Thread: Your latest projects

I know I've already done a thread like this recently, but it's gone now so I'll start another one.

Just a thread to share what you're working on currently (at home, work, don't care) or will be working on soon.

My biggest programming-related project recently has been making a scalable Modbus data logger for work. For those of you unfamiliar, Modbus is a data exchange protocol traditionally used over serial and ethernet connections. Made a GUI for the user to input all the parameters for a certain connection (either IP and port or serial port and settings), and then the devices on that connection (Modbus serial allows for multi-drop, can have up to 247 devices on one pair of lines). The register maps are predefined by manufacturer and model code so that once the logging begins, the program can then display what each data register is (i.e. flow rate, temperature) in a neatly displayed table. The project was pretty straightforward since I already know about Modbus and what I wanted it to do was pretty well-defined, but it was a good learning experience because I got some more experience with IPC (I spawned additional processes for each connection). For whatever reason, I seem to enjoy writing programs that involve serial comms or sockets too, so.

New home project is to pair up a spare wifi router with an Arduino and relay board to intelligently control my sprinklers. Going to load OpenWrt on the router and write a program to get the weather forecast, water right before sunrise, scale timings based on temperatures, etc. The Arduino itself will just be for the digital I/O, not doing any local sensors (although I'm sure it'd work better with some, more interested in seeing how well it can work without sensors).

Also, guess I'm starting to drink more Maker Kool-aid, going to fool around with an ESP8266 board to see what I can accomplish with it in terms of connectivity.

Very nice, surprising there isn't a decent one out there already.

Interesting - I also recently created a data logger for a specific application at work, though much less sophisticated than yours. The hardware is basically a box with a microcontroller and analog voltage inputs, and is driven by a custom GUI I wrote. It uses gnuplot to plot the data, and data for a particular test can be coupled with a "reference" data plot for comparison. Assemblies are tested under extreme environment stress conditions, so comparing their output with a reference lets us know if any soft failures have occurred.

My current work project is pretty ambitious. We have a circuit board that controls and monitors various aspects of a system, as well as other high level functions (serial comm, digital display, etc). All of the monitoring is done via analog circuits, e.g. charging caps, comparators, potentiometers, etc. I am attempting to design all of this functionality (including all the monitoring) into code for a single microcontroller. This will not only make fault monitoring more precise and repeatable (instead of relying on pots), but will reduce the board size, component count, assembly time, and test time by more than 70%.

At home, I think I've finally come up with a marketable invention idea that, as far as I have seen, does not yet exist. While it will rely somewhat on electronics, its main function utilizes a discipline I am unfamiliar with, so I am currently on a learning curve.

I am also wrapping up a simple project, a "mister". Basically, I installed garden water misters in a PVC pipe to help keep cool during hot summer days. I'm mounting it on 6' wooden legs so I can sit beneath it and beat the heat this summer.

For a home project I'm designing an evolution simulator. In a 2D world, the shapes and actions of the animals will evolve. Some will be herbivores, some carnivores, some a combination. The shapes and motions of their limbs will evolve as would their overall patterns of movement. The herbivores will probably just eat dots that are scattered every now and then. When I get that working I may add some senses.

I want to do it in C++, which I don't know very well, and SDL.

Originally Posted by Matticus

Interesting - I also recently created a data logger for a specific application at work, though much less sophisticated than yours. The hardware is basically a box with a microcontroller and analog voltage inputs, and is driven by a custom GUI I wrote. It uses gnuplot to plot the data, and data for a particular test can be coupled with a "reference" data plot for comparison. Assemblies are tested under extreme environment stress conditions, so comparing their output with a reference lets us know if any soft failures have occurred.

My current work project is pretty ambitious. We have a circuit board that controls and monitors various aspects of a system, as well as other high level functions (serial comm, digital display, etc). All of the monitoring is done via analog circuits, e.g. charging caps, comparators, potentiometers, etc. I am attempting to design all of this functionality (including all the monitoring) into code for a single microcontroller. This will not only make fault monitoring more precise and repeatable (instead of relying on pots), but will reduce the board size, component count, assembly time, and test time by more than 70%.

At home, I think I've finally come up with a marketable invention idea that, as far as I have seen, does not yet exist. While it will rely somewhat on electronics, its main function utilizes a discipline I am unfamiliar with, so I am currently on a learning curve.

I am also wrapping up a simple project, a "mister". Basically, I installed garden water misters in a PVC pipe to help keep cool during hot summer days. I'm mounting it on 6' wooden legs so I can sit beneath it and beat the heat this summer.

Nice. Interested in what MCU you're using for this. I recently did some looking into how we could make an RTU to aggregate our instrumentation signals, and found that at least for cheaper MCUs, their built-in analogs varied too much as other signals (digitals etc) changed around them. Then again, I'm naive mostly to noise considerations. I was looking into using separate chips via SPI to do all the 4-20mA signals.

Originally Posted by algorism

For a home project I'm designing an evolution simulator. In a 2D world, the shapes and actions of the animals will evolve. Some will be herbivores, some carnivores, some a combination. The shapes and motions of their limbs will evolve as would their overall patterns of movement. The herbivores will probably just eat dots that are scattered every now and then. When I get that working I may add some senses.

I want to do it in C++, which I don't know very well, and SDL.

Sounds fun. Reminds me of when I took my first C class in college. Extra credit assignment was to implement Conway's game of life (10 x 10 grid).

Originally Posted by Epy

Nice. Interested in what MCU you're using for this. I recently did some looking into how we could make an RTU to aggregate our instrumentation signals, and found that at least for cheaper MCUs, their built-in analogs varied too much as other signals (digitals etc) changed around them. Then again, I'm naive mostly to noise considerations. I was looking into using separate chips via SPI to do all the 4-20mA signals.

That particular application used Silicon Labs C8051F124.

Separating analog and digital ground usually help reduce noise on the analog lines, as well as routing power lines, digital lines, and analog lines away from each other. Separating ground planes is not something I'm very studied in, since PCB layout is up to our Drafting department, but I have found some good white papers on the net.

I've seen digital pulse signals cause currents that wreaked havoc on ground before. Earlier on in my career, I buffered a control pulse to eight separate outputs, and the resulting noise as each pulse dropped low was bad enough to throw other circuitry out of whack. Using a DC-to-DC isolator helped mitigate this by isolating the power/ground lines between key points in the circuit.

Originally Posted by Matticus

That particular application used Silicon Labs C8051F124.

Separating analog and digital ground usually help reduce noise on the analog lines, as well as routing power lines, digital lines, and analog lines away from each other. Separating ground planes is not something I'm very studied in, since PCB layout is up to our Drafting department, but I have found some good white papers on the net.

I've seen digital pulse signals cause currents that wreaked havoc on ground before. Earlier on in my career, I buffered a control pulse to eight separate outputs, and the resulting noise as each pulse dropped low was bad enough to throw other circuitry out of whack. Using a DC-to-DC isolator helped mitigate this by isolating the power/ground lines between key points in the circuit.

Gotcha. I was using the Arduino Due (feel like a tool admitting that), the ARM chip on it has comparable specs as far as the ADCs go. Speaking of which, just looked at their site and they just (as in today or yesterday) retired the Due and Yun.

Originally Posted by Epy

Sounds fun. Reminds me of when I took my first C class in college. Extra credit assignment was to implement Conway's game of life (10 x 10 grid).

It's a lot more complicated than the Game of Life! I could code that in 15 minutes.
Even an initial basic version of this project will probably take me at least 15 hours.
I'll try to motivate myself to code something up so you can see the basic idea.

Originally Posted by Elkvis

Digital logic switching transients often reach well into the RF range.

Sure, but you explicitly mentioned "RF", which typically refers to radio circuitry, not the presence of radio frequency oscillations.