Theory Of Operation
When I wrote this code, I tried to emulate what I did during
a normal Couder Mask, Foucault test.
During this conventional test, a series of very defined steps were
followed to record the null positions for each zone pair. Asides from some user interpretation
problems, and the amount of time it took to do an accurate test the process was
fairly simple.
Consequently I went about teaching my computer to do what I
did during a Foucault test (Pretty simple yea J, I wish!)
So, Lets start with the procedure often taught for a Manual Foucault
test
- First
we set the mirror on the test stand
- Second
we set the platform somewhere near the ROC of the mirror
- Then
we try and “AIM” the mirror, so the return beam hit the sharp edge of our
knife
- Finally
we put our EYEBALL down by the knife, to see what we see J
- Odds
are, that either the mirror was fully illuminated, or it was completely
dark.
- If
fully illuminated, we’d cut the knife edge in until we saw a Foucault
shadow
- If
dark, we’d cut the knife edge out, until we saw a Foucault shadow.
- In
either event, the first thing to happen was often the lateral movement of
the knife edge. Well RTAFT does
the same thing J
- Back
to a manual test. If we saw a
Foucault shadow, we’d probably look at it for a bit to see how smooth it
was
- After
that, odds are we’d walk down range to the mirror, and hang our couder
mask on the mirror.
- A
large portion of the time, the test was conducted starting with zone #1
and working out from there. RTAFT does the same thing
- So,
were back at the test stand, with our eyeball at the knife edge. Staring intently at zone #1
- Often
We’d find that Left and right zone pairs were either:
i.
Really Bright
ii.
Really Dim
- Again,
the first thing to happen was a Lateral knife edge motion, to get a Shade
of gray in one of the two hole pairs.
- Now
the tricky part. If the hole
pairs are not equally the same shade of gray (Very rare on the first
look) We move the platform longitudinally towards or away from the
mirror!
i.
But WHICH way to move it?
ii.
Well, If you KNOW what side the knife edge is on, there is no
need to guess, as its simply the following:
1.
If the zone hole opposite the knife is Bright, and the zone
hole on the same side of the knife is Dark, Move further away from the mirror!
2.
If the zone hole opposite the knife is Dark, and the zone hole
on the same side of the knife is Bright, Move Closer to the mirror!
iii.
Simple ehh J
- Once
we get both hole pairs of equal intensity for zone one, we record the
position, and start all over again for the next zone!
- We keep
repeating this for each zone, and for the number of repetitions we think
is reasonable for the complete test!
- Of course, its not quite so simple as
this, as there are lots of variables, such as:
- What
shade of gray?
- How
much do you move the knife Longitudinally when the zones aren’t even?
- Is
there backlash? If so, do you
always approach a zone pair null from a certain direction?
So that’s an overview of the basic Foucault test, and the
variables involved. Before we get any
further, let me explain what RTAFT doesn’t and does do J
- RTAFT
doesn’t use a paper couder mask.
- RTAFT
doesn’t do a FULL SURFACE Test
- RTAFT
works with fixed zone radiuses, moving the platform about to find each
radius’ null (just like a paper couder mask!)
- With
a 5 zone “MASK”, RTAFT only tests 10 points on the mirror!
- RTAFT
does, in effect, a BRUTE FORCE attack for finding zone nulls
- Instead
RTAFT finds its way to zone 1, gets a null, then finds its way to zone 2,
gets a null, and keeps repeating until its out of zones.
- The
whole process follows a very simple set of rules
- Set
the shade of gray
- Look
at the zone under test
- Decide
which way to move the platform longitudinally to get closer to a null
- Repeat
step A until zone null found.
- Select
next zone, repeat Step A until all zones found.
(See its simple! I wish!)
Shades of Gray
As you know, the Foucault test is all about finding equal
shades of gray for a zone pair of equal radius from the center of the
mirror. RTAFT does the very same
thing! However, to make this process
automatic, I need to make one of the Zone radius sides a “Reference” and the
remaining side a “Comparative” measurement.
This “reference” zone is the side of the zone pair OPPOSITE
the knife-edge. So, starting with zone
1, the code does the following:
- Set
the reference side of zone 1 to the “Shade of Gray” level
- Compare
the two sides of zone 1 and make a decision,
- If
the reference side is greater than the Comparative side, move away from
the mirror and start all over
- If
the reference side is less than the comparative side, move closer to the
mirror and start all over.
- If
the reference side and comparative side are equal, then this must be the
Zone null!
i.
Record longitudinal position, start working on the next zone
The only problem with the above summation is RARELY will
both zone pairs be EQUALLY illuminated!
Often they are close, but not perfectly equal! The result was the code would get stuck in an infinite loop
moving a bit forward, then a bit backward.
To solve this, I did a little trickery. I told the code to keep moving towards the
mirror, until the zone null was behind the knife-edge some 0.01” or so. Once that flag is set, the code will start
moving slowly away from the mirror, towards the zone null. At this point 1 of 2 things can happen:
- The
nulls become perfectly equal! (That’s nice)
- The
Reference side becomes less than the comparative side, thus we over-shot
the target. However, if the last
move was only 0.001”, then what’s the big deal J I know the null is pretty close to this
current position!
BTW, just as an aside.
If you charted the platform motion during an entire mirror test. The platform would move in a Z shaped
plot. Once zone 1 is found, it slides
sideways to set the shade of gray for zone 2, then starts moving towards zone
2. Once zone 2 is found, it slides
sideways to set the shade of gray for zone 3, and so on.
Not very efficient, but it’s the only way I can be certain
to have the code find its way to each zone’s null J
Well I guess that’s it.
Any questions feel free to ask!
Take care,
James Lerch
jlerch@tampabay.rr.com