Improved laser digitizer accuracy

[Karl_T]

I bought an Omron laser displacement sensor with plans to make a laser digitizer or poor man's CMM on my mill with Camsoft control. This sensor outputs the very common 4 - 20 milliamp standard. Galil boards accept 0 - 10 volt for their analog input.

The first suggestion was to just use a 500 ohm resistor in parallel with the Galil input to generate the voltage. This works but was plagued with electrical noise of up to 0.03 volts. This unit measures a range of about 700 thou, the noise converts to an accuracy of +/- 2.5 thou for an individual reading of a static sample

So, I talked to my friend, Don Foreman, that is a retired electrical engineer. He suggested, correctly, that a lot of noise could be reduced by isolating everything. Doing this improved the statistical accuracy of a static sample to +/- 1.25 thou.

Don offered to build a custom circuit board to convert the 4 - 20 milliamp signal to the 0 - 10 volt for the Galil input. I just installed this board and tested the statistical accuracy of a static sample at +/- 0.5 thou. The unit takes 60 milisec. to get a reading. Plots of 500 consecutive samples shows that the data variation is random. This is important because it means consecutive readings can be averaged to further improve accuracy. Staying over an individual point for 500 milisec and averaging the results will give a statistical accuracy of 2 tenths. That's good enough, even for an anal engineer.

Of course, the rest of the machine can't hold this tolerance. Table flatness, backlash in ball screws, sample preparation, etc. will determine the ultimate accuracy of using the laser displacement sensor for digitizing.

I'm attaching a complete set of instructions on how to build Don Foreman's magic converter unit. One can be built for under $50 and can be used to convert anything that outputs 4 - 20 milliamp to 0 - 10 volt.

Karl