motion control

The problem then becomes one of dust and/or grease obscuring any track unless enclosed. Or wires and pulleys and rotating stuff on bearings etc

I'm starting to think about using concentric insulated metal cylinders and measuring the capacitance. Very similar to Bill's system. All they have to do is slide in and out, plus (and this is just a moderately educated guess...) they do not have to maintain a constant concentricity to maintain constant capacitance for a given length. As a side effect I might be able to use one of the tubes as a structural support.

They are capacitive, not magnetic, just as Paul wrote, using two multiple-plate electrodes with fairly wide spacing and interpolating to get the required resolution. The mass-market calipers use an ASIC to do most of the work. The original patents are from a Swiss company IIRC, and the above type of Chinese clones are slightly different from the Swiss and Mitutoyo designs.

If they are really moving at 1500mm/second you'll need thousands of readings per second to get 1mm accuracy.

Can you just use a simple optical linear quadrature pickup? Neither the 1mm nor the speed are anything like a problem. You could add an index channel to keep it pseudo-absolute rather than incremental.

Best regards, Spehro Pefhany

Google for wire potentiometers, they also come equiped with up/down pulse digital potentiometers,an can be put inside a save/clean housing. They come with a travel path of 10cm to 3 meters.

And cost

I dont think that you will find any solution for

The capacitative one looks good. I only need a resolution of around 1mm Interestingly, nobody makes the kind I'm thinking about that I can discover.

First, martin griffin suggested the LVDT. Second, IME an LVDT is a "cheap" part

Because I looked up manufacturers of LVDTs and saw prices of $60+, without signal conditioning. Anyway, I now think I can see a way to include position measurement into the structure of a linear motor, using the same uP controller that does the control loop.

You weren't asking the OP - "Ebi" - but Dirk Bruere at Neopax, who is a completely different person in a completely different country, asking for a solution to a rather different problem.

Do pay attention ...

Any old "ball" mouse should be fine for ~1mm accuracy.... re-mount and interface to the wheels and IR LED interface on the platform of choice, and hey presto... a position sensor. PIC should easily cope with reading. The old serial (D9) mice possibly better, since more 'hobbyist' info available. Very cheap and simple. With a simple driver / rx chip, you could skip mouse interface IC , and go from PIC direct to LEDs yourself.

Dave M.

plate

so

solution to

other

with

which

Zygo

go.

which

"cheap"

An optical mouse might be an option. However, I'm a bit wary of using optics in such a potentially dirty and inaccessible environment. BTW, what sort of surface is best for an optical mouse?