hold Mountain Models responsible for any of these issues. The plane was far from "stock," in particular having a much more powerful motor than the recommended setup, a rather over-built wing, a heavily modified camera mount etc. Had I built the plane "stock", I'm sure it would have performed better. Indeed many people have reported having no trouble with it whatsoever and being delighted with how it performs.