Dear microscopists,
thank you to everybody who gave a reply on the pixelshift in videomicroscopy.
Here are some details of our setup:
we have a Leica DMIRBE, the filterwheel and the camera is mounted on the top
port.
I tested the pixelshift with oil immersion lenses: 20x NA07, 40x NA1,25, 63x NA
1,32, 100x NA1,4 (all lenses are PL APOs). The pixelshift was always in the same
direction. I tried to screw out the 40 x lens 1/4 turn; the pixelshift did not
change.
I did not change the dichroic, because it is a dualbandpass dichroic for CFP/YFP
and FITC/Cy3, the only moving part is the filterwheel. Excitation was changed by
the monochromator without touching the system.

I also tried to put the dichroic out of the microscope (this excludes the
dichroic from the beampath), using transmission white light instead of
fluorescence light and a objective micrometer as a test sample. I was only
switching the emission filters by turning the emission filterwheel. This gives a
RGB-image when using a red, green and blue filter in front of the camera. The
pixelshift is still there.

We use filters from AHF-Analysentechnik (as far as I know theses filters are
made by Chroma). The beamshift of the filters was  measured with a special
equipment of the company. The beamshift of all filters is in x-axis = 0'' in
y-axis= between -2 and -8,5 ''.The company told me that this could not cause a
pixelshift.

Thank you for your suggestions
Annett