Hi Clemens,
your suggested description of positive rotation doesn't remove the
ambiguity.
My intuitive way to define positive rotation (in 1994 or so, when
building the SBC) was:
Looking onto the sample, rotation of the sample in mathematically
positive direction
Only later I was told that in HKL (and I believe also d*trek) the
definition was mathematically positive direction for an observer
standing behind the rotary table looking through the base to the sample.
Not the way a user approaches the goniostat's rotary table.
But no big deal. It's a -1 in one of the axes definitions. There are, as
you stated, many more axes definitions and they depend on the geometry
of the end station. E.g., at the SBC, the orientation of otherwise
identical goniostats is mirror-image between the ID and the BM for
reasons of layout of the two beamlines relative to each other. The idea
that the axes definitions should be uniform is not practical.
Regards,
Gerd
On 12.05.2015 16:24, Clemens Vonrhein wrote:
Hi Kay,
On Tue, May 12, 2015 at 08:39:24PM +0100, Kay Diederichs wrote:
According to
http://homes.mpimf-heidelberg.mpg.de/~kabsch/xds/html_doc/xds_parameters.html#ROTATION_AXIS=
the definition of a positive value in the ROTATION_AXIS= line is:
"When looking along the axis, the crystal would rotate clockwise
when proceeding to the next data image."
I find the even better description in that part of the XDS
documentation is
The direction of the axis is chosen to describe a right-handed
rotation.
which should follow the right-hand rule a la
http://en.wikipedia.org/wiki/Right-hand_rule
The "looking along the axis" doesn't clearly define if it is (a)
looking from the crystal towards the base of the rotation axis or (b)
from the rotation-axis base towards the crystal.
There is of course no right or wrong when it comes to choosing the
direction of rotation. However, conventionally the sense of rotation
is positive; only a small minority of beamlines needs a -1 ("reverse
phi").
Yes: one can always define the rotation axis without the need for a
'-1'. But this has an impact on the chosen lab coordinate system and
therefore might require a change of INCIDENT_BEAM_DIRECTION= and/or
DIRECTION_OF_DETECTOR_{X,y}-AXIS= ... and there might be good reasons
for having those defined in a particular way (eg. to avoid a negative
value for DETECTOR_DISTANCE= or to have them aligned with the
fast/slow changing axis of the image array as X and Y).
The problem is that
a) beamlines do not usually document this on their webpages, and
sometimes change it without notice
Indeed.
Most beamlines are quite good in providing up-to-date XDS.INP
templates that are known to work with data collected on that
beamline. Ideally, the {X,Y}-GEO_CORR files for Pilatus detectors
should also be placed in a public space (and everything else that
might be required). All so that users can process the data again once
they are back in the home lab with more time and less stress - trying
to reproduce what happened by the automatic processing systems
installed on most beamlines (and through that task especially new
users will actually learn what entails good data processing practice).
Personally, I wish that beamline designers would be aware of the
potential problem for users; I suspect they often are not. Life
would be easier if all beamlines would use the same convention, and
I'm pretty sure that spindle motors can be
produced/bought/programmed for both directions.
Sometimes there are restrictions upon beamlines regarding the choice
of coordinate system to be used: this often has to be identical for
everything and all beamlines at a given synchrotron.
Reaching perfection in an imperfect world ;-)
Cheers
Clemens
