I got a few suggestions for alternatives to the old spring arrangement. Pretty popular was the idea of using various forms of rubber in various arrangements. A bit of scaddery, followed by experimentation with rubber O-rings and various methods of winding rubber bands, I tested the above design. It works really well! There’s a good, strong centring force, very smooth response and it’s entirely silent.
I think longevity will be the biggest challenge with this design; over time the rubber will atrophy. Perhaps I could look at using some other kind of elastic band rather than a garden-variety rubber band? I also need to round off a few sharp edges to help prevent abrasion.
Following from my general dissatisfaction with the assembly process of the Mk. 1 gimbal, I redesigned it from scratch to be much easier to put together, with lower chance of breakages through over-tightening various bits. The new design is bigger and uses more bearings, but I think it’s better overall.
I’m currently working on refining the spring system for providing centring force on the stick. The current arrangement works well, except over time the spring steel piano wire develops a permanent bend, and the stick develops slop around the centre. If anyone has any suggestions for a centring mechanism which can be built into this design, and doesn’t use very particular spring components, let me know.
At the cost of seeing this datasheet every time I close my eyes, I’ve got EEPROM writing working. In order to avoid writing bogus data to the memory the chip makes you look up a table of codes and solve a maths problem before it’ll let you write. Feels like copy protection circa 1995.
Being able to tweak the interal parameters of the chip means that we can use the dedicated signal processing grubbins in there to clean up the signal before delivering it to the microprocessor doing the USB interfacing. Next up I’m going to look at modifying the gimbal base to support the PCB to which the MLX chip is mounted. This means I can do proper testing and determine what size magnet will be best for this design.
The Thrustmaster Warthog (an incredible and incredibly expensive) joystick uses a very similar chip but it moves the magnet around a hemisphere centered at the chip. My gimbal moves the magnet in a hemisphere centered at a rotation point above the chip. Both modes of movement are supported by the built-in maths according to the documentation, which is great for me, but I’m not going to sit comfortably until I’ve tested it myself.
Hello, I'd like to know if you could help me make to access the MLX90363. I understand it can be programmed by SPI without an external programmer. Is that correct? What experience do you have with that? I'd like to learn about its capabilities. Have a nice weekend over there. Greetings from Argentina!
I’ve begun work on an interface to a new 3D hall effect sensor for detecting the deflection of the stick. Price-wise it’s a little bit cheaper, but the main appeal is that its EEPROM settings can be altered via a standard Serial Peripheral Interface (SPI) bus, rather than using Melexis’ super-expensive programming box. This means that linearisation and calibration can be done in-chip, making things much simpler.
In a bout of webtwopointoery I’ve signed up on Trello and created a board for OpenJoystick for tracking progress on various aspects. Looks mildly useful.
I’ve built the plates for the back and right sides of the right hand throttle quadrant. The two hats on the back (the slew stick and the coolie hat) rest underneath the index and middle fingers, whilst the hat (radio switch) and switches on the right side are manipulated by the thumb.
The thumb switches are going to get a bit trickier than they currently appear, since only one of them is an actual three-position switch. The others are sprung from some positions. I imagine I’m going to have to work out some kind of spring mechanism to enforce positions in which the switch cannot rest.
If you’re curious as to what these switches are used for, there’s a handy guide to the A-10C HOTAS layout here. Be warned: It’s not for the faint of heart.