Beware, not CCS: brass proximity sensor

[PrinceNai]

Dear All,

My friend and I got a task to find out how to detect a brass object (the object in question is a 3,5cm x 4cm brass cylinder, namely dolly marker for a roulette). Our best guess was a 10x10cm planar antenna, Colpitts oscillator and to measure the change of frequency when this object rests on the antenna. Our problem here is that brass doesn't cause much of a change of frequency, maybe 5%, and it is hard to insure that the "base" frequency will stay in this range. Of course we could measure the rate of change over time. Does anybody have a better idea how to do it? Rfid is another option.

Mind you, I'm fully aware that this is not a CCS related question, even if the final product will be coded in it. But you guys are a fountain of knowledge, so...

Best, Samo

[Mike Walne]

The early metal detectors I played with had two oscillators operating at close to the same frequency.
One used a large coil (your 10*10) and the other a compact coil.
Any change in one oscillator was detected as a change in beat frequency between the two oscillators.
The beat frequency was in the audio range and should be easy to measure with a CCS coded PIC.

Mike

[temtronic]

I looked up 'dolly' and most, if not all, are made of acrylic so I assume your brass one is not for 'official' gaming.
Now if you have control over the playing surface, it'd be very easy to place a 'sensor mat' under the table that could detect where the dolly if located. since the squares are rather large ,it makes detection easy. Any number of 'proximity' detection methods will work.
You could for instance have a series of 1/4" wide copper strips laid out 'x' and more as 'y'( or 'rows and columns'). send a signal out the X, rcv via the Y. The brass dolly will change the signal strength. Easy for the PIC to send and decode.
just one idea for you...
a lot depends on what you're allowed to do to the table.

[PrinceNai]

Interesting, where wrong assumptions can lead you. In my 20 years working in a casino I never saw an acrylic dolly, we use metallic ones, so I thought everybody does :-). I guess that rules out any metal detectors because of future compatibility. The problem we are trying to solve is very simple. Speed of the game. We do not need to detect the dolly on every number, just on one predetermined spot on a table where the croupier leaves the dolly while the game is in progress. Every time the dolly leaves that spot, we count one game.

Right now we have a tracking system that doesn't measure the speed of the game, it uses estimates eg. "slow", "normal", "fast" and the supervisor must change that on the fly based on his feeling. Given the fact that we track the players and that their volume of play is "bet x number of events/h x time of play" any mistake in "number of events/h" directly relates to wrong tracking results for the players. The problem with that is that all the agents are paid based on those results and we don't have a clue if we are overpaying or underpaying them and how much.

Regards, Samo

[temtronic]

Well that simplifies it a whole lot! I'd thought you needed to detect the winning 'square' (where the dolly goes ?). A simple, single channel 'proximity' detector is all you need. Any PIC would work, though one with a capacitive module makes coding easier. Think of an elevator push button, same principle.

[PrinceNai]

Detecting a winning square this way would be a real challenge, because of the presence of chips on that particular square (higher denomination chips are also metallic). Every "normal" system gets all the info info from the wheel (winning number, speed), whereas ours is as stupid as you can get. Floating in void and waiting for an input from users. We all know how accurate that is. Give the user a foolproof system and only fools know how to use it.

Back to the original question. Capacitive proximity sensor would work. Any hints on how to make this capacitor large, let's say 100x100mm? This dimension is ideal because of the cheap Chinese PCB's and it is also big enough for a croupier not to miss the target. I've been searching and so far I only got capacitive touch sensors and designs.

BTW, we are able to do whatever we want with the table. Sink the sensor into the wood, drill the holes, whatever. It gets covered with the felt, so it doesn't matter.

[temtronic]

You just need to R&D the sensor. Having a solid brass dolly means a lot of 'metal mass', esp. compared to the chips. If the squares are 100x100, consider making the sensor 50x50, probably a little bigger than the footprint of the dolly.
Any PIC with the capacitive sensor peripheral is where I'd start. I don't know what the 'matrix' of a roulette board is but it's not that hard to create the code, just think of it as a keypad of X by Y, or several smaller keypads.

Jay

[Gabriel]

Magnets are not an option?
_________________
CCS PCM 5.078 & CCS PCH 5.093

[PrinceNai]

Hi,

Magnets on a gaming table, no :-). Once again, I only have to read only one spot on a table, not the whole board. So far the two options are rfid and capacitive sensor. Rfid would be trivial to detect, but a modification of a marker would be necessary to incorporate a tag into it. That is why we are exploring the capacitive way. So far I was unable to find out exactly how the capacitive sensor looks like. Is it possible that it is only a copper plate with one wire leading to it? Microchips AN1478 suggests so :-)

Regards, Samo

[PCM programmer]

Omron E2EY-X8C1. A Proximity Sensor for Aluminum, Brass and Other
Non-ferrous Metals. Iron Is Not Detected. (8 mm range).
$284.40 USD

Data sheet:
http://www.mouser.com/ds/2/307/e2ey_ds_csm451-12428.pdf

Link:
https://www.mouser.com/Search/Refine.aspx?Keyword=E2EY-X8C1

[temtronic]

PIC16F1829, as an example PIC with CPS. Also has UART and a lot of other 'sturr'.. Section 27 is the CaPacitive Sensor section. I assume Microchip have several application notes on the design of sensors though I'd just grab a 50by50 section of dbl sided PCB, add a couple wires, breadboard a PIC and 2 LEDs, add ICSP and start coding....
There's no reason why a $2 PIC can't do what that $300 sensor (and PIC) can do for your project.
The key to your project is to get reliable 'there-not there' operation, so grab some parts and start !