Argonne National Laboratory

Advanced Photon Source
Sector 7

Argonne Home > Advanced Photon Source >

MHATT-CAT Monochromator stability tests FY01 run2-4


Eric Dufresne, Dohn Arms, MHATT-CAT, first started in March 2001
(www.mhatt.aps.anl.gov/Sectors/Sector7/Operations/FY01_run2-4/)



Table of Content:

Introduction.


The work shown below is being done to improve the stability of our LN2 cooled 
HHL Monochromator. It contains mostly time serie analysis of data taken 
with the new first crystal mount. This continues the installation notes of the
High Heat Load Monochromator First Crystal Mount. 
The data will be taken and reported on an ongoing fashion and starts on 02/28/2001.

Monday, February 19, 2001: Post Scriptum.


P.S.  Note that the intensity fluctuations that were seen on January 2001 have 
pretty much disappeared following an orbit correction performed on 2/15/2001.
During the last alignment of the mono, some mistakes were made (possibly the 
offset is a little too large) which thus sent the beam a fraction of a mm to 
high. After the micromono exit Be window in the 7ID-B hutch, the beam had a 
sharp horizontal edge blocking the top part of the beam. We've known for a
while but we did not have any diagnostics in the beamline until last December
2000. On 2/15, we looked at the beam in 7ID-A just after the mono and found it
was not appertured at all. So it is certain that the edge is either caused by 
the P4, or an aperture in the micromono (a Be window maybe?). This will have
to be investigated further in March 2001. 

Because the edge cut off the beam on axis so much, what I aligned on was 
a low energy wing of the undulator spectrum. It was shown by Sandy et al. 
(JSR (1999) 6, p1174.) that the beamprofile at an energy about 0.5 keV below 
a 7.66 keV fundamental at the APS has a double peak profile with less intensity 
right on the undulator axis. What I did turn out to do was just to align our 
first crystal at 10 keV and peak the total integrated flux at 10.8 keV thus 
the bottom peak on the double peak profile was not apertured and it peaked 
up for a gap of 10.8 keV. The horizontal edge really blocked the on-axis 
flux. This explains the poorer flux that was seen by 7ID-C experimenters during
their experiment and likely the low gain seen with the new mirror. 

On 2/15, Dohn Arms, Steve Dierker and I performed a 40 microradian orbit 
steering correction and we were able to clear the aperture. We now have a 
10 keV X-ray peak in the flux with a gap set to 10.05 keV (like before), 
the beam is on axis. The total flux is now as much as 2.5E13 ph/s/100mA, which 
is about a facor 2 better than the best we've done before. So this is good news
for future experiments with the HHL monochromator. 

Below is a time serie of the total flux in 7ID-B. The total flux is now stable
to about 0.35 % peak to peak over one hour. Note that the time fluctuations of
the total intensity still have some periodicity, but they have improved by an
order of magnitude almost by resteering the beam. The beam is now far more 
stable than it has ever been. 
Latest time series showing 0.15% stability in the flux!

The total 10 keV flux measured in 7ID-B with the white beam slits set to 2.5 mm (V) by 1 mm (H).
The total flux is limited by the L5-20 to 1.7E13 ph/s/100mA here and the peak to peak fluxtuations
are only 0.35 %! We are now far less sensitive to the EMW than in January, and indeed the APS
improved their orbit stability implemented by Louis Emery. The LN2 cryocooler filled 20 minutes
before the end of the time series showing that we are now pretty insensitive to it.

Wednesday, February 28, 2001: Commissioning period til 3/12.

Today, I switched back the operation from white to monochromatic mode. Some 
of the problems I had during this mode was that the ORNL group had not moved 
the tank back into white beam passthrough mode. Also the white beam slits 
database had been modified with the following offset between the user and 
dial values, the dial coordinates not having been modified: 

m5 offset=-0.2665 (mm)
m6  ""   =-1.109   ""
m7  ""   =-0.3335  ""
m8  ""   =-1.071   ""

The FEV also had to be reopened, and it closed typically once a day during the 
ORNL run. John Tischler explained further that the APS added vertical 
corrections to their feedforward algorithm. The ORNL group complained about 
beam instabilities also and after several consultations on Friday 2/23, the APS
realized that there could be substantial vertical motion due to the EMW. In the
end, Louis Emery added vertical correections which John and Ben think fixed the
problem. I took data during the orbit correction which I will show on tomorrow's
entry. 

Below is a long time serie (Fig1) taken late on 2/28. Three
LN2 fills are seen as dips, the first started just before the beginning of
the time serie. The data from time t = 20000 til t = 27000 is a drift which
I suspect was due to the ring which dumped suddenly and stopped the time serie. 
Ignoring the last part of the data, the overall stability is at the few percent
level. In the first 20000 seconds (5.5 hours), the intensity rises by 1.8 %. 
Latest time series!

Fig 1. The total 10 keV flux measured in 7ID-B with the white beam slits set to 2.0 mm (V) by 3 mm (H).
The time serie is cut by a beam dump, and the drop after 20000 seconds is likely related to the dump.

Thursday, March 01, 2001: Orbit correction data.

Fig. 2 shows the total flux during two separate orbit
corrections. I used one of the 7ID BPM (vertical angle FE:07:ID:SR:VANG:CC)
on the X-axis. The data for 02/28 saturated at 17.8x10^12. I retweaked the 
second crystal shortly after and got more flux, but this also steers the beam
slightly. On 03/01, the APS did not restore the orbit after the beam dump, so 
the orbit correction was repeated, and to my surprise did not overlap with 
yesterday, likely because of the retweaking. On today's curve, three edges are 
obviously cutting the beam. Wow. 

Fig. 3a shows a time serie after the orbit was recentered.
The total flux is drifting by 2.2 % over 2.8 hours, and the LN2 fills reduce the
intensity by as much as 0.8 % during the fill, nothing compared to the previous
mount. Fig. 3b shows the stability of the same time serie
over one hour. The short time scale fluctuations are at the 0.1 % level, the
drifts being higher.
Orbit correction data 02/28 and 03/01.

Fig 2. The total 10 keV flux measured in 7ID-B with the white beam slits set to 2.0 mm (V) by 3 mm (H)
during two different orbit corrections, one on 02/28, one on 03/01. One would expect the two curves would
overlap, but on 02/28, the mono was retweaked after the orbit correction, thus this steered the beam as well.
There seems to be 2 or three edges in the beam at various positions...

Time series, 03/01.

Fig 3a). The total 10 keV flux measured in 7ID-B with the white beam slits set to 2.0 mm (V) by 3 mm (H).
The time serie is cut by a GV3 closure. The drop due to LN2 fills are 0.3 % and 0.8 % deep.
The total flux drifts by 2.2 %.

Time series, 03/01.

Fig 3b). The total 10 keV flux measured in 7ID-B with the white beam slits set to 2.0 mm (V) by 3 mm (H).
The time serie is one hour long and expands Fig. 3a from t=2000 s.

Friday, March 02, 2001. Realignment.

Today, I realigned the white beam slits to be centered on the white beam, after
its orbit correction. I forgot to move the mono chamber down. It is also too 
high. I still have to do this.

Fig. 4 shows the dependence of the total monochromatic flux
versus the L5-20 opening, scanning either the horizontal or vertical opening.
Note that one gets half the total flux when  the vertical opening is 0.3 mm, 
and 97 % of it with a 1 mm opening. Thus the 2 mm vertical opening was too large
since it adds twice as much power for 3 % more flux! I decided to set the L5-20
to 3 mm (H) by 1 mm (V) for tonight's data acquisition. In Fig.4, a 0.85 mm 
horizontal opening yields about 50 % of the total flux.

Note that my formula for normalizing the flux in Fig 1-3 did not subtract the
dark current of the ion chamber. This was about 900 counts per seconds, or about
0.3 to 0.6 % of the signal depending on the beam current. One can see that it
becomes difficult to measure stability better than 0.1 % with an ion chamber
without some careful dark noise corrections. When the beam current decays,
my scaling was possibly introducing some small intensity drifts given that the
dark does not drift over time.  I thus decided to restart the data collection,
balance the noise, and subtract it in the formula. 

Big news too (I learned this at CAT-CHAT), the EMW wiggler will not be pulsed
this run, and it was not pulsing today. This will be good for experiments this
month. There will also be top-up operation this month.

Fig. 5a) shows a time serie lasting 4.7 hours, with a 1.1 
% increase in flux, and about 0.7 % dips during LN2 fills. A finer time 
scale lasting 500 s is shown in Fig. 5b). The intensity 
fluctuations were measured at roughly 2 Hz, and the total range of variation 
is about 0.2 %. Fast fluctuations are obvious, cohabiting with slower 
oscillation on a few minutes time scale. The flux is overall quite stable over
this time scale. Over the next days, I will try to calculate the spectrum
of these fluctuations.
Total flux versus L5-20 opening.

Fig 4. The total 10 keV flux measured in 7ID-B versus the white beam slits opening.

Time serie 03/03 of the total flux.

Fig 5a). The total 10 keV flux measured in 7ID-B versus time, starting in the middle of a LN2 fill.

Time serie 03/03 of the total flux.

Fig 5b). The total 10 keV flux measured in 7ID-B versus time for 500 s.

Weekend, March 3 and 4, 2001: long time serie.

Saturday, I took a long time serie overnight, starting shortly after the beam 
was reinjected in the machine around 7 pm. Dohn opened the shutter once the 
beam was delivered between 7 and 8h30, and I started the time serie at 8h30.
The mono was cold for several hours, so this time serie is useful as it shows
thermal drifts over long heating cycles. This was repeated a bit more carefully
on Sunday 03/04. The photon shutter was opened a few minutes before the time
serie was started. 

The data from 03/03 shows a fast (2 hour relaxation, ~5% effect) followed by
a slower relaxation (5 hours ) starting about 4 hours in the time serie and
lasting til the end or 9.4 hours. The data from 03/04 shows a more pronounced 
effect (8.5%) with most of the intensity gain occuring within 40 minutes of the 
start of the time serie. 

These heating effects should be more obvious at closed gap and smaller energy
because the deposited power is larger. For most experiments at the moment, 
smaller beams, typically 0.5 mm by 0.5 mm, are used thus the power is about a
factor 10 smaller than in these tests. One would hope that these heating effects
will be far less pronounced for typical experiments. I recall also that heating
effects were also noticable with the original HHL first crystal mount.
Time serie 03/03 of the total flux starting at 20h30.

Fig 6. The total 10 keV flux measured in 7ID-B versus time on 3/3 (black) and on 3/4 (red).

Monday, March 5 2001: Setting up the new LCL-903HS with a YAG.

Today, I spent most of the afternoon and evening setting up a WATEC LCL-903HS
with a YAG fluorescent screen to image the full monochromatic beam. The new 
cameras are nicer than the old one, i.e. higher resolution, sensitivity, and
settings which makes it a nice addition to our cameras. See this for details.
This screen is needed to align the mirror filter, but I though I would also 
see how stable the beam is now. We still see some motion. During the day, I 
also took some time series, showing similar warming up curves as yesterday.

Tuesday, March 6 2001: Machine study day til 4pm.

Tonight after dinner, I closed the white beam slits to 0.1 mm by 0.1 mm in 
preparation for the Mirror Filter commissioning and I am taking a time serie 
with a small beam to see how stable things can be. I also improved the optics
for observing the pink beam.

Wednesday, March 7 2001: Mono maintenance.

This morning, while completing the time serie from yesterday, I noticed that 
the mono needed maintenance, so Dohn and I had to pump again the vacuum jackets
of the tank and cryocooler, and fill the LN2 Buffer. It was quite low at 40%. 

I continued the time series started last night and finished it this evening at 
10:50pm. The time serie started one half hour after the L5-20 had been cut 
from 3 mm x 1mm to 0.1 mm x 0.1 mm. So one would expect that significant thermal
drifts of the first crystal mount would occur which may explain the intensity
increase from t=0 to t=43300s. There are two APS fill-on-fill in this time
serie, one at t= 20200s one at 63300s. These fill-on-fill are followed by a
flux decay lasting about two hours. Note that the flux decay in about 2 hours
also when the beam is on again after t = 52600s. This thus seem to indicate that
some thermal drifts causes an angular mismatch of the beam which relaxes in 
roughly two hours. The effect can be as large as 30 % or so, which may be 
worsened  by the fact that 100 micron is comparable to the APS beam size in 
7ID-A. Note that two hours is also roughly the time period noted during the
time series of the full APS beam. 

The sudden dip then increase at t=37400s is a tweak of the second crystal.
There was a gain of 5 % in flux after the tweak. The 7ID-A hutch was reopened 
for some monochromator maintenance which made me concerned because the closed
loop pressure was quite below the set point, and the Buffer level was too low.
Fig. 8 shows a time serie started at the beginning of a LN2 
fill. A small dip is noticable during the fill, but a large 15 % drifts occur 
during the time serie. 

For a small beam, the monochromatic flux shows surprising slow dynamics which
for XPCS is a nuisance since it would require good coherent flux monitor data
recorded simultaneously with the CCD data. A monitor will be needed for sure 
if we use the monochromator.
Time serie 03/07. with a small beam (0.1mm)^2.

Fig 7. The 10 keV flux measured in 7ID-B versus time with the L5-20 set to 0.1 mm by 0.1 mm.

Time serie 03/07. with a small beam (0.1mm)^2.

Fig 8. The 10 keV flux measured in 7ID-B versus time with the L5-20 set to 0.1 mm by 0.1 mm.
The time serie is started just at the beginning of a LN2 fill.

Thursday, March 8 2001: Correlation between BPM and flux data.

Today, a lot of visitors were around so it was fun, but it took a while to get 
back to commissioning activities. Of interest, I analyzed the time serie from 
3/8/01 started at 01:30. It was again interrupted by GV3 closing around 7am. 
I noticed drifts of 0.93 microrads in the BPM vertical angle with a time scale
of 2.23 hours. A drift is present also in the intensity, with also a 2.13 hours
time scale, so it appears that the BPM tracks the flux increase. 
Time correlations of the BPM vertical angle shows two decay processes, a quick
one which decays in 45s, and a slow one which decays in 2000 seconds. A similar 
quick decay in the flux is observed (54s), followed by long decaying 
oscillation that peak again at 1055 s, so the flux has slow oscillations on
the order of 18 minutes which are not present in the BPM signal. The beam moves
on the L5-20 by roughly 25 microns which is not sufficient to explain the gain
in intensity. Today I fit the data from Fig. 4 with an error function to
determine the beam size on the L5-20. It is 222 um in the vertical which
compares reasonably well to the expected 175 um. The beam center position moving
by 25 um can only accout for possibly 1% of the flux variation, not the 12%
observed. I don't undertand this yet. I have to input figures here when I get 
a chance. 

Friday, March 16 2001: Intensity stability in top-up mode.

Fig. 9 shows the total flux in 7ID-B versus time for five
hours during top-up operation on 3/16/2001. The top-up operation was interrupted
temporarily at t=6273s, and the beam decayed from 102.05mA to 100.65 mA until
7127 s when top-up was restarted. Then the full current was ramped up and
restored by t=7942s. The policy for top-up is to slowly ramp the current up if
the decayed current is above 90 mA, but to operate at the decayed current if it
is below 90 mA. The intensity in Fig. 9 is stable to 0.29%!.
This is excellent. Note that the beam intensity drops by up to 0.075/4.4=1.7%
during a top-up fill, which implies a 17 ms loss of flux due to top-up given
that the scaler had a one second counting time and the SR-570 had no filter set.
For more details on top-up, see my December 2000 summary on top-up.
A lot of the drifts seen earlier in March have disappeared in top-up operation.
See for example Fig 5a or Fig. 6.
Someone after my TWG talk mentioned that the heating effects I mentioned 
earlier are likely due to the second crystal heating up from Compton scattering
off the first crystal. During shutdown, we should think of setting a Cu
braid on the second crystal to cool it down to LN2 temperature.

Time serie on 03/16.

Fig 9. The 10 keV flux measured in 7ID-B versus time in top-up mode.

Saturday, March 17 2001: Energy stability.

David Reis and 7ID-D coworkers wanted to investigate whether the monochromatic
beam energy could fluctuate with time in a significant way. They set up a GaP
absorber in front of the ion chamber in 7ID-B, set the monochromator energy
to the Ga edge and monitored the intensity fluctuations. On the Ga edge
scan(not shown), the slope is 2.5x10^4 cps/eV. The Ga edge was not very sharp 
though, perhaps because of the chemical bounding to P or the thermal bump on the
first crystal?. Fig. 10 shows that average intensity is stable
to 0.133x10^4 cps, thus multiplying with the Ga edge slope, we get 53 meV
stability! This had not been done before, so it is nice to have this proof of
energy stability. 
Ga edge time serie.

Fig 10. The 10.363 keV flux measured in 7ID-B versus time with a GaP thin sample in front
of the 7ID-B ion chamber.

Tuesday, May 9 2001: Summary of our commissioning activities
for the April 2001 shutdown.

In the last three days of the shutdown, I was able to reopen the monochromator
and perform the following activities. The monochromator chamber was reopened. 
I noticed that the crystal surface now is much cleaner than what it was when I 
installed the crystal in the chamber in January. A new beam footprint  has been 
made uncovering beautiful mirror finish Si. With the monochromator tank open, 
I rechecked the alignment of the crystal. The horizontal slide and the first 
crystal surface are off by 0.6 degree, the first crystal being 0.2 degrees from
the horizontal while the horizontal slide is 0.4 degrees from the horizontal.
Chi is only off by 0.1 degrees. This misalignment occured when we retightened the 
seal one night at 5 am in January 2001, causing the crystal to move in its mount.
This misalignment resulted in an offset of 36 mm at a Bragg angle of 15 degrees,
thus 1 mm larger than the usual. This is why when we resteared the beam down, we
got the flux back. In the future, the orbit can climb up again although the beam
obstruction, 1 mm above the nominal height,  is still present and should be
investigated... I fixed the offset to be 35 mm at 15 degrees. We should still 
realign the first crystal in the tank when we have a chance in August.

An RGA was installed and mounted on the HHL Mono tank in 7ID-A. With this RGA, we 
now  have a tool to investigate vacuum problems in the mono and specifically leak 
detect a LN2 seal. Today, I used the SRS RGA100/12 to leak test the crystal LN2
seal and found no evidence of any leaks. On 05/07, I also found no leaks with a 
borrowed leak detector from the APS-AOD division. The leak detector works great!
Fig. 11 shows two scans of the RGA proving how good of a 
cryopump the first crystal and its LN2 lines are! The total pressure almost 
dropped on order of manitude from 1.6E-7 to 2E-8 Torr. 

Before closing, I took some measurements for adding a Cu braid to cool the second 
crystal and finally we prepared and cooled the first crystal. In summary, to 
address the daily vacuum trips that are occuring since January 2001, I installed
a RGA on the mono chamber which can be used to provide additional information to 
solve the problem, and we leaked checked the In seal again but found no evidence
of a leak.
Data from the new HHLM RGA.

Fig 11. Here is data of the HHL partial pressure with the first crystal at 77 and 300 K.

Sunday, September 9 2001: Test of the monochromatic beam stability
using the new X-ray BPM in 7ID-C.

For the last two days, since Friday night, we are running long time series
using a 2 mm (H) by 1 mm (V) beam incident of the monochromator. The 
monochromator is set to 10 keV. The On-Trac Amplifier is used for this time 
serie to amplify the BPM signals. The gain setting is #5 i.e. about 4 uA/V.
Fig. 12 shows the output X-ratio and Y-ratio of the Ontrac
amplifier versus the X position of the X-BPM which is scanned +/- 1 mm. The X
ratio varies linearly with a slope of -1.14 V/mm. Some slight cross-talk is 
observable on the Y-ratio but is fairly small. Scanning the Y position of the 
X-BPM shows a similar behavior with a slope of -1.22 V/mm. The difference in 
slope is believed to originate from the different beam size in the vertical
and horizontal direction.

Fig. 13  shows the ion chamber signal in 7ID-B, 7ID-C and
the diode sum signal all normalized to their maximum values. Each ion chamber
signals has been normalized to the beam current decay thus scaled to 100 mA. 
We would expect all the signals to overlap, but they don't. The APS fill-on-fill
occur at 8.3 and 20.2 hours. Note that the ion chambers in B and C only track 
3 hours before a fill. This might be caused by the large beam motion seen in 
Fig. 14. The vertical and horizontal motions are 
respectively 0.45 and 0.1 mm peak to peak over 24 hours. This is significant
and most likely due to thermal heating effect on the second crystal. We may
be able to reduce this motion by careful cooling of the second crystal with 
a LN2 Cu braid. This has been done at PNC and BESSRC.
Fig. 15 and Fig. 16 shows the 
remaining data taken on 09/10 until the beam dumped for a two days machine
maintenance. The data clearly continues the earlier data.
Data from the new XBPM on 090701.

Fig 12. Calibration data of MHATT-CAT Xray-BPM on 09/07. Displayed is the
Xratio (dash), and Yratio(solid).

Data from the new XBPM on 090901.

Fig 13. The ion chambers signals, and diode sums. The time serie is
started on 09/08 at 23h31 and last until 09/09 at 22h00

Data from the new XBPM on 090901.

Fig 14. The beam position, 49 m from the source, or 19 m from the
High Heat Load mono in 7ID-C during the same time serie as Fig. 13.

Data from the new XBPM on 091001.

Fig 15. The ion chambers signals, and diode sums, started on 090901 at 22h06 and lasting until 8 am on 9/10.

Data from the new XBPM on 090901.

Fig 16. The beam position, 49 m from the source, or 19 m from the
High Heat Load mono in 7ID-C during the same time serie as Fig. 15.

Wednesday, September 12 2001: Test of the monochromatic beam
stability using the new X-ray BPM in 7ID-C in the
low-emmittance top up mode.

During this new time serie, the APS is running in top up operation with a
smaller horizontal emmittance of about 3 nm-rad, a coupling of 2%. In effect
this reduces the horizontal source size and divergences by the the square root
of 7/3=1.53, thus it increases the horizontal coherence length by 1.53 to about
12 microns at 6.9 keV from 8 um, it decreases the horizontal divergence by the
same factor, thus increases the coherent flux by a factor 7/3 or 2.33.  This is
learly appreciable.

Fig. 17 shows a 24 hours long time serie of the intensity using
a 2 mm (H) by 1 mm (V) L5-20 opening. I tweaked the second crystal at t = 8 
hours. The intensity increases by 10 % after the tweak and remains stable to 
better than 2% after t = 10.5 hours. It is not clear what the sharp increase at 
t = 10.5 hours is due to.  The On-Trac Amplifier is used for this time serie to
amplify the BPM signals. The gain setting is #5 i.e. about 4 uA/V.

Fig. 18 shows the beam position for the same time serie.
The beam moves down by 0.3 mm after a tweak, and after 10.5 hours the beam is 
stable to within about 50 um over 14 hours. This is nearly a factor 10 better
than for the data while running in regular fill-on-fill mode as shown in 
Fig. 14. The motion was as much as 0.5 mm then!

Top up operation provides a much more stable beam as many of you have
experienced (See also Fig. 9). The beam is far more stable when the ring
current is stable and thus the heat load on the monochromator is stable. To
improve the stability while in fill-on-fill operation, or to reduce the
monochromator heating effects, we plan to cool down the second crystal and
possibly stabilize the temperature of the first and second crystal mounts.
Data from the new XBPM on 091201.

Fig 17. The ion chambers signals, and diode sums, started on 09/12/01
at 09h37 and lasting about a day while the APS is in its new low-emmittance
top-up mode. Click on the figure to see a wider vertical scale.

Data from the new XBPM on 09/12/01.

Fig 18. The beam position, 49 m from the source, or 19 m from the
High Heat Load mono in 7ID-C during the same time serie as Fig. 17.
Click on the figure to see a wider vertical scale.

Thursday, September 13 2001: Test of the monochromatic beam
stability using the new X-ray BPM in 7ID-C in the
low-emmittance top up mode.

On 9/13, I ran a time serie from 9h49 am for about 13 hours, and found some
interesting beam motion caused by the APS top-up process. During the day,
at around 10 am, the APS reinitialized top-up operation because the BPM bucket
was not receiving any charge anymore. This suddendly returned the vertical 
emmittance to what it should be. (The vertical emmittance was slowly drifting
up). Also, around 3h19 pm, some beam motion was observed by our APS X-ray BPM.
Fig. 19 shows the various ion chambers signals, and diode
sums. They track fairly well.
Data from the new XBPM on 091301.

Fig 19. The ion chambers signals, and diode sums, started on 09/13/01
at 09h49 and lasting about 13 hours while the APS is in its new low-emmittance
top-up mode.

Fig. 20 shows the vertical and horizontal position.
Note the nominally 1 hour periodic motion of the horizontal position. This 
is caused by the top-up effect. Essentially, because only one bucket can be
refilled every 2 minutes, it takes 2min*23 buckets~46 minutes or so (Glenn
Decker actually measured on 9/13 52 minutes) to refill entirely all the buckets
in the ring. This corresponds to the time scale observed in Fig. 20.
The full explanation of the beam motion is more involved but its timing source
originates from the new mode of operation. Note also the vertical beam position
change around 10 am or t = 0.3 hours. This was caused by the restart of the 
top-up programs, and it was confirmed by the APS, and is also clearly observable
in the APS X-ray BPM.
Data from the new XBPM on 09/13/01.

Fig 20. The beam position, 49 m from the source, or 19 m from the
High Heat Load mono in 7ID-C during the same time serie as Fig. 19.
The time serie is started on 9/13 at 9h49 am, and lasts for 13 hours or so.
Note the 15 micron horizontal motion with a 1 hour period. This is observed
as well in the 7ID X-ray BPM angular PVs. Also, all vertical motions are
correlated to our APS FE X-ray BPM. For example, the spike around 10 am(0.3 h)
and the step around 3h19 pm(5.5h) are observed in the APS X-ray BPM(not shown).

Friday, September 14 2001: Test of the monochromatic beam
stability using the new X-ray BPM in 7ID-C in the
low-emmittance top up mode.

Fig. 21 shows the total fluxes measured on 09/14/01 starting
from 01h02am. Fig. 22 shows the X-ray BPM positions during
the same time serie. The same horizontal motion is clearly observable.
Data from the new XBPM on 091401.

Fig 21. The ion chambers signals, and diode sums, started on 09/14/01
at 01h02 and lasting about 13 hours while the APS is in its new low-emmittance
top-up mode.

Data from the new XBPM on 09/14/01.

Fig 22. The beam position, 49 m from the source, or 19 m from the
High Heat Load mono in 7ID-C during the same time serie as Fig. 21.
The time serie is started on 9/14 at 1h02 am, and lasts for 13 hours or so.
Note the 15 micron horizontal motion with a 1 hour period.

Saturday, September 15 2001: Test of the monochromatic beam
stability using the new X-ray BPM in 7ID-C in the
low-emmittance top up mode.

Fig. 23 shows a time serie started on 9/15/01 at 2h16 and
lasting for 13 hours. Today, I modified Dohn's gnuplot routines to plot the ring
current and the LN2 level on the same graph as the intensity. Note that the LN2
Dewar filled every three hours or so. The top-up operation was interrupted at
the beginning of the time serie, explaining the ring current decay around 
t=2.5h.
Data from the new XBPM on 09/15/01.

Fig 23. The ion chambers signals (7ID-B:green, 7ID-C:navy), and diode
sum (purple), the ring current (red) and LN2 Dewar level (light blue) started
on 09/15/01 at 02h16 and lasting about 13 hours while the APS is in its new
low-emmittance top-up mode.

Fig. 24 shows the beam position in the same time serie.
Note the 5 micron beam motion every 3 hours in the vertical position. It is
caused by the LN2 cryocooler refilling. Substantial beam motion occured while
top-up was paused. The beam motion caused by the monochromator is 3 times 
smaller than the variations in the horizontal position caused by top-up.
Data from the new XBPM on 09/15/01.

Fig 24. The beam position, 49 m from the source, or 19 m from the
High Heat Load mono in 7ID-C during the same time serie as Fig. 23.
The time serie is started on 9/15 at 2h16 am, and lasts for 13 hours or so.
Note the 5 micron vertical and horizontal motion with a 3 hour period. It is
correlated with the LN2 fill as seen in Fig. 23. The APS top-up operation
was paused from t=2.5 to 4 h in the tserie and more substantial beam motion
occur during this period.

Tuesday, September 18 2001: Test of the monochromatic beam
stability using the new X-ray BPM in 7ID-C in the
low-emmittance top up mode.

On 9/17/01 at 13h00, the APS machine physicists made few changes to the DC orbit
configuration which seems to help the 46 minute horizontal motion at ID07.  On
Tuesday overnight, I ran another time serie to test whether these changes 
had improved the situation.

Fig. 25 shows a time serie started on 9/18/01 at 1h34 and
lasting for 6.5 hours until the beam dumps at 8 am. Please note the LN2 level
of MHATT-CAT's cryocooler (it cools the Si (111) monochromator) which 
periodically must refill every 3 hours or so.
Data from the new XBPM on 09/18/01.

Fig 25. The ion chambers signals (7ID-B:green, 7ID-C:navy), and diode
sum (light blue), the ring current (red) and LN2 Dewar level (purple) started
on 09/18/01 at 01h34 and lasting about 6.5 hours while the APS is in its new
low-emmittance top-up mode.

Fig. 26 shows the beam position in the same time serie.
Note the 5 micron beam motion every 3 hours in the vertical position. It is
caused by the LN2 cryocooler refilling. The 55 minutes periodic variations with
15 micron amplitude in the horizontal position have disappeared! The 
beam is a lot more stable now. The improvement in the DC orbit configuration
have improved the beam stability. For comparison with earlier data, please see
Fig. 22. It is also very interesting to point out that our
coherent flux in Fig. 25 (the ion chamber signal in 7ID-C)
is now extremely stable compared to the yesterday's data (not shown yet).  This
may be due to the improved APS orbit control.

In conclusion, I would like to make a comment on beam position stability 
required for XPCS experiments at the APS. Several beamline at the APS use the
transverse coherence of the X-ray beam to perform coherent diffraction
experiments (sector 7,8,34,etc). In the current mode of operation, I selected
the coherent flux with a 12 x 12 um aperture to illuminate a sample with a
coherent beam. 12 um is the horizontal transverse coherence length at 7 keV in
the 7ID-C hutch. For many reason which are beamline related (Be windows,
optical elements, etc), it is well known that a spatial speckle pattern exists
on the beam profile of the undulator beam with huge spatial variation on a 10
um length scale (See Fig. 27 for beam profiles.).

If this fine structure moves with the center of mass of the beam, then intensity
variations in coherent flux can be quite severe if the beam position variations
are as small as 10 um.  For my experiment, I need to keep the beam stable to 1
micron or less...  Obviously, my monochromator is causing motion of 5 microns
or so every 3 hours. So, I can improve this further.  But 15 microns motion of
the APS beam will definitely be noticable in my coherent flux as intensity 
variations possibly as high as 10-50%. As food for thoughts, I would say we have
to aim at 1 um stabilty. The changes made by the APS yesterday afternoon made
the beam stabilitgy pretty close to one micron!
Data from the new XBPM on 09/18/01.

Fig 26. The beam position, 49 m from the source, or 19 m from the
High Heat Load mono in 7ID-C during the same time serie as Fig. 25.
The time serie is started on 9/18 at 1h34 am, and lasts for 6.5 hours or so.
Note the 5 micron vertical and horizontal motion with a 3 hour period. It is
correlated with the LN2 fill as seen in Fig. 25. The 55 minutes horizontal
oscillations have disappeared.

Beam profile in 7ID-C in low-emmittance mode.

Fig 27. The horizontal beam profile, i.e. the coherent flux though a
0.012 mm x 0.012 mm aperture versus the horizontal position. Click on figure
to see the vertical beam profile. Speckle is clearly present on the X-ray beam.


Back to my homepage (retour vers ma page maison).
Récemment modifié le 08/15/2003, par Eric Dufresne, Bldg 432 Rm D007 630-252-0274
U.S. Department of Energy UChicago Argonne LLC Office of Science - Department of EnergyOffice of Basic Energy Sciences - Department of Energy
Privacy & Security Notice | Contact Us | Site Map