Okay, so, this thing has been running silent for the better part of a year. Might as well walk you through what has happened, and where we are (hint: more progress than the zero updates here would lead you to believe).  I've been working on several iterations of a base, and have actually cooked up something functional!

Box Designs

So, the first thing I really took a crack at was designing a physical enclosure for the base of the collective. One requirement I settled on that was not part of the original project spec was the ability to clamp it to the underside of a desk, since I have this inexpensive laptop desk from Amazon functioning as a left-side extension to my desk, and holding my TM Warthog throttle too. That particular desk is not extremely stable, which may become a problem later, but it does handily solve the "how do I get this thing in approximately the appropriate place relative to my desk chair" issue.

Design 1

Box #1

My first design was, in an attempt to avoid machining operations, 100% 3d printed. Rigidity was the big challenge, as I needed to hold the fulcrum of the collective pivot stable, as well as the mount point for the gas strut. First design used a rectangular box, with the fulcrum pivot through the bottom right and the gas strut pivot through the upper left. In hindsight, this was a gloriously wrong-headed design. When at minimum (close to the horizontal), the gas strut was parallel to the collective; when at maximum (roughly 45 degrees), the strut was more or less perpendicular. 

(I should really insert an image here but I was lax in taking screenshots).

Box #2

The next iteration was smarter. The big insight was a position for the gas damper that kept its axis of motion roughly perpendicular to the motion of the lever arm throughout the range of motion. This moved the pivot point for the strut from above the fulcrum to below and forward.

The other insight was that 3d printing is really horrible for structure, and I'd rather vacuum aluminum shavings off the floor of my flat than have a wobbly, crappy system. I purchased a 1"x1/2" x 1/8" wall aluminum rectangular tube from McMaster, and went through a few iterations of boring out holes in the appropriate places with a step drillbit and a cordless electric drill. Turns out hand drills are awful at precision work (who knew) but since I don't have access to a drillpress or mill, here we are. 

The rectangular tube is sandwiched between two 3d printed plates in the housing. I'm still not 100% satisfied with the rigidity of this setup, but it's vastly better in every way than the prior attempt.

Sensors

I knew I wanted contactless sensors for all the rotary axes in the box, since I've replaced the potentiometers in my old Thrustmaster Cougar enough times to be over that particular task. 

My first thought was a TE Connectivity hall effect pot. However, in my infinite forethought I neglected to check operating voltages, still being in the mind of pots as voltage dividers. Turns out the sensors I purchased required 5v in, and my Teensy setup is 3v3.  My initial fix used a voltage divider with an opamp buffer for impedance matching, but I couldn't smooth out some periodic noise in the signal. I need to revisit this problem again, and test out these sensors with a 5v ADC (probably an Arduino Pro Micro 5V) to try and isolate the noise I was experiencing. Analog design is challenging, especially if you're an armchair EE at best.

Next option was an I2C rotary sensor from AMS, the AMS AS5048B, which is available in an eval board. There's an Arduino library available that's compatible with Teensy, so that made my choice there easy.  The part has 14-bit resolution over a 360 degree range of motion. Teensy's HID joystick library allows 10-bit precision, so with the magnet directly coupled to my fulcrum's shaft and an approximately 45 degree range of motion, I've still got about one more bit of precision available than I can use (and that gets into noise real fast). I'd initially worried that I might need to use gears to convert the range of rotation around the fulcrum to a range that's compatible with my sensor, but I suspect any gearing would introduce more slop and backlash than I've got now in sensor noise.

Downside, I've got a 5v multi-turn hall effect pot from Bourns that I'd wanted to use for the twist throttle, but I still need a 5v ADC plan for that.

Button Boxes

One of my design goals is to make the control box on the end of the collective swappable, to support different simulated helicopters' switchology. I'm primarily building this out for Digital Combat Simulator (DCS), with XPlane11 as a distant second, so that means the Ka-50, Mi-8, UH-1, and SA-342. The design I'm working on uses a combination of a keyed fitting and a threaded joint to physically connect the box to the collective shaft. Electrically, I'm implementing I2C slave on an Arduino Pro Mini 3.3v clone. I chose I2C over using a pile of 74HC165 parallel-in shift registers directly so that I could support analog hats and/or indicator lights (or really, anything that's not digital switches). Also, I've written I2C master code plenty of times but never implemented both sides, so that's of educational value.