Operating principle of a spatial auto-correlator. The input beam is split into two signals and they are recombined on a detector plane at an angle. At each point along the detector, the wavefronts of the two beams interfere with each other with a fixed time delay set by the angle. This is equivalent to a classical autocorrelation signal. The left figure shows the spatial autocorrelation of optical laser pulses using a detector array, while the schematic on the right shows the spatial autocorrelation using the broadband coherent radiation (CR) signal from an electron beam.
The top figure shows the engineering design of the RTI with the optics. The lines trace the rays of the radiation through the optics. The bottom figure shows the actual experimental setup of the RTI. The input beam is split by a wedge splitter (c) and then sent to mirrors (a, b). One of the split signals travels to the movable stage (d) and then is reflected back through the mirrors (e, h) to the cylindrical mirror (i). The other signal from mirror (a) travels to mirror (g) and is reflected by a angle-adjustable mirror (f), which is sent to the flat mirror (h). After both signals arrive at the cylindrical mirror (i), they are focussed with the cylindrical mirror (i) and directed to the pyro-detector array with the flat mirror (j). The signal from the pyro-detector array (k) is sent to an oscilloscope.
Experimental setup of the A0 photoinjector facility. The electrons generated by the RF gun are accelerated by the booster cavity to 14 MeV. The beam is then focused using the quadrupoles (Q1, Q2, Q3) before it is sent through the emittance exchange (EEX) beam line. After going through the dogleg section of the beamline, the beam hits a metal screen at X24 and generates coherent transition radiation (CTR) which is transported to the RTI. The quadrupoles are marked as ovals and the diagnostic stations are marked as diamonds. D1, D2, D3, and D4 are dipole magnets used to bend the beam.
The raw oscilloscope trace obtained from the 32-channel detector array of the RTI showing the central part of a typical autocorrelation. The x-axis is time (1 ms/div) and the y-axis is volts (200 mV/div). Each element in the array is illuminated depending on the strength of the autocorrelation signal at that element. By counting the number of pixels in the central (FWHM) width, for a given mixing angle, the bunch length can be estimated. The trace was obtained from the coherent transition radiation, when a 13.4 MeV electron beam hits a foil. The number of pulses in the bunch train was 40 bunches with each pulse carrying 180 pC.
Autocorrelation trace obtained from the real-time interferometer (RTI compared with the autocorrelation trace obtained from the Martin-Pupplett interferometer (MPI). The agreement in the central part of the autocorrelation – which determines the bunch length – is good while the sides of the autocorrelation differs due to the low-frequency response of the detectors in MPI and the RTI.
Fast Fourier transform of the autocorrelation trace obtained from the RTI and MPI. While the high frequency cut-off between the MPI and the RTI agrees well, the low frequency response is different. The difference in the size of the detector might explain the low frequency response.
A comparison of the bunch length (FWHM) obtained from the streak camera and the RTI for various quadrupole settings of Q2AX06 (labeled Q3 in Fig. 3). At longer bunch lengths, the discrepancy between the RTI and the streak camera is due to the poor low-frequency response of the RTI. At the shorter bunch lengths, the streak camera approaches its resolution limit and hence a larger uncertainty. Moreover, the simple estimation of the bunch length from the RTI autocorrelation trace excludes the form factor or the shape of the bunch, which could play a significant role as well.
Article metrics loading...
Full text loading...