Beam-based phase monitoring and gradient calibration of Jefferson Laboratory RF systems Page: 4 of 5
This article is part of the collection entitled: Office of Scientific & Technical Information Technical Reports and was provided to Digital Library by the UNT Libraries Government Documents Department.
The following text was automatically extracted from the image on this page using optical character recognition software:
determinations, we may write the relation above as:
(18&\2 Eioc(+) -EIack(-) t ~ 2O
The uncertainty for a single cavity cresting measurement
is equivalent to approximately 20 square degrees. Using
a 10 degree shift in the cavity phase results in an over-
all energy shift (presuming a starting value near crest) of
only 10-4, and an anticipated error in determination of the
crested phase setpoint of approximately 2 degrees. These
estimates are confirmed by monitoring of successive deter-
minations of the crest phase of given RF systems and by
monitoring of the total system energy.
The most frequent problem with this algorithm is the in-
fluence of other components in the RF system, which also
affect the energy lock gradient setpoints (other cavities go-
ing out of gradient lock, manipulation by operators, etc.).
2.3 AutoKrest Implementation
Implementing AutoKrest as a Unix process accomodated
the need for for frequent modification during development.
However, the other Unix control processes (energy and or-
bit locks) had user interfaces limited to the single screen
from which they were invoked. To allow multiple users
to control or monitor the progress of AutoKrest without
spawning interfering processes, multiple instances of Au-
toKrest were prevented by using lock files as flags in the
file system. The lock file contents provide CPU, process
ID, and certain other useful information for later queries
of the system to verify the status of the process. This has
proven to be very useful and robust. A control interface us-
able from multiple operator screens was implemented us-
ing the Tcl/Tk language as extended to communicate with
the EPICS database fields of the control system. To speed
implementation, we began writing a Unix-based data ac-
quisition, sequencing, and analysis task and an operator in-
terface to connect the Unix program through a set of EPICS
records accessible to a computer interface screen.
Because there was no other easily monitorable parame-
ter to determine the status of the energy lock, we chose to
monitor the gradient setpoints used by the lock to stabilize
the overall system energy. These were updated at a nominal
0.2 Hz rate. We determined that it would be sufficient for
AutoKrest to acquire data until the standard deviation of the
last three data points was less than a threshold value. If a
cavity or the rest of the RF system were particularly noisy,
or if some other stimulus were being applied to the beam
energy, the series of energy lock setpoints would not sta-
bilize. This was the primary protection against AutoKrest
introducing phase setpoint errors into the RF system.
The master oscillator system has been stabilized since
AutoKrest was introduced, and this protection is marginal,
as we have found examples of RF system behavior which
can mislead AutoKrest. Several enhancements are planned
to improve the performance of AutoKrest, notably the in-
corporation of an AC coupled operation mode (as used in
),and algorithmic improvements to allow use of all avail-
able data for consistency checks.
3 RF GRADIENT CALIBRATIONS
The fine energy resolution in the arcs has also been used
to check the gradient calibration of the RF systems in both
linacs last summer and again this spring. The gradient cal-
ibration for each RF system was obtained from the careful
measurement of many independent pieces of hardware dur-
ing cavity and RF commissioning, and was subject to many
different sources of error. Calibration with beam yields an
overall system response which is not subject to cumulative
error buildup. In the first calibration of the linacs, one en-
tire 8-cavity cryomodule was found in which the gradients
had been overestimated by from 25% to 30%. Errors of this
magnitude were few, and some have been found to be due
to aging of RF modules left in service for two years before
3.1 Protocol and Errors
The path we have taken to date is based on a null technique:
with a reference cavity turned off, adjust the linac energy
to the desired value; then successively turn each cavity in
the linac off while supplementing its energy gain using the
reference cavity. After a relative calibration of each cav-
ity against the reference cavity, the absolute calibration is
obtained from the known total beam energy.
The dominant error source is the variability of the to-
tal beam energy from random variations within the linacs.
With the energy lock off, the energy is sometimes stable
at the 5 x 10~5 level, but at other times may oscillate at a
level as high as 103 with a period of a few minutes. We
have not yet traced the source of this behavior. Comparison
of each cavity against the reference takes approximately 20
seconds, and we periodically monitor the total linac energy
to allow correction for drifts. We now track the background
beam energy to about the 1-2 - 10-4 level, so this source
introduces errors of about 2% of a typical gradient for the
first linac and 3% for a typical cavity in the other linac.
Each cavity in the 400 MeV linacs gives 2 to 4 MeV to
the beam. The 2.5 MeV average is roughly 0.6% of the
total energy of the beam in arc 1 and 0.3% of the total en-
ergy in arc 2. Turning each cavity off to make a differential
energy measurement is possible, but can introduce errors
from exceeding the linear field region of the bpms and of
the quadrupoles. The errors from this type of measurement
appear to be greater than those involved in the null mea-
surement described above. Adjusting the arc dipole current
to maintain the beam position would be reasonable but for
the need to check for drifts in the total energy from the rest
of the linac and because the solid iron dipoles need exces-
sive (on the time scale of these measurements) amounts of
time to settle in field after changes in current.
One weakness of the technique used is that it relies upon
linearity between the gradient setpoint and the actual cavity
gradient. As this is one of the primary bench tests for the
Here’s what’s next.
This article can be searched. Note: Results may vary based on the legibility of text within the document.
Tools / Downloads
Get a copy of this page or view the extracted text.
Citing and Sharing
Basic information for referencing this web page. We also provide extended guidance on usage rights, references, copying or embedding.
Reference the current page of this Article.
Tiefenback, M. G. & Brown, K. Beam-based phase monitoring and gradient calibration of Jefferson Laboratory RF systems, article, July 1997; Newport News, Virginia. (digital.library.unt.edu/ark:/67531/metadc698225/m1/4/: accessed October 24, 2018), University of North Texas Libraries, Digital Library, digital.library.unt.edu; crediting UNT Libraries Government Documents Department.