I'm trying to build a chess board which can automatically record a game without any help from the players. I don't need anything as fancy as actual tags on the pieces (I'm going to figure it out just by dumb dead reckoning), but I'm not sure of the most effective sensors to use for the squares. Preferably, I'd like the price to be low, but here are some of the ideas I had.
Attach magnets to the bottom of the pieces and lay a grid of wires underneath the board (it's very thin). When the pieces move up or down off the wires I should see some kind of current induced from the magnetic flux.
Second idea, was to use magnets in combination with a bunch of reed switches. This is basically the exact same idea except it's using an actual sensor which is probably going to do a way better job. The price obviously goes up pretty rapidly though ( I need a reed sensor per square, $1.50 * 64 + 1.50*24 for the magnets).
Lastly, I could use the FSR sensors listed here: Force Sensitive Resistor 0.5" - SEN-09375 - SparkFun Electronics but they're obviously a lot more pricey at 6.50 a pop. Since there are 64 squares, the price is rocketing up to a cool 416 (or somewhere in this ball park). I considered maybe using square FSR sensors to bring the cost down and having them share squares. I'm not sure if that would work well.
This is my first time ever building up a full project, and I don't really have any experience with these ideas. I was hoping you guys could act as a sounding board, maybe come up with some ideas
Yah I thought of the light sensors as well but I want the board to look as normal as possible with no cuts or perforations or anything. So cuts in the board I think won't work.
Try ebay or something - Last time I bought them they were around £1.50 for 10...
Hall effect sensors would be much better though (and still less than $1.50 each).
I used to know a really cheap place to buy neodymium magnets but I can't find/remember it. (They sold the packs of spherical ones and cube ones before they became 'cool toys' and became very expensive items from toy stores - they were really cheap too)
Yes... I did read the bit about you wanting a normal board, no cuts.
But... for DEVELOPMENT of this project, I would recommend that you compromise, build a "board" that uses something "awful", but reliable, for sensing the presence of a piece on a square... some system of little pegs protruding from the base of the piece and getting stuck into a hole in the board comes to mind as cheap, do-able, and reliable.
Without the software (costs only your time), the board, however elegant, is meaningless.
Get the software written... that's the fun bit anyway... and if you get that done (oops... WHEN you get that done!) you'll know a lot more about how the board has to be made, and you'll make fewer false starts.
Another option: A web-cam above the board. This would require a "big PC" to catch and analyse the image, but you're going to need that to manage the data from the board sensors anyway. Although, eventually, the data capture would be a good job for an Arduino, and would get rid of the unsightly web cam over the board.
I want to do the same thing, record a chess game while people are playing it.
But I want to keep track of which piece is which to catch errors (black bishop goes across the board and is put on a white square by accident).
For sensors, I thought of 2 ways.
First, barcodes on the bottom and 2 stepper motors traversing and scanning a grid after each move to update the board. It could use either a low res camera to capture an image and read the barcode on the bottom of the piece, or an old barcode reader to scan the barcode at each point.
this may give physical height to the base of the board though.
Secondly, Dallas Semiconductor makes id chips with unique serial numbers. No idea of cost, but the idea is to mount a chip in the bottom of each piece, and a reader chip in the base of the board,
one for each square, to tell which piece is resting on each square. You could probably limit the range by mounting the reader in a short cylinder of metal or foil so it only sees the piece on the square. You might also explore the idea of using a pair of steppers to move one reader around the columns and rows to read the id chips.
A third variation might be to use a low res camera and use a grayscale code on the bottom of the part to identify pieces as you scan the board from underneath (glass board, maybe?).
I suggest unipolar Hall effect switches, such as http://www.digikey.co.uk/product-detail/en/AH337-WG-7/AH337-WGCT-ND/2122985. They are smaller than reed switches, so it will be easier to fit them in the space available, and they don't bounce. You can surface-mount them on a PCB that goes under the board, which will save you a lot of wiring. Make sure you put the magnets the right way round on the bottom of the pieces.
To connect them to an Arduino, as they have open drain outputs, I think you could multiplex them. Use an active high output 8-way demultiplexer chip (74HC238) to power all the sensors in a single column. In each row, connect all the outputs together, with a common pullup resistor. Then use an 8-input multiplexer to select which row to read (or 8 input pins).
dc42:
I suggest unipolar Hall effect switches, such as http://www.digikey.co.uk/product-detail/en/AH337-WG-7/AH337-WGCT-ND/2122985. They are smaller than reed switches, so it will be easier to fit them in the space available, and they don't bounce. You can surface-mount them on a PCB that goes under the board, which will save you a lot of wiring. Make sure you put the magnets the right way round on the bottom of the pieces.
To connect them to an Arduino, as they have open drain outputs, I think you could multiplex them. Use an active high output 8-way demultiplexer chip (74HC238) to power all the sensors in a single column. In each row, connect all the outputs together, with a common pullup resistor. Then use an 8-input multiplexer to select which row to read (or 8 input pins).
I'm working on a similar project, except I only need to register the piece's placement. I want to use the hall effect sensors because of their increased accuracy, so your idea sounds great. However, I don't much experience with circuitry. I understand the names, (demultiplexer, pullup, multiplexer, etc.) but have idea of how to build them into the board. Do have a example or diagram, or a possible link? Any help will be appreciated.