Can someone help me understand this circuit? | Chessboard Piece Recognition

[lp1234poha]

Hi im new to the forum and im looking for some help to understand a circuit vagueIy described in a patent i found online.

I was thinking about building a chessboard that can detect piece positions on the board. There are several comercial products but all of them are quite expensive thats why i wanted to try to build my own. The most well known company that produces electronic chessboards with piece recognition is called DGT. DGT held a patent for their method of piece recognition which has expired by now. In the patent they describe the method used to detect chess pieces located on the squares on the 8x8 grid of a chessboard.

The idea is the following:
A transmit coil is located under each of the 8 rows of squares and a receive coil is located under each of the 8 files (columns) in such a way that for each square on the board there is exactly one combination of transmit and receive coils intersecting underneath that square. Inside of the each chess piece there is resonance coil. The patent states that the transmit coil is connected to the output of an amplifier and the receive coil is connected to the input of that amplifier. By selecting a specific transmit and receive coil the circuit oscillates with a frequency of the resonance coil located on the square where the two coils intersect. Piece detection is therefor achieved by selecting a specific resonance frequency for each piece type and the position of all pieces is detected by scanning through all combinations of transmit and receive coils.


Can someone maybe explain to me how this amplifier circuit could look to achieve this?
Or maybe hint me to some resources where i can try to understand this concept?

here is a link to the patent: https://patents.google.com/patent/US6168158B1/en

thanks in advance

[Doctorandus_P]

Why not simply use capacitive sensing in a matrix? You would need the bottom of the pieces to be conductive, but that can be simple steel washers or aluminum foil.

[macboy]

Why not simply use capacitive sensing in a matrix? You would need the bottom of the pieces to be conductive, but that can be simple steel washers or aluminum foil.

How could that accomplish uniquely identifying the position if each individual piece?

[lp1234poha]

Yes thats the problem with many methods i looked at. Recognizing the specific piece type is important. I saw suggestions to use rfid tags for the pieces but this seems much more expensive and complicated because you would need 64 coils that need to be selected individualy.

For the "DGT-Approach" i looked at many oscillator circuits that contain a transformer in the feedback path. But i cant really wrap my head around how a third resonance coil would determine the frequency of the oscillation.

My knowledge of the theory of resonance circuits (and electronics in general) is quite limited. When i researched a little bit i stumbled on the phrase "resonant inductive coupling". The wikipedia page about this topic contains a picture that seems to show something similar to what i want to achieve. But i dont really understand how it works or how an oscillator circuit would work in that configuration.

https://en.wikipedia.org/wiki/Resonant_inductive_coupling#/media/File:Magnetic_phase_synchronus_coupling.svg

[mariush]

Can you have tiny holes in the wood base?   

I was thinking to have some kind of lens under the wood so that when you put a piece on the square, the piece will sit above at least one dot shaped lens.  You could have an infrared led under each square and have each square constantly send the unique code of the cell once a second or something like that.

Each piece could have an infrared sensor ... when you put the piece down on the square, the sensor picks up the infrared signal and if it's different than previous one, it could "broadcast" its new position, like for example using a very tiny fm transmitter...  doesn't need to have a lot of power, just a meter would be plenty, to be picked by a receiver under the board.

Alternatively, maybe have the infrared led in the bottom of the piece and the sensor in the board, and a magnetic switch or sensor in the piece , and a magnet on the square. When the piece sits on the board, the magnet is close enough to trigger the magnetic switch/sensor, which makes the piece pulse its unique id a few times and that gets picked up by the board.

As the piece only pulses the id when it's placed on the board, a tiny lithium rechargeable battery (like in ecigs) would last several games.

[Zero999]

Is it necessary?

Are you really worried about the user cheating?

[lp1234poha]

Thanks for your idea. I was also thinking about different possibilities to solve this issue.
But this approach seems much more complicated and expensive to me. Each piece would need a battery and some sort of logic device to handle the id pulsing. I liked the DGT-Approach because you only need a coil around a ferrite rod and a capacitor in each piece which would be much cheaper and you wouldnt need to recharge batteries in the pieces to keep playing . And it would look better if there are no holes for lenses in the squares.

[lp1234poha]

No im not worried about cheating. It is also possible to not recognize specific pieces but only if there is a piece or not and keeping track of the movement from a specific starting position. The downside of this is that you couldnt start from a different position without telling the software side and there are also cases like promotion of a piece that would be hard to handle for an approach like this.

[Zero999]

How about inductively powering all of the pieces and using infrared to send the code? Each piece could have a six pin micro controller which sends out a code using an infrared LED, which would be picked up by an infra red diode on the board.

[Kleinstein]

The matirx with send and receive coil could work with different pieces, e.g.  a piece of metal to block the coupling some ferromagentic (e.g. ferrite) core to increase the coupling or some resonant circuit to get a frequency specific coupling. With different resonant frequencies one could tell apart different pieces. This would be a kind of analog RFID of the simplest kind.

For very sharp resonances one could use a ferroresonance similar to some the theft protection stickers: magnetistriction excites a mechincal resonance, that can have a rather high Q (e.g. some 1000), which would be tricky with a simple RC element.

[lp1234poha]

The matirx with send and receive coil could work with different pieces, e.g.  a piece of metal to block the coupling some ferromagentic (e.g. ferrite) core to increase the coupling or some resonant circuit to get a frequency specific coupling. With different resonant frequencies one could tell apart different pieces. This would be a kind of analog RFID of the simplest kind.

For very sharp resonances one could use a ferroresonance similar to some the theft protection stickers: magnetistriction excites a mechincal resonance, that can have a rather high Q (e.g. some 1000), which would be tricky with a simple RC element.

Thanks for all your answers

Yes i think that the idea is to get a frequency specific coupling. I would be happy to achieve this with only a single transmit and receive coil as a first test. Do you know how to construct a circuit to achieve this? Or can you explain with a little bit more detail how this works?

Regarding the ferroresonance:
Is the working principle the same as with a LC resonant circuit? Googling this only led me to some very technical explanations.

[Kleinstein]

The ferro-resonance is a mechanical resonance, at little like with a quarz crystal, but with magentic couling. This is used quite a lot with anti theft tags as it can be cheap for the tags.
The tag contains a strip of suitable material (usually some amorphous alloy), some 2-3 mm wide, some 50 µm think and about 30 mm long. To turn the tags off there is also some 2nd / 3rd metal strip to create a DC magnetic field. These extra strips are not needed here. One may need an extra DC magnetic field (the earths field my be suffivcient, but not sure) to get a good response.

At a specific frequency (AFAIR in the 50 kHz range) there is a sharp resonance where the effective µr changes quite fast. AFAIK they are used by sending pulses of this frequency with a coil (or pair of coils) and just after the pulse listen for ringing at that frequency.

One can see the effect by having such a tag inside a coil in a slow sweep. The resonance is quite narrow (e.g. only a few 10 Hz) and thus easily missed.

[TimFox]

You want to avoid needing a battery in each piece, obviously.
Can you use a conventional bar code (minimal size) on the base of each piece with individual readers behind the squares?
Or conventional RFID tags with suitable readers behind the squares?
You only need 32 individual codes, but the queening of a pawn makes an interesting variation, as Alice learned.
Luckily, for Go stones, you only need two distinct codes (black/white) plus "empty".

[lp1234poha]

You want to avoid needing a battery in each piece, obviously.
Can you use a conventional bar code (minimal size) on the base of each piece with individual readers behind the squares?
Or conventional RFID tags with suitable readers behind the squares?
You only need 32 individual codes, but the queening of a pawn makes an interesting variation, as Alice learned.
Luckily, for Go stones, you only need two distinct codes (black/white) plus "empty".

I liked the approach of having only 16 coils in total to be able to scan all 64 squares since it uses a minimum amount of components. I thought about using rfid readers for each square an multiplexing the signals but it would get quite messy and expensive to have 64 rfid readers and multiplex the control signals to every single reader.  maybe it is possible to have an i2c bus an be able to adress all the readers but i discarded the idea because its to expensive. A form of transmission through leds or something similar (bar code) was discarded because the squares would need to be transparent or have some sort of viewing hole.

[TimFox]

You want to avoid needing a battery in each piece, obviously.
Can you use a conventional bar code (minimal size) on the base of each piece with individual readers behind the squares?
Or conventional RFID tags with suitable readers behind the squares?
You only need 32 individual codes, but the queening of a pawn makes an interesting variation, as Alice learned.
Luckily, for Go stones, you only need two distinct codes (black/white) plus "empty".

I liked the approach of having only 16 coils in total to be able to scan all 64 squares since it uses a minimum amount of components. I thought about using rfid readers for each square an multiplexing the signals but it would get quite messy and expensive to have 64 rfid readers and multiplex the control signals to every single reader.  maybe it is possible to have an i2c bus an be able to adress all the readers but i discarded the idea because its to expensive. A form of transmission through leds or something similar (bar code) was discarded because the squares would need to be transparent or have some sort of viewing hole.

Multiplexing is, indeed, a challenge.  However, if you use infrared LEDs and sensors, the squares could be opaque to visible radiation.  Something like the old-style Wratten #87 filter.
With an optical technique, could you use one (or more) area detectors (as in digital cameras) with a simple optical chain to image the squares onto it with some appropriate digital image processing?

[ataradov]

DGT boards are surprisingly expensive even when broken.

Here is a high-level overview of how they work https://www.chessprogramming.org/DGT_Board

I think the piece simply contains an inductor and a capacitor. The values are different for each piece. Soldering SMD inductors and capacitors back to back would make the piece part tiny. Something like they do for wireless LEDs (https://www.aliexpress.us/item/3256804431028018.html).

The amount of coupling to the receive coil would depend on how close to the transmit frequency to the resonance of the LC circuit. So, the scan involves trying all the frequencies in each transmit coil and sampling the response in each receive coil. The algorithm described there says "measureFrequency", but I don't think they actually measure the frequency. It is more likely that the set of frequencies is fixed and they measure the amplitude of the received signal.

They also mentions scanning 35 pieces, but I'm not sure why you would need to differentiate individual pawns.

It would take quite a bit of experimentation to tune all of that, so it may be easier to try that with a signal generator and scope at first. After that works, then it would be time to think about standalone circuit.

[BrianHG]

You only need 1 single powered inductive pulse.  This would be used to power-up / energize all the pieces on the board.  Within each piece, you need a tiny microcontroller, with nothing more than a cap to store the inductive pulse, which knows which piece it is.  Each piece ID will wait a specific time based on their ID before they respond to the global power pulse telling which piece it is and the grid of the inductive coils, most likely just a chess-board sized PCB with printed inductors below each square, will tell you where every piece is located.

Now, there are other ways to do this, but, I am trying to describe a solution here which can be designed to be guaranteed to work as well as designed with test smaller segments of a chess board, then, multiple ones tied together to build a complete chess board.

Note that the receiver and transmitters can also be an identical tiny PCB, one for the bottom of the chess piece with the mounted transceiver, and the one for the chess board.

[ataradov]

This is far more complicated than a simple passive LC circuit in each piece.

DGT boards have figured it out, just do what they did if you are doing it for personal use. Otherwise you would have to work around the patent.

[BrianHG]

This is far more complicated than a simple passive LC circuit in each piece.

DGT boards have figured it out, just do what they did if you are doing it for personal use. Otherwise you would have to work around the patent.

Yes, it is more complicated.  But it is a repetitive scheme which can be tested using an identical circuit for each chess piece and will not rely on specific mechanical characteristics of each chess piece and the precision of their placement which needs to be software interpreted for the DGT boards.  But, the DGT boards are far simpler.

Having a 8 pin PIC tied to an PCB printed inductor with schoctky diode and 0.1uf cap, at each space and each chess piece, with a simple code on the board tied with a chained RS232 port to report status can still be tested and done on a tiny scale with 5$ test PCB from China.  With the close distance of the 2 coils, the general IOs on the PIC should be enough to transmit power and transceive 1 byte of data.

If you need extra cheap, use the 3cent Padauk microcontroller.

[lp1234poha]

DGT boards are surprisingly expensive even when broken.

Here is a high-level overview of how they work https://www.chessprogramming.org/DGT_Board

I think the piece simply contains an inductor and a capacitor. The values are different for each piece. Soldering SMD inductors and capacitors back to back would make the piece part tiny. Something like they do for wireless LEDs (https://www.aliexpress.us/item/3256804431028018.html).

The amount of coupling to the receive coil would depend on how close to the transmit frequency to the resonance of the LC circuit. So, the scan involves trying all the frequencies in each transmit coil and sampling the response in each receive coil. The algorithm described there says "measureFrequency", but I don't think they actually measure the frequency. It is more likely that the set of frequencies is fixed and they measure the amplitude of the received signal.

They also mentions scanning 35 pieces, but I'm not sure why you would need to differentiate individual pawns.

It would take quite a bit of experimentation to tune all of that, so it may be easier to try that with a signal generator and scope at first. After that works, then it would be time to think about standalone circuit.

I already found this page but it doesnt seem to have much more information than what is described in the patent. Although there is a picture of the transmit and receive coils printed on a polyester film. But are you sure that there isnt another way of measuring the resonance frequency except for looking at the amplitude of the received signal of a set of frequencies applied to the transmit coil? The text of the patent states the following:

"The operation of a device according to the invention is as
follows. Control device 11 selects a transmit coil 18 and a
receive coil 19 (see FIG.3). Transmit coil 18 is subsequently
connected to the output of an amplifier arranged in control
device 11 and receive coil 19 is connected to the input
thereof. A resonance will occur in the circuit due to the
induction between transmit coil 18 and receive coil 19.
Because a playing piece 16 with a resonance coil 15 is
Situated above the interSecting part of transmit coil 18 and
receive coil 19, resonance coil 15 of piece 16 will influence
the above Stated resonance Such that a fixed resonance
frequency results."

It doesnt sound to me that the solution involves needing to apply different frequencies but that the circuit just oscillates at the frequency of the resonance coil but i might be wrong and the patent just isnt very clear about the exact process.

This is all just for personal use but i think it shouldnt matter since the patent is expired anyway.

[lp1234poha]

i had an idea of sending a pulse through the transmit coil with a single hard edge. The resonance coil starts oscillating in its specific frequency. Then just listening on the receive coil for oscillation transmitted through the resonance coil. Do you think an approach like this could work?

[AndyBeez]

RFID chips inside the chess pieces and a reader coil on each square?

[lp1234poha]

RFID chips inside the chess pieces and a reader coil on each square?

we already discussed this possibility but there are some downsides to that approach. You would either need a reader ic under each square and multiplex the control signals or have a single coil under each square and multiplex the signals to each coil. I think it would be quite a challenge to multiplex a signal to 64 individual squares. Thats why i think that the approach of only having 16 coils in total to scan all 64 squares is neat since it uses a minimum amount of components.

[magic]

Your patent refers to an earlier one which seems to contain more details on circuit operation.
https://patentimages.storage.googleapis.com/0b/c4/32/e7234e5464c7f2/US5188368.pdf

It doesnt sound to me that the solution involves needing to apply different frequencies but that the circuit just oscillates at the frequency of the resonance coil but i might be wrong and the patent just isnt very clear about the exact process.

I think you are reading it right.

[Terry Bites]

Understand it and then: Tilt your head to one side, raise an eyebrow, adopt a whimsical smile, say hmmm... and close that page.