-----------------------------------------------------------
Minutes of the ABP-RLC team meeting of 10.03.2006
present: AG, EM, WH, TP, FC, ES, FR, RT
web site: http://ab-abp-rlc.web.cern.ch/ab-abp-rlc/
------------------------------------------------------------
WH decides that GR volunteers to take the minutes of this meeting.
No comments on the minutes of last meeting.
(1) Follow-up of TCTV collimator and IPM impedance issues (AG)
Is there a bug in Gdfidl concerning the origin of the TCLIA/TCTV transverse
impedance? The problem was that the BB impedance seemed not to depend on the
tapering angle and was 10-20 times larger than the analytically predicted value,
see the minutes of the ABP-RLC team meeting of 17.02.2006.
It seems that the calculation is indeed correct and there is no bug in Gdfidl
(Warner Bruns agrees).
The full geometry (250x125 mm^2) was simulated assuming smaller and smaller
vertical chamber sizes (Ymax=35,25,15,5,2 mm). The BB impedance does not change
much from the full geometry with 35 mm vertical aperture. Reducing the horizontal
chamber size (xmax=90,60,30 mm) the trapped modes tend to disappear and the
BB impedance decreases (the case xmax=30 mm half with 45 kOhm/m reproduces well
the analytical estimate of 35 kOhm/m and has a completely constant imaginary part
with no trapped modes).
The contribution to the BB impedance from dipole trapped modes can be calculated
using the resonator formula fitting the first 4 trapped modes (obtained with HFSS)
and summing up single contributions. It gives 350 kOhm/m as a result.
The transverse BB impedance therefore comes from jaw tapers (40 kOhm/m) and dipole
trapped modes (350 kohm/m), in perfect agreement with Gdfidl, and since it is
dominated by the trapped modes, it also explains why no or little dependence on
the tapering angle was seen in previous simulations.
Opening the RF fingers to damp trapped modes using ferrite has a detrimental effect.
It increases the BB impedance because trapped modes are shifted to lower frequencies.
What would help is instead:
- increasing the gap
- reducing the transverse dimensions of the elliptical vacuum chamber.
For different gaps Gdfidl and HFSS simulations are needed to establish impedances
and characterize the trapped modes, respectively. This will be followed-up with
high priority by AG. FR will check whether a reduction of the transverse dimensions
of the TCTV vacuum chamber is compatible with Totem Roman Pots requirements.
Another request on the allowed maximum thickness of the resistive coating on the
ceramic plates of the IPM monitor has come in parallel and awaits processing.
A first estimation of 100 Ohm/square by AG is an acceptable coating resistivity
that seems safe for ultimate LHC intensity. The IPM team is asking for a lower
resistivity window for the initial coating, which would then increase after several
activation cycles and ion bombardment. FR suggests to do some simplified scaling
taking into account that in the first 2 years the LHC beam intensity will be
limited below half of its nominal value. FC suggests that an approximate scaling
for the Q of the trapped modes should be with ~1/sqrt(coating resistance)
and suggests an initial window of 30 to 50 Ohm/square that will later increase
to 50–100 Ohm/square. To be taken as an educated hand-waving guess.
=> ACTION => EM will collect existing literature on the effect of resistive coatings
by FC, check with AG the scaling and the damping required for half of the
nominal LHC intensity, and clarify the logics for a correct reply to the IPM team.
(2) HEADTAIL simulations with measured BPM impedance (GR)
GR presents results on HEADTAIL simulations of a single bunch under the effect of the
trapped modes in the BPM’s (H and V). These trapped modes were calculated by Bruno Spataro
and only the 2 main modes for BPH’s and for the BPV’s were considered in the simulations
(as given by Gianluigi Arduini). Since these modes have very high Q’s, FC doubts the result
itself in the first place, and also the fact that such narrow-band resonators could
significantly affect the single bunch dynamics.
=> ACTION => GR will contact GA and/or Bruno Spataro to clarify whether the high Q's
of these modes include the effect of a termination load.
HEADTAIL was modified to handle an arbitrarily high number of resonators. Current scans
in the range of 0.5-2.0x10^12 ppb were done since E. Métral had previously estimated with
MOSES that the TMCI threshold for the mode with highest R/Q was expected to be around
1.4x10^12 ppb. Result: the threshold sits at 10^12 and the tune has a negative shift with
increasing current, tending to merge with a growing m=-2 line.
(3) TMCI instability and possible re-construction of SPS impedance (GR)
GR presented then possible techniques of Fourier analysis of the Delta_y(z) signal
during a TMCI in order to find out from off-line analysis the frequency of the resonator
that caused a TMCI. Pure Fourier transform of the Delta_y(z) traces turn by turn show
the existence of an unstable harmonic. The frequency resolution is very poor for
short bunches (several hundred MHz for the SPS), but is high enough and clearly reveals
the frequency of the resonator for long bunches (PS) even if it is broad-band.
By performing the Fourier transform on a longer signal (completed with zeroes up to filling
a 25 ns bucket) helps to determine with more accuracy the dominant frequency for a short bunch
and narrow band impedance. It still does not help much for broad band impedance.
Looking at the Fourier transform of the signal constructed by adjoining all the Delta_y(z)
traces after one another gives the same amount of information as the previous technique.
It permits us to see definite frequencies for short bunches and broad-band resonator,
but the relation of these frequencies with the exciting frequencies is somehow still obscure.
FR pointed out that the head-tail mode numbers of the coupling modes during TMCI
contains some indirect information about the frequency dependence of the impedance.
GR’s third presentation on the SPS measurements of emittance/life time/e-cloud during
long coasts done in August 2004 has been postponed to next week for lack of time.
FR makes a strong point on the fact that timing of the talks should be respected
by the speakers: owing to the large number of different topics to be dealt with by
the RLC team, technical discussions should take place in restricted meetings before
presentations at the weekly RLC meetings. The latter are meant to steer future activities
and to inform colleagues about relevant results already somewhat "pre-digested".
In a restricted discussion following the RLC meeting, GR and FR concluded that the
ideal way to diagnose the frequency content and possible sources of SPS transverse
impedance is to use much longer bunches. An attempt was done some time ago by FZ
using debunching beams, but with limited success. It would be important to understand
why this did not succed:
- too low intensity? => reapeat in 2006 during high intensity LHC bunch tests from the PS
- too large energy spread during/after debunching? => try dispersive scraping of the beam
- can we inject very long bunches from the PS? => no RF capture would be needed.
=> ACTION => GR and FR will follow-up these ideas with GA, ES, and FZ
(4) Status of coherent beam-beam studies (TP)
In the LHC, multi head-on and long-range beam-beam interactions will couple all
the bunches in the machine. However a bunch at the head of a train will see a
different number of long-range interaction with respect to one in the center of the train.
This will lead to different tune spectra for different
bunches as also simulations with COMBI show. To study the coherent beam-beam modes,
the One Turn Map model(OTM) was improved to evaluate eigenvalues and eigenvectors
of a system of N bunches colliding head-on and long-range for different collision
pattern. For each eigenfrequency one can produce and study the mode pattern and
predict the contribution of each bunch to it. Several examples were shown for the
case of a 5-bunch train colliding head-on and long-range. It was demonstrated that
with OTM we can identify all eigenfrequencies with their degeneracy as well as all
the oscillating patterns associated. Differences in the tune spectra peaks from
bunch to bunch are understood and can be evaluated by the eigenmode analysis.
FR asks whether it is correct to look only at the real part of the eigenvectors or
if we lose information in the process. In the somewhat pathological case of eigenvectors
1 and i, the real parts are 1 and 0, and information on the second oscillator 90 degrees
out of the phase with the first would be lost. WH argues that this is indeed a pathological
case and is ~excluded by exploiting the self-grouping of eigenmodes
around the sigma and pi modes. One of the motivations of this study is to investigate
the possibility of using a (narrow-band and low-noise) feedback system to control
modes which are not Landau damped.
Posted on the web: Slides by AG, GR, and TP
Web site: http://ab-abp-rlc.web.cern.ch/ab-abp-rlc/