Beruflich Dokumente
Kultur Dokumente
However, these should both be giving the sameresult for the length. As shown above, we can
determine the fringe fractions and we should already know the calibrated values for the
wavelengths,so the two items we do not know are the fringe orders.
At first glance, it seemswe have not gained anything despitehaving made two measurements.
Each set of solutions (one for the first wavelength,one for the second)allows us to make any
choice we want for the fringe orders and so we still end up with severalanswersclose to the
expectedsize of the gaugeblock, from which we haveto pick the right one.
Gauge block interferometry page 2118Q
However, things are considerably improved if one examines both sets of solutions
simultaneously.
To illustrate this, imagine the measurementof a really short gaugeblock, say about 3 pm long
(not a standardsize!). We measureit in the interferometerand by chancewe get fringe fraction
readingsof exactly zero, for both wavelengths(this is quite possibleand in fact a quite common
occurrence).
I
Figure 13 - plot ofboth setsofsolutions
The solutions for the first wavelength are plotted along the top row, those for the second
wavelength are below. Notice how some pairs of solutions (top and bottom) line up, whereas
othersdo not. Thosewhich line up are markedwith an arrow. In fact we can discardany solutions
where there is not a match as both measurements(two wavelengths)should agreeon the same
size of the gaugeblock. Now, if the gaugeis of high quality (e.g. gradeK) we expectits length to
be close to nominal. For a 3 pm gaugeblock, using a first wavelengthof 0.6 pm (0.3 pm fringe
spacing),we shouldexpectan answercloseto fringe order 10 (10 x 0.3 pm: 3 pm) - and this is
'-
exactly what we find. The solution at order 10 for the first wavelengthmatchesthe solution at
order 4 for the secondwavelength.This must be the right solution.Generallyspeaking,it is!
If you think that Figure 13 is reminiscentof what one would see on a slide rule or perhapsa
vernier micrometer,then you are right! It's the sameprocessat work- alignment of two scales
with slightly different spacings.Except that on a vender, you only get one solution as the lower
scale (e.g. the 1/10 mm scale) is not very long and there is only one matching solution visible.
The key with gaugeblock interferometry, is making sure you pick the right matching solution.
v
In general , the more wavelengths (or the more accurately one can measruethe fringe fractions)
the easierit is to detectthe correct solution. With the Hilger interferometer,the wavelengthsare
not as stable as those of a laser and the measurementof the fringe fraction by eye is not so
accurate.This is why 3 or 4 or even 5 wavelengthsare neededto ensurethe correct solution is
picked. In the NPL-TESA and FLaP interferometers, the fringe fraction is measured
automatically with better accuracy than by eye and the wavelengths of the lasers are very stable
and well known (if calibrated frequently) and so only two wavelengthsare needed.
In Figure 13, severalpossible solutions are highlighted by arows. In fact there are an infinite
numberof suchsolutions,equi-spaced.It is thereforenecessaryto know the approximatelysize of
the gaugeblock before using the method of exact fractions.The required accuracyof this prior
determination is dependenton the spacing of the set of possible solutions. If, within your
estimatedboundsfor the actualgaugeblock size,thereare two possiblesolutions,then thereis no
ly'
way of choosing between them and you are stuck! For the example shown in Figure 13, the
possible solutions (arrowed) are spacedapart by five fringe orders of the first wavelength.The
fringe spacing is half the wavelength, or 0.6 pm / 2, i.e. 3 pm. So the solutions are spaced
5 x 3 pm or 15 pm apart. So you needto know the approximatesize of the gaugeblock to within
15 pm (i.e. to within +7.5 pm), otherwiseits length is not uniquely determinedusing this method.
,,, !, wavelengtl4xwavelength, I
soluttonspacng=-
- x-
wcmelengtlt, -wavelength, 2
oaqe 23180
O.0lgP*0'543P-- I
solutionspacing-- X_
O.enpm_ 0.543pm 2