some of the initially simpler methods still have a downside,labour.
the inductance method sounds really nice but it would require a fair bit =
of
calibration.
the resistance idea would require quite a bit less calibration but the yo=
u
need to worry about making proper contact, this applies with any contact
method.
RFID would be pretty effective I think but it presents a major problem in
reliably containing the scan field (try holding your wallet up to a prox
card reader with your card buried under 26 layers of cards and other crap
etc, most often it will still work. it penetrates pretty damn well, which=
,
in its normal situation is perfect.

Which leaves us with a lot of creative methods of contact id or IR.

Has anyone actually tried shielding the RFID stuff? or directing the fiel=
d
possibly?
I'd be interested to know how effective it could be done in a 'low cost'
manner.

Cheers,

JJ

-----Original Message-----
From: pic microcontroller discussion list
[mailto:PICLIST@MITVMA.MIT.EDU]On Behalf Of Robert Rolf
Sent: Tuesday, 9 March 2004 7:05 PM
To: PICLIST@MITVMA.MIT.EDU
Subject: Re: [OT:] Chessboard (sensor)


0xDEADBEEF wrote:
>
> Am Dienstag, 9. M=E4rz 2004 05:46 schrieben Sie:
> > So how would you handle the confusion of capturing apiece. For exampl=
e
> > you move a white bishop across the board to capture a black rook.  Yo=
u
> > remove the
> > black rook and set the white bishop in its place.
> >
> > Being a chess players and computer nut for years an engineering frien=
d
> > and I thought this out back in the days of the Commodore PET. For it =
to
> > go beyond having
> > entering the data manually. you would have to as a minimum have 12
> > unique peices identified. for the white side: pawn, rook, knight,
> > bishop, queen & king. and the
> > same for black side.  We did some experiments with the resister idea =
but
> > had some problems with the resistance of the conection.
>
> Well it's still the easiest and cheapest solution!!! If you choose the
> resistance of each resistor the way that even with small mistakes in th=
e
> connection resistance the controller could still detect it reliable!
> I think about building a potential divider with the chess figures as a =
par
of
> it. The controler then detects the potentials and dicides what figure i=
s
on
> the scanned field (therefore the potential must be within a given range=
,
that
> could surely be achieved even with these connection resistance problems=
!)
>
> >
> > What about these micro ID chips they  inject into dogs and other pets=
 to
> > ID them?
> >
>
> This might be the most elegant solution, but the most expensive, too.

What Jinx suggested, a resonant coil under each square
whose frequency is changed by the loading of the piece,
is elegant and inexpensive. If you used RFID you'd have the coils
and multiplexer anyway.

You could drill the bases of existing pieces and insert a
varying length of iron rod to get N discrete frequencies
corresponding to the type of piece. Or possibly various types
of metals in a couple of thicknesses. Their eddy current
decay profiles are all different (hence metal detectors
that can discriminate between brass and iron).

And the idea of using hall effect and having different
field strengths for the pieces is also cheap and simple.
Black would be one polarity. White the other.
This avoids the problem of contact resistance and oxidation.

One could also use an IR transparent base and a reflective
patch with discrete reflectances (not bar code although
a varying duty cycle grid (halftone) would give you something that
is easy to print).

What I don't understand is why you need to know the 'starting
position'. I thought all chess games start with the pieces
in the same position, so all you need to detect is which
squares have activity, assuming that the humans know the
movement rules for the pieces.

Do let us know what you discover works best.
My bet is hall effect with enough variance in the
magnet strengths that position error is within the
12 quantizing levels you'd need. Or IR.

Robert

--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.

--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.