Re: [CAD_CAM_EDM_DRO] Re: DRO linear scales discussion

Ken Jenkins wrote:

> As Doug Fortune pointed out to me, and some others as well,
> the real key once a reasonable resolution is reached in a
> film output scale, is having a HEDS that will match your
> particular scale proportion/resolution (and I would add
> a means of mounting the film scale with some sort of backing
> which makes it "rigid enough" to read reliably) . So since no
> question is too stupid to ask here, can that be done ....
> I mean fabricating a HEDS unit to some exact tolerance or
> altering an existing one in some way ... without buying my
> own silicon fabrication facilities which would be just outside
> the budget (as I so laughingly call it) for this hobby.
>
> WHAT WE'RE AFTER HERE IS DIRT CHEAP DRO'S WITH 0.0005
> READABILTY ... (just to keep the goal in mind).

What you do is take a piece of the scale and place it in a holder
that allows you to tilt it a few degrees from parallel to the main
scale. You put two LEDs on one side, and two photodiodes
on the other side at some separation. When the tilt is adjusted,
you will find a setting that causes bands of light and dark to sweep
across the sensors such that when one sensor is at 50% brightness,
the other is at either 0 or 100%. At that tilt, you have obtained
a quadrature (90 degrees) phase shift.

The real problems are three, I guess. One is that the films are not
terribly accurate over temperature and humidity changes. You might
be able to glue the film to a glass strip with the emulsion out, using
an optically clear adhesive. You would have to make sure the
adhesive did not have bubbles.

Two is that the film is fragile, so that you would have to keep the analyzer

strip only a few thousandths of an inch from the scale strip, but they must
never touch. This is pretty much how commercial glass scales are
built, but the metal pattern on them is a bit tougher.

Three is that the thickness of the emulsion may cause undesirable
effects on the optics. This gets worse as the stripes get smaller.
.0005" resolution requires only 500 stripes/inch, so theoretically a
1000 DPI printer could do it. A 2000 DPI laser typesetter or photoplotter
would be better. I suspect having a film made by one of the PCB
houses would only cost $50 or so, even at this resolution. That film
could then be contact printed to other litho films to make as many as
you wanted.

Jon