Sunday, August 13, 2023

New display works!

 The PCBs for the display I described here came in a couple days ago. The upshot: It works!! Woo hoo!

The only thing I haven't tested so far is the I2C memory, but I'm not too worried -- even if that's borked, I would count this board spin a success! I also got a signal polarity for the LCD enable signal confused in my schematic, but this is not rocket surgery; I figured it out and added some code to set the GPIO properly. All is well.

I'm now working on form factors for the housing. I noticed the CM4 chips (in particular, the CPU and Wi-Fi module) get pretty hot, so I'm experimenting with cooling slots. I'm also working on various ideas for mounting; stay tuned for more deets....

The way I 3D printed the SD card slot is new for me. I programmed a pause at a given layer, inserted a 3D printed support that was covered with Kapton tape, and continued printing. Once done, I popped out the support!

Sunday, July 30, 2023

New probe wired into N291DR

Today, I did the necessary legwork to get the new probe wired into N291DR. The way it's set up, we are tapping into the +12VDC supply to the nav lights, and (still) using Wi-Fi for sending data to the display.

The AoA and airspeed worked really well. The yaw angle differed from the inclinometer ball by maybe a degree or two -- I just set it by eye, so I'll tweak it to be centered. Other than that, it worked like an Airball probe and did its job. Yay!

Of course, no test would be a test without squawks. Here they are:

Flaky Wi-Fi. Sometimes, I connect the display to the probe and the data rate seems low. I power cycle one or the other side, and it self-fixes. I think there may be an issue with the Wi-Fi base station implementation on the ESP32, or something. I hope it's easy to fix. Eek! I will try to repro on the bench.

Weird altitude readings. The altitudes on the display were maybe 100-200 feet higher than the rest of my instruments. I cargo culted some kind of temperature correction in my altimetry calculation in the display; that may be the culprit. I need to clean that up and use the standard conversion formula, and see what that yields.

Probe inflight autozero is a disaster. If the probe is power cycled in flight, it tries to autozero, using whatever large dynamic pressure readings it's getting, and is basically useless for the rest of the flight. I need to program some more cowardice into the algorithm, refusing to autozero if there is significant pressure on the sensors. Extra credit: See if I can store the latest offsets in some NVRAM, if the ESP32 has any.

Ball "trails" don't seem visible. For some reason, I have lost the "trails" / "history" the ball leaves behind. I can't think of why. Need to debug -- might be some stupid mistake somewhere.

USB malarkey. The USB connectivity from the display to the control "knob" remains as gefurfified as ever. Just waiting for that new display with bated breath...

I am hoping that the basic static pressure measurement scheme is solid. Otherwise, I'll have to go back to fabricating some kind of static pressure "spike" like the older models, and that's really a total pain to manufacture. Fingers crossed.

Friday, July 28, 2023

Digi-Key fairy, and sealing

The Digi-Key fairy delivered a new bunch of Newhaven 2.4" displays today. In other news, I finally got a relative low-labor stackup of nose parts to seal together without leaks:

The no-leak seal between the manifold parts was created by sanding the faces of the 3D printed parts on a flat surface with 220 grit sandpaper, then assembling using gaskets out of 3M 468MP adhesive film. This product is used to secure 3D printer bed covers to the beds, and should hold up to extended high temperatures and other forms of abuse. At some point I'm going to show a montage of the full process, but for now, this is the setup I use to cut the gaskets. I clamp these templates onto the material, and cut out the holes with a sharpened stainless tube that's held in a pin vise:

This is one of the aspects of the project I've been working on for many weeks, but has been sort of unsung because progress happens in fits and starts. I have tried many, many options....

One thing I wanted to do was to not have to sand down the "grainy" surface of the MJF plastic. It turns out it's impossible to do this and get a decent seal. So maybe I'll have a quicker way to clean up the surface, or maybe I'll find an economical but smoother 3D printing process, but either way. Needs to be smooth.

I tried to use silicone rubber gaskets of various sorts, relying on clamping pressure. This does not work very well -- they all leak.

I tried silicone rubber foam, about 1/32" thick -- and that works fine. However, the foam "squeegees" out of the sides of the seal, and can interfere with the aerodynamic shape. It requires post-trimming, and never looks really "finished".

I tried Jo-Ann Fabrics thermal adhesive, baked in the oven. But some of my "manifold" parts include some electronics pre-assembled, and a heat soak sufficient to melt the adhesive also fried the electronics.

I did not try anything involving grease. Maybe that too would work, but if the 3M adhesive film works, it's likely ideal. It is not messy and seems to "wet" the faying surfaces pretty well.

Another minor hurdle cleared. :)

Sunday, July 23, 2023

New display form factor mockup

While waiting for parts in the mail, I made a first guess at what the case for the new display would look like, and assembled it to get an idea of the form factor. This is what the CAD model looks like:


This is the actual mockup on my bench. As you can see, it's actually quite small:

The SD card slot is at the top left:

The USB-C port on the back, emerging from the bottom of the unit, is the single "umbilical" for power and peripherals. A user can purchase many kinds of straight-out or right-angle cables to fit their installation:

I'm thinking the holes on the back would be made to fit a RAM ball mount, but I'm not quite sure yet. I certainly want some kind of threaded insert hole pattern on the back so people can mount this thing however they want.

That's all folks! Looking forward to the circuits coming in. Fingers crossed!