Part 2 of Sharing my Flight Quadrant Project

September 29, 2020

So this, as the title suggests is a continuation of an earlier thread I started I while back on a project which I've linked here:

I'll have to spread this out over 3 posts as this forum only allows 5mb per post and I have 24 photos which I managed to reduce down to about 11.1mb. (Hopefully they'll merge.)

This project took a bit of a hiatus due to lack of motivation and other projects, etc. but now it's all finished and I've also made a brake lever for my modded G940 which is integrated into the same system so I'm finally getting round to sharing the WIP photos and the finished products. I apologise for the lack of WIP photos, I didn't take so many when I was working on it, especially of the brake lever.

So to pick up where we left off:

The Hall sensor mount/axels needed to be remade for better sensor positioning (see my last post on the previous thread):

I started designing and building the brake lever. This is integrated into the flight quadrant system so I could get away with just one microcontroller and it worked out very well. The fixing of the wires to the Hall sensor for the brake lever went through three iterations hence the inconsistencies in the photos.

After buying the big (ON)-OFF-(ON) toggle switches I realised they wouldn't fit in my original design so I made a kind of spacer block and a new front panel. This actually worked to be great as it gave me a block to mount various things such as a support for the bottom lever end-stop on the far left (I wasn't happy with how flimsy the M3 threaded rod inside it was), a place to screw in the cable gland, housing mounting holes and a place to mount the calibration switch.

Back in time here a minute, in March I was visiting home in New Zealand (and Australia - nearly got stranded there due to damn Corona) and working on the programming and electronics with my dad. That is to say I watched him be a genius, writing various bits of code for testing and coming up with an amplifier circuit for the Hall sensor on the brake lever and helped where I could. (I will go into more details about coding and circuitry later on.)

Once back in Germany and after having built most of the brake lever I was able to finish off the testing to find out exactly what resistors I would need for that amplifier circuit. Then I built a housing with I/Os and indicator LEDs for the microcontroller.

Parallel to this I was working on design for the flight quadrant housing. I drew out what I wanted and then sat down with my friend Johannes who made a 3D CAD model of it which I then sent to another friend, Lenard who 3D printed it for me. This worked well so I set about painting it. I used nearly an entire can of primer to level it, then got a can of horrible black lacquer (from an actual online paint store in Germany) that never set and nearly ruined the whole thing and finally I finished it with 2-component, matte clear lacquer. Unfortunately that damn black lacquer has recently started causing the clear coat start crinkling so at some point I'll probably have to either sand it back to the primer or get a new housing printed.

While this was going on I made levers and knobs. I had a couple of scraps of various plastics in various colours which worked out nicely. Blue for Water Radiator/Engine Outlet Cowls. Grey for Oil Radiator. White/cream for Prop RPM (shape inspired by some versions of the Spitfire throttle control). Finally the Mixture knob was off a dead machine at work that got scrapped and I saved the knob for a rainy day. I also changed my initial idea of having the levers move in a 90° arc, it seemed a bit much so rather than raising the stop bar (which would have required a huge amount of work because many parts would have had to be remade) I opted to add spacer-angles to the levers. (There is actually a lot more to this but I'm already going into a lot of detail.)

And that was it for machining parts and painting so then came wiring all the switches and Hall sensors in the quadrant and fitting the housing.
The 3 toggle switches are there to correspond to the axes above them (because using an axis to control the radiators in say the P-47 for example just feels wrong) plus for any aircraft that for use buttons for radiators (such as the Bf-110). Finally the button underneath the mixture control is for engine start/stop.

So if I were to do the mechanical side again there are a couple of things I would do differently, mainly the bearings, they have a little lateral play and this can allow the levers to scrape the sides of the slots in the housing, not a big deal but could have been done better. Other than that, there are a few minor things that could be optimised, I kinda designed and built it at the same time which is obviously not a sound practice but it was way more motivating and worked out in the end. Also I think the friction system could use a little tweak, the springs do basically nothing so there's a fairly narrow margin where the thumbscrews start tighten at the end of their travel that is usable. Having fine threads did help though and it does work okay, just not as nice as I planned. Also, if anyone is wondering why I used a compression spring for the brake lever return instead of a tension spring the answer is it was easier for me to make it this way with spring lengths I had access to than to research what tension springs would work, then order it and hope I got it right, etc.

Coding, etc.

The microcontroller we used was a Teensy++ 2.0 which was nice to work with and worked very well. The only problem I had with it was the analogue resolution which was only 1024 counts and would (theoretically) be the full 180° of rotation of the magnets around the Hall sensor. This wasn't too much of a problem for the 4 axes on the flight quadrant (using roughly 80-85°) but the brake lever was only using maybe 20-25° (and I initially thought it would be less, like 10-15°). This would actually probably be fine for a brake lever as I would argue you don't need a high degree of precision in its position however we had planned for it and made an amplifier circuit so it was a fun little addition that ended up working perfectly.

Another feature I wanted was a calibration switch so if any of the axes drift over time I can correct for it. Plus LEDs to indicate what's going on with calibration. What I came up with for that was a recessed switch underneath the quadrant that, when flipped puts the system in calibration mode and turns on an amber LED (there's a green one for power). Then you either push and hold a switch down or the button below the axis you want to calibrate, the red LED starts blinking, you move the axis to each end of its travel, then let go of the switch and that axis' end points are written to the Teensy's onboard EPROM. To calibrate the brake lever you push and hold the left two switches up. Once you've finished, flip the calibration switch off and you're done. Again, all the real credit goes to my dad who did the coding for this.

At one point I noticed I occasionally bumped the brake lever while flying, etc. which causes a pneumatic hiss in the sim, rather more than what I think it should be for the tiny movement and this kinda bugged me a little bit so I got my dad program in a small deadzone at the beginning of the travel which solved the problem.

Finally the biggest issue we encountered was that the signal coming from all five Hall sensors had a small amount of jitter, about 1 count (out of 1024) with an occasional 2 count spike and this would cause the axes to constantly move ever so slightly - no way near as bad as the Logitech/Saitek Flight Quadrant however but this was still one of the main reasons I built the thing in the first place. The trouble is in, to use our previous example, the P-47, to open say the oil radiators with the switch (how I intended it) any input from an axis with the same binding will override the position set with the switch and if your axis is constantly moving the switch becomes unusable. The solution my dad came up with was to make the Teensy ignore any movement less than two counts. The 2 counts are such a small increment that it's completely unnoticeable when using it. This completely solved the problem, it's now currently working perfectly and I'm very happy with it.

It was also really cool to be able to work on a "Father/Son project" being on the opposite side of the world from my dad. I've been in Germany for ten years now and this has been a great way to keep in contact (not that we really needed a project for that). Special thanks to him.