Go up to 5 Common problems with wavefront reduction.
Go forward to 5.2 Problems with the program.
5.1 Problems affecting the the reconstructed wavefront.
If you have not selected manually set extra-focal image diameter, the
expected size of the extra-focal image is computed from the f-ratio and
the extra-focal distance. In either case the CCD camera pixel size
is then used to compute the number of pixels expected across the
extra-focal image. This is one of the most critical numbers needed for the
program, and you should try to specify parameters so that the diameter
is known within 1% or better.
An error in this diameter (a scaling error) will lead to errors in
determining the wavefront.
Unfortunately it is not possible for the program to detect in an general
manner when this happens.
Often you will have little indication that the wavefront reconstruction is
incorrect when the program uses an incorrect extra-focal image size.
When the data reduction has finished, both corrected extra-focal images
should be the same size.
If this is not the case and the reconstructed wavefront contains a large
focus term, then you probably have either problems background subtraction,
or the integrated flux in each extrafocal image is different.
To see the corrected extra-focal images, carry out a data reduction
sequence, then invoke the Displays :: Display buffer menu item.
In the resulting Display image buffer window perform the following
operations:
- Select the Display destination radio button labeled Left.
- Select radio-button labeled Corrected extra-focal image 1.
- Press the Display button.
- Select the Display destination radio button labeled Right.
- Select radio-button labeled Corrected extra-focal image 2.
- Press the Display button.
You should now have both corrected extra-focal images displayed, one in each of
the display buffers. The images should be the same size and circular.
The images should be uniformly illuminated, except for high frequency
structure that should appear positive in 1 image and negative in the other.
The image radii should be roughly that specified in the
Desired radius of output wavefront field of the
Edit :: Algorithm parameters form.
If one or more of the above conditions is not true, a problem with the
data or the data-reduction is strongly indicated.
There are two main causes for an incorrect focus value, either a bad
backgroud subtraction, or different different integrated intensity
in the two extra-focal images.
Verify that bad background subtraction is the problem by loading your images
in the usual way, and checking the background values, in comparison to the
values inside the extra-focal images.
If the background values are all of the same sign, and exceed a few
percent of the image value, then you have inadequate background subtraction.
Sometimes you may find that the background varies with focus position.
In this case you can solve the problem by taking one dark image for
each extra-focal image position. Since Dark images are subtracted
when the extra-focal image is first loaded into the program you can
apply these dark's to the appropriate ef image by following the
following sequence of operations:
- Load first dark frame.
- Load first extra-focal image.
- Load second dark frame.
- Load second extra-focal image.
If this does not work, you need to improve the baffling of your
extra-focal image camera, or use a separate image processing package
to pre-process your images.
If you wish you may modify the image loading code (found in the
file ef_file_menu_open.tcl) to apply more sophisticated processing.
If the residual from background subtraction is not uniform
over the extra-focal image, you will also see spurious higher order
aberrations generated by the program.
If the background subtractions appears to be satisfacory there are
still focus errors in the reconstructon, you should suspect
that the extra-focal images have different integrated light levels.
To test this try enabling theNormalise extra-focal images toggle in the
Edit::Image pre-processing form. Now re-load the extra-focal
images and try the data-reduction again.
The normalistion fuction is normally disabled, because it can introduce
problems if the images are too close to focus, or if there is any
structure other than the extra-focal images themselves. Do not use
this function unless it is really necessary.
In general the curvature technique has lower SNR for aberration terms
which have no curvature. These terms include tip/tilt, astigmatism
and Quad astigmatism, Zernikes 2,3,5,6,14,15 respectively. Zernike
14 is particularly prone to error, because it has the
same symmetry as the image pixelation. If the reconstructed wavefront
map contains a cross pattern which aligns with the x,y axis, you
should carefully inspect your extrafocal images for any of the
following problems.
- Large defocus.
If the extra-focal images have a large differential defocus, one of them
will have to be enlarged and the other reduced in size during the computation.
It is very difficult to perform a large change in scale without introducing
some artifacts from the original pixelation. This will in turn lead to a
spurious value for Zernike 14. For this reason it is best to obtain
extra-focal images as symmetrically about focus as possible.
- Differing intensities.
If the extra-focal images have differing integrated intensities you will
observe a spurious focus and possibly Zernike 14 term.
Try enabling the Edit::Image pre-processing
Normalise extra-focal images toggle and re-loading the extra-focal
images.
- Inadequate background subtraction.
If the background outside of the extra-focal images is more than a few
percent of the extra-focal image intensity, a better background
subtraction is called for.
- Rolled edge on wavefront.
Having a severely rolled edge on the wavefront to be measured makes the
use of curvature sensing less accurate.
See chapter
on optical configuration
- Evidence for nonlinearity.
This program only works correctly if your extra-focal images are
outside the caustic zone (region where rays cross). If the
modulation in your extra-focal image is within +/- 30%, your
images should be safely outside the caustic zone. If your
image modulation exceeds 30%, then your images may have parts which
are within the caustic zone. This reduces the accuracy of the
reconstruction, and may cause Zernike 14 to grow. Since wavefront
reconstruction degrades quite gracefully as your extra-focal
images approach the caustic zone, you may choose to accept
less accuracy in return for working closer to the caustic zone.
You should find that the residual wavefront aberrations for those
Zernike terms that are corrected by the algorithm are less than
10 percent of the total aberration. If this is not the case, and
aberrations do not fall rapidly with continued manual iterations,
then you may have a problem with large high order aberrations.
Usually this will involve large aberrations near
the edges of the wavefront.
Evidence for this will appear as very bright arcs, and dark areas near
the periphery of the extra-focal images.
Essentially these parts of the extra-focal images may be near to
non-linearity, while the rest of the image is good.
Under these circumstances, there is likely to be some cross-coupling
between high order aberrations an low order aberrations, with
quite poor accuracy on the reconstructed wavefront.
If the quality of the data is good, you may try using the
highorder reduction sequence, available under the
Edit :: Select reduction sequence. This sequence will
correct higher order Zernikes and potentially give more
accurate results. However with data of questionable
quality, or with a very large central obscuration,
the algorithm may become unstable in this mode.
Instability is indicated if the two corrected extrafocal images
are not circularly symmetric and/or significantly differ from
each other.
Laplacian Optics Inc. Email: laplace@laplacian.com