[Author Prev][Author Next][Thread Prev][Thread Next][Author Index][Thread Index]

[ATM] A Null Test



I thought I'd present some early findings...

I started tweaking a null test design using Melles Griot lenses, so designs
will be set up for those lenses: LPX309 300 mm f.l. planoconvex (D = 95  
mm),
LPK065 -600 mm f.l. planoconcave (D = 94.2 mm) and LPX303 250 mm f.l.
planoconvex (D = 101.6 mm).  I later found out that the MG site doesn't seem
to list LPX309 -- luckily, there may be several alternatives that include
(with some reoptimization):

    Ross           L-PCX357 R=155.0388 t=7.74  D=76.2
    Newport        KPX232   R=155.04   t=7.754 D=76.2
    Ross           L-PCX376 R=155.6178 t=10.4  D=95
    JML            CPX10458 R=155.6178 t=10.4  D=95
    Melles Griot   LPX309   R=155.62   t=10.4  D=95
    Optosigma      011-4570 R=155.7    t=8.2   D=80
    Optosigma      011-4770 R=155.7    t=11.2  D=100
    Linos          312391   R=156.22   t=10.7  D=100
    Spindler&Hoyer 312391   R=156.2256 t=10.7  D=100

The basic layout of the test is similar to a Ross Null Test in that the
light passes through the three lenses on both the forward and return
directions  (See Peter Ceravolo's article.).  For these setups, I
selected a 500 mm diameter mirror with conic constants of -0.8 to -1.2.
This range of conics constants cover most of several designs:
Ritchey-Chretien (both Cassegrain and Gregorian), Classical Cassegrain,
and Newtonian.  The LPX309 planoconvex lens is nearest the source/knife
with the LPK065 planoconcave lens in the middle.  An example prescription
is:

Surf      Radius   Thickness    Glass    Diameter   Conic   Note

 OBJ     298.476     76.2331                    4       0   source
   1    Infinity        10.4      BK7          95       0   LPX309
   2     -155.62    29.31369                   95       0
   3    Infinity           3      BK7        94.2       0   LPK065
   4    311.2356    35.82609                 94.2       0
   5    Infinity        14.1      BK7       101.6       0   LPX303
   6     -131.75    2035.169                101.6       0
 STO    Infinity    12.49376                  500       0   stop at 
mirror edge
   8       -2500   -12.49376   MIRROR         500    -1.2   mirror
   9    Infinity   -2035.169                  500       0
  10     -131.75       -14.1      BK7    98.14704       0   LPX303
  11    Infinity   -35.82609             96.33898       0
  12    311.2356          -3      BK7    75.90638       0   LPK065
  13    Infinity   -29.31369             74.31415       0
  14     -155.62       -10.4      BK7    59.47545       0   LPX309
  15    Infinity    -76.2331             56.18785       0
 IMA     298.476                         4.009802       0   knife

Note: Leaving the diameters computed (by ZEMAX) on the return path
      was for my benefit.

This combination is sufficient to test f/2.5 to f/3.3 mirrors:

       f/2.5     f/2.5     f/2.5     f/3.3     f/3.3     f/3.3
       K=-0.8    K=-1.0    K=-1.2    K=-0.8    K=-1.0    K=-1.2

t0    52.49215  66.03618  76.23310  74.68975  87.94469  99.42131
t2    51.76884  39.16714  29.31369  31.31073  18.07115  5.712305
t4    46.85110  40.98195  35.82609  35.26693  27.28308  19.10841
t6    2044.477  2039.650  2035.169  2847.064  2843.051  2841.557

d10   95.53217  96.89402  98.14704  76.80393  77.98929  78.76043
d12   66.53048  71.45829  75.90638  58.56477  63.47730  68.16789
d14   40.13300  50.74207  59.47545  45.78434  55.19209  63.94494

The field selected for this design was +/- 2 mm, allowing for some
misalignment of the forward and return paths.  Even if a pellicle
is used to physically separate the source and knife and allow
tighter alignment, some tolerance will be needed.

On axis, the worst case error (f/2.5, K=-1.2) is better than lambda/46
P-V (green e-line) -- in all cases, the longitudinal error changes
direction twice across the semi-aperture.

Off axis, astigmatism seems to dominate.  It seems to be fairly well
corrected here, as the worst case error (also f/2.5, K=-1.2) is better
than lambda/27 RMS at the field edge.

Faster than f/2.5, the restriction, seen at K=-1.2, is the available
aperture of the lens closest to the mirror.  Slower than f/3.3, the
restriction, also seen at K=-1.2, is the separation between the
first lens and the planoconcave lens.

Now for the cute part;-)  Interchange the two planoconvex lenses.
This combination is sufficient to test f/3.3 to f/5.0 mirrors:

       f/3.3     f/3.3     f/3.3     f/5.0     f/5.0     f/5.0
       K=-0.8    K=-1.0    K=-1.2    K=-0.8    K=-1.0    K=-1.2

t0    40.14796  55.92810  67.65114  76.38614  110.1226  104.7744
t2    81.40512  65.04342  52.67993  42.80672  5.426498  16.79347
t4    74.80451  68.78164  62.72810  55.39374  17.09104  28.63592
t6    2705.824  2702.497  2699.466  4418.633  4463.230  4397.879

d10   96.76945  97.86856  98.84198  69.25888  66.30842  73.15745
d12   60.15706  64.11848  68.02961  49.67239  59.81393  62.38797
d14   28.16121  38.27889  46.73764  38.11838  57.30619  56.95432

Again, the faster speed has aperture restrictions.  In fact, the
full 2 mm field exceeds the 95 mm diameter lens at f/3.3;  however,
the field center is still covered.  Perhaps the full field is
covered with one of the 100 mm diameter alternative lenses.

Performance on-axis is good, better than lambda/250 P-V; two
longitudinal error direction changes are still present, except
at f/5.0,K=-1.2.  Also with the exception of f/5.0,K=-1.2, off
axis correction is good, better than lambda/79 RMS at the field
edge.

What about f/5.0,K=-1.2?  That's the bugaboo.  From about f/4.3 to
f/5.0 at K=-1.2 and from K=-1.0 to K=-1.2 at f/5.0, there are two
solutions.  The solution for the rest of the region does not meet
the interlens space requirement in that corner.  For that corner
solution, the situation fortunately improves away from that corner.
At the "boundary" between the two solutions, two longitudinal error
direction changes are again present.  Performance off-axis at the
corner is also not as good, lambda/19 RMS at the field edge.

At the "boundary" between the two solutions, the interlens spacing
gets small.  I could not prove that there were any points that did
not have a valid solution.  There also appears to be some overlap
in coverage; a lot seemed to depend on the (working from an existing
solution) desired amount of change in selected parameters (f number
or K value).

Of course, of one the problems with this test is the number of
lenses that must pass muster.  They must also be kept in alignment
over the changes in spacing -- about 75 to 80 mm.

_______________________________________________
ATM mailing list http://www.atmlist.net/