Testing for Parallelism and Flatness
In theory, if an optical flat is in perfect contact with a perfectly flat anvil, then no light bands will be visible. Make sure the optical flat and anvil are clean. Attempts to wring the optical flat will only scratch the glass. Get a good fit by gently squeezing the optical flat onto the surface. When finished, lift off without sliding. If the light bands (rainbows) you see are evenly spaced and in straight lines, then your surface is flat.
Ideally you'd use monochromatic light and if you're doing calibration full time, it's probably a good idea to invest in such a light bulb. Regular room lighting works fine for us. When looking through the optical flat, look straight down: avoid looking at an angle. Check this out for yourself and you'll see that the image changes dramatically as you increase the angle of vision.
If you see many light bands, then press the optical flat a little harder. You'll probably see fewer bands and that makes it easier to interpret the results.
The degree to which the light bands arch can be used to calculate the flatness. The ideal micrometer anvil is flat to .000012".
What this means is that, when you're looking at the arches, the top of one arch just touches the bottom of the next arch. See how the imaginary line indicates the bottom of the next arch in the image on the left?
one light band (left) vs. two light bands (right)
At this point you have a flatness error of one light band, or .000012". This is exactly what you want. If, on the other hand, the bottom of the light band touches the top of the second arch over, as in the image on the right, then you have a flatness error of 2 light bands, or .000024" and it's high time to have your anvils lapped.
Performing this test on each anvil will determine the degree of flatness of each anvil; but, by using a flat which has parallel sides you can close the micrometer anvils on the flat and also determine the degree of parallelism.
Mitutoyo suggests this procedure: wring the optical parallel to the micrometer anvil (the stationary part of the micrometer) so that only one interference fringe (light band) shows. Now close the micrometer spindle onto the parallel. This should occur exactly at .500" when using the optical parallels in our calibration kit. At this point count the number of fringes (light bands) on the spindle by looking through the optical parallel from the other side. Then apply this formula:
Number of fringes on spindle side x 0.32µm = parallelism of the anvils with the spindle in that position
For example: 3 fringes x 0.32µm = 0.96µm which is the ideal parallelism of a 0-1" range micrometer's anvils.
You may want to convert this metric result to inches using a scientific calculator or your own gray matter (equivalent to about .00004"). Since the optical flats themselves are parallel to .00002", you'll have to take this possible deviation into account. Your result would be expressed as .00004" ± .00002"
It's important to take note of the phrase: with the spindle in that position. If you rotate the spindle a bit, the surfaces may no longer be parallel. For that reason the calibration set shown above includes two parallels. Perform the same test using the second parallel. Now your reading will occur at .5125" instead of .500" This puts the spindle at 180° from the first reading. If the anvils still are parallel, then you're set to go. If the anvils are now out of parallel then the spindle isn't running true and we're in trouble. A real stickler for details would even use 4 optical parallels to measure every 90 degrees, but for our purposes that may be going a bit too far. A qualified calibration lab can perform that procedure for you if the need arises.
A somewhat easier method for checking parallelism requires the use of a gage ball. Any diameter under 1" will do. Close the micrometer onto the gage ball and take the reading. Do this in 5 different places on the surface of the anvils. If the anvils are parallel, then the readings will all be the same. It proves parallelism but doesn't actually give you a numeric value. This method can also locate high spots or dips on the anvil surface, which should lead you to have them serviced and lapped.
Parallelism on larger micrometers
Using gage balls as described above will work fine. You can use a gage ball larger than 1" for this purpose, or you can use a gage ball in conjunction with a certified gage block, although this will be a tricky procedure if you're normally "all thumbs."
You can use optical parallels instead. You must have optical flats with parallel sides. The ones in your micrometer checking set are parallel to .00002" but these optical flats are only good for checking parallelism on the 0-1" range micrometers. Larger ranges need larger and very expensive optical parallels (upwards of $1000). It will be more cost effective to have a calibration lab check these for you. Without investing in more expensive equipment, you may have to resort to the gage ball technique mentioned above.
Checking the Zero Setting
When the spindle is screwed closed on a 0-1" micrometer, the reading should be zero. Use the spindle ratchet to obtain a light, even pressure, if your micrometer has one. If the zero is slightly off it can be adjusted by turning the barrel into position. A special wrench is usually provided for this procedure. On micrometers with ranges above 1" you will have to insert a gage block or micrometer standard equal to the lower value of the micrometer's range and set the zero as above.
Calibrating the Micrometer
Use micrometer standards or gage blocks for this procedure. Be certain that the blocks are properly wrung and take special care with carbide tipped anvils so that you don't damage the gage blocks. The micrometer is calibrated at several points throughout its range. Arbitrary readings are considered better than evenly spaced dimensions. The "lead" error will be the difference, plus or minus, between the actual and the observed readings. Lead errors should not exceed .0001" or possibly .0002". If errors are found, keep track of them and you can always add or subtract the lead error when you use the micrometer at that particular range.
Indicating (dial) micrometers pose other problems. Since the micrometer is not used to make direct measurements (it is a comparator) and, since the anvil is movable, we can not calibrate the spindle using gage blocks or micrometer standards. Of importance here is the repeatability of the indicating pointer, which should be less than one-half graduation. Close the spindle and lock it in place. Now check for the pointer's ability at repetition. The accuracy of the indicating mechanism is then verified by sequentially inserting gage blocks with a difference of .001" to verify that the pointer registers the correct reading. Ultimately, flatness and parallelism are of paramount importance (see notes above).
Refer to specifications for Etalon indicating micrometers.