"A millionth of an inch"
Chapter XIV of Moving Forward
By
Henry Ford and Samuel Crowther
Author note: The material that follows in this chapter is based on Mr. Johansson's description
of this most interesting science and has been read and revised by him.
[From footnote in chapter - JRH]
The
making of things to a high measure of accuracy is not just a test of
workmanship. It is a fundamental
to service production. In such production there can be no
fitting of parts in assemblies or in repairs. Every crankshaft must
be exactly like every other crankshaft and so on through every
part of the automobile or anything else that one may be making
by this method.
Of
course, no two parts are ever absolutely alike except by accident, for
it does not pay to try for accuracy beyond
a certain point. But any kind of a machine which has moving
parts must be accurately made or there will be an amount of vibration
through play that will shorten the life of the machine and also
decrease its running efficiency. A designer always plans a
machine made to absolute measurements, and then he sets the
limits by which each part can be allowed to vary from the measurements.
In volume production the measuring has to be done quickly, and
for that reason each part is judged by two gauges of whatever
designs prove most suitable for the part. These two gauges
represent the allowable limits of size. If the part passes through
both it is too small, and if it passes through neither it is
too large; to be within the limits it should pass through one
and not the other.
Some
parts do not require accuracy in any high degree,
while others must have limits of one ten-thousandth of an inch.
If a gauge is going to check to one one-thousandth of an inch—which
is not extreme accuracy—then that gauge must itself be accurate
to at least one ten-thousandth of an inch. And it must
be kept accurate. Thus there must be a master gauge to test
it—and finally a gauge to test the master gauge. All of
which presumes that somewhere on the premises one has some
absolutely accurate measurement with which to test the master gauge.
In
point of fact, it was not always possible to
know what an inch was until we had C. E. Johansson join us. And
today the functioning of our industries depends upon what
goes on in a small room in the Dearborn laboratories. And so
also does the functioning of many other industries, for in this
room are tested the precision blocks not only for our own use
but also those that we sell to other industries, for it would
not be in line with our policy to keep the means of accurate
measurement exclusively with us. These measuring
devices are available to anyone who wants them, and they are
the standards of all accurate work in this country and in most
parts of the world. The obtaining of a standard of accurate
measurement is not so simple as it might seem to those who measure
with a foot rule and think they are measuring. The finest steel
rule made is so inaccurate that, if we had to use it, we should
wreck our production. That gives some idea of our necessity for
accuracy.
Until
recent years there has been no way of attaining a real accuracy
in machine production. For we have had no standard inch for
shop work. It is service production that has forced the making
of standards of accuracy. The English foot of twelve inches was
legalized about the year 950, but fought for more than three
hundred years before it supplanted the Belgaic foot of the
Tungari, which was 13.22 inches, and was brought to England at
the time of the Belgian influx before the Tenth Century. The
origin of the Belgaic foot is lost in the obscurity of ancient
Asia and Greece.
The
metric system is the result of a demand by French scientists that the
unit of measure be based upon some measurable
and immutable distance in the universe. The French National
Assembly chose a distance equal to one ten-millionth of a quadrant
of the earth through Paris. This distance was duly computed,
and the resultant metre-a-traite, or measure of distance, known
as the metre, was made the
compulsory national standard in 1801. In 1875, the International
Bureau of Weights and Measures was constituted in Paris to
furnish prototypes of the metre to subscribing countries. The official
French metre is represented by the distance lying between two
microscopic lines on an iridio-platinum bar (now reposing in
the archives of the Bureau), at zero centigrade.
The
alloy and the cross-section chosen for this rod are least susceptible
to change. Temperature is
a vital factor in computing the metre, since, as the
temperature of the rod goes above or below zero centigrade, the distance
between the scribings becomes less or greater than a metre. The
consideration of temperature was one of the great problems
previous to the advent of the Ford-Johansson gauge.
The
metre is measured at zero centigrade; the English inch at sixty-two
degrees Fahrenheit, and the U. S. inch, by
Act of Congress, July 28, 1886, at sixty-eight degrees
Fahrenheit. Considering these temperatures it is evident that, to bring
the three standards into accord, the units must be held at
three different temperatures. If a standard (French) 100 mm. steel
gauge, at zero centigrade, having a coefficient of expansion of
.000,011,5 per one degree centigrade, is measured at the English
standard temperature, it is .019 mm. longer than 1 decimetre.
Measured at the U. S. standard temperature, it is .023mm. longer
than 1 decimetre. Conversely, a U. S. standard decimetre gauge
and an English standard gauge become correspondingly less than
a decimetre when measured at the French official (metric
standard) temperature.
Since
these standards of measure have been adopted, another method
of checking the accuracy of measures has been discovered. The
instrument used is called an interferometer. Its operation is
based upon the constant length of the rays of light in the
spectrum. Should the official metre gauge be destroyed, it might
be reproduced within one ten-millionth of its theoretical
length with the aid of the interferometer. This instrument has shown
that the metre is equal to 1,533,164.13 wave lengths of the red
ray of the spectrum of cadmium in the air at fifteen degrees
centigrade, and 760 mm. barometric pressure. But of what avail
is such a method of measuring in a machine shop?
Less
than forty years
ago most of the difficulties experienced in commercial
measuring in mediaeval times still existed. Each country, each large
industry, and even each toolmaker had individual measures.
While these measures all conformed to either the metric or the
English standards, no two agreed unless made from the same
material, in the same shop, and at the same time. Each shop had
its own set of master gauges from which measurements were
sometimes taken directly, but more often these gauges were used
to make other individual gauges for specific work. Since gauge
making was slow, expensive work, it followed that a shop doing
any volume of business had an accumulation of gauges that
represented an investment out of all proportion to the amount of
use which they received. The matter of the investment, however,
was only incidental to the gauge difficulty.
When
a gauge is made
of steel it must be hardened to withstand even moderate use
without appreciable wear. When a piece of steel is hardened tremendous
internal strains are set up within the surfaces of the gauge
which cool and set first as the gauge is quenched after heat
treating. These strains are gradually relieved by slow
expansion of the metal—a process that may continue for a year
or more after the gauge is completed. Also if a gauge is not
used at the same temperature as that at which it was finished
its size will vary with the difference of temperature.
Thus,
what with a variation of standards, a variation of materials, and a
variation of temperatures, the paths of
the machine and tool manufacturers were anything but rosy.
Arguments between a mechanic on the warm side of a shop with one
on the cold side often became legal battles between seller and
buyer which involved broken contracts, months or even years
of litigation, and often not inconsiderable damage awards by
courts. The court decisions could never establish precedents,
for they were rendered arbitrarily for lack of any universally
recognized standards of measure.
C.
E. Johansson began his career as a mechanic in
Eskilstuna, Sweden, in 1885, and as tool-room foreman of the
government's arms factory he first faced the problems arising
from an inaccurate system of measure. Although well equipped
with the usual quantity of gauges, scales, and measuring devices,
the arsenal did not adopt the tolerance system of measures
until 1889, when they undertook the manufacturing of arms, in the
present-day sense, in place of the highly individual method of
arms manufacture which obtained from 1867 back to the days
of the stone axe and arrow makers. During the period 1867-1889
the arms were manufactured by machines, and although the quality
was good the quantity production was limited because the limit
system was not used.
One
of the first points which impressed him in those
days was the preference of toolmakers for solid blocks of steel
finished to a specific size corresponding to some specific
dimension to be imparted to the manufactured part. Scales
varied in their calibration and micrometers varied in the leads
of their threads, but once a machinist measured a dimension to a
steel block, and it was accepted, he worked speedily and
confidently to the block until it was worn out. Consequently,
Mr. Johansson specialized on these solid gauges. Soon there
were hundreds of them in the arsenal and hundreds more being
made all the time to take their places. At each change in
specifications,
from two to a hundred of these became obsolete—most of them
long before they were worn out. Though inches, half inches,
quarter inches, and the like did not change, parts could not be
designed to suit the existing gauges—gauges must be
made to suit the designs.
Mr.
Johansson, from his experience, knew that two gauge blocks, if finished
with the most painstaking care, would
adhere slightly when wrung together. A close study of the
finished surfaces of such blocks held a promise that several gauges
might be made so accurately and so perfectly finished that they
might be used separately or in combination to produce from
one to a number of measure values limited only by the number of
gauge blocks used. He began work along this line at once,
but almost immediately another difficulty was encountered. No
method for determining the size or the parallelism of the surfaces
of these gauge blocks was accessible. Hours of work were
necessary to measure any gauge, and even then the margin of error
was great. Eventually it was solved in a manner that permitted
the instant gauging not only of the flatness and the parallelism
of the opposite surfaces of a gauge but the determining exactly
how far these surfaces were apart. However, as this end was
approached, a new difficulty became increasingly great; namely,
the stabilization of the steel itself from which the gauge
was made, a difficulty which seemed insurmountable.
Unless
a gauge block is heat treated and hardened, its usefulness is soon
greatly impaired by wear.
In the case of a gauge the accuracy of which is required only
within one one-thousandth of an inch—and it is now a quick
and simple process to divide this accuracy by ten thousand,
measuring to one ten-millionth part of an inch—this accuracy
would be completely lost with very little use unless the gauge
were rendered glass hard by an adequate heat treatment. It
became evident, when working in such small dimensions, that a
block which was heat treated and finished to within one ten-thousandth
part of an inch of the measure value desired, would begin to
expand, very slowly, after it was finished, until at the end
of a year it might measure several ten-thousandth inches more
than when finished. The larger the block and the greater volume
of steel it contained, the greater the growth after finishing.
By
the year 1899 gauge blocks could be made by Mr. Johansson so
flat that they would adhere when wrung together. Their surfaces
could be placed at a predetermined distance from each other
within a margin of a few hundred thousandths of an inch. The
surfaces of the blocks could be rendered so nearly parallel that,
when placed in combination, the cumulative error in several
blocks was no greater than the error in a single block of the
same length as the combination. Selection of the steel best
suited to the manufacture of gauge blocks and the discovery of
the heat treatment which would render this steel sufficiently
stable required experiments over a period of nine years. Then
a universal gauging system passed from an ideal to an
accomplishment.
The
first set produced contained 102 blocks. Years of observation
had shown that the temperature of the average machine shop
varied between fifteen and twenty-five degrees centigrade; therefore
twenty degrees centigrade (sixty-eight degrees Fahrenheit) was
selected as the standard temperature at which these and subsequent
blocks should be finished and used, except blocks made for use
at specified temperatures other than sixty-eight degrees Fahrenheit.
Sets
of these gauges
were introduced into the United States in 1907, and were
received with some hesitancy. Purchasing agents at first considered
them too fine for their requirements. Even if they allowed
themselves to be persuaded by shop superintendents to purchase
sets, they frequently placed them in the office vault and
permitted their use only when a difference of opinion on measures
arose. While American manufacturers were accustoming themselves
to the possession and use of gauges, the same gradual acceptance
of this new standard was going on in all foreign countries. As
the gauges became known their possibilities became increasingly
evident. The limit system of manufacture, that is, the process
of holding the dimensions of each part of each assembly between
two limiting dimensions as determined by these new gauges,
became universal. Even the limits became somewhat standardized.
Thus, one standard permits a round rod to fit a round hole in
four degrees. These are: running fit, push fit, driving fit,
and forced fit. In the first class, the rod slides in the hole
easily; in the second, it may be pushed into the hole but fits
snugly; in the third, a hammer is needed to drive the rod into
the hole; in the fourth, a mechanical or hydraulic press is
needed to force the rod in. In the first, or running fit class,
standard practice requires the rod—supposing it to be
more than 1" and less than 2" in diameter—to be less than the
diameter of the hole by not less than .0015",
and less than the diameter of the hole by not more than .0025",
or a limit of variation of .001", the difference
between the two tolerances. In the case of a driving fit, the
limit changes from less to greater; the diameter must not be
less than .00075" greater than the diameter of the hole, nor
more than .002" greater, or a limit of plus .00125";
the limit on the case of a forced fit is plus .002".
If
rods similar to the one described were to be used in quantities,
a gauge for measuring their diameters would be made. For the
rods the gauge would consist of four pieces of steel, set in
a horseshoe-shaped holder, two far enough apart to allow the
rods to pass between them in their diameters were not greater
than the maximum tolerance specified, and the other two close
enough together to keep the rods from passing between them unless
the diameters of the rods were below the low limit prescribed.
Such a gauge is called a "go-and-no-go," or snap
gauge, and the points may be spaced to the specified distances
apart by setting them to gauge blocks, or combinations of them,
of the measure value desired.
The
accuracy of practically every manufacturing operation carried on to-day
may be traced directly to
the use of these gauges. The Ford Motor Company has one hundred
and forty complete sets in daily use, besides a quantity of
single blocks for purposes where complete sets are unnecessary.
This number does not include many pocket sets individually
owned and used daily to test the accuracy of micrometers.
The
alliance of the C. E. Johansson Company with the Ford Motor
Company furnishes an interesting chapter in the development of
the gauges. Considering the importance to the Ford Motor Company
of high accuracy at a low cost, the gravitation to it of the C.
E. Johansson Company, with exactly that combination, was inevitable.
Also, by means of the resources and the manufacturing talent
made available the price of the gauges has been materially decreased
since the Johansson Company was taken over.
Thousands
of gauges and fixtures must be manufactured in our tool rooms, and by
using these blocks it
is possible to make them to the necessary limits. Without the
Johansson blocks and tools our tool rooms would be blind. By
using them the gauge inspectors also can tell at any moment the
condition of the gauges and fixtures in service.
How
important this
is may be realized from the fact that, when one of the gauges
is worn to the point it no longer checks the limits properly,
one hour's run may mean several thousand pieces of scrap, tying
up of the assembly line, or the holding up of some branch
in a remote part of the world.
If
extreme accuracy is to be obtained economically — as mass production
demands — the master
standard gauges must be just as near to size as it is possible
to make them. Any error in the master is multiplied several
times before the working gauges reach production. When
production machines or the assembly lines are held up for minutes,
or the fraction of a minute, the cost is very high. Every error in a gauge must be paid
for in the slowing down of machinery and the loss of parts.
Take,
for example, the snap gauge used in the manufacture of the
push rod in the car. These cost approximately four dollars each
to repair; twenty are used on each eight-hour shift, and their
life is approximately four hours. On all gauges twenty per cent
of the tolerance is the wear limit. That means a wear limit
of one ten-thousandth of an inch or one hundred-millionths.
Hence, it costs one dollar for each twenty-five millionths of
an inch wear.
At
least three sets of Johansson gauge blocks are needed to pass these
gauges every time they are sent to the tool
room for repair. One is used by the man who repairs; another by
the tool-room inspector; and a third is used in checking them
every four hours. If an error of many millionths exists between
the three sets, more gauges would be returned to the tool
room than actually wear out. Production would be retarded and
costs would run high. Therefore, we have determined that it
pays to hold the master gauges to the extreme accuracy of not
exceeding plus or minus two millionths of an inch. Thus the
inspection gauges for the tool room and for inspection have an
accuracy of plus or minus four millionths, and working gauges
can be supplied the production line at an exceedingly close
limit, well within an accuracy of fifty millionths of an inch.
All of the close-limit
gauges used are checked once every day, many of them three times
every eight hours. Under the guidance of Mr. Johansson himself,
the workmen are able to control degrees of accuracy at a saving.
He believes that it is no harder with proper means to hold to a
close limit than to the old accepted accuracy.
We
have searched practically
the entire world for special gauging equipment. Much of this is
reconditioned by the Johansson division after it is received.
And it is a large undertaking to get dies, jigs, and fixtures
together to build accuracy. A large proportion of the machinery
we use must he rebuilt to meet the accuracy standards.
One of the chief requirements in the gauging system is a definite manufacturing—wear tolerance,
established and maintained. The
first thought, of course, must be accuracy; the second,
speed. The third is the original cost, plus the amount of
repairs and reconditioning necessary. There must be some swift,
accurate way to determine when it is time to retire a gauge
from use. For this, our men use the Johansson block.
The
manufacturing of
these blocks is carried on in the Dearborn laboratory, and they
must be finished at a temperature of sixty-eight degrees Fahrenheit.
This requires a special room. The constant-temperature room in
which the finest gauges are finished and tested is seventeen
by eleven and one-half feet. Its walls are of plaster to a
height of three feet; above that they are of glass. The glass walls
are double, with a space of twelve inches between, which serves
as an air duct. The glass in the outer wall is of triple thickness,
that of the inner of but one. Throughout, the glass is set in
heavy felt and hermetically sealed in its frames. The doors,
of which there are two, are fitted with weather strips. The
walls and ceiling are insulated with two layers of cork, laid
directly on masonry surfaces, and the floor is insulated by the
same material, laid in asphalt cement. All walls have a four-inch
base of cement.
A
constant temperature in the room is maintained by circulating air at
the proper temperature in the air duct formed
by the double walls of glass. A number of electric elements
supply warm air to the duct whenever it is needed. A refrigerating
machine supplies cold air. Thermostats at three points control
the flow of warm and cold air. The presence of workers in the
room reduces the amount of oxygen and adds to the amount of
moisture in the air. To make up for the first an ozonizer, having
a capacity of six grams of ozone per hour, is used. With this
are installed two dehydrators, capable of supplying this room
with thirty litres of dry air per hour, by means of which the
moisture is counteracted. Mercury tube lights are used for illumination.
To prevent their heat from interfering with the temperature of
the room, they are separated from it by a case of glass, and
the resistance coils with which they are equipped are placed
outside the room.
The
making of these gauge blocks is an intricate and highly skilled
process. Some of the gauges used for shop work are very
ingenious, and they must be tested by special apparatus. Take the
wrist pin hole in the piston of a Model A motor. It is checked
by a special Johansson internal gauge, which in turn is set
by a ring gauge. The wrist pin hole and the pin must fit within
limits of .0003 of an inch. To test the ring gauge we have
a machine that will check the inside walls of a hole. The
machine is known as an "inside optometer." It looks like
an elaborate microscope. The ring gauge is placed in the centre
of the machine on a tiny platform, which can be swung in any
direction. It is brought to position beneath two jaws, or
measuring parallels, that extend downward toward the floating table.
These parallels are set with Johansson blocks, and the distance
between is thus known accurately. On the tips of the jaws
are jewels, placed there to provide a hard, unyielding contact
against the inside walls of the ring gauge. After the parallels
have been set against the inside walls of the gauge the
operator, by looking into an illuminated tube on the right of the
machine, is able to detect whether there is any variance. If
the walls are within one half of one ten-thousandth of an inch,
the operator can see the difference in greatly magnified degree
as through a microscope. It is enlarged so much that the operator
can halve the size just mentioned—which is one hundred times
finer than the size of a human hair.
This
instrument may be used to detect not only a
tapering in the walls of the hole but also the exact inside and
outside diameters. It can also tell whether a tool or part
is out of round, and by how much.
Here
is another example. In the use of screw threads many dimensions must be
checked to insure proper
fits when the threaded parts are assembled. For example, a bolt
and nut must fit together. Both the external thread on the
bolt and the internal thread in the nut must be measured. That
can be done with master gauges made and threaded like the parts
to be fitted. With these the parts can be tested during
manufacture.
But
how to measure the accuracy of the master thread gauges? The
answer is found in an intricate-looking machine in the
gauge-inspection room. It is known as a "universal measuring
microscope."
With the microscope in this machine, the screw threads can be
magnified to such proportions that they can be easily measured.
It is the second machine of its kind to be brought to America;
the first is in our Dearborn laboratory.
Checking
is accomplished by use of the knife-edge
method. The knife-edge is a narrow blade of steel having an
angle on one end. On the knife-edge a fine line has been engraved
at a given distance from, and parallel with, the edge formed by
the angle at the end. In the microscope are lines that have
been spaced apart the same distance as that between the
engraved line on the knife-edge and the edge of the knife. Two blades
of steel are used. The knife-edge of one blade is placed in
contact with the flank of the screw. The other is placed on the opposite side of the screw thread so that it comes in contact with the opposite flank
of the screw.
By
means of the microscope one of the lines in the eyepiece is
superimposed upon the engraved line on one of the
knife-edges. A reading is then taken on a magnified vernier
scale. The carriage of the machine is then moved until the other
line in the microscope eyepiece is superimposed upon the
engraved line of the other knife-edge. A reading is again taken on
the magnified vernier scale, and the difference between the two
readings is the correct pitch diameter.
There
are three different phases of a screw, all
of which will affect its accuracy, and which all must be
measured. These are the pitch diameter, which technically is the
diameter of an imaginary cylinder, the surface of which would
pass through the threads at such points as to make equal the
widths of the threads and the width of the spaces cut by the
surface of the cylinder; the lead, which is the distance a screw
thread advances axially in one turn; and the angle, which is
the angle included between the sides of the thread measured in
an axial plane.
All
of these phases can be checked on the universal measuring microscope.
In fact, since the machine was installed,
the accuracy of thread gauges in the plants has improved almost
one hundred per cent. The machine will determine whether a
surface is absolutely straight and square. It will measure
either round or tapering surfaces equally well. It is especially
useful in checking templates and in laying out holes. It has
been said to measure as fine as one ten-thousandth of an inch.
With
a dividing head, this machine can be used to
determine angles or indices. The head is placed in position in a
specially constructed groove. Through its magnifying lens
the operator can look down on a circle, the degrees of which
have been so much enlarged that each is divided into three parts
and each third into twenty. The operator can thus read the
minutes as well as the degrees of the angle and can check within
limits of one minute.
These,
of course, are not manufacturing gauges. They have to be so nearly
absolute in their accuracy only because
the gauges in use must be what they purport to be, else the
parts turned out in production will not fit. An error at the source
is multiplied many times when it reaches the part.
One ten-thousandth of an inch is a workable accuracy, but to insure that degree of accuracy one
must be able to control to perhaps a millionth of an inch. That we can do.
+++
If you find this interesting or wonder why I
do, I welcome discussion. Drop me a line at johnhenry@changeover.com