Electronic Chess Board: Arduino Source-Code

[dkl]

I've built my own electronic chess board based on reed switches, similar to this project https://sites.google.com/site/bergersprojects/home

However compared to his solution, I am using an ATmega32u4 based Microcontroller (i.e. Arduino Leonardo, Micro or other clones) and my source code is freely available @ https://github.com/asdfjkl/ArdEBoard

So for those who are into such kind of projects this might or might not be useful; hence this announcement.

[lithander]

Wow, that's awesome! I want something like that!

Stupid question: do you need to connect it to a computer and it's just working as an input device or do you run the engine on the Microcontroller?

Edit: I found the video. So it's more more or less an input device and the engine runs on a PC, right? Have you considered making it a stand-alone chess computer? If yes: why didn't you?

[mvanthoor]

Wow, that's awesome! I want something like that!

Me too. I actually do have the programming and electronic skills to build something like that... but not the woodworking skills. That'd become one ugly chess computer.

[dkl]

Wow, that's awesome! I want something like that!

Stupid question: do you need to connect it to a computer and it's just working as an input device or do you run the engine on the Microcontroller?

Edit: I found the video. So it's more more or less an input device and the engine runs on a PC, right? Have you considered making it a stand-alone chess computer? If yes: why didn't you?

I've designed it as "just" an input board. Simply to reduce the workload for myself. For example, it can be connected to Art Bik's "Chess for Android" to a mobile phone via an OTG adapter and then is pretty similar to the experience of playing with a chess computer.

However it would be easy to extend it to a full chess computer. Just get a Raspberry Pi (Zero) + a touch display or connect some display/control buttons to it. Then connect the board via USB to the Raspberry and communicate to and from the board via the built-in USB to serial bridge (e.g. python-serial). The "computer" part is then on the raspberry, and the board acts as the input device (but can also receive commands from the raspberry). Then, using some library like python-chess it should be easy to extend this into a "full" chess computer with arbitrary chess engines. One could even go so far as replace the whole Arduino with the raspberry, and use the GPIO pins on the raspberry to read out the reed switches. I think this is what they do with the Pewatronic Grandmaster....

Due to time constraints I skipped that part, though.

[dkl]

Wow, that's awesome! I want something like that!

Me too. I actually do have the programming and electronic skills to build something like that... but not the woodworking skills. That'd become one ugly chess computer.

I had the same problem, and therefore I am not really satisfied with the result. I envy those who have such woodworking skills, especially working with veneer to create nice looking boards... Still, this is better than paying 160 Euro for a plastic Chess Genius Pro, I think.

[dkl]

I am bumping this: In the meantime I've simplified the design and cost (~50 Euro will do), using tactile switches. Similar designs have been proposed before (DIY USB Chessboard, "Sish"-Board), but not based on an Arduino. Source-Code is freely available again (but this time, dead simple).

If anybody is curious, an image and source code is available @ https://github.com/asdfjkl/ArdEBoard. There is also a video available @ of playing against the excellent "Chess for Android" by Aart Bik.

[Ras]

I am bumping this: In the meantime I've simplified the design and cost (~50 Euro will do), using tactile switches.

Are you interested in a way how to decounce them while still staying insanely responsive?

[dkl]

I guess it's a typo and you mean debouncing the switches? Always interested to improve the code! But of course, one needs to make sure that if a user presses a switch accidentally for too long, that the field information is send only once (otherwise the GUI will be confused if it receives, say a2a2a2)

btw extending the CT800 to a board-version should be straight forward with this approach. Just extend the current input buttons to the tactile switches on the board...

[Ras]

I guess it's a typo and you mean debouncing the switches?

Oh, yes, that's what I meant. What I found works much better than the classic "three times the same state when sampled every 10ms" approach: You can cample much faster than that. I'm using a matrix keypad approach where I switch rows/columns in a 1ms raster, hence there is a reading every 2ms. That's done right from the timer interrupt.

If a key is "pressed", I increment the "active" counter, else I decrement it, while clipping it to a max value of 8 and min value of 0. Now, if the keystate has been "not pressed", and it is pressed, I recognise it as pressed if the counter reaches 6. Or, when it has been pressed and goes down, I recognise it as "not pressed" when the counter goes down to 2. That way, the keypad is already responsive even before the bouncing is over.

In fact, it is so fast that I'm limiting it to one input on the same key every 100ms by bumping up the active counter to 50 if a key has gone from inactive to active. That's meant for people with limited motor control in their hands, and it would make even more sense for a chess board because there's little reason why you would want to tap the same square twice in a row.

On top of that, I have a small keypad queue where the timer interrupt puts in pressed keys, and the main application can read it so that no key press will be lost even if there should be some occasional delay in processing.

But of course, one needs to make sure that if a user presses a switch accidentally for too long, that the field information is send only once (otherwise the GUI will be confused if it receives, say a2a2a2)

Sure, it only triggers when a key state changes from inactive to active.

btw extending the CT800 to a board-version should be straight forward with this approach. Just extend the current input buttons to the tactile switches on the board...

That would need 16 IO pins for the board, and my H405 mainboard doesn't have that many pins left over.

[dkl]

If a key is "pressed", I increment the "active" counter, else I decrement it, while clipping it to a max value of 8 and min value of 0. Now, if the keystate has been "not pressed", and it is pressed, I recognise it as pressed if the counter reaches 6. Or, when it has been pressed and goes down, I recognise it as "not pressed" when the counter goes down to 2. That way, the keypad is already responsive even before the bouncing is over.

That's a really neat approach and so easy to implement. Updated my code, and it works like a charm!

In fact, it is so fast that I'm limiting it to one input on the same key every 100ms by bumping up the active counter to 50 if a key has gone from inactive to active. That's meant for people with limited motor control in their hands, and it would make even more sense for a chess board because there's little reason why you would want to tap the same square twice in a row.

That's something I did not include. The style of the switches that I used are such that even someone very clumsy would not accidantly press a button twice if only one field should be activated. On the other hand, it is sometimes a need to quickly press a field twice in a row. Think of the Ruy, where White plays against a computer, plays Bf1-b5 (press on b5), then very quickly moves a6 (computer move, no press), and then presses b5 again for Bb5-a4. Actually happened to me in a test game

btw extending the CT800 to a board-version should be straight forward with this approach. Just extend the current input buttons to the tactile switches on the board...

That would need 16 IO pins for the board, and my H405 mainboard doesn't have that many pins left over.

ah, that's a pitty.

Is your code portable? In particular, would it potentially run on an Arduino Due (that's a Atmel SAM3X8E ARM Cortex-M3). Not an STM32, but with 84 Mhz should have enough power. And lot's of ports...