So… what actually happened over the last 14 months?

 When it comes to the cockpit build, four main things happened (roughly speaking):

  1. I finished several additional instruments for the dashboard
  2. Changed the core software architecture
  3. Brought the Stewart platform up from the basement and reworked it both mechanically and electrically
  4. Started building a full cockpit on top of the motion platform
Even though this list is roughly in chronological order, I’ll start with point 2, since it ended up influencing point 1 as well (at least towards the end).

Integration Hell (aka Cable Chaos) 

Changing the core software architecture


Even though it’s really convenient and quick to hook everything up using 3rd-party solutions like Air Manager or MobiFlight, things get messy fast once the project grows beyond a certain size.

It’s not just that discrete wiring for stepper motors, encoders, buttons, and so on quickly turns into an impenetrable mess of cables. What really stood out to me is how a slightly more complex setup can turn into a maintenance nightmare - especially when the central system (in my case, Air Manager) pushes software updates with planned or unplanned breaking changes.

I don’t have first-hand experience with MobiFlight, since I wanted to keep the cockpit platform-independent and MobiFlight is Windows-only. But based on my experience, the core issue comes from the centralized hardware and software architecture these systems rely on.

So I decided to move away from that and switch to a more modular approach. One where I can develop and test individual components, instruments, and controls separately, and then combine them later like building blocks.

Luckily, I’m not the first to run into this problem. There’s already a well-established, widely used, and relatively inexpensive solution for exactly this: the CAN bus.

Enter the CAN bus

Other than the fact that the CAN bus has been used for over 40 years to connect components in cars (and is basically indispensable in modern vehicles) I had no prior experience with it.

I’d read that you need special transceivers, that the network architecture is a bit different, and that some builders use it for DIY projects like more complex cockpit setups.

But after doing some more in-depth research on the topic and running a few breadboard tests, it quickly became clear: this could (and probably should) be the solution to my problems.

At a basic level, it takes me from this 

and this

to this

The number of components and the overall complexity of each instrument doesn’t just magically disappear. In fact, the component count actually increases, since each instrument (or CAN node) now also needs its own power supply, one of those CAN transceivers mentioned earlier, and a dedicated microcontroller on top of the existing parts:


The key point, though, is that this complexity is local to each instrument. It can be developed and tested independently, then neatly enclosed and protected inside its own housing.

What you get for that price is a self-contained, testable, and replaceable unit that can be connected to the overall system with a simple point-to-point daisy chain.

And that price isn’t all that high - especially these days, when you can get an Arduino Nano clone, a DC-DC converter, and a CAN transceiver for under €8 (around $10).

I’m planning to write a separate post about the CAN bus architecture I’m using, where I’ll go into a bit more detail about the hardware and software behind my setup.

Finalizing the 6-Pack (and More)

I finished several additional instruments for the dashboard

I actually managed to finish the HSI instrument. It was a tougher ride than I expected, but after two or three complete rebuilds of both the mechanics and electronics, I finally ended up with a working electromechanical HSI.

I’ll definitely write a separate post about the HSI. I filmed the entire final assembly, so there might even be a companion video for this project—if I ever find the time to edit the gigabytes of footage.

"Smaller stuff"

To complete the full six-pack, the only thing missing was the altimeter. I’d already struggled with that one about two years ago, trying to drive it with a clockwork-style gear mechanism and just a single stepper motor. After two full redesigns and a couple of non-working prototypes, I ended up shelving it for a while.

Now that it was the last missing instrument in the six-pack, I moved away from the single-stepper approach and - taking inspiration from Captain Bob’s design - built a version of the altimeter using three stepper motors (plus one servo for a 10k flag).


And then there was still an RPM gauge (with a nice little OLED odometer featuring digit-roll animations), a fuel gauge (the first CAN bus instrument in the dashboard), a (mechanical) trim indicator, a small EGT gauge, and a transponder (the second CAN bus instrument).

 
 


I’ll probably write separate posts about some of these instruments as well.

As for the Stewart Platform…

Yes that's point 3. and 4.!

That topic will have to wait for the next post. This article has already gotten way too long, and I’d like to do a few other things today besides just writing blog posts.

So, see you next time!


Comments

Post a Comment

Popular posts from this blog

Setting up a new (blog-) home for my build log

Image
Hi, After starting a build log for this project on another platform a few years ago, I ran into issues with how the blog was organized there. So I decided to move my build log for my DIY flight sim cockpit here. I’m hoping this move will make it easier to post updates more regularly and keep track of the project’s progress. I’ll be moving my existing posts over from the old build log and then share an update on where the project currently stands, along with the next steps. So, here’s what’s happened so far… The Beginning (January 2024) Piper Arrow III Sim Cockpit Originally, I set out to build a dashboard for a Piper Arrow III for use with flight simulators. The idea was to keep it modular so it could be set up and taken down easily on a desk, since I didn’t have space in my apartment for a full, stationary cockpit. The setup was meant to be centered around my CH Products yoke and rudder pedals, which I had already owned for years. Since then, the scope of the project has evolved quite...
Read more

The GNS 430 GPS

Image
Originally published 1. May 2024 The first device I built was the GNS 430 GPS unit. Here's the finished thing: I wanted this to have with backlit buttons and real dual rotary encoders. After searching the web for a while it became clear, that I had to design this myself. The screen content comes directly from the flight sim. But since I'm currently running the sim on a Macbook Pro, and thus do not have an option to connect too many external displays, I use AirManager for grabbing the GPS' display content directly from XPlane and sending it via a network connection to a Raspberry Pi 4. This works surprisingly well, even over wifi. Unfortunately the Raspberry doesn't have enough GPIO pins to also drive all the buttons and encoders. I originally planned to build two GNS 430 units anyway (as there are two of them in the simulator) and it's no problem for the Raspberry Pi 4 with its 2 HDMI connectors (and the AirManager/AirPlayer combo) to provide the display content for...
Read more

Merry Chistmas…

Image
Originally published 23. December 2024 I already finished the Attitude Indicator. I plan to do a more detailed writeup and a video on this one, so for now, only to images. From the front, it looks more or less like any other DIY Attitude Indicator. However, while it is almost never needed during normal flight, I have always disliked the fact that most DIY Attitude Indicators have a roll limit to avoid tangling the wires. One solution could be to use a slip ring for the cables of the pitch motor or servo. However, my idea was to over-engineer the entire thing by designing it with coaxial motor shafts, placing both motors for the two axes outside the moving cage. The truly challenging part was getting the software to work. Over the last two weeks, I’ve been designing and building an HSI from scratch. I couldn’t find any DIY versions of an HSI. Since a typical HSI requires (at least) five axes to operate (six in my case), I quickly decided to keep the mechanics as simple as possible and u...
Read more