To: atm{at}shore.net
From: Jim Burrows 
Reply-To: Jim Burrows 


At 22:49 2003-03-26 -0400, Alfredo Neves wrote:

>I've been using a test which is a cross between the caustic test and a
>hartman test. Using a mask with holes over the mirror and a caustic
>tester, I perform two sets of measurements. The first just a bit inside
>ROC, where I record the lateral (side ways) positions where each hole
>darkens to a minimum. Next I move a fixed amount to the outside ROC and
>also measure the position where each hole darkens. With the two sets of
>measurements and the distance between the two ROC, I calculate the ROC
>of each zone as you would with a hartman test. The advantage is no grey
>matching and no complex numeric computation as in the caustic test.

The true Hartmann test doesn't work with curvatures.  Suiter, p. 285, says,
"Finally, the corresponding dots in the two pictures are connected
mathematically."  Essentially, you're doing a backwards ray-trace:  a
line through three points (the two dots and the center of the mask hole), a
line from mask hole to source, plus angle of reflection = angle of
incidence gives you the normal to the mirror patch seen through the mask
hole.

I would recommend not processing your data in this fashion because your
short base-line (distance between the two fixed longitudinal distances)
combined with lateral measurement errors gives large slope errors.  A
better way would be to regard the two sets of measurements as two LWTs
(Lateral Wire Tests) and gain the benefit of averaging.

At 08:49 2003-03-27 +0100, Nils Olof Carlin wrote:

>You say "no complex
>numeric computation as in the caustic test" - I don't see that there
>would be much difference, and if you have a computer that can run
>SIXTESTS, you can let it do the computing for you.

If we regard this as two LWTs, as recommended above, there are two ways to
use the data in Sixtests: 1) just cram both sets of measurements into the
input file (being careful about the signs of the lateral numbers:
positive/negative for inside/outside focus) (I think this would work - not
sure), 2) make two runs and do an eyeball average of the results.  Probably
1) would show a rougher surface (more but smaller peaks and valleys).


         -- Jim Burrows
         -- mailto://burrjaw{at}earthlink.net
         -- http://home.earthlink.net/~burrjaw
         -- Seattle N47.4723 W122.3662 (WGS84)

--- BBBS/NT v4.00 MP