Why does my flapping wing plane lose height after a couple of minutes?

The symptoms you describe point away from a simple power issue and more towards dynamic efficiency or a change in the motor's load profile once the system is fully engaged. That gradual descent suggests the power output is dropping faster than the required thrust at higher flap rates, but the mechanism sounds consistent. When an ornithopter transitions from climbing to sinking, it usually means the power-to-weight ratio is being overwhelmed, but since you have full throttle, the primary culprit is likely a mechanical or aerodynamic breakdown rather than just battery sag. Flapping mechanisms, especially with nylon gears, are prone to backlash. If the drive system has slack, the motor isn't delivering continuous, consistent torque to the wing at all times, but only when the linkage catches the drive point. As the structure vibrates and expands during flight, that backlash can grow, especially if the linkage mounting points are stressed.

I'd start by checking the linkage slop rigorously. Try placing a thin piece of paper or card between a key contact point on the flap mechanism and the adjacent part. If that paper slips even slightly during the high-frequency movement, you have slop that is compounding the power loss. You might also need to check the motor brushes or the ESC setup, specifically the voltage curve under load. Even if the battery looks fine, if the ESC is throttling too hard to manage the instantaneous current draw of the flappers, it could be limiting the available torque.

Also, have you considered the dynamic change in wing stiffness? As the carbon spar heats up and the fabric stretches slightly, the wing's aerodynamic profile might be changing in a way that increases induced drag disproportionately. Look into the motor magnets and winding quality. If the brushes are starting to wear down, the output efficiency will drop off dramatically after what is likely a short warm-up period, even if the sound of the motor itself seems fine. Check the torque output specification against the load of your full-flap cycle. This is what will tell you if the power supply is even adequate for the full performance envelope.

You mentioned the mechanism sounds the same at minute one and minute five, wich rules out obvious mechanical binding. My experiance with high-frequency actuators is that sometimes the issue isn't wear, but timing. In one project I handled, we found that if the motor peak torque was delayed by even 1.2 milliseconds from the flap's primary stroke, the whole system choked, no matter how powerful the setup was. Check the phase alignment of your drive train. It's a small thing, but that timing can make all the difference in sustain.

It sounds like the motor delay thing is super complicated. I ussually just assume that if it sounds right, it is right, but I guess that is not always true for a flapper. Someone talked about checking the connection where the flap hits the frame. I checked all the screws on my setup alredy, and they all seem tight enough. If the problem is that slack in the mechanism, is there a simple way to see where that wiggle is happening without taking apart the whole thing?

Update: swapped in a fresh pack and watched the voltage during a flight, and it was sagging way harder than I thougth. With a higher-drain pack the sinking basially stops for most of the flight now. Still fades a bit at the end though. That timing idea from the reply about the motor torque peaking late, how woud I even check for that on a gear drive like this?