[getting old punched cards read]

<True. If you write the program to keep careful time (and reject cards wi <fewer than 80 chars), it _might_work. There is, of course, a brute-forc <approach - 960 phototransistors. Additionally, one could build a 4 dpi <scanner (12 pixels and a stepper motor). With the steppper from a 5.25" <drive and either a mechanical amplifier (lever) to pull the 7" stroke or <either a) multiple 12-bit sensors or b) one 80-bit sensor w/3" stroke, <it wouldn't be that hard to make, not that I have the time to make a <contraption of that complexity. Ah gads you guys do it the hard way. Take a stepper from something and put a roller on it with a diameter such that one step will move the card forward one column. Then all you need is one rows worth of phototransistors. I forget the punched card orgainization but a parallel port (printer) off a PC should be enough bits. For hand pulled it's a row of phototransistors for reading the column and one phototransistor for each row. V--to read column =============================================================guide rail + + + + +++++++++++++++++++++++++++ <<<<<<<direction of motion (96 needed) ^ + | + / + | + |=============================================================guide rail ^optos to read the leading edge of the card for column clock, spacing is such that each one is obscured as the column is over the column leds making it self indexing. It's possible to go very fast. Oh the only difference with hole vs mark sense is transmittance or reflective sensing. Allison