41f1a623   
   From: lugnut808@yahoo.com   
      
   nickec wrote:   
      
   > I have made a simple tachometer using old roller ball mouse parts.  I   
   > record the output using a soundcard.  Works well.   
   >   
   > What I want to do now is print a ribbon, attach it to the rim of a   
   > disk, and point a laser mouse at the vertical lines running around the   
   > rim.   
   >   
   > Basically I hope to sample the speed of the disk as often as   
   > practical.   
   >   
   > I have been thus far stymied in finding out about the data stream of   
   > the mouse - both inside the mouse (DSP chip), and external to the   
   > mouse (USB).   
   >   
   > Anybody got any ideas?   
      
   Go to the Avago website, there are datasheets for the chips there.  Avago   
   was formerly the Agilent Semiconductor division, which was formerly part of   
   HP before they specialised in ink cartridge and their accessories.   
      
   Unfortunately, the Avago mouse chips seem to be not quite what one would   
   like for a rotary or linear encoder.  The chip does not know that the   
   pattern of lines is repeating and regularly spaced.  If you make a large   
   movement, it does not count the lines and then multiply by the line   
   spacing, but instead it measures how far each line moves past, and then   
   adds up these movements, getting the result somewhat wrong.  Remember, you   
   never got to tell it what the spacing of the lines was in the first place.   
   If the lens is a little bit closer to the pattern, the image will appear to   
   move faster because of the greater magnification, and the result will be a   
   larger apparent movement.  There are graphs of that effect in the   
   datasheet.   
      
   You could just use the mouse chip as a camera (image sensor) since it is   
   possible to read images out of the chip in a debug mode, but then you would   
   have to do the processing in another microcontroller or FPGA, with a more   
   suitable algorithm that knows that the pattern is regularly spaced.  I   
   suspect that the main problem with this approach is speed: I am not sure   
   that it is possible to read data out of the mouse chip at the full frame   
   rate, and even if it is possible, the external image processing chip will   
   need to be pretty fast at processing the data if you want it to work as   
   fast as it does in the mouse.   
      
   A more attractive option would be for someone to reverse engineer the   
   firmware of the Avago mouse chips.  It is possible to upload new firmware   
   to these mouse chips, according to the Avago website.  If the hardware is   
   flexible enough, then someone could re-write the firmware so that the chip   
   understands that the pattern you will be putting in front of it is   
   regularly spaced, so it can work out large distances much more exactly by   
   counting the lines, and only interpolate the fine movement between the   
   lines.  This way it would be a useful encoder.   
      
   These devices would be particularly useful as 2-D encoders, because it is   
   relatively easy to buy a 1-D rotary or linear encoder, but then if you want   
   2-D you need two 1-D encoders as well as a way of making them move at   
   exactly 90 degrees (or another known angle).  If you started out with a 2-D   
   pattern, then there would be no further mechanical linkages with associated   
   backlash, angle errors etc.  This could be used for example on the X-Y   
   table of a hobby CNC mill.  You could also make a two dimensional linear   
   motor stage with neodymium magnets and supported by aerostatic bearings,   
   and the mouse chip would be a nice feedback encoder (maybe not quite as   
   fast as one would like, but very convenient).   
      
   Chris   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)