Here's a photo of the work in progress. This is the underside of the table from my cylindrical grinder. I am mounting a ground steel strip, 3mm x 12mm x 680mm, which will carry a Renishaw RGS-20s optical scale strip. The scale strip is 6mm x 0.2mm with a 20 micron grating and read by a separate optical read head. The distance from the read head is 0.8mm (0.032") and this spacing (ride height) should have less than 0.1mm (0.004") variation. Getting this to fit and avoid interference is quite a hassle. The machine is from a pre-DRO era, so it's hard to find space for even a scale strip and separate read head. No way to get a classic enclosed scale in there.
To do this with my undersized mill requires many steps:
(1) Get the strip in place with an accuracy of about 0.1-0.2mm. That's what I am doing now, indexing off some of the machined faces of the table (I have checked these already for truth before removing the table from the machine).
(2) Lay the table upside down (as you see it) on my surface plate, which is long enough to support the entire length. Run a dial indicator up and down the length of the strip and see if I got it level within 0.1mm.
(3) If not, either add shims to or scrape the mounting pockets to get strip level to better than 0.1mm. (I could also selectively grind away the back of the strip where it rests on the mounting pockets but am not sure if that will telegraph to the other side of the strip. Does anyone know?)
(4) Test in place on the grinder (with an indicator tip running on the strip) to make sure it's parallel to the travel to better than 0.1mm
(5) Finally mount the 6mm x 0.2mm optical strip scale onto the 12 x 3mm ground strip shown in the photo. I'll have to make an application jig for this that references the scale position from the side of the table, because it needs to be laid straight within 0.1mm and it's self-adhesive so I only have one shot at it.
The ground metal strip will be held in place with epoxy in the pockets and epoxy and M5 screws on the two ends. I'll mount it (and eventually the scale) when the table is upside down on the surface plate and equalised in temperature. The steel has a coefficient of thermal expansion that's very close to cast iron so they should move together. After carrying out single point calibration (also called "slope correction") the scale/head combination should be accurate to within 3 microns (0.00012") over the entire travel and better than 1 micron (0.00004") over travels less than a few inches. Just for comparison, a five Celsius (nine Farenheit) increase in temperature will cause the scale assembly to increase in length by 30 microns (0.0012").