This paper discusses the implementation of abort gap monitoring at the Tevatron. There are two somewhat independent monitors which measure the intensity of the synchrotron light emitted by particles in the abort gaps. One system uses an optical focusing system, gated image intensifier, and CID camera to image the beam over many revolutions; and the other system uses a gated photomultiplier to measure the light intensity.
During operation of the Tevatron in colliding beam mode, a small amount of the beam diffuses out of the bunches and spreads around the ring1. The presence of beam in the abort gap can have a serious effect on superconducting magnets and a devastating effect on the silicon detector of CDF2. During an abort, the kicker magnets ramp up during the abort gap. Beam passing through the kickers while they are ramping sprays into magnets and into the silicon detector. Until now, the only measure of the amount of beam in the abort gap was the count rate in a set of photomultipliers surrounding the beamline at CDF  that were gated on the abort gap. The measurements from these counters suffer from the fact that they are measuring the beam leaving the abort gap, not the beam still in it. This document describes two partially independent methods for directly measuring the beam in the abort gap using synchrotron light: CID camera, and photomultiplier tube.
A charged particle that undergoes transverse acceleration emits radiation in a cone around its velocity vector. This radiation is called synchrotron radiation after its first observation in a synchrotron. See , , , and  for the full theoretical background to synchrotron light. The Tevatron has 1113 RF buckets and typically contains 1013 protons in 36 bunches arranged in 3 trains of 12 (the antiproton intensity is ~1/10 the proton intensity).
If one wants to observe a DC beam of 1 part in 104 (109 protons), then,
The synchrotron light apparatus is sensitive only to wavelengths in the 400-500nm region. From , the intensity in the vicinity of 450nm is
shows the optics of the synchrotron light apparatus which is located near the short warm section at C11.
Figure 1: Diagram of SyncLite system. The optics through the beam splitter are shared by both the camera and PMT systems. The top drawing is the logical diagram and the bottom drawing is the physical layout (100:1 filter is in PMT module).
The light is picked off by a mirror in the beam pipe  and directed out a quartz window  to the light box. Inside the light box, the light traverses a 1500mm focal length lens  and another mirror before hitting the beam splitter . The synchrotron light is clearly visible on the wall of light box in Figure 2.
Figure 2: Video stills from a CCD camera mounted in the light tight box showing synchrotron light impacting the side of the light tight box. The smaller white specs are radiation damaged pixels.
After the beam splitter, the PMT system and camera system follow separate paths. Just before the beam splitter there is a 4% neutral density filter that can be inserted into the light path to facilitate calibrations.
The observed number of photoelectrons, making use of (3), is tabulated as follows:
The SyncLite system  functions by using a gated image intensifier to act as a fast shutter and amplifier for a generic CID camera. This allows for the accumulation of many short-duration ‘frames’ during one 1/30 sec camera frame. The intensifier is operated at a gain of ~1000. The number of times the shutter is opened during a single camera frame is adjusted by the DAQ system based on the measured intensity. In the case of the abort gap, this is typically every 4th turn (~12 kHz). A LabView DAQ system  collects the camera frames and fits horizontal and vertical projections of the beam profile to obtain the integrated intensity. Each abort gap measurement is the sum of 200 camera frames, or 8 x 104 abort gaps. Camera data of the abort gap are displayed in Figure 3 and Figure 4. The bump corresponds to a DC beam intensity around the ring of ~5 E9.
Figure 3: Camera image in abort gap. This is after pixel by pixel background subtraction, but before horizontal line subtraction. The peak corresponds to a DC beam intensity of ~5 E9. See Section 6 for details about background subtraction.
Figure 4: This image is after both pixel and horizontal line subtraction. The peak corresponds to a DC beam intensity of ~5 E9. See Section 6 for details about background subtraction.
A 9-stage side window photomultiplier tube is attached to the SynchLite optics box and observes the light from the beam splitter. Between the beam splitter  and the PMT, there is a 1% neutral density filter which can be moved in or out of the light beam.
To avoid saturating the PMT with the light from the main bunches, there is a gating circuit attached to 2 of the dynodes of the PMT. The gating circuit holds the dynodes at a potential below the previous dynodes effectively turning off the tube (Figure 5). When the gate is on, the dynodes are pushed up to their nominal operating voltage. Figure 6 shows the schematic for the PMT base.
Figure 5: PMT behavior for gated on and gated off modes. the gated on mode looks like an ordinary PMT, while the gated off mode turns around the field lines between several dynodes effectively shutting off the multiplication.
Figure 6: Schematic of the gated base for a 10-stage PMT. A single gate pulse drives 2 FET circuits that are AC coupled into dynodes 1 and 4. The DC state has dynode 1 held ~30V below the photocathode and dynode 4 held at the nominal potential of dynode 1. With interdynode voltages of ~100V, dynode 1 is pulsed with 130V and dynode 4 is pulsed with 300V. These voltages push the dynodes to their nominal values.
To study the behavior of various gating circuit / PMT combinations, a system was set up with two LEDs – a blue pulsed LED to simulate the nominal bunches and a green LED with a tiny DC level to simulate the DC beam.
Three variations of dynode gating were studied: only dynode 2, dynodes 2 and 4, and dynodes 1 and 4. The studies focused on both the stability of the gain immediately following application of the gate (Figure 7), and on the sensitivity to light incident on the photocathode immediately before the gate is applied (Figure 8). Tests of the gating combinations demonstrated that gating dynode 1 is essential. Both of the dynode 2 versions exhibited unacceptable sensitivity to pre-gate light. Thus the final choice was to gate dynodes 1 and 4.
Two PMTs were studied: an end window tube (Electron Tubes, 9902B), and a side window tube (RCA or Burle or whoever, 4552). The tests of the PMTs showed that the gain of the side window PMT becomes stable 200ns into the gate. This time is dominated by post-gate ringing on the anode signal. The end window PMT also has the post-gate ringing, but in addition, has a long transient where the gain slowly increases until it reaches a plateau. Both PMTs are resistant to pre-gate light until the light reaches 5 or 10 times the typical bunch light intensity. The reason for this sensitivity to light incident before the gate is unknown. At present bunch intensities, this sensitivity does not affect us. Because of the transient in the gain of the end window tube, the 4552 was chosen. Figure 9 shows the flat response of this tube using the dynode 1 and dynode 4 gating circuit.
Figure 7: Scope trace showing gain transient at beginning of gate observed in the end window tube. The PMT is observing an LED which is always on. The turn-on transient has a time constant of ~20μs. The source of the transient was thought to be photocathode resistivity but was seemingly ruled out by selectively covering the central part of the photocathode. In any case this tube was not used.
Figure 8: Scope trace demonstrating sensitivity to light present before the gate is applied when gating dynodes 2 and 4. This test uses a pulsed LED to simulate light from bunches. The size of the transient depends on how close the gate is to the last bunch and is not present if the bunch-simulating pulsed LED is not on.
Figure 9: Side window tube with dynode 1 and 4 gated. There is no turn-on transient present in this case. Note that this is the combination that is in use in the abort gap monitor.
The DAQ system  consists of an MVME board running VxWorks talking to a COMET 12-bit ADC board and a VRFT board for beam timing. The PMT anode signal is brought upstairs to a fast integrator which feeds the ADC board. The integration gate is typically 1.4 microseconds (2/3 of the abort gap). The DAQ program on the MVME performs a ~60ms readout cycle once every second (1000 samples of each abort gap). Figure 10 and Figure 11 show the timing of the gating for a number of turns. Only one gate happens every turn, so over 3 turns each abort gap is sampled once.
Figure 10: Scope trace showing timing of PMT gates. The anode signal is into 50Ω and is displayed on a 2mV scale. The first gate from the left occurs at abort gap 1. The second gate is abort gap 2, one and one third turns later, and the third is abort gap 3. The spacing between gates is necessary to keep the duty cycle low enough for the gating circuit.
Figure 11: Zoomed view of Figure 10. Each abort gap has a different portion of it sampled, i.e. abort gap 1 is gated in the middle, abort gap 2 is gated at the beginning, and abort gap 3 is gated at the end.
The SyncLite system uses its calibration of the proton bunches to calibrate the abort gap signal. It calibrates bunch measurements using the bunch intensity obtained from the Fast Bunch Integrators (FBI). Background subtraction is twofold. First, the constant CID pixel background is subtracted pixel by pixel from the measured signal. This pixel pedestal is measured at the beginning of every store with the image intensifier voltage lowered. Secondly, a flat background subtraction is performed line by line where the value is determined by taking the average of 10 pixels at each end. The final calibration step is to scale3 the value to represent beam intensity around the entire ring in units of 109 protons.
The PMT is calibrated by inserting neutral density filters and moving the gate such that it coincides with a bunch. The scale factor is then FBI / signal * filter. Background subtraction is done manually by inserting the optional optical filters and taking a measurement in the abort gap. It is subtracted before the scale factor is applied. The final calibration step is to scale to units of 109 beam around the ring.
The optical filters were calibrated by positioning the gate over a bunch and taking measurements with and without the filters in place. This of course assumes that the PMT is mostly linear with input intensity. Using this approach, the filter values were measured to be 4.2% and 1.04% (4% and 1% nominal).
A study has been done by turning off the Tevatron Electron Lens (TEL) and observing the increase in the abort gap beam (see Figure 12). The result of this study is that the systems agree with each other and correlate well with the amount of beam in the abort gap inferred from the total beam current measurement4. The absolute scales for the abort gap systems are ~50% larger than the inferred measurement from the total beam current. One candidate for this discrepancy is the remaining left sideband of the IBEAM measurement which is not flat, possibly indicating an erroneous sideband subtraction.
Figure 12: Plot of total beam and beam in the abort gap. The IBEAM measurement is smaller than the abort gap measurements. These plots have all had sideband subtraction done. The IBEAM subtraction is a line rather than a constant since it is a falling spectrum.
7Summary and Future
There are now 2 abort gap monitors based on synchrotron light. They are both calibrated and exhibit reasonable agreement with the IBEAM measurement during studies where the TEL is turned off for a period. The SynchLite system is slower and can not measure below 0.5 x 109 protons. The PMT is faster and has more sensitivity, but is subject to some amount of noise induced by the pulsing of the SynchLite system. Figure 13 shows the behavior of the PMT version compared with the SynchLite version. Figure 14 is a sample from after raising the voltage on the PMT to increase the signal to noise.
The PMT system has a number of bugs which need to be worked out. In addition to the aforementioned SynchLite crosstalk, the pulsing circuit loses steam over the course of the 60 ms, due in part to the RC constant of the pulsing output, and in part to not enough current flowing through the low voltage zener diode circuit. These will be investigated during the 2004 shutdown.
Figure 13: Plots of both the PMT and the SynchLite system. The PMT system is the AGIGIx devices. The SynchLite system is the SLPAH device. One can see the shift in backgrounds for the PMT system before and after the store.
Figure 14: After raising the voltage on the PMT, S/N is remarkably improved. The PMT calibration is wrong in this plot however and was redone shortly after.
Note: Around 8/3/04, the gated PMT described above was replaced with a high speed gated MCP-PMT from Hamamatsu (R5916U-50) loaned to us from LBL as part of LHC studies. It has a gain which is a factor of 10-20 lower than the side window tube that was in use and as such needs external amplification. However it seems to be nearly completely insensitive to pre-gate light unlike the other PMTs and has virtually no gating noise. As such it can be gated on immediately following the bunch. Figure 15 shows the behavior of the new tube.
Figure 15: Behavior of the new LBNL MCP-PMT. This setup is using an external gain of ~100 to compensate for the shorter gate and the lower intrinsic gain of the tube. As such there is more sensitivity to the SynchLite pulsing. Not only can you see the discrete pulses which show up as dim bands of points, but one can see a background level shift as well (just before the store started).
I would like to thank Stephen Pordes for continuing to push for the evolution of SynchLite, and Eugene Lorman and Tom Meyer for the present DAQ systems.
Stephen and I would also like to thank these people from PPD for their help: Alan Hahn, the original builder of the Synclite system; Harry Cheung, who worked on the Synclite system in the beginning of Run II; Jim Fast, Ken Schultz and Mark Ruschman who provided the beam-splitter and mechanical modifications to the Synclite box to mount the abort gap monitor PMT; Jim Fast (again), Carl Lindenmeyer and Ron Miksa who built the filters and filter controls; Sten Hansen and Heide Schneider who provided the gated base for the abort gap monitor PMT, and Morris Binkley who designed the high duty-factor pulser used in the Synclite system.
Sasha Valishev from the Tevatron department was a great help in understanding many aspects of the Synclight system and Dale Miller and Carl Lundberg of the Instrumentation department were key in the actual installation.
Finally, very special thanks to Vladimir Shiltsev and Roger Dixon for recognizing and rewarding our efforts.
BEAMS-DOC-976-V1. Several documents on CDF abort gap counters and such.
Aluminized Mirror. Loss is typically 10%.
MDC Vacuum Products quartz window #450023 or #450024. The transmission of the quartz window is curve number 2 below (90% for 400-500nm).
Oriel (now Spectra-Physics) 2” DIA plano convex lens #40825 made of BK 7 glass with 1500 mm focal length and 1495.6 mm back focal length. Transmittance for BK7 glass is shown below (93% for 400-500nm).
Thorlabs non-polarizing beam splitter #BS013 constructed of two BK7 glass prisms with antireflection coatings on the entrance and exit surfaces. The left plot below shows the transmittance, T, of forward beam for the two polarization states through the beamsplitting coating between the prisms. The right plot shows the reflectance, R, for a typical broadband antireflection coating. So the total transmittance for the forward light path would be (1-R)*T*(1-R).
A. Hahn, H.W.K. Cheung, et al., BEAMS-DOC-185-V1, BEAMS-DOC-186-V1, BEAMS-DOC-587-V3, BEAMS-DOC-466-V1. Background and details of the existing Synchrotron Light system.
R. Coisson, “Angular-spectral distribution and polarization of synchrotron radiation from a ‘short’ magnet”, Phys. Rev. A 20 (1979) 524.
J.D. Jackson, Classical Electrodynamics, John Wiley and Sons, (1962), Ch. 14.
J. Bosser et al., “Proton beam profile measurements with synchrotron light”, Nucl. Inst. and Meth. 164 (1979) 375.
A. Hahn, BEAMS-DOC-418-V1, Synchrotron radiation spectrum for various Tevatron energies.
T. Meyer, BEAMS-DOC-1219-V1, PMT DAQ reference.
E. Lorman, BEAMS-DOC-1262-V2, SynchLite DAQ reference.
1 A consequence of this DC beam is the presence of beam in the abort gaps.
2 The DØ silicon detector is at less risk since CDF acts as a collimator between the abort kickers (at A0) and DØ.
3 The scale factor is the ratio of 1113 (number of buckets around the ring) to the number of buckets sampled.
4 The amount of beam in the abort gap is inferred from the IBEAM measurement by comparing the beam loss rate with the TEL on to the beam loss rate with the TEL off and assuming that the decrease in the loss rate is because the DC beam is not being removed by the TEL.