Thanks again Dick for your excellent information! Good to know about the original airplane as it should always be the yardstick for scale modeling.

Funny poor chap, this errant raven, but those bird-model stories are also moving and heart-touching for me. Maybe one has to see such a model to realize why. [link=http://www.das-nurfluegelteam.de/fred_ludwig/storch.html]Here[/link]'s a stork model that looks and flies very realistic, too. Once it was thermalling together with four real storks (they came along) for nearly 10 minutes and nobody could tell them apart.

Now, after testing spalex's model in the simulator, I can verify all what otrcman and BMatthews recommended. The simulator model isn't completely realistic but it shows it's characteristics rather clear.

To begin with performance: For 4.5 lbs weight, use the small 2208/34 motors, with 2s2200 LiPo and 9x6 prop or an equivalent drive with bigger props. The 0.37 thrust/weight ratio gives reasonably scale-like take-off and climb and is well enough, even for grass runways. Less than half throttle is enough for parkflyer-like cruise speed. Landing roll is too short in any case unless you set some power.

Behavior: You just can't avoid tip stall anyway, neither by washout nor by flaps. Both don't hurt, either. I'd pass washout but build flaps because they make for more scale-like (slower than cruise and steeper) landing approaches (much power provided). With or without flaps, no three-point landings are possible due to stall. Just do it the other way.

Don't fret about tip stall, you won't suffer it unless you set out to it. It's easy to avoid having only 10 oz/sqft wing loading. Recommended control throws: aileron 10, elevator 12, rudder 20 (split flaps 60) degrees. If you think that's too little set dual rate to 15/20/30, but you'll see that's even worse. Set 50% (or more) aileron differential and activate 100% aileron-rudder coupling. Roll response is good.

Setup: 0 degrees stab incidence, 3 degrees (or even more) wing incidence, no right or down thrust. That's what the 3-view drawing shows, and it's good. Balance point (C/G) should be 6" behind the center section leading edge (or even more, 7" is not bad, neutral point is at 7.62")).

Great flyer! Still too fast cruise, too small turns, too short landings, too much control response. As otrcman said, it's a compromise. You're on the controls to let it look more scale-like, as BMatthews described. Keep us posted!

The other day the current owner of the DC-3/C47 was at the field with the plane I had mentioned earlier.
He made the comment that he added a gyro on the rudder not because of the engine out performance but rather that it was a real bear as far as ground handling, as well as takeoff and landing.
Just some information I though the original poster might find useful.
Apparently the rudder is not all that effective at low speeds...
He also added that he is going to add the split flaps as well, now he flies multi-engine aircraft all the time including a p-38 so I would heed his advice.

Gosh, can your simulator tell you all that? Guess I haven't given simulators much thought.

I'm working out the control laws for a twin model right now, but doing it by successive iteration in flight. So far I've got 5 flights and am nearly where I want to be. The model is a P-61 Black Widow. It has pretty nasty stall characteristics, so I've been slowly reducing the elevator travel to see whether I can reach a compromise where elevators still give adequate control and yet don't cause much of a stall.

My approach is to set two different elevator travels in the transmitter, one on HI rate and the other on LO rate. I take off in HI rate, then evaluate LO rate once airborne. So far on each flight I have found LO rate to give satisfactory handling so I have done the remainder of the flight in LO. Then on the next flight I set the new HI rate to whatever was LO on the previous flight, then set the new LO rate still lower. On each flight I also evaluate the stall at a safe altitude. On the first flight, the stall occurred at less than 1/2 aft stick and it was horrific. I'm now down to about half of the original elevator travel With this travel I occasionally get a mild snap at full back stick but sometimes I have to further aggravate things with a bit of rudder to cause a snap.

On the next flight I will reduce elevator travel a bit more. I'm hoping that I will be able to find a point where the stall is pretty much nonexistent while still having enough elevator travel for satisfactory flight. Typically there will be a difference in elevator travel required to stall power-on versus power-off. My transmitter isn't sophisticated enough to schedule elevator travel against throttle position, but hopefully the day will come when I have such a transmitter.

The man who built the raven models published a very nice article in RCM about 5 or 6 years ago. Lots of pictures and a very detailed explanation of the control system.

Dick

Yeah Dick, the simulator is that good! But seriously, it gets you in the ballpark and gives a good impression and feeling of a model before building the real thing. Sort of proof-of-concept. Because it's not quite accurate in the first place it would be of no use in your case. Finding control throws by test flying seems to be the best idea. Once you have the real model, though, you could calibrate a simulator model to be very accurate and "virtually" practise to cope with the nasty flight behavior. We once did that with a very stall-prone F5B hotliner, which was as hard to fly as a helicopter.

iron eagle, the DC-3 rudder is huge and quite effective, but the problem might be that it's not really hit by propwash, in contrast to the stab. It might help to pull full elevator and have a steerable tailwheel. (That's pure guessing, though.)

The heavy glow-powered (as well as electric) versions are rather different models. I tried a 14.5 lbs version with .40 4-strokes in the simulator. It looks exactly like in the video of Charly Binder's electric model. Flaps are really advisable to get the thing to the ground properly. Of course there's no three-point landing as well. Only a tad more control throws needed. By the way, the C/G position given in the other thread (135 mm / 5.3") is quite far forward (as usual) and would require more wing incidence (4 degrees) or give a rather fast cruise flight. It's a starting point but 155 mm / 6.1" or even 175 mm / 6.9" is better.

A heading-lock gyro would be not a bad idea in case of engine failure. The engines are not far outboards and there's no problem in cruise flight. But disaster is looming if you add power for whatever reason, to fly faster, to climb, or even to deploy flaps. You know who you are, but I'm swamped in this case (too slow reflexes). I'd recommend smaller engines like .20 2-strokes (even the old .19 would still suffice, really) to make things easier. They surely give enough power and a more scale-like flying in addition.

The low-wing-loading / low-power electric version discussed here is child's play even in case of engine failure. Little rudder needed, even with full flaps. Full flaps possible even with only one motor running. Fault-tolerant if you're mistaken which motor quit.

Dick, it sounds like you're forcing the stall rather than nosing up to maybe 5 degrees nose up then use just enough added elevator to hold it there and just wait for it to loose enough speed on it's own and stall naturally. From all the stalls done this way that I've played with there's not enough energy left in the airplane to snap. At best with some rudder added just pre-stall I'll get a fall off on the pro rudder side into a spiral dive that may or may not develop into a spin depending on the model if I hold the rudder input.

Now if I were to pull the nose up strongly to something like 15 or 20 degrees nose up then yeah, I'll get more funky stuff happening and I may well get a snap if I add in rudder just before it slows down enough to stop flying.

If anybody cares, here's the result of my sunday afternoon relaxation, trying the [link=http://xflr5.sourceforge.net/xflr5.htm]QFLR5[/link] program mentioned in [link=http://www.rcuniverse.com/forum/fb.asp?m=9070389]the other thread[/link]:

Only wing and empennage are rendered because I was interested in the tip stall problem. Fuse and nacelles would obstruct the view of the wing airflow and have been omitted. The wing center section has NACA 2215 airfoil, the outer wing panels are tapered to 2206, just like the original. Only the elliptical wing tips have washout but not the swept panels with the ailerons, just to elaborate the problem. Size of the model is 1:12 (95" wing span), weight 4.5 lbs, c/g 6.9" behind center leading edge.

With all due caution, the program shows the general problem of the DC-3 wing, even if not exactly.

The first picture shows the case aoa=6 degrees, what is twice the incidence angle. So the fuselage is not exactly in flight direction. Speed is 22 mph, the lift coefficient 0.66. That is the cruise flight tested in the simulator. The wing's upper surface is blue indicating low pressure. The orange grid behind the wing indicates induced drag, the pink grid behind wing and tail indicates viscous drag. The red line on the wing from tip to tip shows the point of transition from laminar to turbulent flow. It's slightly aft of the maximum thickness, as normal. The green lines behind the wing are streamlines showing the decent tip vortices.

Now compare the second picture, aoa=8 degrees. Bigger induced drag, very noticeable tip vortices, and a more forward transition line. Most interesting is that the transition line comes close to the leading edge outboards near the wing tips. Only 1 or 2 degrees more aoa and the transition reaches the leading edge and stall begins. (At least that's my interpretation.) That's what happens also in the simulator. So the wing's lift coefficient seems to be limited to 0.85 or 0.9 due to early tip stall.

Now the tapered and swept wing panels have 3 degrees washout. Still the elliptical wing tips have additional washout, like before. Of course, the tip vortices are smaller at aoa=6 degrees and the laminar/turbulent transition is more aft near the wing tips. Lift coefficient is only 0.60, as is due.

Gain is 2 degrees more aoa (10 degrees) and a 0.97 or 1.0 lift coefficient - not too bad. What the pictures don't show is that the stall is now more abrupt and severe. The complete outer wing stalls now at once while the stall progressed from the tips inboards when the wing had no washout.

Neat! (Sorry for the metric units.)

As I said before, I would pass washout and build (and use) flaps. But maybe the models have different airfoils and wing layout and behave differently. Compare this interesting [link=http://www.rcuniverse.com/forum/fb.asp?m=1604561]post[/link]. Or the calculations are just wrong...

Ah, but while with washout the stall may be more abrupt this is a far better state of affairs than to have the tips stall first. A "nice" stall is where the center section stalls first and the tips, that keep the plane stable through the sort of stall manage to hang on bitterly so you still maintain control rather than falling off to one side or the other.

Deploying flaps around the center not only increases the camber of the area with the flaps but it also severly shifts the angle of attack of the flapped section in a positive way. In effect it results in extreme washout but makes the plane more stable by doing so.

Granted a lot has to do with each designer's ideas but I'd rather see a healthy amount of washout and avoid the tips stalling before the rest of the wing.

Yeah I know that very well, that's what I always loved the good old single-engine Cessnas for. They have less taper, the same airfoil thickness near the tips (and uncambered or later models the drooped nose), and don't need that much washout, though.

That's what I meant in my former posts. But at least in the simulator that doesn't cure the problem, still don't know why. I could try it in QFLR5 when I have some time.

So even 3 degrees of washout aren't healthy because still the tips stall first, even the whole outer wing at once. I could try 6 degrees washout, we'll see. A different airfoil might be better, though, and not really perceivable on a scale model.

Have to correct: It's not the elliptical wing tips which stall first, it's the tapered and swept wing panels. To make it blatantly obvious, here's how it looks if this panel has even 6 degrees washout, progressively from the square center panel to the elliptical wingtips. At the same time the airfoil is tapered from NACA 2215 to 2206, as before.

Nearly no lift at the wing tips (look at the orange grid for induced drag) and laminar flow (the red transition line) for aoa=6. One more degree of aoa before stall begins: aoa=11. Very little lift at the wing tips and mostly laminar flow. But strange things happen, there's much viscous drag (the pink grid) in the middle of the tapered wing panels. Like before, the transition line comes close to the leading edge near the wing tips.

At one more degree aoa (12) the tapered wing panel stalls except a small outer part. Again one more degree aoa (13) lets the whole panel stall, and even one more (14) also the wing tips. The square center section is still unaffected. Only at aoa=17 the center panel stalls all at once.

Well, using the 2215 airfoil for the whole tapered wing panel does change something but doesn't really make things better.

In both cases, 3 and 6 degrees washout, there's now a real tip stall: the elliptical wing tips stall first. The stall is not abrupt but smooth and progresses slowly - just in the wrong direction. The stall starts at the tips (aoa=12) and progresses inwards until it reaches the root (aoa=17). Duh.

I think it's simply the high taper that makes for a "bottleneck" in the midlle of the tapered wing panel. There the chord is too small while it is enough at both ends, so to speak. Seems the designer knew what he was doing.

OK, here it is with flaps. Original configuration with 2215/2206 without washout. There are no split flaps in the program, only plain flaps, here set to 45 degrees deflection. The classic VLM method didn't work, so it's Quads. Don't know if the results are comparable now, but they look reasonable.

It's as expected: smooth stall beginning at root and progressing to the tips, over a span of 4 degrees of aoa. But: These are rather small aoas. First picture is for aoa=3, meaning the fuselage is level. The second picture is for aoa=6, just before stall, which begins at aoa=7 and is complete at aoa=10. That's not bad at all. It confirms that the tip-stall problem vanished and the plane stays controllable as well as that no three-point landing is possible.

The pictures look rather messy, showing some strange (?) things. There seems to be a negative induced drag where the ailerons are. Much uneven viscous drag behind the flaps. Strong vortices at the flap-aileron break. Doesn't look wrong, though.

The aoas look similar in the simulator, but the stall seems to be somewhat indifferent and looks like a tip stall. The model is controllable though, like a 3D pattern model which can do no steady harrier. I wonder which simplified stall model the simulator uses to get this behavior. Oh well.

An afterthought: The rear view with streamlines clearly shows any vortices. The "clean" wing at aoa=8 also looks clean, no vortices except at the tips. The "dirty" configuration also looks dirty with many vortices, especially of course at the flaps-aileron break. But the whole outer wing panels with the ailerons have small vortices as well. Maybe that's why the airplane seems so wavery, not knowing whether to fly left or right. Over-interpretation?

UStick,
Thanks for the link to the software.
The graphics you posted for the wing performance of the DC-3 is very interesting.

Here's a [link=http://www.youtube.com/watch?v=3pKk9KCVpIA]video[/link] of the maiden flight of a Ziroli 1:8 DC-3 - turned out to be a sad day. I really feel sorry for these guys, but have to say that the too short flight looks exactly like a stalled flight in the simulator, that wavering and uncontrollability and tilting over to one side or the other.

So the simulator can't be completely wrong, that's one lesson learned. Another one is to never forget that there's something like down elevator. As was said before in this thread, we can not feel what happens to our model, but we can see.

Here's a good source of all sorts of [link=http://www.douglasdc3.com/]DC-3 information[/link], including flight beahvior, landing procedures, ...

Yes, a sad end for that Ziroli DC-3. But what great pictures ! Having spent part of my flight test career doing stall/spin research, that is really classic stuff. Notice the nose slice (yawing in the direction of the roll) signaling the onset of each departure. That is some of the best footage that I've ever seen.

At wing stall, most people think in terms of loss of lift. But the other and usually more dangerous effect is the huge increase in drag that accompanies flow separation. The drag increase, if it occurs on one wing more than the other, precipitates a yawing motion which exacerbates the flow separation. The greater flow separation causes more drag and more loss of lift, and more yaw, and on and on. Truly an unstable divergence.

In retirement, I have been doing (full scale) flight instruction in advanced skills & techniques. Mostly what I do is take Cessna flyers and make them into competent vintage airplane pilots. Vintage airplanes typically have tailwheels, no artificial stall warning, and lots of adverse yaw. One of the things I usually demonstrate is that you can fly straight and level just above stall speed using only a whisper of power. Then a little additional up elevator and the airplane stalls. Rather than recovering immediately in the conventional FAA approved fashion, I hold the stick aft, keeping the airplane fully stalled, and work the rudder briskly to avoid rolling off in either direction. At this point the student is directed to look at the rate of sink. What he sees is that the airplane is coming down like a load of bricks while still using the same amount of power that kept us nicely airborne when just a few knots faster. Then the stick is allowed to go forward slightly so the wing is no longer stalled. With no change in power, the sink rate goes away and you are back to level flight. A truly graphic demonstration of drag increase due to stall, and also a demonstration of the treachery of an unsymmetrical stall. I also warn the students not to try this unsupervised unless there is plenty of space between the airplane and the ground, because one false move precipitates a spin.

Dick

The technique described above, is commonly called the "parachute" by 3D model flyers.
When I got into small electric models ,especially the things I could build from flat sheets of foam, Iwas able to proove/disproove a number of of rules about flying
The most interesting were concerning very low speed flight .
The Lightplane rule was to simply avoid low speeds - they can kill you.
Seen that happen too often.
Learning how to control a deep sink takes a fair bit of practice
In the little foam models -if you screw up -maybe some bent foam and a broken prop.
In Light planes "stretching th eglide" or dragging th plane in on landing are risky -ditto for typical models of the lightplanes .they simply are not good at "departure " flight. One hard rule is to leave the ailerons alone and concentrate on rudder corrections. It's counter to the feeling that you must "pick up the low wing".

Here's another video that seems to suffer from the same sort of tip stalling issue.

http://www.youtube.com/watch?v=LTUoG...eature=related

It looks like the Ziroli crash could have been avoided with a little more airspeed. On takeoff they survive one tip stall oscillation and the nose lowers and it seems like all would be well if they had let the model accelerate some more. But the trees on the edge of the field obviously had the pilot's attention and he tried to pull the nose up before he had the right airspeed and we saw the result. On the second video it seems like a combination of an overweight model combined with the same tip stalling issue. Except in this case it occurs at the bottom of a dive pullout due to, presumably, the weight of the model and the small size combining to form a high G loaded stall.

A large amount of washout could possibly have avoided both of these instances or at least made the departures less radical and controllable. And without knowing the flying weight and wing loading it's hard to say but it's entirely possible that the first was overweight or at least on the edge enough that it required a more knowledgable hand at the controls to work towards avoiding the issues. He tries to climb away immediately on takeoff rather than letting it accelerate in the ground effect. A little more speed and letting the model reach trim speed and nose up naturally into a climb would have avoided this issue. Granted it was a first flight and the trim condition was unknown but in any event a little more speed would have helped. I don't know about you guys but models on grass fields often get bumped into the air when they are all too close to the ragged edge. And a heavy wing loading combined with a "scale" amount of power means that such a model needs a bit of time to reach a safe climb speed.

Added to this issue would be the likely use of large aileron deflections to try to pick up the wing. The adverse yaw and aileron induced stalling on such a critical wing shape flown at a near stall speed would make such inputs disasterous. This may have been a factor in both videos.

All in all an interesting thread and some excellent studies on all counts. It points out that the venerable DC-3 design may be a long lived one but that it's not without its foibles that pilots need to come to terms with before they can expect a long and healthy career.

A super light version will avoid SOME of the issues but the tendency will still be there. An initial takeoff that lift off and then flies level at first would be a wise precaution. As the model speeds up to a safe speed it should establish a natural climb with any stick pressure being held at the same setting that provided the level flight in the ground effect. If it doesn't want to climb out naturally then only give it some additional up pressure after a couple or three seconds. Given what we've seen in these videos and simulator studies this is how I'd approach it until the model's charactaristics are well known.

Couldn't agree more! There's even more information about that jinx. It's mentioned in the [link=http://www.rcuniverse.com/forum/m_1136542/anchors_1604561/mpage_5/key_/anchor/tm.htm#1604561]thread[/link] I linked to earlier. Go to the last page (#23) and you'll find a video of the rebuilt model, now electric. There are also some specs in earlier posts of this thread including some advice by Nick Ziroli.

Turns out to be a nice flier even at rather high wing loading and little power, just with more luck. I don't think the original version with the gas engines was underpowered, I think it was still overpowered. I had the same idea as you that the plane bumped into the air. To me it seems the pilot was surprised and overwhelmed by the unexpected, both the bump and the flight behavior. He's said to be an experienced twin-engine pilot but in this case he seems to be in panic as he doesn't realize what happens.

The plane is rather heavy and trimmed rather stable. They say c/g something like 7.5" behind LE what means 20% static margin - overly stable. You'd have to "hold" the plane with up elevator till it reaches high flight speed what is no problem if the model is overpowered at the same time. That's what model fliers are used to, unfortunately. All the good advice, for instance yours, comes from full-scale pilots. They are not necessarily better pilots but sometimes they have learned to know their reflexes and to overcome them if they are wrong. The up-elevator reflex, the pick-up-with-aileron reflex, the forget-rudder reflex, all mentioned by you, are clearly seen in the video. Stop it at a certain moment and view it full-screen and you'll see all these reflexes in one single picture.

Very common choice of the average modeler who believes the more nose heavy the "safer" and that sluggish elevator response is an indication of same.

Try this
IF the setup is at 20% cg =it requires more downforce at the stab for the SAME angle of attack on the wing
if weight is the same
then
lift is the same
but
drag profile (trim drag increase) shifts aft
providing better stability
True/False.

This discussion seems to have sorta died out, but I came into some information today that might be worthwhile.

In a prior post I related that I had talked to a former co-worker who had flown the full sized C-47 / DC-3 and I asked him about stall characteristics. In that discussion I was told that the full sized airplane has very strong buffet which warns the pilot against further angle of attack increase. Well, my friend called another of our former co-workers, who was chief pilot. That gentleman said that he had gone a bit farther into the stall regime and the full scale airplane exhibits a strong roll off when sufficiently provoked.

So there's the bottom line. The DC-3 models do the same thing as their full scale counterparts they drop a wing quickly. The only difference is that the model pilot doesn't have the benefit of feeling buffet which might warn him not to pull the stick back any further.

Dick

Bump.

After nearly one year, Vichineu, a fellow modeler from Belgium, has made a visual/3D model of a DC-3 for the Reflex simulator. That caused me to revisit my virtual DC-3 models which were mere parameter sets with a dummy appearance.

Only recently I found out how to correctly set the wing incidence in the simulator. Now I removed the washout from the DC-3 models and set the wing incidence to 2 degrees geometric (true to original), what is about 4.2 degrees aerodynamic incidence with the NACA 2215 aifoil. Both 1:8 Ziroli versions, that with 43 lbs weight and the other with the ambitious 33 lbs weight, are docile flyers if only some caveats mentioned above are observed. Elevator throw is even reduced to 15 degrees, but still full throw should be used together with full flaps only. With no flaps the nasty stall described above will result.

One good effect of this setup is a scale-like level fuselage (zero deck angle) at cruise speed, which is only 20 m/s or 17 m/s, respectively. Two .91 four-stroke engines are well enough even for climb with full flaps and for a 50% to 60% cruise power. Take-offs look very scale-like, and landings do if done with approach power till full stop. Approach speed must be only slightly less than cruise speed.

That's all nice but not really worth mentioning. I'm writing here again because the 2 / 4.2 degrees geometric / aerodynamic decalage requires the C/G set where Nick Ziroli recommends, that is for a 20.7% static stability margin. I was surprised because I dislike so much stability, but in this case it seems to be just true-to-original, considering the wing incidence and level fuselage. A crash like shown in the video might simply come from too much elevator deflection. After all, 15 degrees seems extremely small as long as one doesn't know that the airplane's AOA range and hence speed range is so small. Even one dead engine isn't that critical knowing that and minding airspeed. Maybe model pilots are just not as smooth on the sticks as full-size pilots.

Now the C/G is quite near to the landing gear so the airplane is easy on the tailwheel when taxying. One issue described was that the model refuses to turn on the ground. To some extent I can copy that in the simulator and full up elevator really helps. But even better could be full down elevator and quite a bit of power, even if not full power. That will lift the tail and the rudder can swing the airplane around. I don't know if that is scale-like.

I failed to offer the simulator models so far. If someone happens to have the Reflex simulator and would like to try the DC-3 models, just PM me and I'll send an installer. The same holds for the file for the QFLR5 program (now called XFLR5).

Thanks again to the posters in this thread for several insights, even if I still don't get the point in Dick Hanson's last post.

No Jim, it is nonsense. There is no such a thing as scale speed.

Yes, I suppose that a model covering its own length in the same time as the full size could be regarded as scale speed, but ONLY in a straight line and in zero wind. How long does that last?

In manoeuvres (turns loops etc) it would look stupidly slow and floaty, and of course it would stop in a wind. That's even before you approach any of the other aerodynamic considerations mentioned.

I wrote a series of 4 articles in RC Model World (June to September 2009) detailing what is wrong with the concept of Scale Speed and how we can reach a practical compromise. It culminated in a Spreadsheet that runs in Excel, and you can download it (and the instructions to use it) from
http://www20.brinkster.com/gvmac/
look on the Alasdair's Aerodynamic page

The Ziroli DC-3 crash... everyone does those!
Several factors....
Inexperienced pilot... evident on the taxi out, with no elevator to assist the tail wheel authority.
After takeoff.. speed too slow.
Gyrations.. aileron induced tip stalls.. the plane goes opposite the control input, as the downgoing aileron stalls the outer wing on that side. The corrections are then extreme, and merely exacerbate the situation.
A slightly forward c.g, and more speed before taking off, and a lot less aileron authority would be nifty.

Scale speed is intriguing. Time is linear in 'our world' so it stands to reason that a 1/4 scale model should fly 1/4 as fast as origional. However, scaling 'volume' and 'area' adhere to different factors. All of this added to Reynold's Numbers poses unreasonable demands to scale fidelity regarding flight characterisitics. This is one of those academic questions, for sure! The 'Moon Illusion' is one easily proved. Fullscale airliners appear to be flying v-e-r-y slowly but we all know that they are moving along at a good clip! The question of scale speed seems never resolved!