From: "James Lerch" 
To: 
Cc: "ATM List" 
Reply-To: "James Lerch" 


Greetings All,

I've found an interesting phenomenon when using Robo-Foucault while testing
a 12.5" F/4.5 mirror for Carl Zambuto.  (BTW, Carl makes an incredible
product, on which I shall comment later in the day :)

Observation:
Keeping everything constant and changing ONLY the exposure speed of the
digital camera, I get a consistent systematic change in knife edge null
readings. This change in knife edge readings is fairly dramatic as well,
resulting in a spread of Strehl readings from 0.99 to 0.43!  (Dim images =
0.99, Bright images = 0.43)

Background Data:
Carl set me this mirror for the purposes of evaluating Robo's abilities.
Prior to testing this mirror I HAD the highest confidence in Robo, but now
I'm very concerned!!  Robo is still incredibly consistent, with a standard
deviation (n-1) hovering around 0.0008" across 7 zones and 4 knife
edge readings per zone for each of the tests I will detail later in this
document.

After Robo's initial disappointing results,  I created a Couder mask that
replicated Carl's.  I then took the mirror and the mask down to our ATM
lab, dusted off the 'Old School' eyeball knife edge tester and had our most
eagle eyed Foucault testers take several sets of readings.  The results of
our eyeball tests concurred with Carl's Foucault results, giving the mirror
a Strehl of 0.99.

Since our Couder mask Foucault results concur with Carl's results, and I
have the highest confidence in Carl's ability to test mirrors, it was
obvious Robo was in error.

Next I set about trying to find the source of Robo's error.  After
eliminating such items as:
    A) Optical / Mechanical platform alignment
    B) Camera vignetting
    C) Non-Uniform mirror illumination
    D) Camera Pin Cushion resulting in software zone radius positional
errors
    E) Mechanical errors in the platform (travel accuracy, motor slippage,
etc)
    F) Camera exposure changing.

After systematically eliminating all the above, I stumbled onto the
following experiment.

Test Setup:

#1 Default install of Robo-Foucault
#2 Slitless X-Y Foucault platform, using Blue LED
#3 Sony DSC-F707 5MP digital camera
    A) I choose this camera for its large lens set (avoids Vignetting)
    B) Has Full Manual Control options for Focus, F-Ratio, and Exposure
speed
    C) Has Analog Video Out.
#4 Camera Video out was captured by a Hauppauge USB video capture dongle
#5 All tests results were conducted in less than an hour, in temp controlled
environment, after allowing several hours to ensure stabile mirror temp.
#6 All tests were conducted across the same mirror diameter
#7 ONLY change during the test was to gently adjusting the exposure speed on
the digital camera!

Below are the results for total knife edge travel from zone 1 to zone 7
over various camera exposure speeds (Complete knife edge readings are
available as separate downloads at end of document)

Camera Speed               Total Knife edge Travel    Strehl   Correction
1/200                               0.147"                          0.988
Over
1/100                               0.141"                          0.994
Under
1/50                                 0.137"                          0.801
Under
1/25                                 0.134"                          0.708
Under
1/13                                 0.132"                          0.662
Under
1/6                                   0.130"                          0.429
Under

Classic Couder mask Results:
Carl Zambuto                  0.141"                           0.996   Under
Mirror Lab                      0.144"                           0.986
Under

The above results are troubling with regards to Robo-Foucault!  What's more
interesting is I can obtain similar results by the following two methods
(Keeping everything constant other than the below changes)

#1 Changing light source intensity
        A) Brighter source = brighter image = Less total knife edge travel
        B) Dimmer source = dimmer image = More total knife edge travel

#2 Changing the 'Shade of Gray' the software uses to set lateral knife
position
        B) Brighter shade of gray = Brighter image = *More total knife edge
travel
        C) Dimmer shade of gray = dimmer image = *Less total knife edge
travel

After sitting and thinking about this for a while now, I've come to the
conclusion that all three variables share one common denominator.  The
WIDTH of the virtual slit created by the lateral knife edge position on the
slitless tester!

For instance,  for the software to set a consistent gray value for null, as
the image gets brighter (via increased light source intensity, increased
camera sensitivity) to keep the same shade of gray, the software has to
command the knife into the return beam to create a smaller virtual slit.
Conversely as the input image intensity is decreased, the software has to
command the knife OUT of the return beam more, creating a wider virtual
slit (to keep the same shade of gray as with a brighter input image)

Now, if the input image is kept constant, and the software shade of gray is
changed to a higher value, the virtual slit width must be made larger to
get that higher shade of gray for null evaluation.  The converse is true
here as well, a software setting for a darker shade of gray means the
virtual slit width must decrease to lower the shade of gray seen by the
software.

In all three cases, the following rules seem to apply:

#1 Larger virtual slit width = an increase in total knife edge travel
#2 Smaller virtual slit width = a decrease in total knife edge travel

I've run some preliminary tests using a Slit and Pinhole source for the
test.  So far the results show less change in knife edge readings as a
result of image intensity, however the result is considerably 'Wrong' when
compared to classic Couder Mask test results...

Can anyone help me explain WHY this is happening, or independently confirm
my results on a mirror of similar size and F-Ratio??

Here's a link to an Excel spread sheet containing the knife edge results
for all the various tests.

http://lerch.no-ip.com/atm/ZOC.xls


Take Care,
James Lerch
http://lerch.no-ip.com/atm (My telescope construction,testing, and coating site)

--- BBBS/NT v4.00 MP