This memo describes beamline improvements and problems encountered commissioning the liquid surface spectrometer in the C hutch. This memo is being written partway through (6,13 jun 00) this commissioning run (3-27 jun 00).
Let me start by mentioning how hard everyone worked, how much support there is here, and how much I appreciated finding that all the LSS tools and gizmos were right in the toolbox or cabinet where I expected to find them. I appreciated that a LOT. Thank you!
Spec 4.05.08 will be delivered to scooter.cmc.aps.anl.gov, probably installed around 13 june 00. This includes a fix to the previous problem of which Epics motor record Spec needed to monitor. Until it's installed, we still need to do "reconfig" whenever a parameter is changed in Epics. The deadband problem also still exists until then.
Upon trying to set up two scaler devices, with the second one in the C hutch, we found that Spec only allows one Epics Generic Scaler to be defined. (Workaround: long BNC cables going from the amplifiers in the C hutch to the scalers in the B hutch through the holes in the walls.) This would be a trivial thing to change in Spec except that one has to decide how to do the gating. The option we discussed is to allow a second scaler to be configured, but not allow that device to have a timer. We would then run a cable from the gate output on the first unit to the gate input on the second. We will test this configuration when the upgrade is done.
Found and fixed bugs in mirror.mac, slit.mac, surf.mac, and opco.mac; added filter-box.mac and sem_util.mac: see macro notes. Still to do:
Three-phase motors have been installed and tested on the LSS, and all LSS motors now live on the crate "9idcs1". For now, channel assignments and driver current listed in the motor info page are totally out of date. An updated version will be created at some point here, and this will include documentation on the Spec/Epics motor parameter utility macros in sem_util.mac. Motor record fields for all three crates were saved in Surf, and for the B crate in Fourc (to get the six-circle parameters). The recovery files are 13jun00.sem and 13jun00_fourc.sem respectively.
For the most part, the 3ph motors work very nicely with the default resolution of 1000 steps/rev. For the phi motion at 19 keV, I went to the second coarsest microstepping mode (4000 steps/rev). Here, I encountered problems with reproduceablilty and had to slow down the motion, especially the backlash. It's worth mentioning that my setup is sensitive to steps of 0.0002 degrees and is much less forgiving than what you'll get working at reasonable x-ray energies of 8-10 keV.
I encountered some sign that sample height was not behaving within 0.05 mm, but am not satisfied that I really know what the problem was: more later.
Two four-phase motors remain in the system: phix, which works fine, and the filter wheel. The filter wheel is not losing steps, but occasionally the holding current seems to glitch so that in the middle of the night, the wheel slips to block the beam. We plan to replace this motor for the liquids run.
A problem reoccurred that I encountered last run: the XIA slits don't seem to like being scanned. By XIA slits I mean the Huber slits attached to the XIA serial controller. I find no problems with individual motions, but when I do a scan (especially two blades at once), the slit jaws move to apparently random positions within the scan interval. It does not seem to help much to make the step sizes bigger (as if it's a resolution problem) or to count longer (as though commands are coming too fast for the serial port to handle). After the scrambled scan, the jaws go back exactly where they were. So it looks like the command telling them where to go just gets screwed up somehow. Because all motions of the slit gap can be scanned with other motors on the LSS (the rotation stages for example), and then the gap can be easily repositioned, I didn't spend much time trying to solve this problem. Someone should do it though.
A note about counting to monitor using a negative argument as the count time: Epics looks for which counter was designated a "monitor" in the spec config file. If for some reason you want to change the monitor without going through the Spec config, the function counter_par(mne, "monitor") (where mne is the mnemonic of the counter you want) will do the job. When the config file is read again, the monitor will revert to the one defined there.
This is an aluminum coated mirror. At 19 keV, we set the grazing angle to 1.4 mrad, which illuminates essentially the whole thing. It appears to be doing some unauthorized focussing with the focal point in the B hutch somewhere. Just before the mirror, the vertical beam size as gauged by slits of 0.6 mm. In the front of the B hutch, the beam is focussed to 0.25 mm or so. In the C hutch, we can resolve some structure in the reflection, which has a batman shape and is about 0.6 mm high, again. We satisfied ourselves that this was not caused by detuning the undulator gap.
For the 19 keV run I decided to cut the beam to 0.3 mm before the mirror, which gets rid of some of the goofy shape. At lower energies it may look better since we can go to a larger grazing angle. We did not find evidence for anything really awful on the mirror surface. Because the ion chambers are not normalized perfectly on the two sides of the mirror, it's hard to say exactly what the reflectivity is, but it's certainly in the range of 50-70%.
The mirror actuators do not work very well. This makes it frustrating and difficult to really study the mirror position, since it does not return reproduceably from a move. Improving this should be a high priority.
The spectrometer was moved on airpads into position about a foot from the front wall of the C hutch. We rediscovered that fact that it sits too high to operate with the SOE mirror out. With the mirror in, though, there are no height limitations: at 19 keV, the crystal stage is about 10 mm above its bottom position.
Tracking at 19 keV is extremely sensitive to the horizontal crystal position, the orientation of the theta stage, and the beam tilt parameter. By iterating and repeating certain tracking steps for a couple days, I got pretty good results, and expect to do better for the lower energy run. As far as I can tell, beam is stable, important motions are reliable, and it should work fine. I will be happier, though, when I find a really good recipe for determining the horizontal crystal position.
Due to the length of the new 3ph phi motor, the chi motion is limited to 39 degrees or about 2 inverse angstroms. A right angle coupler would fix this problem and prevent me from jamming it against the theta stage as I did last week.
Add these to my wish list:
Naturally there will be a highlight and a fine web page. We (ED and J O Fossum from NTNU) are looking at phase separated clay gels, where the flat clay particles are suspended in aqueous salt solutions. The LSS allows us to distinguish between particles oriented in different directions in each phase. The unique capability we have here, scattering in both horizontal and vertical planes, scanning our strata of layered liquids while keeping them fixed in orientation, lets us distinguish between changes in orientation and changes in particle density among the phases. We definitely have found evidence that populations of horizontal and vertical platelets are different in different phases; what we hope to find is evidence for liquid crystal ordering.