When displayed on Google Earth, the track of N233NN looks almost perfectly straight, so I wrote some code to take the long, straight segment across Georgia and Alabama and compare it to a perfect great-circle route. Viewed this way, there is a very sudden 75m deviation to the right followed by a recovery. This seems like it could be consistent with losing a winglet.
If my observation is correct, the winglet most likely fell off between the labeled point and Millerville which corresponds to the maximum deflection on the graph.
I spent half an hour flying over the area and was unable to locate the winglet.
]]>
Here is the platform in operation. Notice the 3d printed phone holder attached to the eyepiece. I also use a DSLR with a 2x barlow.
The north segments are the only potentially unique part of the design. Inspired by the common Vertical North Sector (VNS) design, these are “bent” elliptically so they contact the drive wheel in the same point for the complete rotation. This was an attempt to prevent the rotation rate from varying with time. I haven’t done the math to see how much this matters and it seems there are other errors that might be more significant in my build. The bottoms of the sectors are the surface of a cone.
The electronics are very simple. I bought a 1:100 planetary gearbox stepper motor from Amazon, the Pololu Tic T834 Stepper Motor Controller, and the Arduino Nano with Bluetooth Low Energy (BLE). The Bluetooth allows me to precisely adjust the velocity command during observation. All of this is powered by 4 AAA batteries. The batteries seem sufficient for over 5 hours of operation. AA batteries would probably be a little better.
For friction, the drive wheel is wrapped in a rubber end cap with the end cut off. The cap is 22mm and I sized the wheel to be 23mm to give it enough friction not to slip off. I also printed a wall at the outside edge of the drive wheel to retain it from getting extruded out. The wheel is also slightly cone shaped because this is actually a conical drive.
The south bearing is a simple pivot angled to point along the rotation axis (towards the north star) of the platform. I’d like to make the platform more square, but that would require a more complex south bearing and I just wanted to see if the idea would even work first.
Pictures taken using the platform
The ignition spark is generated when the current drops sharply to 0. There’s a period of almost zero current lasting 2.5ms while the magnet is still close to the trigger. This period is longer at slower RPMs and I believe it gets shorter at higher RPMs. I wasn’t able to test higher RPMs with my drill setup, but I did move the magnet far enough from the trigger that it was barely triggering and saw periods of ~1ms at the same RPM.
Using this data, I simulated the expected waveform at a range of RPMs to show how the average current draw decreases as RPMs increase. It’s only about a 13% decrease in current and probably not enough to be readily noticeable.
While investigating this, I noticed the triggers de-energize the coils after about 2 seconds without a trigger. This greatly reduces their drain on the battery while the engine isn’t running.
]]>
I suspect the dips until 1 second are due to higher torque during the compression strokes before engine started running. I continued to hold the start button until ~2.5 seconds. You can see some periodic noise that I believe is due to the secondary ignition coils charging.
]]>I wondered if a current transient could be the problem and found an article by Bob Nuckolls that describes how the starter solenoid uses two coils so there’s an initial high current to quickly actuate it and a lower current to hold it closed.
Then the builder mentioned he didn’t have the heavy battery cable connected to the starter and another builder wondered if that might be preventing the initial current transient from decreasing.
I thought it would be interesting to know exactly what is happening, so I measured the current with and without the battery attached using a 75mV per 5A current shunt in series with the start button. Turns out, I had a 10A fuse on that circuit and ended up blowing it just like the original poster when the battery was not attached.
I found the following chart in the datasheet for Littelfuse that agrees with the ~0.15s time to blow the 10A fuse at 34A.
The landing was one of my better recent landings. I used full flaps and crossed the dune at about 60-65mph and used maybe a third of the runway. Philip slept through the entire landing. The ground was slightly damp in places and I had to keep my rpm up around 1500 with occasional bursts to 2000 to maintain a constant speed while turning the airplane around and taxiing to the “ramp”.
Parked in the ramp area. The grass/sand here is typical of the entire runway.
I imagine larger wheels would be desirable for flying regularly to Dog Island. For takeoff, I used 10 degrees of flaps in an attempt to get off the ground and free of the drag from the grass sooner. I imagine my takeoff roll was a little longer than at Moontown or other grass runways, but not by much. I turned and flew down the coastline while gaining altitude for the 3 mile hop back over Saint George Sound.
The runway and the beach area we explored by Pelican Inn.
Philip was bundled up in two layers of clothes and a coat, which he quickly removed to splash in the waves. The sea shells were pretty impressive–I imagine they aren’t already picked-over like the ones on the mainland beaches.
The Pelican Inn looked like it’s probably not currently available for lodging. I believe it was damaged by Hurrican Hermine in 2016. While we were there, a group of official-looking people were walking around the property taking pictures of the foundation–maybe work is under way to fix it up again.
On the return flight, we landed at Weedon Field, a nice little airport in Eufaula, Alabama. I mention it because of the amazing airbrushed murals on the wall.
This was Philip’s longest flight, so we planned to spend a night with my parents near the half-way mark on both the flight out and the return flight. Strong winds were a big part of the entire trip. On the first leg the air was so bumpy we found ourselves at 10,500 feet before the ride smoothed out. The time spent climbing and winds combined to cause the trip to take about 20 minutes longer than normal, but the smooth air seemed worth it.
On the second day (Launch Day), we arrived at the airport around 08:00 and found the forecast tailwinds I had noticed several days in advance. We saw the ocean from the Sonex for the first time. This part of the Florida coast line is pretty remote and doesn’t have the sandy beaches that I’m used to seeing along the panhandle.
We could have made it all the way to Space Coast Regional in a single leg, but not knowing what to expect transitioning the Orlando Class B airspace, we refueled at Marion County/Dunnellon X35.
After departure, I contacted Jacksonville Center to inform them of my intention to fly direct to Space Coast Regional at 3,500 feet. He chuckled when I mentioned being capable of 120 knots and said he’d see what they could do for me. When I switched over to Orlando Approach, they immediately sent me down to 2,000 feet–below their airspace where I could have flown anyway. Oh well. The controller made it feel more pleasant by referring to it as “begin your descent now” despite us both knowing it doesn’t take 60+ miles for me to descend from 3,500 feet :-)
Philip was very excited to get to the beach. I read online that Playalinda beach is a pretty good location to watch launches from. I emphasized the beach part of the trip to him since I was a little worried he might be scared of seeing the actual rocket launch. The water was the perfect temperature. He had a great time splashing and jumping over the waves for the hour or so we waited before the launch (and for an hour after the launch and most people had left).
The beach ended up feeling closer to the launch than I expected. The rocket was clearly recognizable as it lifted off. The most impressive part about watching the launch in person was how brilliantly red the exhaust gases are. On computer screens it tends to look white, but the brightness and sharpness of the red flame are impressive in person. I was able to track the rocket until the first stage separated before I lost track of it.
Shortly before the launch I let Philip know that the launch was imminent so he knew what to expect. He happily watched the launch and seemed to understand what was happening, mentioning that it was going into space.
Anticipating strong winds and turbulence, we planned our departure for early the next morning. I told the control tower that I was planning to fly over to do some sightseeing at the Shuttle Landing Facility. They had me talk to Orlando who assigned me a squawk code and had me switch to the control tower’s frequency that was acting as a CTAF since the tower was closed.
The headwinds on the way back to Enterprise were strong and turned what had been a < 3 hour flight the previous day into a 4 hour one. The first couple hours were pretty smooth, but it started getting bumpy as we descended for our fuel stop. We refueled at Quincy, Florida 2J9 where the runway is an island within a sea of well-maintained grass that is used as a taxiway and a runway. We stayed long enough to watch a couple loads of skydivers land.
The runway in Enterprise is oriented exactly perpendicular to the headwind we fought all morning so I fought one last battle against the wind while Philip slept. The wind caused me to go around for a second try, but the second try resulted in what felt like the best landing of the trip. Philip was happy to spend the rest of the day with the complete attention of his grandparents. The morning’s flight back was met with a slight tailwind, smooth air, and a very light rain shower right at our home airport.
]]>The winds during takeoff were a little gusty but almost directly down the runway. I got to pattern altitude sooner than normal, but around 2000 feet started to have doubts that the forecast headwind was correct. The air smoothed out and around 3000 feet I was easily moving 10mph backwards. I zigzagged a bit side to side to keep an eye on what was happening on the field.
I began my descent north of the runway while still flying backwards and zigzagged intentionally thinking it would allow me to keep making progress backwards while descending into slower air. Once back below about 2600 feet the air was turbulent again and difficult to comfortably keep the airplane close to a stall. By this time I easily had the field made and flew a normal full flap approach at 70mph down to the runway.
Once above 3000 feet, I was able to maintain altitude easily between 2200-2400 rpm. Oil temperature rose as high as 215F, much higher than it normally would in cruise. The highest CHT stabilized around 350F, also higher than I would normally expect.
]]>The simplest solution seems to be cutting a sawhorse a little taller than the bottom of the fuselage and lifting the tail while someone slides it into position. The sawhorse is either positioned at the spar tunnel or as far forward as possible on the fuselage. I typically work on the airplane alone and briefly considered and abandoned a trick method using ropes to position the sawhorse. I have a mental image of somehow getting stuck holding the tail up while the sawhorse is awkwardly positioned to dent my wing.
A popular method on the vansairforce forum involves these hydraulic jacks from harbor freight. Unfortunately, they are 24″ tall and the bottom of the Sonex fuselage is 19″.
Scissor jacks seemed like a good solution, but I kept finding small ones intended for cars. Securely mounting two of them to 2x4s seemed like it might be a challenge and would be dependent on the exact design of the jack. I stumbled across these 24″ tall “leveling scissor jacks” on amazon. They come with all the hardware needed to attach them to 2x4s and mine came with a hand crank rather than the drill attachment they advertised. I much rather have the hand crank. 
