Over May and June 2011 I had a chance to think about what was going wrong in previous flights.

Since I suspected the GPS was getting confused by the swinging under the parachute (see my previous post), I decided the best way to debug the navigation system was to manually steer the parachute from the ground, while letting the navigation system run and try to figure out the correlation between servo position and turn rate, and the system lag.

So I built my “universal controller” gadget I described earlier – this has a knob that lets me manually control the servo from the ground.

I also used the time to install an “808” video camera into the electronics bay, looking upward at the parachute.  I hoped that a view of the parachute in flight might be useful in understanding what was going on with the navigation system.

Last, it seemed that my recent flights had all deployed the parachute a bit past apogee (late).  That’s not good because it means the parachute is coming out at a higher-than-necessary airspeed, which leads to much larger deployment forces and possible zippers.  So I modified the apogee detect algorithm to be more sensitive and trigger earlier.

On July 16 I got a chance to try these improvements.  The first flight was my 4″ SuperHorizion on a Skyripper H124 hybrid motor:

Parachute deployment was a bit on the early side.

Again, the parachute did not fully inflate, so I didn’t get a chance to try manual steering, or get any useful navigation experience.  I was lucky – there was no damage at all after the hard landing.

Here is the view from the onboard video camera – I think it’s kind of dramatic:

It’s pretty obvious from the video that the parachute lines are all tangled up, which is why the chute didn’t inflate.  This just verified that I still didn’t have a packing method that resulted in reliable deployment.

I tried again the same day with my 3″ Thrud rocket, again on a Skyripper H124 motor:

This was much worse.  The new apogee detection algorithm was much too sensitive, deploying the parachute too early, while still ascending.  The deployment forces broke the Kevlar cord attaching the tailcone to the rest of the rocket and resulted in a very severe zipper (the body tube was totaled).

The parachute did deploy OK, but again it just spun around in one direction for the whole descent.  I took over with the “universal controller” and seemed to be able to control the rate of turn, but I couldn’t get the parachute to fly straight.

The video from the onboard camera didn’t capture a view of the parachute (it’s just out of the field of view), but the audio track was unexpectedly interesting:

The “cycling” sound on the onboard video is the servo attempting but failing to move to the commanded position.  So part of my problem was that the servo wasn’t even moving when it was supposed to! (This affected only the servo on the Thrud rocket, not the one on SuperHorizion. But still.)

It turned out that I was sending pulses to the servo that were outside the range of what it could properly handle – if I sent a command for a large movement the servo would work, but commands for small gradual movements were ignored.

This wasn’t a good day.

Before the next flight I needed to fix the servo pulse problem, the apogee dection algorithm, and repair the broken Thrud rocket after the zipper and crash of the tailcone.