Saturday, January 21, 2012

Sky Angel MiG-15 Build and Flight Review

UPDATE 5:  Added landing gear with a steerable nose.  See Z8RC Mods, below.

UPDATE 4:  Added a rudder and thrust vectoring.  See Z8RC Mods, below.

UPDATE 3:  Uploaded the HD version of the maiden video, linked below.

UPDATE 2:  Flight grading complete.

UPDATE 1:  While the wind blows, I added a segment on fan balancing.

I couldn't think of a reason not to buy this cool $64 Sky Angel MiG-15, so here it is!  According to their website, Sky Angel is part of J-Power Group maker of the Z8RC favorite foam parkflier warbird, the P-38 Lightning featured in the sidebar.

I was going to do a build thread followed by a flight review, but the plane was finished before I knew it.  Great job Sky Angel!  The level of ARF subsystem completion actually make it "almost ready" to go.  Refreshing.
I like that the nose is not painted, it is 
wrapped in a form-fit red plastic ring.

- Wingspan: 25.6 in
- Length: 25.6 in
- Wing Aea: 171 sq in
- Wing Loading: 10.8 oz/sqft
- Wing Cube Loading: 9.9
- Flying Weight: 12.4 oz w/800 mAh 3S
- 50 mm Ducted Fan w/Brushless Outrunner
- 20 Amp Brushless Speed Control
    The EPO airframe ships in six large foam pieces, but all the electronics are factory installed. Required assembly consists of gluing the foam building blocks together, running two push rods through pre-installed tubing, then tightening down two elevator control arm adjustment screws.  If you subtract think time, it's a 5 minute build with a few shots of CA accelerator.
    The cockpit doubles as a battery hatch.  It is solid 
    and looks great. The magnets needed a bit of depth 
    adjustment for maximum click.  I changed the ESC battery 
    connector to match my 800 mAh 3S battery collection.
    I only tweaked the jet with two, minor airframe mods:

    First, the mini MiG has a traditional flat-bottom airfoil cross-section.   Like a trainer, the wing should generate  a lot of lift, but also sharp stall behavior once Critical Angle of Attack is exceeded.  I dialed-in about an 1/8th inch of permanent spoileron (up aileron x 2) to keep stall development as wings-level and gradual as possible, by forcing the wing root section to reach Critical AOA sightly before the wing tips.
    A little cranked-in spoileron approximates washout.  
    Both push rods are for elevator.  There is no rudder control 
    as shipped.  The jet has independent aileron servos which 
    add some yaw control options given a computer radio.
    Next I added three carbon fiber landing skids from a thin woven sheet of amazingly light and tough carbon fiber, shown in the pic below.  The aerodynamic effect of the skids should roughly cancel each other, as there is as much nose skid area in front of the Center of Lift (roughly the weighted geometric centroid of the lifting surfaces) as the two wing skids behind it.
    Although there is plenty of power to push the plane on the skids, the jet has no rudder channel so its probably still best to hand-launch.  At least landing on pavement is an option.

    Speaking of the lack of rudder, I decided to partially simulate rudder by making the rudder stick lift a single aileron on the inside of the desired turn direction.  Should be simple, right?  Wrong!  Unfortunately, I used an old $8 OrangeRx receiver as a suitable match for the low price of this MiG, and that decision forced me onto a Spectrum radio standard.

    Spektrum's crude way of issuing control commands made the required mix difficult to implement and near impossible to conceptualize, because Spectrum mixes "flight control" channels, not simple servo channels, this often generates major logic problems.  This desired mix is but one problematic example: with a dual aileron setup, the flap channel always moves both aileron servos in the same direction so it's impossible to control only the servo plugged into the selected flap channel. 

    A proper radio standard logically controls each servo as a discrete, unlabeled channel which keep mixes pure, simple, and powerful.  This fundamental Spektrum design flaw causes endless empirical trial and error gyrations to find the just right anti-kludge to counter the baked-in faulty logic.  Just awful!

    After about an hour of thinking it through, combined with a methodical process of isolating variables and empirically determining input effects, I finally stumbled upon an effective anti-kludge.  It makes no apparent sense, but it works (almost):

    WINGTYPE: Dual Aileron
    MIX 1:  Rudder > Aileron ACT L +100% , R +100%
    MIX 2: Rudder > Flap ACT D +100% , U -100%

    Miraculously, this enables the aileron stick to work as pure aileron control while the rudder stick lifts only the inside aileron while leaving the other motionless, presumably to induce some yaw.  It also mixes the two concepts when the sticks are used together.  Interestingly, when ever the two sticks are moved opposite one another with the same deflection, you always get pure spoilerons (both ailerons up) of the degree of stick deflection, regardless of which direction the sticks are deflected.

    Unfortunately, because Spectrum's flight control logic is fundamentally flawed, this complex kludge makes rudder trim impossible so it must be deactivated.  Because the rudder is mixed to the "flap channel" instead of to a servo-as-a-channel concept, the rudder trim rocker always moves both ailerons in the same direction.  This is poor control logic, nothing more than inferior intellect permanently embedded in the DSM-X standard. Uhg.

    A proper radio, like my Hitec Aurora 9, simply requires mixing left rudder to +100% of Servo 1 (= the left aileron servo) and another mix to map right rudder to +100% of Servo 5 (= the right aileron servo).  How friggin hard is that?

    The MiG's power is good for a basic ducted fan model with a measured Thrust:Weight ratio of [9.5 oz / 12.4 oz], or about 0.77 running on a fresh 800 mAh 20C 3S.

    When completed with a battery laid into the bay, the MiG-15 balanced exactly at the 95mm point from the root leading edge, as outlined in the decent instructions.  The spot is coincident with the rear edge of the painted black wing walkways.  Nice.

    Overall, I'm very impressed with the quality of this simple and tough EPO foam model.  Even with a battery installed it feels light when handled, which is usually a good sign.  It is such a quick and easy build, it makes you wonder why it even costs $64.  In my view, that is goodness.

    I hope the flights go as well as well as the ground assembly!

    -----

    Since the winds are out of limits for the time being, and Sky Angel bought me some time with their high level of completion, I decided to do a few upgrades that I sometimes don't have time to do or forget to do.
    Over time, I've found these upgrades well worth the time, some are obviously just for foamies:
    1. Taped the wing leading edge with clear glossy mailing tape for wing strength and foam protection.  The MiG's top side wing fences and leading edge antennas required a trim-around.  There is no problem a new xacto blade can't solve.
    2. Taped the bottom center fuse, gun, and wing tanks in front of the skids.
    3. Taped the square fuse panel under the rear canopy to preserve the finish while popping open the canopy at the magnet. 
    4. Taped the wings roots inside the first wing fence for a place to grab the plane.
    5. Embedded two plastic servo arms in the vertical stabilizer to serve as push rod wire guides.  The push rods are thin and bow under pressure.  This should help lock elevator response to stick position.  After cutting each servo arm and installing it separately, I realized I could have embedded the entire uncut servo arm through the rudder and it would have been stronger.
    6. (Not shown) Embedded two carbon fiber reinforcing rods in each side, where the wing tank mates to the wing.
    7. (Not shown) Embedded a carbon fiber rod down the barrel of the gun.
    8. (Not shown) Cut a thin air duct running from the front of the battery bay down into the nose intake, to get some cool air around the battery and ESC.  The back of the bay is already vented down to the motor.
    9. Balanced the fan.
     
    Perfect fan (or prop) balancing:

    A fan is as easy to balance as a prop, and this one needed it badly producing a loud harsh vibration at full throttle.  All you need is a set of drill bits, a vice that can open a few inches, and two strong magnets capable of suspending the drill bit between them.  Any magnetic prop balancer can also work, but most are too big for a small fan, so you'll need to adjust the distance between the magnets to accommodate the length of a small drill bit. 

    Removing this fan is simple. Stand the jet on its tail to avoid losing the washers between the fan and the motor.  Remove the bottom access panel.  Hold the fan still with an index finger while rotating the rear motor bell counter clockwise (looking into the jet exhaust tube).  The fan will unscrew forward then fall off without applying any force.  Make sure the washers stay on the shaft behind it.  I had to trim away a small bit of foam to get the bullet nose of the fan through the access panel.  The other option is to let it fall out the nose intake, but there are washers still on the motor shaft and I didn't want the fan to get stuck.

    With the fan free, insert the right size drill bit as a substitute motor shaft.  It needs to be a sharp drill bit and a snug shaft fit (shrink tubing can be your friend).

    Suspend the fan on the drill bit between two magnets stuck to the jaws of the vice.  Open the vice wide, so the drill bit is barely suspended in the magnetic field.  Place the drill bit so the sharp end is the one touching its magnet.  Slowly close the vice.  The fatter metal on the flat end of the drill bit will become more attractive to its magnet as the vice closes, eventually the flat end of the drill bit will pop over to its magnet.  Back the vice open a little, and put the sharp end of the drill bit back on its magnet.  You want the vice open just the right amount so the sharp end of the drill bit barely dances on it's magnet, at that point you'll get a virtually friction-free fan balance.

    Apply bits of tape or dots of glue inside the hub of the fan to balance it, until a light nudge spins it for a long time without wobbling.  Don't mess with the fan blades.   Now you'll have a smooth running fan as long as the motor bell is balanced, and most are.

    Good thing the wind is blowing, I probably should have done all this before flying anyway!

    In the Air:

    Finally today dawned with a beautiful calm morning, I definitely need more of those.  I don't even know where to start with this MiG, what fun.  Fun like a wooden rollercaoster is fun, because you aren't quite sure if it is going to fly off the tracks.  This was, without a doubt, my craziest maiden flight ever.

    The adrenalin rush was mostly my fault.  I dialed in some spoileron to keep the MiG from tip stalling.  That was a good thing.  To my credit, I even pre-positioned 5 down-elevator clicks from what looked like neutral.  That was, of course, assuming the balance was good.  I used a 1000 mAh battery, thinking it'd be a little nose heavy and all good.

    Bzzzzt.  Not so much.  To my discredit, I heaved this baby with way too many variables dialed in and out and up and down. The net result was a strong tendency to climb, coupled with decent right roll..  That's no big deal (well, it is a bigger deal with a ducted fan that isn't as responsive as a prop), the real problem was that it ate all of my DX6i's down elevator trim without so much as a burp.  Now what?

    After managing to get some altitude (pretty easy with too much up elevator), I decided it would be better to try to sub-trim down with the airplane up high, than to try to land holding the nose-down with the airplane down low.   So I clicked my DX6i over to the subtrim page.

    I didn't remember if I reversed the elevator servo or not, so I wasn't sure which way to trim.  I tried trimming in the arrow down direction first.  Oops, that just popped the nose way up and the MiG entered a fast spin (2:40 mark).  I threw all the power, down elevator and outside spoiler that I could muster at it, and the plane recovered without a rudder.  Whew!

    Now holding the stick even farther forward, I realized the flaw in the DX6i's right-side roller, making it very hard to fly and modify settings.  The DX8 fixed the problem by making a lot of the menus inaccessible unless you turn the radio off, so you have no choice but to crash right away.  With some thumb switching while doing the lobster eye, I slowly beat back the wrong radio design and reversed the sub-trim, and then...  .....ahhhhhhhhhhhhhhhhhhhhhhhhhhhhhhh...

    Hey, I thought, this thing really flies great, it's me that sux!

    The MiG definitely flies like a 1st gen jet fighter, and I love RC planes that fly like the airplane they attempt to model.  What fun!  The nose track is soft and the usable pitch range is expansive, just as one would expect.  That enables a broad airspeed envelope.

    At high speeds that plane is totally under control and the flight control response feels well tailored.  The scale size of the small surfaces prefer airflow.  High speed passes are point and shoot, and the and the small elevator allows a lot of trim resolution (a place the DX6i normally falls short).  The swept wings are stable in a scream.  The small ailerons keep a lid on over-controlling the delta on the longitudinal axis.  

    Like all early-gen MiGs, this airplane also makes a surprisingly capable low speed adversary.  The plane demands careful attention at slow speeds, both the pitch and the roll axis become mushy.  But there is plenty of room to slow down while ratcheting the nose higher and higher, as long as you project your flight path far enough forward and so you can limit the big changes to pitch.

    I didn't want to flirt too deep into the stall on the maiden and spin the jet, but I managed a few low and slow passes before the battery gave out.  On one or two occasions I felt a few light, pre-stall pitch oscillations.  The plane stayed well composed and seemed to porpoise mildly on the pitch axis only.   Promising.

    The built-in flight timer surprised me, not because of the number of minutes coaxed out of the 1000mAh as revealed by the video counter, but because the first 3.5 minutes of my airborne buffoonery must have passed like 30 seconds.

    Zzzzzz z z z zzz z zz     z z z      z     z       zt.   No motor.  No good runway options.  The MiG-15 glided like a champ and set down softly in the grass.  +1 for the Soviets.


    All things considered, this jet might be the most fun I've had RC'ing for under $70 (OrangeRx included).  No, it definitely is. 

    As much fun as this MiG is, I might alter  a few things.  One, landing gear would be useful, imagine that.  The tin can w/motor design isn't the best hand-launcher, but now that it is trimmed up it might be easier. I'm going to look at some gear options.  Two, my simulated-rudder mix worked great.  It effectively provided a way to trim out yaw and might have saved the jet from the spin.  But if I add landing gear, I'll probably need a rudder for nosewheel steering too.

    Updates to follow.  Or... maybe... this was so much fun it's already time for the bigger Sky Angel MiG-15?  Either way, more to come...

    Z8RC Mods:

    Landing Gear with a Steerable Nose 

    After learning that hand launching is better way to test the durability of Sky Angel's EPO foam (outstanding) than to get airborne, I decided it was time to add landing gear.

    What could be easier than drilling a 5/16" hole in the bottom of the nose to insert a Parkzone Bf-109 steerable tailwheel?  Running a pushrod from the rudder servo (see mod below) through a hole in the cockpit to the tail (nose) wheel's reversible control arm was a piece of cake.  The curvature of the fuselage perfectly matches the curvature of the mounting plate.  The steering works perfectly.
     
    Maybe I'll add some gear doors?
    The mains were almost as easy - 0.25" aluminum tubing with a hole drilled in the bottom to accept a screw to act as an axle.  I drilled a hole halfway into the foam wing to accept the aluminum strut and covered with a square of balsa with a matching hole.  Gorilla Glued it all in place.   I poked a few holes in different directions into the foam so the glue has place to expand and form roots.
    Maybe I'll add some gear doors to finish it off.

    Rudder Mod with thrust Vectoring

    This was an easier mod than I expected.  Installing a rudder is as simple as sliding a second servo into the empty holder next to the elevator servo, then threading a piano wire push rod through the existing hole for the motor wires into the exhaust tube.  I sliced the foam rudder off on the scale lines, in two parts above and below the elevator, then ran a carbon fiber rod along that same line to act as a hinge.  I continued the CF hinge rod all the way down through the exhaust tube and anchored it in the bottom of the aircraft.  That made it easy to add a control horn inside the jet exhaust.  I also added a thin carbon fiber flap to the hinge rod, at the back of the jet exhaust for thrust vectoring.  Incorporating thrust vectoring in a jet is a good thing since there is no prop to blow the tail surfaces (= thrust vectoring).  Pix and flight test to follow when the touch up paint dries.
    A carbon fiber rod forms the new rudder hinge. 
    The control horn is inside the exhaust tube, as 
    is a carbon fiber deflector for thrust vectoring.

    Z8RC Grades for the Sky Angel MiG-15:


    Appearance: A+
    A big red star on a smoking tin can.  Awesome.

    Airframe:  B
    Wide flight envelope. Built for speed.  No rudder or gear.

    Power System: C-
    Decent power.  Fans aren't responsive.  Needed a balance.  Loud. 

    Build Quality/Durability:
    A+
    Super tough EPO foam!  Smooth finish.  Nice canopy/hatch.

    Value:  A+
    Jet around in style on any budget.

    Overall Grade:
    A
    Looks fab.  Flies fab.  Tough to get airborne.  Huge fun for the money.