MQTT and NodeRed on an airplane?

I've had several posts about using NodeRed for home automation, communicating with various nodes over MQTT. Could I do the same in the airplane? I think I could, and there would be some benefits to it. Certainly less wiring, and maybe more reliable and customizable.

The whole point of this blog originally was to use an Arduino and be able to monitor the engine status. Originally, I was going to use a Mega, and a bunch of shields. The old eco-system caused me to lean that way. Mega on the bottom, a custom shield next, and a Bluetooth shield on top. The world has moved on, and now I can use ESP-32 and 8266's with built in Wifi and Bluetooth. The boards only cost about $10, instead of the $20+ for the Mega and Bluetooth shield.

The sensors can mostly be wired to the ESP-8266 board with SPI (3 wire) communications, through adapter boards. If I need an analog or digital input, it is available on the board as well. There are very inexpensive thermocouple modules available, that would take care of the CHT and EGT measurements. The oil pressure and temperature would be analog inputs if I stick with the stock probes, but there is the Dallas18b20 probe that could be used to measure temperature, but there only one temperature measuring point on the engine. The Dallas 18B20 can be used for OAT measuring.

The oil pressure transducer is off on a flex pipe, with a manifold for to hold the hobbs switch. (theoretically, the oil pressure measurement could be combined in the computer somewhere to enable the hobbs timing measurements). There are nice solid state pressure sensors that may be appropriate for fuel and oil pressure, I need to do additional research.

Volt and Amp measurements can be done using the analog inputs, using the existing shunt for current measurements, or adding another shunt. There are also hall effect sensors that could be used to measure current. Using various current sensors would allow measuring all the important values at once, charge rate, drain rate, etc.

The tach measurement can come off the mag or EI unit. A transistor or optoisolator should work when connected to a pin that accepts interrupts.

Other sensors could be added or deleted anytime using additional nodes. If I want to measure the airflow through the baffles, I could add a node with multiple pitot tubes to see where the air might be leaking.

Then the question becomes, how many nodes should I use? It would be tempting to build one for each sensor, I want to limit the nodes to one function (I've said that before). How gray to I want that idea, one function could be all the thermocouples. Another node for all the oil information (pressure, temperature). Another for the fuel pressure, levels. Convenience may dictate some other arrangement, like thermocouples left side, thermocouples right side, that way the wires don't have to be so long.

Each node will still need 5V or 3V connected. It would be tempting to use the micro-USB connectors on the boards to feed the power. I think, the constant vibration would make the connectors fatigue very quickly, and the power would be falling out. It would probably be best to solder the power to the boards.

A $10 raspberry PI zero could be the NodeRed/MQTT hub. The raspberryPI should be mounted inside the cockpit. I am sure the WiFi signal can get around the firewall on a plastic airplane. Even a metal aircraft should have WiFi from the engine compartment to the cockpit. I'd have to try it out though.

In the cockpit, PLA printed enclosures should work fine. In the engine compartment, there may be problems with PLA or printed enclosures (constant vibration, heat, and rain). 3D printing my own enclosures may be compromised quickly. It may take some experimenting, certainly to make appropriate mounting choices, considering the sensor wiring stresses and heat.

NodeRed should allow nice cockpit displays of the instruments, and allow logging in various formats including CSV. Using the same CSV layout as the COTS instruments (like J.P Instruments) would allow people to analyze engine performance using off the shelf tools (excel, etc). The file could be scp'd from the Arduino, right to a tablet or phone, taken home and analyzed.

I don't know. I see a few worrying things, but nothing that is really scary about doing it. Can you talk me out of it?

Eggbot; filling in some missing information

I guess it is that time of year, people are building Eggbots. I built one, and needed to learn a bunch. Some of the information was vague, or not connected to the original design, or anyone elses design.

I started with a Russian frame that I liked on Thingiverse. This was mostly lacking any real documentation. Elliot from Hackaday has a page with lots of information, including how to make a CNC shield move a servo, and how to get G code out of inkscape.

The Russian frame had a bunch of options for the pen holder. I chose the simple one, and am having success using that with a sharpie fine point. Being an American, it was easier to get 1/4-20 5/16-18 threaded rod than 6mm, so that is what I used. I cut 3 1ft pieces. For the rest of the bolts and screws, I used metric (m4 and m3 mostly), since the motors assume metric.

One tool that I will call essential is the taper cutter. Cutting threaded rod is easy, but the threads are suddenly all mucked up. Either the taper goes away, and threading on a nut has to be aligned perfectly, or the threads are bunged up so badly nothing will thread on anymore. This tool will make the trouble disappear, in about 2 seconds!. Sure you could file a taper in using a file, or grinder, but for less than $5, why?

The Russian  frame also assumed some 16mm bearings, that I didn't order. Another similar frame used skateboard bearings in a holder with the same bolt pattern as the stepper motors. I just reprinted the left frame from the Russian frame and two of these bearing holders for the end.

I couldn't find O rings that would hang onto the egg good. One end or the other would slip, and the drawing wasn't real good. I found some #17 bath diverter washers that seemed to be a better shape, and fit the end caps well. They still slipped.

I've started playing with some table foot rubber pieces, like on the bottom of a crutch. They are more grippy. I'll update when I know more.

Some page, maybe another one on Thingiverse suggested I use a CNC V4 Shield. A great design, with the Arduino Nano right on the board with the stepper drivers. The trouble is, the common Chinese clone makers messed up the board. A couple others have figured out the changes needed, and published the wiring changes to make this board work.

What I have seen though, the inkscape G code output doesn't quite make the pen up/down command that the GRBL expects. The delays are about 100 times longer than they say (seconds instead of milliseconds) as well  (Update, in the save dialog, it lets you set delay to 0, that seems to work). Long delays with the pen over at an angle almost always messes up the continuation of the lines. I end up hand editing the G code output. I should probably build a python or perl script to fix the output, or just figure out what unicorn changes are needed.

Changing the GRBL source code, it is important to remember you moved the code to the Arduino directory, so change it there, not where you expanded the code. If the motors are going backwards, reverse the plugs on the CNC V4 board.

For test eggs, I have been using the 18 for a dollar plastic eggs that I can get at the Dollar Tree. I like hard boiled eggs, but I didn't think I'd eat 24-36 in the next couple of weeks. I should look into dry erase markers to see if I can erase the test eggs.