Identifying chess pieces on a board

Thread Starter

4everJang

Joined Jul 21, 2023
13
I want to create a chess board that detects where each piece is located. Existing technology from DGT uses small coils and capacitors, but this solution apparently does not work when metal is involved. My chess pieces have a small magnet at the bottom as they are going to be used on a 3-dimensional chess cube. Any ideas on what type of technology to use to make this work would be very welcome. I have looked at RFID but that will not allow the precision I need. Distance between the piece and the sensor would be a couple of millimeters, with metal and magnet between them.
I need to distinguish between black and white pieces and 6 types of pieces, i.e. 12 in total. Also I would need to combine the identification of pieces on each field with a sensor that tells me if a player is touching a piece. If some pressure is required to detect this that would be acceptable. I guess I need to combine several technologies here but if there is one approach that would handle both requirements (touching and identification of the piece) that would be ideal.
Once the identification / location and sensing which piece is being touched are known and transformed into digital signals, a micro processor can handle calculation of target fields that can be reached, making small LEDs on each of those fields light up.
 

BobTPH

Joined Jun 5, 2013
9,259
Good luck. We had a long thread before on this topic, without the restriction of the metal board and magnetic pieces, so it was far simpler. And no conclusion was ever reached.

This is not an easy problem!

(Sorry if my assumption about the magnets is wrong. If it is, please clue us in.)
 

AnalogKid

Joined Aug 1, 2013
11,196
If you take a different approach, there is hope.

Each chess piece has two concentric rings on the bottom, or a center dot and a surrounding ring.

Each location on the chessboard has matching concentric conductors.

The plastic, wood, or whatever material of the board squares and piece bodies is recessed slightly (0.020" or 0.5 mm -ish) so the pairs of contacts make decent contact.

Each chess piece has an internal resistor with one of 12 values. Or 32 values.

That's it. The chess board now is a scanning ohmmeter. This is relatively simple electronics, with no RF or microprocessor involved in the basic scanning; conceptually, just a counter and some analog mux's. Of course you will need a uC to read the values and drive whatever kind of display you envision. Once a uC is involved, it can handle the mux addresses and eliminate the discrete counters.

The board can be scanned as 1-of-64. The 64 rings are all tied together to a current source, and each center dot is brought out on an individual wire. A 64x1 analog mux array delivers each board square in sequence as either open circuit or with a piece value.

Another option is row-and-column scanning, as in a computer keyboard. This takes fewer wires, but is a bit more complex. Both methods have pros and cons. I would prototype with the 64x1 array because it is so much easier to debug (both hardware and software) and has fewer inherent possible flaws.

With 12 different resistor values you have unique identifiers of each *type* of piece. For example, all white pawns are the same type, different from any black pawn. If you want a unique identity of each individual piece, then you need 32 different resistor values. With a 1960's linear ohmmeter circuit, this needs only 4-bit (or 5-bit) A/D to return the unique piece codes. That is the theoretical minimum; an 8-bit converter would be better for decent noise and component-tolerance immunity.

I have no idea if this type of thing already is on the market. If not, it is totally patentable

ak
 
Last edited:

Thread Starter

4everJang

Joined Jul 21, 2023
13
Good luck. We had a long thread before on this topic, without the restriction of the metal board and magnetic pieces, so it was far simpler. And no conclusion was ever reached.

This is not an easy problem!

(Sorry if my assumption about the magnets is wrong. If it is, please clue us in.)
I do know that DGT is selling kits with 12 different coil-and-capacitor sets that should be built into your chess pieces to make them recognisable on their electronic chess boards. I have too little knowledge of electronics to understand that approach, but I am guessing the metal board and magnets on my pieces would not combine well with that. Still, it would be nice to know how that approach works, so if anyone can explain it to an electronics newbie... ?
 

BobTPH

Joined Jun 5, 2013
9,259
Each will resonate at a different frequency, so a frequency sweep of a coil in the square could identify the resonance. But as you say, the magnetic metal board would not be helpful.
 

panic mode

Joined Oct 10, 2011
2,867
don't add anything into board or chess pieces. use overhead camera. then you can track movement of pieces since begin of game - even without differentiating between shapes of pieces (just using color). the whole thing could be a phone app.
 

Thread Starter

4everJang

Joined Jul 21, 2023
13
don't add anything into board or chess pieces. use overhead camera. then you can track movement of pieces since begin of game - even without differentiating between shapes of pieces (just using color). the whole thing could be a phone app.
Thanks for the suggestion, but that will not work for me, as the 6 (!) boards are mounted on a cube that can revolve around two axles (or rather one axle mounted in a wheel that can also be rotated). There is no (fixed) overhead position here...
 

wayneh

Joined Sep 9, 2010
17,501
You beat me to it. The app could show its version of the game on screen and allow for manual correction of any moves it may have missed. The players' hands might obscure some moves. This problem could be mitigated by putting an identifying mark on the top of each piece.
 

wayneh

Joined Sep 9, 2010
17,501
Thanks for the suggestion, but that will not work for me, as the 6 (!) boards are mounted on a cube that can revolve around two axles (or rather one axle mounted in a wheel that can also be rotated). There is no (fixed) overhead position here...
Oh wow, that is indeed more complicated. Could the boards be transparent, like stained glass, allowing for a visual solution to operate inside the cube?
 

Thread Starter

4everJang

Joined Jul 21, 2023
13
Each chess piece has an internal resistor with one of 12 values. Or 32 values.
Your suggestion gave me another idea that could work and would remove the need for all the extra work in creating concentric rings and recesses to get the circuitry required to measure the resistance of each type of chess piece. With the number of pieces (I am using 4 full chess sets for this) plus the added complexity of having a metal board and magnets on the base of all the pieces that would be a huge task.
So I was imagining a simpler solution: If I supply a wired glove to each of the two players, this would provide one pole of the circuit, meaning I would only need to electrically separate the metal fields and make sure the top section of the pieces connects to the other side of their internal resistors. Then touching a piece would cause a small electric flow, from which I should be able to determine the type of piece that was touched. And the same applies when the piece is put down again on another field.
Would there be a better method than using internal resistors when I work out this idea?
 

BobTPH

Joined Jun 5, 2013
9,259
Thanks for the suggestion, but that will not work for me, as the 6 (!) boards are mounted on a cube that can revolve around two axles (or rather one axle mounted in a wheel that can also be rotated). There is no (fixed) overhead position here...
Ah, the first new piece of info that invalidates the proposed solution. More to come, just like the previous thread.
 

Ya’akov

Joined Jan 27, 2019
9,264
Instead of IDing the pieces, ID the positions. When the board is first set, call it “home” for everything. Then, when players move they indicate the starting and ending positions (with pressure, a discrete switch, a light sensor normally covered by the piece, something else). The tracking is done by checking the piece that was previously resident on that square and at the completion of the move, registering it at the new position.

You could use a game clock for the move complete signal, and if you used light sensors for the “piece moved” signal, it could be automatic. A quiet, audible indication that a selection was made by the player might be helpful. The ID of the position could be based on RFID, since it will be the last covered position that is of interest.

Each square could have a bistable latch that is changed by the light sensor and an RFID scan. When piece is removed from the square, the bistable is toggled to indicate it is empty. When the light sensor is covered, it switches to occupied. all of the unoccupied squares would not respond to the RFID scan, the occupied ones would and the new square is the landing position for the move.

Or, you can include a timeout (maybe even just a small cap) so any square that has been covered for more than N seconds goes into a state where it will not respond to an RFID scan, leaving only the new one.

There are a lot of problems and details to be working out with this, but changing to IDing the immutable squares and simply tracking the pieces makes a lot more sense to me. There is a lot of logic needed to make gameplay transparent, it will need to tolerate changed moves. Visual and audible confirmation would be good, too.

What you want to do is complicated, and is going to be expensive both in terms of money and of time. There is no simple, reliable way to do it. The most effective way is going to be independent sensing on each square, but that is almost certainly cost prohibitive, hence my suggestion of making the squares know when to respond to an RFID scan so just one main RFID transponder could do the job.
 

AnalogKid

Joined Aug 1, 2013
11,196
My chess pieces have a small magnet at the bottom as they are going to be used on a 3-dimensional chess cube.
(I am using 4 full chess sets for this) plus the added complexity of having a metal board and magnets on the base of all the pieces that would be a huge task.
Please clarify this. Are you purchasing chess sets that already have magnets on the bottom of each piece. My take of the first post was that you were making things from scratch, such as buying ordinary chess pieces and drilling a hole in the bottom where you insert a magnet, resistor, coil, capacitor, whatever.

And, again for clarity, 4 sets of pieces positioned across 6 boards?

Separate from that, after the ID and location tech has been resolved, then what? Is the information crunched in some level of computer and put up on a color LCD? It sounds like a fun project. What is the end use of the information?

ak
 

Thread Starter

4everJang

Joined Jul 21, 2023
13
Please clarify this. Are you purchasing chess sets that already have magnets on the bottom of each piece. My take of the first post was that you were making things from scratch, such as buying ordinary chess pieces and drilling a hole in the bottom where you insert a magnet, resistor, coil, capacitor, whatever.
Purchased standard chess sets and glued flat round magnets to the bottoms.

And, again for clarity, 4 sets of pieces positioned across 6 boards?


Black and white pieces set up as you would normally do, but two sides of the cube are empty. Hard to describe, so here is a picture of the prototype.

CubicChess.jpeg

Separate from that, after the ID and location tech has been resolved, then what? Is the information crunched in some level of computer and put up on a color LCD? It sounds like a fun project. What is the end use of the information?
All fields are going to have small LEDs in them, so that the fields to which you can move a piece light up. A player can continue to move across the edges of the board, so it does become a challenge where you can move. Apart from being helpful the lights showing available positions for a piece also looks very cool. The cubic chess is as much an art piece as it is a game.
 

neTC

Joined Jan 12, 2022
18
I guess I would challenge some assumptions with respect to the presence of magnets, a metal board, and near-field RF (i.e. RFID and NFC).

Assume that ...
  • a passive NFC tag and an antenna coil is embedded in the bottom of the chess piece such that it is co-planar to the metal board
  • a (small) magnet is located in the bottom of the chess piece at the center of the NFC antenna coilune
  • the playing board has an NFC antenna coil at each square of the metal chess board
  • the antenna for each square is multiplexed by analog switches to an NFC reader associated with the chess board

Note that NFC is 13.56MHz. This is easy stuff to deal with. Don't be freaked out that it is RF.

It is true that the the presence of metal affects the tuning of the NFC antennas but you can tune the antennas for this. Note that the metal chess board is magnetic and will concentrate the NFC RF fields. The RF performance will be dictated by the antenna tuning, the magnetic properties of the chess board, the design of the coils, the spacing between the coils in the chess board vs. chess pieces, the Tx power of the NFC reader, and the Rx sensitivity of the reader.

This could NOT work without careful custom design and effort. But could it work? I think yes. The one thing that I don't have experience with is the (small) magnet at the center of the coils. But those are static fields when the piece is in position on a square and these magnets have a great benefit in assuring repeatable alignment of the chess piece to the board... or in other words, repeatable alignment of the antenna coils.

Some additional comments.. I think the NFC tag in the chess piece could be a standard NFC tag you can buy on-line (pick your size and shape). Based on my experience with NFC on metal surfaces...the tuning can be done with a single cap (when single-ended) or two caps (when differential) on the NFC reader antenna (no modification of the NFC tag itself).

So, as a concept... think of a PCB the size of the chess board. On the PCB is an array of NFC antenna coils, one per board square. Connect them via a series of analog switches to an NFC reader. Multiplex the analog switches, scan the board, and read the tags for each square. Some squares will be empty. Some squares will have pieces. When a piece is detected its NFC tag will tell you what it is.

Because the NFC reader can be located on the same board I think everything can be done with differential signals / routing. I think that will be better than trying to do everything single-ended. And remember... we are talking 13.56 MHz sine waves. This is not rocket science from a signal integrity perspective.
 

Alec_t

Joined Sep 17, 2013
14,405
Now that we have a better idea of the overall system, wouldn't it be simpler to give each player a device on which they could enter the positions of their chess pieces, rather than sense the positions: then illuminate the positions to which the pieces could be moved accordingly?
 
Top