DESIGNING THE TOP FLITE DOUGLAS DC-3
By Ernest LeClair
The Douglas DC-3. What a great aircraft it was...and is! I've had the pleasures of flying the full size aircraft, and of designing the Top Flite model of that venerable ship. I'm a retired U.S. Air Force pilot, and have flown many of the great aircraft of our time, but the DC-3 has an aura like no other.
There are many stories and memories associated with this airplane, and it's still creating memories more than fifty years after it first flew. I still vividly remember a flight in 1991, as we were taking off from Ft. Meyers in A Boeing 727. I was flying for Eastern Airlines then, and it was my leg. Below us was a formation of five DC-3's spraying for mosquitoes. They were in a banking left turn, and the urge to join them was almost overpowering. What a sight!
There are many stories I could relate concerning my experiences with the DC-3, but this article is about how the Top Flite kit was designed, and how it came to be.
Top Flite decided to produce a model of the DC-3 at about the same time I joined the company as a model designer. Nothing happens quickly at Top Flite, so the DC-3 project didn't reach the top of the project list until after I completed my first design, the Beechcraft Bonanza. The Bonanza was a tough first project, and the DC-3 looked even harder. As I was out of things to do, though the R&D Manager wanted to know if I felt comfortable enough in my new job to handle the DC-3 project.
What can I say? I lied. "Sure I do", I said. "Piece o' cake". (I'm a fighter pilot after all. I can do anything!) Little did anyone realize just how difficult the project would be, or how many problems we would encounter along the way.
It was decided that this would be a Gold Edition kit for .40 size engines. This meant that the model would be big, in the 80-100" wing span range. To be IMAA legal it needed to be at least 80", and for shipping purposes, we needed to keep it in our standard size box. Another consideration was transportability, because anything bigger than 85" become difficult for the average modeler to transport as a one-piece wing.
Once those decisions were made, the project then belonged to the designer. He is responsible for virtually all aspects of the project except marketing. This was an important and costly project for the company, and it was now mine. One way or another, I was going to make a lasting impression. Chuck Yeager's pre-flight prayer came to mind, "Lord, just don't let me screw up".
From the very start I felt that a good design would have to be not only strong, but light as well. This is no easy task for a scale twin-engine model! I used several techniques to keep the aircraft light, to include removing as much as possible from the fuselage. This meant that the fuselage structure could be designed not for strength, but lightness. In the wing I spread the components out across the wing as much as possible. This distributed the load more evenly across the wing, allowing a lighter structure. I would have to carefully analyze each and every part that went into this model! The end result was an impressively large twin-engine model that weighed only 10 lbs., with flaps and retracts.
After documentation from several sources was obtained, I began by scanning several views into the computer. We use auto CAD, no exactly the easiest drafting program to learn, but I was quite comfortable with it by this time. After the scanning, I began several days of tracing over the scanned images. That's when I noticed that the three sets of documentation we had didn't come very close to each other. This was just what I needed! Now I had to guess at which documentation package was the most accurate! Then I remembered that a local aviation museum in Rantoul, IL, had a USAF C-47 in its collection. Living in Rantoul, it was easy to drop by the museum in what turned out to be the first of many visits. With note pad and camera in hand, I spent many ours measuring and photographing the aircraft, while more memories began flooding back into my mind.
With accurate scale outlines in the computer, and piles of pictures in hand, it was time to design the fundamental structure for the model. This is not as easy as it may seem. A good, clean, structural design at this stage of the project is important to keeping it on the time line. Screw up now and there will be many weeks wasted in correcting the design. When I first started this job, I thought my engineering degree would be important. It is, but not as much as I'd expected. What really helps here is experience. I have years of experience in modeling (well, okay, decades if you must know) and that experience is really important! The TLAR theory (That Looks About Right) really does work. And you know that old saying..."If it doesn't look right, it probably isn't". Even if it were right, who'd want it if it looked wrong?
So, where to start? The wing sound fairly easy. Besides, I hate designing fuselages. Yeah, I'll start with the wing. Let's see, big straight center section with split flaps and engine nacelles, and outboard panels with a substantial amount of sweep on the leading edge. I'm going to need some creative engineering to join the spars, especially with that dihedral. And those split flaps...joining them to the center section should be fun! And the engines...and fuel tanks. Gee, where are we going to put those? Landing gear? They're right in the middle of those engine nacelles, and right where the tanks belong! Retractable landing gear too, no less! We're already running into problems here. Maybe I should call Robart and see if they have any clever ideas about retracts.
Bob Walker is a very creative person, and yes, he has some ideas. He'll get back to me in a few days to allow the creative juices to digest what is shaping up to be a vry difficult wing design.
The fuselage? Well, the fuselage looks pretty easy now! Let's see, it's a round, cigar-like fuselage for the most part. Looks like some formers, stringers, and balsa sheeting should work well. But then I notices that, "for the most part" line. There are some pretty unique curves where the wing joins the fuselage, and a big fairing! And the bottom of the fuselage, aft of the wing, is flat!
The location of the leading and trailing edges of the wing define the placement of two formers. Let's start at the leading edge.
Our first line was the location of the former at the leading edge of the wing. It'll be lite ply, 1/8" thick. Our second line gets drawn. There will be dowels in the leading edge of the wing to hold it in position. More lines for the doubler on this former. Ah, we're smokin' now!
Let's lay down the formers up to the nose of the aircraft. For that we need to know the distance from the LE to the nose, and for that we need to know the scale of the model, and for that we need to know the exact wing span. How'd we get back to the wing already?
It was time for a decision. With a full size span of 95', and 80" model would be 1/14.25 scale, not a very round number. Scale modelers deserve a good round number! An 81" span would be a scale of 1/14.07. That's better! But how about 81.5"? That's an easy number for a modeler to remember, and that means a scale of 1/13.99. That's it! We have it! The span is 81.5", and the scale is 14:1. (Later this will change to 82.5", but that's another story)
It's now time to scale the drawings in AutoCAD to the actual dimensions of the model. With that done, it's easy to find the distance from the LE to the nose. Now we can determine where the formers forward of the LE will go, taking into consideration the curve of the nose section which will require those formers to be spaced closer together. The location of the formers aft of the wing can then
be drawn. And pretty soon, it was beginning to look like a fuselage!
Then it was time to figure out how to mount the horizontal stab and the wing. Good designers don't re-invent the wheel whenever a new car is designed. Checking our past designs, it looks like the tail structure for the Cessna 182 is very similar to the DC-3. And the wing mounting system of the P-47 looks like it can be easily adapted as well.
Now for the fun... laying out the actual parts.
See, I figure that is I say it enough, maybe I'll start to believe it, because that's exactly what I'll be doing for the next couple of days. The traced formers from the documentation are in the computer, but they seldom match the former locations in the model. Measuring, calculating, and blending the formers can often take several hours for just one former. This job would be easier on my old drafting board! After the parts are laid out, the notches for the stringers are added, as well as notches to interlock the parts together. Now this was a lot easier to do on the computer than a drafting table.
Before we can build the fuselage we'll need the stab. Nothing special here, as it looks exactly like the stab on many of our warbirds. Of course the DC-3 can be considered a warbird! It's flown in more wars than any other aircraft! I did make a few changes, though. The sheeting is run full span rather than being joined at the center rib. It's slightly harder to install, but this will greatly increase the strength of the stab.
The preliminary fuselage and stab plans, as well as the parts drawing, were turned over to one of our model builders. He would build the model and let me know of any mis-fits, as well as make suggestions on how to improve the design. The scale fin and rudder was the only area of the fuselage he had any real problems with, and he had many suggestions for improving it.
Designing the unusual rudder hinge line of the DC-3 into a model proved to be quite difficult. Building it isn't easy, either. I like the way he phrased it in the instruction manual: "If you ever had aspirations of being a 'neat and tidy' builder, now is the time to exercise those thoughts..."
Now it was time to go back to the wing. The design is actually fairly conventional, except for the dihedral joints of the swept outer panels, and the need to fit seven servos, two engines, two fuel tanks, retractable landing hear, air tank, receiver, and battery! It was immediately apparent that the fuel tanks would not fit into scale appearing engine nacelles. The landing hear mechanism and wheel retract into the nacelle just aft of the engines on the DC-3, leaving no room for fuel tanks. Even without the retracts there is only enough room for 6 oz. tanks!
So I designed an airfoil shaped fuel tank that could be mounted in the leading edge of the wing. But this presented a few problems. It would make the structural design difficult, complicate the location of other components, require fairly long fuel lines, and have a capacity of about 9 oz. I decided to test the idea of using fuel tanks mounted in the center section of the wing. Ground testing on engine test stands was encouraging, and eventually we found this location more than satisfactory.
In order to get the strongest structure possible with the least amount of weight, I decided to use an-I beam spar web design. Several years ago I did some testing on spar design, and found that shear webs with the grain horizontal (span wise) was 40% stronger than webs with the grain vertical! I had studies I-beams in college, and my testing confirmed what we learned. This type of design requires some rather expensive die-cutting tooling, but in this case, the need for strength justifies the cost. This design uses a 1/8" spar web, notched for the ribs, between basswood spars. The web is designed so that the glue joint of the ribs with the web is in compression during positive G loads. This results in a very strong spar.
As time progressed, the design considerations become even more complicated. The shape and location of the ailerons made flutter a possibility. To minimize this, I added more structure than lightness would dictate, in order to stiffen them as much as possible. I also located the servos as close to the aerodynamic center of the ailerons as possible, which was in an area with minimal space! Time consuming complications just seemed to appear at almost every area of the wing design.
The flaps were another area that required a lot of time. I have never liked the design of the flaps on past Top Flite Warbirds, and now I had a chance to improve the design. When I was done, though, I didn't like my design either! But at least my design is easier to build. As it turns out, split flaps just aren't easy to design or build, period. One the DC-3, they are easier to build, they work quite well, and they are very effective. Please take the time to build them!
The engine nacelles went together easily. They are simple boxlike structures that are notched to fit onto the spars. They incorporate down thrust and side thrust, and are really easy to build. Formers are added on the top and bottom to get the nice round shape.
I now had samples of the Robart retracts, and they are nice, quite scale-like, and very light. I was very pleased with them. The air cylinders and pushrods conflict with the spar web, however, complicating the installation a bit. This requires some plywood doublers on the spars to compensate for cutting holes in the sheer webs. The completed nacelles sure look nice, though, and the retracts are so slick, you just have to sit and admire them for a while.
Some of you won't like how I decided to mount the ABS engine cowls. The nacelles looked so nice, I didn't want to lessen the effect with external mounting screws, so I designed some rings positioned with standoff blocks. Internally concealed screws are then screwed into the standoff blocks. It's not all that hard to build and it really looks great.
FLIGHT TESTING THE DC-3
I should mention that the builder of the first prototype DC-3 is known for his museum quality work...truly works of art! It took a long time, but we finally had the most perfect prototype model ever constructed. It looked so nice, we hated to fly it!
It was now January in Illinois. This is such a fun time to do flight testing, especially a twin with two new engines. After waiting several days, the wind finally dies down to only 15 mph, and the temperature was in the 20's. Oh, just PERFECT! We take our museum quality prototype to the field, and despite the freezing temps, we are pleasantly surprised to find most of the snow had blown away. I've had many modelers tell me how much fun my job must be, getting paid to play with models all day, but I didn't see any of them at the field that day!!
With our best guesses at control throws and balance point, and with O.S. FS-52 4C engines dialed in, we were ready for the moment of
truth. Months of work were now resting on our best pilot, known for his excellent reflexes and skills. No, it wasn't me. And I'm glad it wasn't!
The power was advanced slowly, and then, after a roll of just over 10', and with less than half power, the model was airborne! YAHOO! It's flying, It's a handful at first, but it's flying!
This first flight was intended to check control response, throttle response, general handling, and slow flight characteristics. The model was very sensitive to elevator control, and was definitely overpowered with the FS-52 engines. The landing was uneventful. (Thank goodness!)
For the second flight we decided to install O.S. FS-40 4C engines. The balance point was adjusted from 26% to 30%, which is part of the normal process of exploring the balance range. And once again, the little FS-40 engines were more than enough power. But this flight showed that there were a number of trim problems with the model. Power changes resulted in strong pitch changes, due to the placement of the engines well below the centerline of the model. As testing progressed, we would find that 6 degree of down thrust on the engines would minimize these pitch changes. But on this flight, we almost lost the model when power was rapidly advanced on a low approach and go-around. With full power being applied at a low airspeed, the model pitched up so rapidly, it almost stalled. And it did do two of the most spectacular snap rolls I've ever seen that low to the ground! Our very capable test pilot managed to recover with only about 10' of altitude with which to work.
Before the next (third) test flight, we decreased the elevator throw in an attempt to minimize the tendency to snap roll. We also increased down thrust to 8 degree. On this flight we tried some "engine out" flying, by retarding one engine to idle. This resulted in a snap roll after a few seconds, and it didn't seem to matter which engine we pulled back! On landing, we ran out of elevator.
We were becoming very discouraged. With the elevator throw adjusted for good control on landing, we would encounter violent snap rolls at high angles of attack. For the next five flights, we tried adjusting the CG location and elevator throw. Nothing seemed to help. Then, on the ninth flight, the battery failed. Our beautiful, museum quality, prototype crashed into a tree at full throttle. The biggest piece left was a part of one of the engines! Test flying would not resume until the second prototype was ready...and the weather improved. Test flying in 10 to 20 mph winds with temperatures in the low twenties in not fun.
To reduce the tp stalling problems we decided to add 2 degree of washout, increase the tip cord, and blend the tip airfoil to one with better high angle of attack characteristics. On this model we installed O.S. .25 FP engines. The FS-40 engines were more than enough power,so we wanted to see how well it would fly on even smaller engines.
One the first flight with the new prototype, the weather was much better. The temperature was 55 degree and the wind was only 5-10 mph. This model weighed 8-3/4 lbs., which is very light for a model with a span of 82 1/2". Still, that's a lot of weight for two .25's to fly. Power was added and after a roll of 15', the model was airborne with only 3/4 throttle. So much for .25's not being enough! We were very pleased with how well the aircraft performed with these engines, at least for a while. The aircraft was still very pitch sensitive. We lost an engine about half was into the flight, so the other engine was pulled to idle, and the landing was uneventful.
Before the next flight, the elevator throw was reduced again. During this flight we wanted to look at the stall characteristics, and watch for any snap roll problems. This airplane was much better. Stalls were very clean, and we had to force it to snap. At the low elevator rate, the model would not snap at all. We had to use a high rate which was set far higher than we would eventually recommend. It looks like we have the snap roll problem fixed, so it's time to see what happens' with one engine at idle.
Were we pleased or what?? The model flew very well with one engine in idle and the other at full power. It would gradually slow down, yaw into the dead engine, then stall, and then roll over. But we're making good progress! We lost an engine on this flight too, but this time the landing was very eventful, as we again ran out of elevator on landing.
For the next flight we installed O.S. FS-26 engines. These are great little engines and we are sure they will be adequate for this rather large model. And the DC-3 does fly just fine on them, but we can see that this will be the least powerful engines that we will recommend. We continue to fly with these engines for several more flights while we try to find a CG and elevator throw rate that works well. But we can't, and we are stumped. We still need high rate for landing, which is far too sensitive for normal flying.
By now we have accomplished several actual engine out landings, and numerous flights where we continue testing engine out characteristics (with one engine at idle). It handles an engine out well, but with the smaller engines, it slows quickly, yaws, stalls, and then rolls over. But it is gentle enough that we feel the average modeler will have no problems, as long as he follows our recommendations on not getting too slow with an engine out.
But we're still stumped by the elevator throw problem. Our boss, the manager of R&D, suggests that the stab area be increased by at least 10%. My initial reaction is... "Yeah, sure. Just what will that do?" But he is quite insistent, and explains his reasons in some detail. So I start taping some 1" wide strips of balsa to the leading edge of the stab and to the tips. At this time we also decided to try more powerful engines, so I install two O.S. .40FP's. The next flight will be test flight #20 on this aircraft.
And it was a beauty! The model was much less pitch sensitive, even on high rate! And during landings, low rate was adequate. It's a completely different model! We now feel confident that we have a winner. It will take several more flights to arrive at the balance range and throw rates that work just right, but in the end, the model wows us all. By the end of the test flight program, three other pilots will have flown the model as well.
With the model now properly trimmed and flying well, we start some serious engine-out testing. With the .40 FP's now on the aircraft, the test pilot pulls one to idle, and we immediately see...nothing. If he hadn't told us, we wouldn't know it was flying on one engine! The model would yaw slightly into the dead engine, and it slowed somewhat, but it would continue to fly with good authority. We flew the rest of the flight with one engine at idle and the other at full power. Nice!
The model was now flying so well that I had to rewrite the flight recommendation's section of the manual! I was originally going to recommend that rudder be applied into the good engine if an engine failed, to control yaw. This is absolutely mandatory on full scale aircraft. But on full scale aircraft, it is easy to determine which engine has failed. I knew that many modelers would crash their DC-3 if they applied the wrong rudder, something that would be easy to do. But the model flew so well without using extra rudder, I decided to caution the modeler against using it! You can read all about it in the manual.
I'm sure there will be many modelers who will eventually crash their Top Flite DC-3, just as they will eventually crash other models
as well. But I'm equally sure that they won't...if they faithfully follow our flying recommendations. Pilots of any multi-engine aircraft, full scale or model, must always remember that there is an engine out minimum airspeed, below which they must never fly.
Before the testing program concluded, two more DC-3's were built. One was used for construction photos in the instruction manual,
and would later be finished for box-top photography, and as a show model. The other was built from a sample production kit by a modeler outside the company. His suggestions were incorporated into the final production kit. His model was finished and test flown as well.
So, what is the Top Flite Douglas DC-3? It is not a twin engine trainer. The Hobbico Twinstar is about the best model for that (which is something I recommend in the DC-3 manual). But the DC-3 is a great introduction into scale twin engine modeling, which is something we stress on the box and in the advertising.
Oh, yeah, before I forget...Yes, Dick Pettit, it does do impressive rolling circles and thrilling snap rolls. But please, don't try these at home. As part of our testing process, we test the structural integrity of the model. With all of the test flying done, we did all the aerobatics the model was capable of, and even a few high-speed abrupt pull-ups. It even survived a spin or two. But the structure is not designed for this, and it will fail of done often!
I fly my DC-3 just as Dick mentioned...scale-like, with just an occasional roll or loop. This model is impressive just doing a low, slow approach...very impressive!!
Incidentally, one of my winter projects is a contest quality DC-3 that will be fitted with Cobalt 40 electric motors on two 14-cell batteries. But that's a story for another time.
Reprinted with permission.
February, 1999 R/C Report
Editor: Gordon Banks