This is the first real test of model rocket retro-propulsion! At 50m AGL, Signal(the flight computer) commanded release and began computing the optimal retro-burn altitude to "land" the rocket at 14m AGL. Signal missed its landing target by a whopping factor of 2, landing instead at 30m AGL. There are several factors involved, but the most likely candidate is that the pre-flight simulations used to generate the landing math were out of date. Now with real flight data, we can dial in landing simulations more reliably. As with SpaceX and Blue Origin, getting this well-tuned will take several tries.
Stability was quite poor during this test, I believe this was caused by excessive roll on the vehicle. At this time, the current theory for roll accumulation is abnormal motor nozzle geometry. Post-flight motor inspection confirms this as well, and the flight data shows a steady increase in Z axis roll as the burn progresses. This is generally rare, and I've only noticed it occasionally with these motors. Aerodynamic sources seem unlikely as roll continues to accumulate at low airspeeds. The TVC servos have been characterized to handle roughly 180 degrees per second of roll before they can no longer keep up, and this holds true in the flight data as well.
More tests like this will certainly be performed, but before that we need to review a lot of data!
For more info: