Determination of Longitudinal Electron Bunch Lengths on Picosecond Time Scales

At CERN (European Laboratory for Particle Physics) the CLIC (Compact Linear Collider) study is pursuing the design of an electron-positron high-energy linear collider using an innovative concept for the RF (Radio Frequency) power production, the socalled two-beam acceleration scheme. In order to keep the length of the collider in a reasonable range while being able of accelerating electrons and positrons up to 5 TeV, the normal-conducting accelerating structures should operate at very high frequency (in this case 30 GHz). The RF power necessary to feed the accelerating cavities is provided by a second electron beam, the drive beam, running parallel to the main beam. The CLIC Test Facility (CTF) was build with the main aim of studying and demonstrating the feasibility of the two beam acceleration scheme and technology. It is composed of two beams, the drive beam that will generate the 30 GHz RF power and the main beam which will be accelerated by this power. In order to have a good efficiency for the power generated by the drive beam, the length of the drive beam electron bunches has to be shorter than 2 ps rms. Conventional bunch length measurement techniques did not have the required resolution or were destructive. An innovative approach is based on the measurement of the bunch spectrum using non-intercepting pick-ups. The new bunch length monitor measures the frequency spectrum of the electromagnetic field of the bunch. The b unch length is obtained by studying the variation of the signal amplitude with frequency. Due to the small bunch lengths very high frequencies have to be measured, so that the monitor becomes a microwave spectrometer. The electromagnetic field of the bunch is collected by a rectangular RF pick-up connected to a waveguide of the same dimensions that leads the signal to the detection system. The field induced in the waveguide has a temporal evolution directly related to that of the electron bunch. Hence, it is possible to derive the spatial charge distribution from the frequency spectrum of that signal. Detection is made either by a diode detector or by a combination of RF mixers with a RF sweeping oscillator. The diode detector allows single shot measurements and has proven to be a very fast qualitative method to observe variations in the bunch length. However, this method proved out to be very difficult to calibrate and is not suitable for measuring bunch trains due to interference effects. The mixing technique makes use of two RF mixers. The first one, with a local oscillator set at a fixed frequency n, downshifts the frequency spectrum of the incoming RF signal by an amount n. The second mixers has a sweeping oscillator as a source, with each frequency step of the oscillator the mixer down-shifts the RF signal by a different quantity, allowing a digitizing oscilloscope to detect the part of the spectrum within its bandwidth. By sweeping over some given frequency range the oscilloscope will measure the frequency spectrum amplitude in this range. This technique can not only measure the frequency spectrum of a single bunch but also of a train of bunches. Furthermore, an independent self-consistent calibration procedure was developed and successfully applied to this detection system. Two RF pick-ups have been installed in the CTF drive linac. One for the so called WR28 waveguide and the other for the so called WR12 waveguide. The two pick-up together with the proper detection system, allow the measurement of the signal in three frequency ranges, 28.5–40.5 GHz, 58.5–60.5 GHZ and 76.5–88.5 GHz. Experiments performed in the CTF, with a train of 48 bunches, have shown a very good agreement between the values measured with the mixing technique and the values obtained with the simulation programme PARMELA. Moreover, the experimental results demonstrated the capability of the system to measure bunches as short as 0.7 ps rms.