> The other totally crazy idea I have is to take
> a clear plastic rod a meter long or so and wrap it in paper or
> aluminum foil except for a slot along one side.  The detector would go
> at one end and hopefully, any light entering at right angles to the
> rod would bounce around and come out the end.  Is this even possible?

Martin, the use of a lens in front of the receiver can works somehow as
a parallel beam filter, since if you position the sensor right in the
focus point from the direct IR rays.  The IR sensor is around 2 ro 3 sq
millimeter, while a small 20mm diameter lens is collecting and focusing
over the sensor.

Check this out:

A 3mm diameter sensor has only 7 square milimeters of area to receive
the IR.

A 20mm diameter lens concentrating focus over the sensor, has 314 square
millimeters of reception area, so your sensor would be 44.8 times more
sensitive.

A 30mm diameter lens increases the sensitivity in 100 times.

More important than everything is to colimate the transmit rays as much
parallel as possible, a narrow beam. If all the transmitter IR is
spotted over the receiver lens, you have 100% of energy transfered.
However, if the beam opens very few degrees and it projects a disc (at
the receiver plane) bigger than the receiver lens, so it will receive
less energy, the equation (aproximate) is this:

                ET
ER  =   ---------------------
         ( PDD / RLD ) ^ 2

ER = Energy Received
ET = Energy Transmitted
PDD = Projected Disk Diameter
RLD = Receiver Lens Diameter

better formula (I think):

        ET * (RLD/2)^2 * 3.14159
ER =    ----------------------------
          (PDD/2)^2 * 3.14159

that is the same as:

       ET * RLD * RLD
ER =  ---------------
         PDD * PDD


So, as much you could concentrate the IR beam, further you can receive
it, as much you could increase the size of the receiver lens, bigger the
sensitivity.

Hope this can help.
Wagner