The
1.54 GeV S-band linac for the Accelerator Test Facility (ATF)
accelerates multi-bunch beam. The beam has 20 bunches of 2x1010
electrons with 2.8 ns bunch spacing. When multi-bunch beam is
accelerated in the linac, the beam has the energy deviation by
transient beam loading. The 1.54 GeV S-band Linac is an injector
of the damping ring (DR), and the energy acceptance of the DR
is 0.5%. This means that the beam loading compensation system
is necessary in the linac for a successful injection of multi-bunch
into DR. The system consists of a compensating section in addition
to a regular accelerating section. The accelerating structures
of compensating section are operated with slightly different RF
frequencies of 28564.327MHz. This paper describes the principle
of the beam loading compensation system and the results of energy
compensating experiment.
For future linear colliders, one of essential technique to get a sufficient luminosity is the ability to accelerate multi-bunch beam with small bunch spacing. As the pulse length of a multi-bunch beam is shorter than the filling time of accelerating structures, the energy gain of successive bunches drops by approximately linear function due to a transient beam loading in the accelerating structures. The energy loss (Ebl) at time t after the first bunch injection is
,
where tb, r0, 0, tf are the instantaneous current of the beam and shunt impedance, the attenuation parameter and the filling time of the accelerating structures, respectively. The instantaneous beam current is expressed as
,
where e, Nb, and tsp
are 1.6x10-19C, the total number of electrons per
bunch and bunch separation, respectively. There are many methods
to compensate transient beam loading, such as T method, F method
and so on. The T method is to inject a beam before an rf pulse
has filled in an accelerating structure. The F method is to
have one or more accelerator structures running at slightly
higher and lower than fundamental frequency and roughly in
90 degree out of phase from the acceleration.
The F Energy Compensation System (ECS)
compensates for multi-bunch energy spread by keeping a bunch separation
synchronized with an rf frequency. In this compensation system,
compensating structures are installed between the regular accelerating
structures. When a bunch train goes through the compensating structures
driven at an rf frequency which is slightly higher and lower than
the fundamental accelerating frequency, succesive bunches of the
train ride on a different phase of the accelerating field (see
Fig. 1). Due to this phase difference, the energy gain of the
succesive bunches is different. As a result, the multi-bunch energy
spread is compressed to a small value. When a bunch train enters
the compensating structures of +f, the energy gain is lower for
the bunch head and higher for the tail due to the fact that each
bunch accelerates at a positive slope of the part of sinusoidal
wave in the structure. To compensate for this single-bunch energy
spread which is created in the compensating structures, two frequencies
(f+f, f-f) are necessary in order to compensate it by both a negative
slope and a positive slope.
This energy compensation system has
a high flexibility for bunch population changes, the amount of
compensation can be controlled by the input RF power applied to
the compensation structures.
The 1.54 GeV S-band Linac of ATF accelerates
a multi-bunch beam that consists of 20 micro-bunches with 2.8
ns spacing and the repetition rate of 25 Hz. After acceleration
in the 1.54 GeV linac, the multi-bunch beam are injected into
DR which generates extremely low emittance beams. The energy acceptance
of DR is 0.5. As the multi-bunch beam with intensity of 2.01010
electrons/bunch is accelerated in the 1.54 GeV linac and a bunch
train is injected after an rf pulse has filled in an accelerating
structure, the multi-bunch energy spread of a bunch train is
evaluated to be about 9.6% peak to peak without T and F compensation
as shown in Fig. 2. Therefore,
the beam loading compensation system is necessary in ATF for a
successful operation of multi-bunch scheme. With the F
energy compensation
system, the multi-bunch energy spread can be reduced to 0.27%
with beam intensity of 2.01010
electrons/bunch.
Experimental setup
As shown in Fig. 3, the ATF rf system
of the accelerator section consists of 8 regular rf units and
2 ECS rf units.
Fig. 3. Accelerator section of the ATF 1.54 GeV injector linac
(L0-L16 regular sections,
C1, C2 compensating sections)
The regular rf units consists of an E3712 klystron, a pulse modulator, a two-iris SLED cavity, rf waveguides, two 3 m-long accelerating structures and rf dummy loads. The ECS rf unit is composed of a SLAC-5045 klystron, a modulator, and a 3 m-long accelerating structure. The accelerating structures for the energy compensation system are designed for two frequencies (f0f). The rf pulse waveform from the two SLAC-5045 klystrons is rectangular with a width of 1.0 s .
Timing system
In contrast to the regular accelerating section where the bunches are accelerated onto the crest of the RF wave, in the compensation section the bunches enter a phase where a slope of the part of sinusoidal wave. That is, a small jitter results in large uncertainty in the energy gain of bunches. In this reason, a very stable accelerating rf signal is required.
In the ATF, the fundamental frequency
is generated using a 1428 MHz master oscillator, and the other
necessary frequencies are generated from this reference signal
using frequency multipliers and dividers. All the components
are synchronized to this master oscillator. The sideband frequency
for the compensation was selected to be 4.327 MHz signal, twice
the revolution frequency of the DR, and 1/660 of the accelerating
frequency. This frequency was decided by the bunch number (20
banches) and the DR revolution frequency. Two compensation signal
(f0f) are generated by mixing the fundamental (2856 MHz)
and the sideband (4.327 MHz) signal in the special module. Phase
jitter was measured by generating two signals of the same sideband
and feeding them into a mixer. The result of this measurement
was 1.7 ps jitter at FWHM (sigma=0.8 ps)[1][2].
Measurement system of the beam energy
The beam energy of each bunch was measured from the strength of the bending field and the beam position after the bending magnet of the beam transport line. The measurement of the beam position for each bunch was performed by using stripline type BPM. The multi-bunch signal from BPM was measured by the digital oscilloscope of 2.5 GHz sample. The energy difference in a bunch train was calculated from the horizontal beam position and the dispersion function () at the BPM position. In this measurement, the position resolution of the BPM is limited by a sampling resolution and speed of the digital oscilloscope. The position resolution is evaluated to be 22.5 m from the measurement range of the oscilloscope, the signal amplitude and the coefficient of sensitivity of the BPM. This value is sufficient to measure the beam position for each bunch, as the position resolution is converted into the energy resolution of 0.003%.
The dispersion function at the BPM was measured and compared to the calculated value by the program "SAD"[3] in the beam test. The measured value was 14% lower than the calculated value. The discrepancy is small compared to the DR energy acceptance. In the beam test of the ECS, the dispersion function of calculated value by the program "SAD" was used.
The beam profile were observed by a
profile monitor using an optical transition radiation (OTR). A
fast gated camera which has ~3 ns gate width, is used for the
OTR monitor. The beam energy spread of each bunch was measured
by the width (FWHM) of the profile at the beam transport line.
The profile of each bunch was distinguished by changing the delay
of the gated camera timing.
Adjustment of
RF phase for the ECS
An adjustment of the RF phase for the ECS was performed by using the OTR monitor. At first, the gate timing of the OTR camera was set to the center of the bunch train. Then, the current of the bending magnet was adjusted that the bunch profile was seen on the center of the screen. An ECS phase was searched by a phase scan with 20 degrees step. Fig. 4 shows the beam energy dependence on the ECS phase. The optimum phase was decided from the result of a phase scan to find 90 degree apart from the accelerating phase.
The measurement of multi-bunch energy
spread
In this experiment, the multi-bunch of 23 bunches/pulse accelerated up to 1.16 GeV with intensity of 3.21010 electrons/train. The bunch population of each bunch is shown in Fig. 5. After the adjustments of the ECS RF phase, the RF power of the klystrons were set to get a flat energy distribution for all bunches with 1.9 MW for +f and 1.5 MW for -f . The result of ECS on/off is shown in Fig. 6. The energy of each bunch distributed in about 1.5% without ECS, where the calculated energy difference was 2%. The ECS could compress it to about 0.5%. The energy decrease of the bunch train head seems to come from a BPM miss-reading by the beam loss of the collimator in front of the BPM.
Fig. 7 shows a single-bunch energy
spread of each bunch with f ECS and +f only. The single-bunch
energy spread with the f ECS was around 0.3% FWHM.
When only +f ECS is applied on increase of single-bunch
energy spread is expected.
Although, there is no significant difference
in this low compensation voltage. The detail of the single-bunch
energy spread measurement is presented in elsewhere[4].
The beam test of the ECS was performed
by using 28564.327MHz structures in the ATF linac. When the calculated
energy difference by the beam loading was about 2%, the ECS could
compress it to 0.5% by using rf power of 1.9 MW for +f and 1.5
MW for -f with 3.21010 total intensity.
The authors would like to acknowledge all the member of the ATF group and Mr. Morita of E-CUBE corporation for their useful discussion, cooperation and support. The authors also appreciate to Drs. T.Shidara and H.Honma of photon factory division in KEK for the reason that they have readily lent us the pulse modulators for the ECS.
[1] T.Korhonen et al., "R&D of the ATF Timing System". Proc. 1994 Int. Linac Conf., (LINAC94)
[2] T.Korhonen et al., "Energy Compensation System in ATF Linac" .Proc. of the 20th Linear Accelerator Meeting, p308 (1995)
[3] K. Oide, Nucl. Instr. Meth. A335, 1, (1989)
[4] T.Naito et al. "OTR
monitor for ATF". in this conference