TULL'S 13 DECEMBER, 1996 EMAIL


From rgt Fri Dec 13 13:37:39 1996
Date: Fri, 13 Dec 1996 13:21:30 -0600
From: rgt (Bob Tull)
To: observers
Subject: 2dcoude performance
Cc: don, ebakker, esb, pjm, pwk, rgt, tgb, wdc
Status: RO


If you have no interest in 2dcoude you can delete this message.

        The 2dcoude' high-resolution focus F1 has shown resolving power 
as high as 270,000.  However, observers have reported variable 
resolution and I found last February that one quadrant of the collimated 
beam produced an image shifted in wavelength, reducing the overall 
resolving power. Maximum resolution could be obtained by blocking the 
offending quadrant, with a resulting loss of efficiency.  Further work 
in May resulted in improved performance by optimizing the collimator 
focus, but following observers continued to report lower than desired 
resolving power.  

        Recently Bill Cochran found that spectra showed image structure 
which was resolved into doubled monochromatic images in the spatial 
direction.  He suggested it is due to the partial loss of refractive-
index-matching fluid in the interface between prisms.  Ed Barker found 
that the two images were separated by, typically, 10 pixels at the F1 
focus.  He further found that the second image could be eliminated by 
blocking the upper half of the collimated beam, corresponding to the 
region of the cross-dispersing prisms where index-matching fluid has 
been lost.  Don Barry and I inspected the prisms during Ed's coude' run 
in early December.  This interpretation seems to be correct:  the 
variable performance of the F1 focus appears to be related to variable 
area of the index-matching fluid.  This could also explain the 
observation by Mike Marcario that the Ronchi shadow pattern seen 
visually at the F1 focus markedly improved when I tightened a retaining 
screw on the prism mount, squeezing the prisms together. 

        White light Newton's Fringes due to interference in the "un-
oiled" region showed a wedge of about 8 fringes in the air gap, when that 
gap was of order 1/4 the height of the prisms.  We can estimate the 
separation of the images due to the two regions of the prism by use of 
the "thin prism" formula 

        deviation = (n-1) alpha

where alpha is the apex angle of the thin prism.  There are two prism 
interfaces used in double pass, each with similar air gaps forming an 
air wedge of about 8 visible fringes over about 50 mm of prism height or 
about 0.002 degrees in each prism interface.  Applying the thin prism 
formula to 4 prisms in series, with apex angle 0.002 degrees, actually 
does predict an image shift of about 10 pixels in the vertical 
direction.  This seems to confirm the interpretation.  

        Due to the shorter focal length at the R=60,000 F3 focus the 
effects are reduced by a factor 4 and will not normally be a problem at 
that focus. 

POSSIBLE SOLUTIONS.  

        The loss of fluid in the prism interfaces has not been 
explained.  Don and I added some fluid, with the prisms in place, but 
not much actually seeped into the interface, probably due to the 
compression of the prisms together by gravity.  Clearly the fluid does 
not form a good permanent interface and something better is needed.  
Possibilities: 

        (1)  Clean the prisms and send them to Continental Optics for AR 
coating of the interior surfaces.  Re-mount them with an air gap. 

        The reflectance for one coated surface averages 0.5% over the 
spectral passband.  The total loss due to reflectance at these surfaces 
will be of order 4%, equal to the loss at the exterior coated surfaces.  
This is probably the most reliable solution and the light loss is far 
less than that due to masking off part of the beam.  The prisms are 
tilted at an angle that will prevent any reflected light from becoming 
stray light at the camera focus.                        

        (2)  Cement the surfaces.

        These are rather large surfaces.  It requires a professional 
optician to accomplish.  I don't know the probability of success but a 
failure could result in loss of the prisms. 

        (3) Put a spacer shim between the prisms and fill the gap with 
fluid.  This is the cheapest and easiest solution; we could try this 
first, then later consider coating the surfaces.  The only concern I 
have is that any distortion of prism shape, due perhaps to imperfect 
compression of the shim along the prism apex, might result in a non-
uniform fluid thickness resulting in wavefront distortions.   If the 
loss of fluid is due to, e.g., its absorption into the RTV cement used 
to seal the joints, this might solve the problem by separating the fluid 
from the sealant.  We can try this with a couple of pieces of thick 
plate glass. 

        I solicit your comments or alternate suggestions.





Bob Tull