how to fine refine a design?

[Bob Chiang]

Hi all,

I just completed a new design and made the first flights at noon today. Attached is a photo of the model, and the specifications and a couple of construction photos are on this thread: http://www.rcgroups.com/forums/showthread.php?s=&threadid=88700

It seems fast considering the little Johnson 4046 motor for power. In general I'm pleased with the performance, but think it could be made better. Does anyone have a test procedure for optimizing the performance?

It feels kind of twitchy, maybe showing a slight tendency to tuck over and dive in certain conditions. I think the cg and incidence are reasonable because it glides pretty well with the elevator in line with the stab and it is quite responsive to elevator control. I tried to build it with about 3 degrees difference between the wing and horizontal stab.

I think it needs down thrust because it tends to pull up almost into a stall with full power (though it's capable of climbing at quite a steep angle).

It's not very responsive to rudder: I'll try increasing the rudder area, and am thinking of adding upswept wing tips (they are currently square).

Thanks for any suggestions of things to try, observe, and change.

[Sail 'n Soar]

Rudder, elevator & throttle? Test 1, when you give a rudder command, does the plane yaw in the opposite direction, but still keep flying relatively straight, or at most in slight turn inconsistent with what the fuselage angle would indicate? If so, the problem is insufficient dihedral. The upturned tips might help. If not, you might be flying a little too slowly so that the rudder is ineffective. This will also reduce the effectiveness of the dihedral you have. If you are close to CL,max/flying too slowly, adding a bit of yaw will result in an increase in drag on the forward/outside radius wing, counteracting the rudder side load, with little increase in lift on the forward/outside radius wing. i.e., the plane continues to go basically straight!

It also sounds like your CG is too far aft. With reasonable control surface areas a CG between 25% and 30% of the mean aero cord is a good safe bet.

Finally, down-thrust is definitely in order. Are you at ~0 now?

[T. Lyttle]

The photo says it all: not enough dihedral. Either increase the existing dihedral, or add 20deg or so to the last 2 or 3 bays on each wing.

My test for thrustline: advancing the throttle should result in minimal attitude change. If the aircraft rears up, more downthrust is required, particularly on that model (not much difference between motor line and wing line); it won't take much. If the aircraft pulls to the left under full power, more right thrust is required.

The fact that it flew first time, and survived, shows you have been doing your homework; good on ya! :D

[BMatthews]

That's a fine looking model with some nice workmanship. And from the background in the pics from the other thread this isn't your first trip down the scratch built road. Good to see folks that still are willing to accept the challenge of designing and scratch building.

If it has a tendency to tuck by itself and you're sure there isn't anything glitching or flexing then the CG is probably a bit too far back. You could follow S&S's suggestion or try using the dive test prodedure for setting the balance and resulting elevator trim.

But the fact that it noses up so strongly with power suggests to me that you are not that close to neutral stibility as the tucks suggest. The strong nose up is a sign of a forward'ish CG and the extra speed reacts with the elevator trim or decalage trim needed to balance the forward CG and up comes the nose. So it's back to a possible glitch or something flexing that shouldn't. But even with everything set right you'll still need 2 to 4 degrees to kill the tendency to nose up with power.

But try the dive test to be sure. Just leave yourself two mistakes worth of altitude when doing this incase it tries to tuck and you need some room to pull out.

As for the turning I'd say the rudder could be a little bigger since these are very slow flying types but the real culprit is the dihedral, or rather lack of dihedral. Could you not fudge some into that clever wing mount system? That's a very full size glider style of mounting by the way. A suggestion to prevent air leaks that lead to poor center section performance is to add some felt or 1/16 sponge rubber to the wing roots to act as a seal for the wing to fuselage joint.

Other than these little things this is a great effort. It's always a good feeling to see your own design flying. But its even more valuable when you can pull yourself back to reality and study the lessons it provides so the next can be even better. The master craftsmen that you admire are universally their own worst critics. If they weren't then they wouldn't have gotten so good in the first place.

[Bob Chiang]

Thanks for the comments and suggestions. It is rudder, elevator & throttle. This is my first completed rc design since returning to the hobby a few years ago (designed a rudder-only as a youngster). And I am pretty pleased that it survived not only a first flight (including one landing), but was fun enough to run a second battery through.

I found part of the problem (even before reading BMatthews' diagnosis): the elevator pushrod (balsa stick with wire ends) does not act in a straight line, and is able to flex. I should be able to add guides to limit that problem.

The motor is installed with down and right thrust, only the motor knows how much. I didn't notice a tendency to pull left with power, so I think it just needs more down.

I am embarassed to admit that I haven't actually measured where the cg is, just that it looked about right and I started with the battery as far forward in the fuselage bay as I could. I'll check that. I deliberately built it with 3 degrees between the wing and stab, maybe that's more appropriate for a free flight model and is too much for rc? I could adjust it if necessary.

Dihedral is 3 degrees per side. I didn't notice it yawing in the opposite direction, but I was probably too busy making sure it didn't crash to do any critical observation : ) I'd rather not mess with the wing joiner to add dihedral. If I can tell that it yaws properly, what do you think of adding tips of a few inches span at a large angle like 30 or 45 degrees?

Thanks again for your help.
BTW, any suggestions for how to resize a photo to fit the screen better without losing so much resolution? When I resize using MS Photo Editor, they lose their sharpness and are still too big on the screen.

[Karl B²]

Originally posted by Bob Chiang
BTW, any suggestions for how to resize a photo to fit the screen better without losing so much resolution? When I resize using MS Photo Editor, they lose their sharpness and are still too big on the screen.

The issue looks like it could be the amount of jpeg compression, which leaves noticeable artifacts at high detail/low detail boundaries. Try increasing the jpeg quality slider. On MS PhotoEditor, choose "Save As..." then click the "More..." button. I get good results at settings of 80-90%.

PE doesn't show you the loss in detail until you close and reopen the file, and if you make further changes and save again, you might be compounding the jpeg compression losses. Always keep an untouched copy of your original photo/scan, and save changes as new files. Some programs, such as Adobe Photoshop Elements, will preview the new image and file size before you commit to a compression setting, so that you can see what your quality/filesize tradeoff is.

Karl

[Ollie]

You do need to move the CG forward slightly to establish pitch stability so that it doesn't tuck under in a dive. Un fortunately this may degrade turn response. As B Matthews has suggested, more dihedral is the answer to improving turn response in a model without ailerons. If I were you I would double or tripple the dihedral. You really need a lot more dihedral than you have.

The model seems fast because the fasceted airfoil that you are using stalls before it can develop a high enough angle of attack and coefficient of lift to be able to fly slower. It is the lack of smooth curvature between the leading edge and the spar that is the main cause of early stall.

BTW, I think you have built a marvelously light airframe.

[Bob Chiang]

I haven't verified it yet, but I think the tuck problem will be solved by correcting the elevator pushrod slop. So I'll try that before changing the CG.

And I should have been more thorough with some of my descriptions: the model is able to slow down quite nicely with power off. I can land and circle for thermal hunting at a nice glide speed without stalling. But at full throttle, or in a dive, it covers ground pretty quickly (well everything's relative).

I am curious (and will try it unless someone says it definitely won't work) about adding upswept tips. I'm not anxious to rework the central wing joint.

Thanks again for the suggestions, and please keep them coming.

[Sail 'n Soar]

I'd still advise checking the CG position to see if you are in the ball park. If your spar is at a typical ~30% of cord location, you should be able to balance the airplane with two fingers either on or near the wing spar.

[BMatthews]

One option would be to convert this model into a poly design. For this I think I'd look at cutting off the last rib bay and adding a 3 bay extension angled up at 25 to 30 degrees to the existing panel. This would extend the span a little but it looks like you could get away with it. And the extra dihedral would be out at the tips where it has more leverage.

Theres been a few models that just used an upward angled tip plate but these flat plates have the possibility of adding a LOT of drag if they are not done right or even if they are when lots of control is inputed. The yaw may cause the flat plate to stall fairly readily. I don't know this for sure but it's a more likely possibility than the certainty of success with adding a proper tip panel with an airfoil.

But it's your call. Certainly a plate tip sort of like the Graupner Pedro pic'ed below would be easy to add but just watch for bad habits......

http://www.graupner.de/prod_images/elektroflug/4231.jpg

[Bob Chiang]

I did the finger and TLAR CG check before flying. Now I finally measured the CG: it falls at 33%. Unfortunately that's with the battery as far forward as I can put it (there's a former at the front of the bay). When I layed out the equipment, I didn't take the covering into account. The covering on the tail way out on that long moment arm shifted it farther back than I expected. I can move the receiver and esc forward a little bit and will do that before the next flight tests.

And regarding the angled tips, I intend to make short airfoiled panels. I hadn't even thought of flat panels, but thanks for the warning. I'm too lazy to cut the wing up, so I'll try adding the panels to the existing span. I've never seen a model like the Pedro. Does it have any dihedral in the center panels?

Regarding the faceted airfoil: I've seen a fair number of rubber powered models using this kind of wing construction. Does anyone know if the ridges act as turbulators?

[Sail 'n Soar]

33% isn't all that bad, if the tail moment is adequate. Some successful freeflight designs used to balance at the trailing edge, but with VERY large horizontal stabs!

Calculate the following and see what you get: (A,h *l,h)/(A * c) Where A,h is the horizontal stab & elevator area, l,h is the length between the wing quarter cord and the average quarter cord of A,h, A = wing area and c is the mean cord of your wing.

[Bob Chiang]

Hi Sail 'n Soar, is this a calculation for tail volume? I've downloaded some info about that but hadn't gotten around to reading it.

Here's what I have:
horizontal stab plus elevator area ~55 square inches
length between wing & tail 1/4 cords ~ 17.3"
wing area (including fuselage width) ~251 square inches
wing cord ~6.7"

This is the wing before I added the polyhedral tips. Haven't tried flying it with the added tips yet.

comes out to ~0.57
What does that translate to?

[Ollie]

A tail volume coefficient of 0.57 is generous and should allow a more aft CG. I fly gliders with tail volume coefficients between 0.46 and 0.5 with the CG at about 40% of the mean aerodynamic chord. Even allowing for the small destabilizing influence of the propeller you should be able to use a CG of 40 to 42 % of the chord once you get all the slop and deadband is out of the elevator linkage. I suggest that you move the CG back cautiously in small increments over the course of several flights. As you move the CG aft, the controls will become more effective and you may need to reduce elevator throw. Your turning response will improve some.

[Sail 'n Soar]

Bob Chiang

I'm a little more conservative. I design to a tail volume coefficient of ~.5 with a 30-35% CG with reasonable stability and good elevator responsiveness. Of course, I've never flown a faceted foil, so that may make things a little less sensitive to small pitch changes - pure conjecture, no real knowledge on this one.

Gerry

[Bob Chiang]

Interesting. When I was laying out the design, I was shooting for the "stable side of sporty".

Well I made a test flight this morning (snowing lightly, but calm) and it's turning into a slightly different model than I intended, but very nice. I'll have fun flying this one on calm summer evenings.

I fixed the problem with the elevator linkage, and the model does not tuck or feel twitchy anymore.

I added downthrust, but the model still changes pitch with throttle. I don't know if I'll try to add more or just live with it. I am really impressed by the climbing ability on this direct drive $1 motor.

I added tapered and swept back 5" polyhedral tips using the same cracked rib construction. The leading edge has phillips entry to match the main panel, and then drops to a flat section at mid point. The result looks like washout. The turning has improved dramatically, but I think it lost top end speed (or maybe it's just not as scarey to fly at full speed now). It really slows down well now and has changed from a sport model to a glider. Probably better since the whole intent of this model was to have something packable to bring on a vacation. This way I can let my son and his friend fly it.

I'll play with the CG, though it is already very responsive. I haven't tried flying on reduced throws. In general, is it more efficient to use less deflection since that would cause less drag?

Thanks again for all of the comments and suggestions.

[Ollie]

When you increased the wing area with the tip panels you reduced the tail volume coefficient. So that needs to be recalculated and the CG can't be moved so far back without encountering instability.

Reducing control throws will reduce maneuvering drag a little.

It sounds like your design is maturing very nicely.

[Bob Chiang]

Well, it WAS maturing nicely. The electricity went off in my office this morning, so used the opportunity to take a friend out to give him some flying experience. He bought a used .40 sized trainer and has been anxious to start flying.

I flew it to altitude and let him take the controls while talking him through turns and control inputs. We did that several times and I only had to recover from a couple of odd situations. It really has a gentle stall. Then one time it ended up getting low on the other side of some big transmission power lines. I took control and thought I had it climbing back safely but ended up hitting one of the lines!

So one wing is shattered (I tend to build them to fly, not crash) and the nice folding prop blades are broken.

The good news is that he really enjoyed the flying experience and we'll do it again. Maybe I'll use a foamie next time : )