...

U.

S. DEPARTMENT OP COMMERCE

NATIONAL BUREAU OP STANDARDS

RESEARCH PAPER RP1708

Part of Journal of Research of the Nationa.l Bureau of Sta.ndards, Volume 36,

April 1946

A GLOSSMETER FOR SMOOTHNESS COMPARISONS OF

MACHINE-FINISHED SURFACES
By Richard S. Hunter
ABSTRACT
Ai!, shininess is one indication of surface smoothness, a photoelectric glossmeter
was developed for possible use as a production-inspection device for evaluating
the roughness of fairly coarse machine-finished surfaces. A near-grazing angle,
75 degrees, was chosen for the measurement of gloss. Therefore, specimens of
the type having narrow ridges of metal between adjacent tool cuts are rated low
in gloss, or rough. Alt hough the glossmet er essent ially measures the fraction of
the unshadowed surface that is nearly parallel t o mean surface, t he instrument
r esulting from the present development has proved t o be a simple and useful
device for making rapid comparisons of the roughness of surfaces m achined wit h
about the same feeds.

I.
II.
III.
IV.
V.

CONTENTS

Pagf

Introduction _____ ____ ____ ___ __ __ __ ___ ____ _________ ____ __ __ ____ __
The glossmet er ____ _______ ______ ______ ____ ______ __ ______ __ ______ _
Use of the glossmet eL __ __ _ _________ __ __ ____ ____ ____ __ ___ ____ ___ _
Comparisons of glossmeter readin gs with root-mean-square deviations
from m ean surface _________ ________ _____ __ _____ _______ _________
Summary _______ ___ ________ _______________ ____ ____ ____ _____ ____

385
386
387
390
391

1. INTRODUCTION

Recently the National Bureau of Standards was asked by the War

Department to suggest a rapid and simple production-inspection
method for comparing the roughn ess of somewhat similar machinefinished surfaces. The roughness of the surfaces ranged between 100
and 500 microinches root-mean-square deviation from mean surface.
The method was needed for identifying the rougher surfaces in this
range that could not be rendered smooth by the application of single
sprayed films of paint.
As shininess is one indication of surface smoothness, a photoelectric
glossmeter was developed and tried for the purpose. A photoelectric
reading of the gloss of a surface can be made quickly with this apparatus, which is inexpensive and do es not damage the surface.
B ecause the roughness involved is about one-thousand times as large
as that which may affect the gloss of surfaces, every test surface has
to be coated before gloss measurement with a liquid that will fill its
microscopic cracks and pores. Without such a coating, it is impos385

386

Journal of Research of the National Bureau of Standards

sible to separate the gloss effects of microscopic irregularities from

those of larger irregularities. The choice of a liquid for the purpose
is described below.
II. THE GLOSSMETER

In planning the glossmeter, a near-grazing angle of specular reflection, 75 degrees, was chosen for two reasons: (1) low-gloss readings
(indicative of coarseness) would thereby be given to nny specimens
having narrow ridges of metal extending appreciably above mean surface ,~ and (2) the specular reflectance of the required coating liquid is
much higher at this angle than the specular reflectance of the underlying metal surface covered with liquid; therefore, variation of the
gloss of the underlying surface has relatively little effect on glossmeter
reading.
In choosing the parts for the apparatus illustrated in figure 1, a
barrier-layer photocell was selected because of the simplicity of the
electrical equipment required with it. A standard five-cell flashlight
lamp, which possesses a small concentrated filament, was chosen for
the light source. A %-inch lens having a focal length of 4 em was
selected because it provides a narrow beam of high flux density.

1.- Vertical cross section of 75-degree gloss meter showing lens (L), specimen
in posi tion f or measurement (S) , window to photocell chamber (W), and barrierlayer photocell (P ).

F IGU RE

In the glossmeter shown in figure 1, the converging beam from the

lens is directed onto a machined surface at right angles to the tool
marks. The light specularly reflected by the surface proceeds toward
the photocell chamber. The small window at the entrance to this
chamber admits only rays of reflected light whose average deviation
from the direction of mirror reflection does not exceed 6.5 degrees.
The interior of the chamber is coated white to conserve the light that
enters.
.
The body of the instrument is built chiefly of duraluminum. With-

Glo8smeter for Surface Finish

387

out handle, it is 25 by 8 by 4 em in size. The grip handle makes the

unit easy to position on specimens under test.
The control case (fig. 2) that accompanies the glossmeter contains
a pointer-type galvanometer for registering photocell current, a 10ohm rheostat for controlling the lamp current, a switch, and a small
110- to 6-volt transformer. As shown in figure 2, room is provided in
the case for the glossmeter, connecting cable, and a small bottle of the
liquid used to C'oat test surfaces.
The new glossmeter was designed primarily for measuring surfaces
machined by lathe. The beam of light was restricted to small diameter so that test surfaces that are parts of cylinders or cones can be
measured. As shown in figure 1, the plane of the test area must be
placed in the plane of the bottom edges of the glossmeter. With
cylindrical or conical test specimens of like size and shape, this placement may be conveniently obtained by attaching templates of the
proper design to the sides of the glossmeter (see fig. 2).
III. USE OF THE GLOSSMETER

Values of 75-degree specular reflectance are proportional to

deflections of the galvanometer. Polished black glass is a convenient standard to use with the instrument. The 75-degree specular
reflectance of the average piece of black glass is 26 percent. 1
Liquid-coated steel surfaces of the type being studied were found
to reflect from 2 to about 25 percent of the incident beam into the
photocell chamber. In using the glossmeter, an operator first places
it on a clean, polished black-glass standard and adjusts the rheostat
till the galvanometer registers a 26-division deflection. When this
adjustment is made, values of specular gloss are read directly as
deflections of tbe galvanometer.
The specular reflectances of 10 steel specimens machined in 8
different shops on lathes using feeds of from 0.017 to 0.025 inch were
measured with the glossmeter. The root-mean-square roughness
values of the same specimens were measured 2 with a Profilometer.8
The surface of each test specimen was measured with both instruments at four selected areas. The results of these measurements
are compared graphically in figures 3 and 4.
It proved difficult to find a coating liquid that would fill microscopic cracks and pores, without filling the larger cavities. The
liquid necessarily bad to have low surface tension and low viscosity.
Kerosine and Stoddard solvent were first tried. A 90-second
waiting period was required after Stoddard solvent was brushed
generously onto a specimen so that the excess liquid could drain off
and leave a film conforming to the major surface irregularities. As
shown in figure 3, good separation of the machined surfaces according
to their roughness was obtained with Stoddard solvent. Films of
kerosine also performed satisfactorily, but a draining period longer
than 90 seconds was required.
The draining period required after an application of Stoddard
solvent or kerosine is objectionable in a method for production inspecI D. G. Moore and R. S. Hunter, Use of liquid surafces as standards of specular gloss,l. Am. Ceram.
Soc. U, 167 (1941).
J By C. E. Haven, of the National Bureau of Standards Gage Section.
'E. 1. Abbot, S. Bousky, D. E. Williamson, The Profilometer, Mech. Eng. 10, 2M (lg38).

388 J ournaZ of Researoh of the National Bureau of Standards

25
0
0

20

OJ

\
\

01

15

,,
\

Q)

Q)

,0

0
Q)

a::

III
III

0
0

10

0., 0
o "o ~ -..00

.9

(!)

..... ......

100

200

300

400

trms (microinch)
3.-Relation between values of roughness measured with the Profilometer
and those of gloss measured with the Stoddard solvent as the coating liquid.

FIGURE

Measurements were made on steel specimens machined in 10 different shops. Dotted line represents relation
computed for sine'shaped prollle.

389

Glossmeter for Surface F inish

25

\
\
\
0

\
\
\
o \

20

\
\0

0
01

\ 0

15

"0

Q>

a::

...

.!
Q>
E

'"0'"

10

o
100

200

300

400

hrms (microinch)
4.-Relation between root-me an-squ are roughness measured with the Profilometer and gloss measured with mixture of light mineral oil and carbon tetrachloride as the coating liquids.

FIGURE

Data were obtained from tbe same test areas as tbose plotted in figure 3. Dotted line represents relatioll
computed for sineshaped profile.

390

Journal of Researohof the National Bureau of Standards

tion. In addition, there is only a brief period after draining in which

the gloss values of such coated surfaces remain stable. Thus, about
30 to 45 seconds after the excess solvent drains to a thin film conforming to a machined surface, this film starts to disappear because of
evaporation. The problem of timing gloss measurements when using
these liquids is further complicated if specimens are warm, as is often
the case after machining operations. Rates of evaporation vary with
temperature, and draining schedules vary correspondingly.
Because of the obj ections to kerosine and Stoddard solvent, mixtures
containing a volatile liquid and a nonvolatile liquid were tried. Light
mineral oil was used in these mixtures to provide nonevaporating films;
volatile solvents, such as carbon tetrachloride and methyl chloride,
were used to give the mixtures low viscosity and therefore rapid
spreading action. A mixture of roughly 10 parts of volatile liquid to
1 part of light mineral oil was found to give the best results.
The methyl-chloride mixture furnishes a film suitable for measurement 5 seconds after it is applied to a machined surface; the carbontetrachloride mixture furnishes a suitable film 20 seconds after application. The liquid film remaining on the machined surface is nonvolatile. Thus, if not damaged by contact with other objects, or by dust
and dirt accumulation, these oil-coated surfaces remain suitable for
gloss measurement for many hours.

IV. COMPARISONS OF GLOSSMETER READINGS WITH

ROOT-MEAN-SQUARE
DEVIATIONS
FROM
MEAN
SURFACE
The scatter of points in figures 3 and 4 is greater than might be
desired. However, it is not surprisingly large when the variety of
profiles that machined surfaces may possess is considered.
In order to verify in an approximate manner the experimental
findings recorded in figures 3 and 4, a comparison of surface roughness
and gloss was made by computation. For this comparison, different
machined surfaces were assumed to have profiles at right angles to the
direction of tool travel, which can be represented by the equation

h=h max sin

X,

where h is the height above mean surface. When hmax is equal to the
distance between tool cuts divided by 2'11", a profile having the shape
of a sine curve crossing the mean surface at 45 degrees results; as h mxa
approaches zero, the profile approaches a straight line.
The root-mean-square deviation from mean surface of a machined
surface whose profile is defined by the above equation was shown by
Way 4 to be
hrms=0.708hmax.
The value of specular gloss obtained from a machined surface with
the present glossmeter is determined by the fraction of the surface
reflecting light into the photocell chamber. The glossmeter was built
to accept all light reflected within 6.5 degrees of the direction of
S. Way, Description and observation of metal surfaces, Proc. special summer conference on friction and
surface finish, p. 44, Massachusetts Institute of Technology, 1940. (Puhlished by Murray Printing 00.,
Oambridge, Mass.).

Glossmeter for Surface Finish

391

reflection from a smooth surface. Because a reflected beam rotates

twice as fast as the surface which reflects it, the only areas of a test
surface that can thus direct light into the photocell chamber are those
less than 3.25 degrees from parallel to the mean surface.
The fraction of the incident beam that is reflected into the photocell
chamber by a machined surface measured across the direction of tool
travel can be shown from the above considerations to be

R=(1-; cos-10.0064X/hrm.)Rmax,
where >. is the feed, or distance between tool cuts, and Rmn is the
reflectance that the coated metal surface would have if flat. The
tool feeds for the steel specimens used in this study were about 0.020
inch. The glossmeter readings for the smooth steel specimens were
found always to be 25 or 26. For these reasons, >. was made 0.020
inch and Rmax was made 26 in computing the expected relation between
gloss and h rm. This relation is shown graphically by the dotted-line
curve in figures 3 and 4. When the probable differences between
actual profiles and assumed profiles are considered, the agreement
between computed and observed data is considered reasonable.
V. SUMMARY

A glossmeter has been developed for use in comparing the roughness

of machined surfaces. Although this instrument actually evaluates
that fraction of surface area nearly parallel to mean surface not
shaded by peaks, the instrument gives rapid and reasonably reliable
comparisons of the roughness of different surfaces prepared with
about the same tool feeds. Because glossmeter readings depend on
the profile shapes rather than profile dimensions, the glossmeter is not
a suitable instrument for roughness comparisons where tool feeds
differ appreciably from specimen to specimen.
WASHINGTON,

January 7, 1946.

Journal of Research of the National Bureau of Sta ndards

FIG UR I,

2.- Glossmeler on specimen, and control box.

Researc h Paper 1708