Robin

The Farman is sort-of a strange setup aerodynamically. I have some experience with a 3-surface (non-scale) model design, and aerodynamically it is quite interesting. Being a grad aero-student I have the advantage of having access to a lot of aero-software, which helped me to calculate all the correct incidences and most importantly, finding the correct CG and even then it took me a lot of work to get it all right. However, the fact that you have no problems during take-off tells me that you are pretty close. Your model is probably somewhere very close to neutrally stable, and your incidences are probably also not too far off.

The behaviour you describe sounds like a classic dynamic instability. It is probably there during take-off also, but since you are flying slow, you are able to correct it (maybe even unconciously), until the speed goes up and eventually the oscillation starts running away from you. Something that probably also helps initially is that all the horizontal stabilizer area (front and back) cause a lot of aerodynamic damping, which means that although it is unstable, the damping prevents the oscillation from running away immediately.

I recently co-authored a paper for an AIAA conference dealing with computer simulations of the 1903 and 1905 Wright Flyers, and these have behavior very similar to what you describe. Both of those airplanes are unstable (CG behind the neutral points), but both are flyable due to (a) the large amount of damping and (b) the low flight speed. The 1905 flyer however was much easier to fly than the 1903 Flyer because it used a bigger canard that increased the aerodynamic damping to the point where the pilot could easily control the oscilation himself. If he stopped countering the oscilation with canard inputs it would still run away from him though. The way we typically "crash" the Flyers on the research sim is when we let the oscillation run away, and eventually when reaching the top of an oscillation the canard stalls. The canard stall causes a severe nose pitch-down which is unrecoverable as you never have enough altitude to wait for it to start flying again.

OK, that is what I "think" is wrong with your airplane. Of course, it is just a guess based on your description of how it flies. Remember that, since there is a canard, the neutral point of the airplane moves forward from where you normally expect it to be. Most conventional airplanes balance with nice positive stability around 25% to 35% chord (which means the neutral point is at around 35% to 45% chord) - these numbers all depend on the specific airplane of course. However, "normal" canards (not 3-surface airplanes) have to be balanced much further forward due to the canard pulling the neutral point forward. In some cases, when the canards are very big, the correct balance point may even be in front of the leading edge of the wing. Your model has both a conventional tail and a canard. The wing(s) aerodynamic center is at 25% chord. The tail moves the neutral point further back, and the canard moves it further forward again. So, your airplane probably has a neutral point around 25% to 30% chord. Your CG has to be in front of that point or it will be unstable. As you can see, both CG locations you have tried are behind that point...

I suggest you try to move the CG somewhere between approx 18% to 23% chord. I am not sure if you need to adjust your horizontal tail incidence, since you have never actually flown the model in a trimmed condition and therefore you don't really know if it is bad now or not. With the CG where I suggest, you may need to change the incidence on the horizontal slightly nose-down. If it was me though, I would simply move the CG to around 20% and then try to fly it. If you need a lot of elevator to get it off the ground, then you will probably have to adjust the rear horizontal slightly nose-down or you can change the canard slightly nose-up.

I hope this helps. As I said, without actually seeing the model and going through the calculations, I cannot guarantee anything. However, moving the CG too far forward is generally much safer than having it too far to the rear. Also, because of what you observed in regard to the strong pitch-down once the canard stalls, I would suggest that you try never to stall it without having lots of altitude to work with.

Good luck!

Cheers,
Bennie