[Author Prev][Author Next][Thread Prev][Thread Next][Author Index][Thread Index]
ATM Testing Flats ala Waineo
Text of a message sent to me in 1995
the ascii art may be better with a different font.
============================
MAKING AND TESTING FLATS
by
Tom Waineo
The following text was submitted Feb 1995 for publication
in Amateur Telescope Making Jounal for publication.
A SIMPLE FLAT TESTER
One of the easiest ways to accurately test flats require
a source of monochromatic light. With a simple monochromatic
light tester, profile variations between two surfaces can be
measured to a fraction of a wavelength. A good, but simple,
light tester can be made by using replacement fluorescent
tubes made for bug zappers sold by Sears or hardware stores.
Starters, ballasts, and a light switch can be wired to those
tubes which can be mounted in a stand to permit right angle
viewing of a flat. The sketches in figure 1 shows some of the
details. My tester illuminates a 8 by 8 inch diffused area for
a cost of less than $50. Those tubes are blue in color from a
mix of violet and green mercury spectual lines. A plastic green
filter held in front of the eye filters out the violet leaving
only a fairly pure green light which is very suitable for
testing flats.
I used 4 tubes of 15 watt rating with 15 watt rated starter
and ballast wire connected and soldered to the tube pins. The
front glass is a picture frame and the diffuser made from a
white plastic shopping bag. Inside the box is lined with course
sandpaper just like wallpapering and painted flat black. The
lumber used are 10 inch shelfing boards. The tubes are mounted
on a white cardboard and held in place by twist ties.
TESTING FLATS BASICS
When glass surfaces are cleaned with achohol or acetone
and carefully placed together, only a few wavelengths of air
seperates them. The light interfer whenever there are mutliples
of half wavelength spacing between surfaces. This interference
causes dark fringes to appear in the flood of light coming from
the monochromatic light source. The difference in air thickness
between adjacent fringes is 1/2 of the wavelength of the
illuminating light. The 5461 A mercury green spectual line
has a wavelength of 0.0000215 inch or 21.5 microinch. A one
thousands of a inch (0.001) equals 46.5 wavelengths or 93
fringes of interference using mercury green light. The surface
variation between the two surfaces must be less than 1/10 of that
amount in order to have fringes that can be accurately
evaluated. When the two surfaces are flat and close enough
together, fringes appear in the air wedge between thewm. Those
fringes are parallel to each other; the air wedge between them is
at a right angles to the fringe. Pressing down on the top flat
causes the fringes to move to the thicker air space. When the
fringes are viewed at a normal to the surfaces, a eye movement
off the normal in any direction causes a fringe movement to the
thicker spacing.
This system can also be used to test surfaces other than
flats. When both surfaces have equal radii, such as a convex
matching a concave, the interference is the same as if the
surfaces are flat. However if they are unequal in radius, the
air spase is uneven giving contour fringes whenever the air space
are multiple spacings of 1/2 wavelength. With no wedge between
surfaces, the contour fringes are concentric to the center.
With a wedge, they are concentric to the minimum gap between
surfaces even if it is outside the edge. With one reference of
known quality, the errors of the other can be easily seen. When
both surfaces are made to match, a overlap of surfaces helps to
determine which surface has the fault. Error zones usually are
concentric to the center. A overlap makes the zones
idenifiable.
Be aware that the testing of a coated surface
against a uncoated one will make the fringe very hard to see.
The reference master must be coated with a thin but still
transparent layer of reflecvtive material to bring up the
reflectivity higher then the 4% from a uncoated surface. My
guess that at least 20% is needed.
Flats can be made from many materials. When they are
used near the surface normal, a slight amount of curvature
(non-flatness) can be focused out. Curvature make a wavefront
astigmatic when the surface reflects at a angle. When a
temperature change occurs, if there is a temperature difference
between the front and back of a flat, then the surface would
have curvature until the flat reaches equilibrium with the new
temperature. A flat that has a front and back of the same
condition would absorb or emit thermal energy equally from both
surfaces and remains flat. I had a 8 inch pyrex flat in a
Florida garage going thru a 15 degree daily temperature change
and I could not detect any change from flatness. I have made
and used a 24 inch aluminum flat with both sides nickel plated
and polished. That flat is still being used after 15 years.
Quartz and Zerodur are excellent materials for a master
reference flat to test other flats. They cost about 10 times as
much as a blank than Pyrex.
When two surfaces are to be tested, the backs have to be
polished to clearly see the fringes. The backs need not be flat
or completely polished. Often backs are given a felt polish to
save time. The edges should have a rounded bevel. Be aware of
the Twyman affect. When one side is polished and then the
other. The first side will change. It is best to polish both
sides and then figure a side. The two surfaces are cleaned
with alcohol or acetone. The high evaperation of the solvent
chills the surfaces and makes them concave to each other.
Slowly place the edges in contact with a slight overlap and
gently lower in place. The fringes should appear are there are
no large lint or particle that makes a large wedge. Try not to
force the pieces together. It is best to reclean and try again.
When fringes are seen, then pressure can be applied to squeeze
out air and reduce wedge. It does pay to use masking tape
around the outside to make the outside diameter less slippery.
Accidents do happen, so be carefull. Do not slide the surfaces
across each other. Any hard particle can scratch. It is best
to take apart and reclean. Sometimes when placing one flat on
another or in removing the top flat, the top piece will "float"
on a thin layer of air and move sideways with respect to the
bottom flat very easily. Guard against any large slides
sideways on this air layer. When removing the top flat, it
feels like it is stuck down by a vacuum. You have to lift very
slowly with a firm upwards force to allow air to enter between
the flats. Fringes can be seen between a fine ground surface
and a test flat. When you view a ground surface at a shallow
angle to a light source, you can see a reflection. Place a
flat on the ground surface and view the monochromatic light at
this small angle, fringes will appear.
Flats can be made without a known good master. One
method is to make three and test them against each other.
The only surface common to three is a flat. I have tested three
which are A, B, and C. A on B is 1/3 fringe convex. A on C is
1/4 fringe concave. B on C has straight fringes. The three
equations are: A + B = 1/3; A + C = - 1/4; B + C = 0.
Solving those three unknowns gives A as 1/24 fringe convex. B
7/24 convex and C - 7/24 concave. This method works well. The
worst of the three is figured with the other two as references.
A flat can be tested at a angle to a accurate spherical
with no astigmatism and examining the focus using the Foucault
test as shown in figure 2. A knife edge device with two knife
edges at right angles and the ability to move those knife edges
both left and right and up and down must be used. Figure 3
shows a device that permits this kind of motion. A slightly
curved "flat" surface will generate astigmatism at the center of
curvature. There would be two positions giving even light cut
offs using this knife edge device with the vertical and
horizonal cut offs seperated. Due to the viewing angle, a
round flat looks elliptical. A reference measurement is made by
moving the knife edge perpendicular to the major axis of the
apparent ellipoidal flat at the test position. The knife edge
cut off at this reference position should give even darkening.
The cut off in the fold direction will be inside this reference
if the flat is concave or outside if the flat is convex. When
the surface is flat, the two foci are the same. Unfortunately,
this result can also be true if an astigmatic flat is positioned
at the right rotation angle. To guard against this error, the
flat is rotated 45 degrees and tested again. This test is
called the Ritchey-Common test for a flat. It is the only test
for a flat that does not require another flat for a reference.
The knife edge evaluation is the same as a sphere, With a
coated sphere and two reflections of a uncoated flat, the
aberration of the returned light is double that of a single
reflection of the flat. The test is very sensitive to figure
error. If L is the knife edge toi flat center distance, D is
the flat diameter, the flat is positioned at 45 degrees to the
sphere and 5461A wavelength is used, the distance between the
two foci is 0.000486 L^2/D^2 per wavelength deviation of
flatness of the flat. For example, this means a 10 inch flat 100
inches away from the knife edge has a foci variation of 0.0486
inches per wave. With a inclined knife edge travelof figure 3,
the two foci cut offs is easily done. The reflected pinhole
image is placed in the corner of the horizoinal and verticval
knife edges. The inclined knife edge travel allows both edges
to cut into the image for the measurement of focus difference
for flatness calculation.
Diagonals are best done in blocks. Each diagonal is
mounted on a plate with buttons of blocking pitch. It is best
to grind, polish, and finish in one day. If the block left
overnight the diagonals can shift away from their co-planer
position. This requires a regrind. Curvature vary by the
square of diameter. This means a 1/10 wave convex or concave
over 2 inch diameter is 4 times 1/10 or 0.4 wave over 4 inches,
or scaled to 1.6 wave (3.2 fringes) over 8 inch diameter.
Professional flat work are usually done on a planetary
polishing machine. The lap is a large annulus or ring. A large
flat called a conditioner is polished face down on the lap.
It's purpose is to keep the lap in condition so that the work
being made are polished flat. There is no stroking at all. All
polishing is done by rotation only. Moving the conditioner in
towards the center concaves the lap and makes the parts convex.
The reverse happens on move out. Parts are placed in cutouts
inside rotating rings. Some of those machines are huge with
laps as big as 12 feet in diameter. The one I have seen has a
72 inch lap with a 16 inch hole. The 600 lb conditioner is 40
inches diameter. The work are placed inside of one of three 22
inch diameter rings. The machine is so productive that one
operator can do the work done with 16 conventional machines. I
do wish that ATM could watch a planetary machine running and see
the quality of flatness it can produce. Your diagonal mirror is
very likely made on this type of machine.
A spherometer is almost a requirement to make a flat.
An inexpensive spherometer gage can be made by using a pulley.
Try to find a large one with a 3/8 inch bore which is the
diameter of a dial indicater stem since the more comman 1/2 inch
bore requires a adapter to hold the indicator. Wwarp masking
tape tightly around the indicator stem until it fits the pulley
bore. Three acron nuts and bolts at the rim at 120 degree apart
finishes the spherometer gauge. Try to obtain a indicator with
at least 0.0001 inch gradulation. That much is 9.3 fringes in
the green. When two disks are ground against each other, the
spherometer can check the dial gauge measurement of each. If
the same reading is obtained from both surfaces then the
surfaces are flat. With the same sized blanks, the top one
will concave during grinding. With machine grinding, I prefer a
tool 5/6 diameter. This allows controlling flatness without the
need to flip blanks. Stroking more in center will make the
glass concave and moving out to stroke more over the edge
convex. A setting can be found to hold the sagitta close to a
flat during grinding.
Polishing is best done with same size lap. Barrel walkers
will find it is hard to convex even with lap on top. It is
easier to stroke lap on top with mirror rotating. For a 6 inch
5 to 10 rpm is a good speed using a short stroke out over the
edge and return to a centered position while holding the lap.
Let it slip a short turn and repeat. Small flats requires a
faster speed. Something 2 inch diameter best done at 60 to 80
rpm. At a temperature of 72 to 78 degrees, I like GUGOLZ 64
pitch. I do not find that a harder pitch is needed for flats.
It is the same pitch that I use for telescope mirrors. Machines
are a great aid for working flats. The following is a list of
machine setting that usually works for flats with lap on top.
Start of stroke with work centered and stroke out over the edge
and return back to start and centered.
To concave To convex
More strokes per rotation Less strokes per rotation
More pressure Less pressure
Shorter stroke Longer stroke
Scrape off some pitch at the edge Scrape out the center
Let lap dry out Keep wet
Let pitch channels close Open channels
By trial and error, it is possible to find a setting
that the curvature would not change after hours of polishing.
The 8 inch flat was grayed out in 7 hours polishing and only
changed by one fringe concave. The machine run at 11 rpm, 20
strokes per min, 1 3/4 inch stroke, with GUGOLZ 64, and 10 lbs
of weights. The machine had only one motor and the ratio of
stroke to rev is fixed. It does have a variable pitch pulley
on the motor. It was run on the slowest speed.
I did finish by hand. On a 1 rpm rotating table, I
could not make the flat go convex. Working at 7 rpm gave me
control of both the figure and flatness. I am sure that a
barrel walker would have the same trouble that I had. Who
knows, with the proper technique it can be done. I am just too
old and lazy to do any more barrel walking. I hope that this
article will inspire someone to give flats a try. With this
tester, you can test flats and leave them under test for many
days while you decide if they are good enough. Flats are used
in transmittion in lenses, windows, and Schmidt correctors. In
reflection, they are used as diagonals, fold mirrors,
coelostats, and heliostats. They are used to test paraboloids,
optical systems, and other flats. Making them I find is fun
and I hope that others will find it also. Wanta make a flat?
() () () () () () () () 4 Bug Zapper 15 watt U tubes
----------------------------- white shopping bag diffuser
/
/
/
/
/
/
/ document frame and glass at 45
/ degrees
/
/
/
/
/
/
/ ____________
l__________l flats
l__________l
Flat monochromatic Light tester
__________
______________l l
l _____________l starter l
l l l_________l
__ 0 0 __
tubes l__l l__l
0 0
_______________l l
line __________ l
voltage ___l ballast l____l
l_________l
End View of U shape bug zapper tube showing connections to the
4 pins
/\
_____ / \
l l \ \
l l \ \
l l \ \
l l \ \
l l \ /
l___l \/
spherical flat at 45 degrees
mirror to sphere
/l\ light directed to flat
l
___
light l___l --- knife edge
source
___
l l l__ horizonal and vertical knife
l l l\ ll edges movement on a 45 degree
l l l \l angled slide
__________l___l___l____\_______
Ritchey Common Test for a Flat with a Foucault tester
l l