Linear accelerator system with stable interleaved and intermittent pulsing

Abstract

A linac-based X-ray system for cargo scanning and imaging applications uses linac design, RF power control, beam current control, and beam current pulse duration control to provide stable sequences of interleaved pulses having different energy levels, for example alternating 4 MeV and 6 MeV pulses or other energies where the difference in levels is at least approximately 1 MeV and less than approximately 5 MeV. The pulse repetition rate can be 100-400 pulses per second or more. In an embodiment, a cool down calculation is combined with automatic frequency control to provide stable energy and dose per pulse even upon restarting of pulsing after an off period of indeterminate duration.

Claims

1. A linac system for providing pulses with stable dose per pulse when pulsed intermittently, comprising: a power source for supplying power, an RF source with a controllable output frequency for supplying RF pulses at a selected frequency, a supply of electron pulses, a linac responsive to the RF pulses and electron pulses for generating pulses of accelerated electrons, the linac having a resonance frequency that changes at least in part due to changes in operating temperature, where the frequency of the RF pulses is selected and controlled to substantially match the resonance of the linac, and a controller for enabling and disabling power to the RF source for periods of time as desired, wherein at least the period of disabling can vary indeterminately, such that the RF source supplies RF pulses only when power is enabled, where during periods of pulsing, RF feedback techniques are used to tune the output frequency of the RF source in order to substantially match the linac resonance frequency, and where during periods of non-pulsing when RF power is off and without RF feedback, compensating algorithms are used to tune the output frequency of the RF source in order to substantially match the linac resonance frequency, resulting in stable dose per pulse when pulsing is resumed after a period of non-pulsing.

2. The system of claim 1 wherein the period of non-pulsing is variable.

3. The system of claim 1 wherein the period of non-pulsing varies between one second and one minute.

4. The system of claim 1 further comprising: a target for generating X-ray pulses when struck by the pulses of accelerated electrons, and a pulse-to-pulse dose detector configured to detect a portion of the X-ray pulses and providing an output signal proportional thereto as an input to the controller for adjusting the duration of the electron pulses for the purpose of controlling dose on a pulse to pulse basis.

5. The system of claim 4 wherein the pulse-to-pulse dose detector comprises at least one scintillator crystal.

6. The system of claim 4 wherein the pulse-to-pulse dose detector comprises an ion chamber.

7. The system of claim 5 wherein the X-ray pulses form a fan beam and the dose detector is placed proximate to the fan beam in a position that is not obscured by an object being scanned by the X-ray pulses.

8. The system of claim 1 further comprising: a target for generating X-ray pulses when struck by the pulses of accelerated electrons, and a pulse-to-pulse dose detector configured to detect a portion of the X-ray pulses and providing an output signal proportional thereto as an input to the controller for adjusting RF power for the purpose of controlling dose on a pulse to pulse basis.

9. The system of claim 1 wherein the period of non-pulsing varies between one second and time required for a tuner for adjusting frequency of the RF source to return substantially to the home position.

Description

THE FIGURES

(1) The foregoing summary of the invention, as well as additional aspects and features, will be better understood from the following detailed description, taken in conjunction with the appended Figures, in which:

(2) FIG. 1 is a block diagram of a linac-based X-ray system in accordance with an embodiment of the invention.

(3) FIG. 2 illustrates in cross-section a linac in accordance with an embodiment of the invention.

(4) FIG. 3 illustrates the relationship between RF pulses and electron beam pulses in accordance with an embodiment of the invention, including the use of beam pulse duration to achieve pulse-to-pulse dose control.

(5) FIGS. 4A-4E are timing diagrams showing an interleaved sequence of stable pulses in accordance with an embodiment of the invention, including dose control.

(6) FIG. 5 is a block diagram of a dual mode gun driver in accordance with the invention.

(7) FIGS. 6A-6D are timing diagrams illustrating the cool down cycle which results from intermittent scanning operation, and the improvement resulting from the cool down compensation system and method of the present invention.

(8) FIG. 7 illustrates in block diagram form an alternative embodiment of the invention which offers dose-to-dose pulse control.

DETAILED DESCRIPTION OF THE INVENTION

(9) Referring first to FIG. 1, an X-ray scanning system according to an embodiment of the invention is shown in block diagram form at 100. In such an embodiment, external power and signals 105 are received by a control processor 110. Included among the external signals are, typically, one or more trigger signals indicating that the user desires to scan an object, for example a cargo container on a vehicle passing before the scanning system. The control processor 110 controls, directly or indirectly, the operation of the remaining functional blocks shown in FIG. 1 by virtue of signals sent on internal bus 115, which, for simplicity, is shown combined with internal power.

(10) In response to the trigger signal(s), the control processor 110 sends, depending upon the implementation, a plurality of signals to initiate generation of an X-ray pulse. In particular, the processor 105 sends control signals to a high voltage power supply 120 and an associated modulator 125 which receives the output from the supply 120. The supply 120 can be, for example, a Lambda LC1202. The output of the modulator 125 supplies a high voltage output to a pulse transformer 130, typically immersed in an insulating tank for purposes of electrical isolation. An aspect of the modulator is that can vary the voltages from one pulse to the next, and can operate at pulse durations of 2.5 sec or less, to permit operation at 400 pulses per second. The modulator may incorporate a pulse-forming network or PFN. A heater power supply 135 is associated with the tank and supplies the magnetron 140 or other suitable RF power source. The pulse transformer 130 supplies high energy pulses, for example 30-50 kV at 100-110 amps, to a magnetron 140 or other suitable RF power source. One suitable magnetron is the e2V model MG5193, which has an output of 2.6 MW at the normal S band frequency of 2.998 GHz. Another is the MG7095, also from e2V. Still other similar magnetrons are available from NJRC. The specific magnetron frequency is controlled by a mechanical tuner 145.

(11) The magnetron 140 outputs an RF power pulse, indicated at 150, at the frequency determined by the tuner 145. As explained in greater detail below, the pulses received by the magnetron can be of different, pre-selected voltage and currents, thus causing the magnetron to output pulses of different, pre-selected RF powers, for example, pulses of 40 kV and 45 kV at 100 amps and 110 amps, respectively. Because of the nature of the magnetron, the different powers of the RF pulses also affect the frequency of the output pulse, again as explained in greater detail below. The RF power pulses pass through an arc detector 155, an isolator 160, and then to a linear particle accelerator (sometimes linac hereinafter) 165. Suitable isolators are available from Ferrite Incorporated. Conventional S-band waveguide 157 is used between magnetron and linac. Depending upon the requirements of the particular linac, for example of the type shown in FIG. 2, the pulses received from the transformer 130 can be in the range of, for example, 35-50 kV. The linac 165, which typically has an effective Q in the range of 2000-4000, but in any event less than 5000, receives the RF pulse. The tuner 145 is adjusted so that the RF pulses from magnetron 140 are within the resonance bandwidth of the linac 150. The pulses from the magnetron are, in an embodiment, substantially in the range of 2.5 MW, or between 2.0 MW and 3.0 MW.

(12) In the embodiment of FIG. 1, the control processor 110 sends a control pulse to the modulator 125, and it sends a synchronized control pulse to a dual mode electron gun driver block 170. The timing of these control pulses may be individually optimized. The dual mode electron gun driver block 170, explained in greater detail hereinafter in connection with FIG. 5, drives an electron gun 175, the cathode of which is within the vacuum envelope of the linac 165. The gun 175 can be a triode gun design. In an embodiment, the pulses of beam current from electron gun 175 launch electrons into the cavities of the linac. The cathode voltage is substantially in the range of 10 to 20 kV. In an embodiment it is 20 kV. By timing the electron pulses at the linac to optimally synchronize with the electric field created by the RF pulses from the magnetron, as discussed in greater detail hereinafter, the electrons are accelerated by the linac to a desired energy level, typically in the range of two to ten or more MeV with, for at least some embodiments, a separation between the energy levels of approximately 1 MeV or more between sequential pulses in a rapidly pulsed ABABABAB pattern. The pulses are directed toward a target 180, for example tungsten, which, when hit with the pulse of accelerated electrons, emits pulses of X-rays. As discussed below in connection with FIGS. 3 and 4A-4E, the RF pulses are, in an embodiment, somewhat longer in duration than the electron gun pulses, such that the RF pulses can be thought of as creating an envelope within which the beam current pulses occur. To control dose, the duration of the beam current is selected by the control system 105, or can be pre-set during manufacture.

(13) Because the transmission characteristics of the linac and magnetron vary with temperature and other environmental factors, an AFC circuit 185 detects forward and reflected power from the linac, using dual directional couplers 190, and in turn controls the tuner 145 to ensure a continuing match between the linac and the magnetron. in a manner known to those skilled in the art.

(14) In another aspect of the invention, in some embodiments, the AFC circuit 185 also includes cool-down compensation, discussed in greater detail hereinafter in connection with FIG. 6. Because X-ray scanning systems operate intermittently, and the duration of the off periods is essentially unpredictable, the cool down compensation of the present invention operates the tuner during periods when the scanner is off to maintain a good match between the linac's resonance bandwidth and the output frequency of the magnetron, thus providing improved operational stability.

(15) In addition, ancillaries 195 connect to the arc detector 155 and an ion pump 200 that feeds the linac 165, both in a manner known to those skilled in the art. Finally, a cooling system 205 cools portions of the system in a manner known in the art, for example, the modulator, the pulse transformer tank, the linac, the target, and the isolator, as indicated by the dashed line 210. Suitable cooling systems are available from OptiTemp, and can be chosen dependent upon temperature and cooling requirements of the linac system.

(16) With reference to FIG. 2, a linac 165 in accordance with an embodiment of the invention can be better understood. The linac 165 is configured to provide X-ray pulses of different and controllable energies and doses in an alternating fashion. The linac comprises an electron gun 175, which can comprise a gridded electron gun. The beamline 220 is comprised of the main cylindrical section, and is comprised of cavities 225 that have been designed in order to accelerate electrons from the cathode of the electron gun up to speeds that correspond to energies in the MeV range. The cavities of the beamline can be inline coupled or side-coupled, and the cavities are also designed for resonance with the RF source, such as the magnetron 140 in FIG. 1. A water-cooled target assembly 230 is attached to the linac beamline, and houses a target 180, which is made substantially of tungsten, or other suitable material. In an embodiment, the linac is designed to accelerate electrons up to 6 MeV in energy and can also accelerate electrons to 4 MeV in energy on an alternating pulse, interleaved basis, for the purpose of generating X-rays with similar maximum energies per pulse. An ion pump 200 is connected to the vacuum envelope of the linac in a conventional manner. An RF window 235 is attached to the linac beamline, for the purpose of input coupling of RF power. Between the RF window 235 and the beamline 220 is a coupling iris 240 or opening, the size of which can be adjusted in order to assist in tuning the effective Q of the linac assembly.

(17) As discussed above, one aspect of at least some embodiments of the present invention is to generate interleaved sequences of X-ray pulses of at least two differing energies, for example 6 MeV and 4 MeV. In an embodiment, and with reference to FIGS. 3 and 4, the interleaved sequence of pulses rapidly alternates between the higher energy and the lower energy pulses, e.g., ABABABABAB, although other embodiments can provide different sequences of interleaved pulses, such as AABB or any other desired sequence. For purposes of clarity only, the alternating, or ABABABAB, sequence will be described in greater detail hereinafter, as those skilled in the art will understand how to provide other sequences, given the teachings herein. It will also be appreciated that more than two energy levels of X-ray pulses can be generated with the current invention as long as the frequency shifts of the magnetron pulses remain within resonance bandwidth of the linac and are properly matched to the beam current. For purposes of illustration, use of the present invention to generate an AB sequence is explained below, with reference to FIG. 1 and the timing diagrams of FIG. 3.

(18) To generate an alternating sequence of lower and higher energy X-ray pulses, the control system 110 instructs the modulator 125 to supply a different level of power, to the pulse transformer 120 for each different energy level of pulse. This, in turn, varies the power supplied to the magnetron 140, such that the alternating higher and lower RF pulses are supplied by the magnetron to the linac, as shown in FIG. 3 at 300 and 305. By varying the power to the magnetron, the frequency of the output of the magnetron changes, for example by approximately 10 kHz per amp, or 100 kHz when the current supplied to the magnetron varies from 110 amps for 6 MeV to 100 amps for 4 MeV, as an example. As discussed above, the Q of the linac is configured to provide a resonance bandwidth of sufficiently more than 100 KHz, so that both the higher and lower frequencies output by the magnetron remain within the operating range. Again as discussed above, in an embodiment, the Q of the linac is preferably in the range of 2000-4000, and a Q of 3000 has been demonstrated to operate well. In any event, the Q is less than 5000.

(19) As shown in FIG. 3, within the pulse duration of the magnetron pulses 300 and 305, the electron gun driver also generates current pulses of different energies, as shown at 310 and 315. The A pulse is of a desired electron and X-ray energy and X-ray dose, and the B pulse is of a different electron and X-ray energy and X-ray dose that is at least 1 or 1.5 MeV different from the energy associated with the A pulse. The magnetron current pulses are delivered by the pulse forming network and modulator of a preferred embodiment of the invention, and have a duration that is set by the design of the pulse forming network within the modulator 125 shown in FIG. 1. Other modulators could be used where the magnetron current pulse duration can be changed. In an embodiment, the duration of the current pulse is between 2 and 2.5 microseconds for both A and B pulses, and the repetition rate of the pulses is on the order of 100 to 400 pulses per second, or whatever is allowed by the maximum average power dissipation and delivery that a given magnetron can tolerate. For a faster magnetron, a rate of 1,000 pps, and, at the other end of the range, a rate of 1 pps can also have utility in some applications.

(20) The higher current magnetron pulse is typically correlated to the higher energy X-ray pulse, and the lower magnetron current pulse is typically correlated to the lower energy X-ray pulse. The high and low currents cause a higher and lower RF peak power to be delivered to the linear accelerator. The difference between these currents is limited in order to limit the magnetron frequency shift that occurs as a result of different voltage and current applied to the magnetron, in order that both the A and B magnetron frequencies are both well matched to the resonance bandwidth of the linac while still obtaining at least 1 MeV of difference between the X-ray energy of the A pulse and that of the B pulse, when pulsing in an interleaved sequence.

(21) As discussed in connection with FIG. 5, below, the gun driver can be controlled from the control system to alter amplitude, duration and timing of the beam current pulses. The gun current magnitudes are controlled by the gun driver, so that different beam currents for the A pulse and the B pulse are controlled. The combination of different RF power and different beam currents allows control of the energy of the A and the B pulses, in accordance with the method of the present invention. In an embodiment, the energy difference between the A and B pulses is 2 MeV, although in other embodiments the difference can range from 1 to 5 MeV. Further, the gun current pulse durations of the A pulses and the B pulses are independently controlled by the gun driver in order to control the X-ray dose on a pulse-to-pulse basis. The duration of the voltage pulse to the grid substantially controls the duration of the gun current pulse, and therefore the duration of the X-ray pulse. If all other parameters are fixed, and only the pulse duration parameter is adjusted, then the dose per pulse will be adjusted while the X-ray energy for a given pulse will remain the same. In an embodiment, the dose of the A pulse and the dose of the B pulse are substantially equal. In another embodiment, the ratio of the A dose and the B dose is adjusted to a desired ratio other than one.

(22) In addition to the adjustment of gun current pulse duration to control the dose per pulse, the timing of the start of the gun current pulse with respect to the RF pulse can also be adjusted. In a preferred embodiment, the leading edge of the gun current pulse begins early in the flat part of the RF waveform. In other embodiments, the leading edge of the gun current pulse can be started later with respect to the flat part of the RF waveform.

(23) The overall AB sequence can be appreciated from FIGS. 4A-4E, in which FIG. 4A illustrates two sequences of X-ray pulses with an off period of indeterminate duration in between. FIG. 4B illustrates the dose per X-ray pulse for the higher energy X-ray output, while FIG. 4C illustrates dose per pulse for the lower energy X-ray output. Both FIGS. 4B and 4C illustrate a range of dosage amounts with a variance which represents acceptable stability, typically about 10 percent or less, or 2% or less, or 1% or less. FIG. 4D illustrates the current supplied to the magnetron for each of the lower and higher energy pulses, while FIG. 4E illustrates the beam current supplied to the linac for the higher and lower energy pulses. It will be appreciated that the beam current is lower for the higher energy pulses than for the lower energy pulses in an embodiment. In one embodiment of the invention, the peak beam current is controlled to between approximately 25 mA and 125 mA when high energy pulses are desired, and the peak beam current is controlled to between approximately 125 mA and 250 mA as associated with the low energy pulses. Other beam currents are selectable with the linac system of the invention. In an embodiment, the beam currents can be precisely tuned in conjunction with the RF power in order to precisely provide a desired X-ray energy, and thus the X-ray energy can be alternated on a rapid pulse to pulse basis in an ABABABAB interleaved manner.

(24) As discussed above, the pulses generated by the electron gun driver are adjustable in amplitude, duration and timing, which permits the beam current pulses to be synchronized and matched with the RF pulses to generate X-ray pulses of different, controllable energy levels, with consistent, controllable dose per pulse. FIG. 5 illustrates an embodiment of a dual mode gun driver suitable for generating pulses of 2.5 microseconds or less, suitable for cargo scanning functions which typically operate from 100 to 400 pulses per second, in accordance with an embodiment of the invention. Higher repetition rates are possible if standard electrical and thermal parameters are considered, as will be appreciated by those of skill in the art.

(25) The gun driver 500 shown in FIG. 5 essentially comprises two independently controllable driver modules, one for the high energy pulses, indicated at 505, and another for low energy pulses shown at 510, where each module connects to the grid of the gun 175 (FIG. 1) at the appropriate times through control of high voltage switching transistors, and supplies the appropriate pulse at appropriate but different times as shown in FIG. 4E. The high energy module 505 comprises a heater supply 515 and high energy grid top power supply 520, with the output of the grid top supply 520 connected to the grid output through a switching transistor indicated at 525. The low energy module 510 comprises a low energy grid top power supply 530 and a grid bias supply 535, and the low energy grid top power supply connects to the grid via a switching transistor 540 and a diode 545, while the grid bias supply connects to the grid through a switching transistor 550. The gun driver also comprises a high voltage filter module 555 and a high voltage power supply 560 in a conventional manner.

(26) The performance of the gun driver modules, including the high voltage switching transistors, is controlled from the control system 110 and bus 115, as more generally shown in FIG. 1. Feedback signals from each module are also used, as is common in the art of electronics. The grid modulating modules 505 and 510 are referenced to the cathode voltage, as shown, and apply voltage to the grid of a triode gun that is part of the linac, shown at 175 in FIG. 1. As is known in the art, the control of grid voltage is a common method for controlling the gun current emitted from the cathode of a triode or gridded electron gun. In the preferred embodiment, one grid modulator is used to control the grid voltage associated with the A pulse, and the other grid modulator is used to control the grid voltage associated with the B pulse. In another embodiment, a single grid modulator can be used if the grid voltage can be changed accurately to desired values on a pulse-to-pulse time basis. The grid voltages are controlled in order to control the gun current that is launched into the linac, as the magnitude of the beam current in the linac is a function of the gun current. The independent control of the amplitude of the beam current, combined with the independent control of the RF power, is used to control the X-ray energy on a pulse to pulse basis.

(27) The duration of the grid pulse is used to control the dose of a given X-ray pulse. As discussed above, a shorter pulse yields a lower dose per pulse, whereas a longer pulse corresponds to a larger dose per pulse.

(28) Referring next to FIGS. 6A-6D, a different aspect of the present invention can be appreciated, by which pulses of consistent energy are provided even with a cold start or after the linac system in the scanner has been off for an indefinite period. More specifically, FIGS. 6A-6D depict both the effect of classic AFC techniques as well as the novel cool down compensation technique of the present invention that provides stability from the first pulse in an intermittent pulsing application. Since the invention outlined above uses standard prior art AFC techniques, the interleaved ABABABABAB etc. output can be stable over long time periods, for example, seconds, and minutes. Longer durations are possible as well. Prior art AFC techniques require at least one feedback signal proportional to the reflected RF power from the linac to insure proper mechanical tuning of the magnetron, as in the case of a preferred embodiment, and as is well known. In the case of a klystron, tuning is accomplished via an input RF driver, but the AFC still requires feedback representative of the output of the RF source with respect to the linac resonance. Since feedback techniques are used, many RF pulses (and therefore X-ray pulses in most systems) are required before the AFC subsystem has sampled enough information and adjusted the magnetron mechanical sufficiently that optimal and stable X-ray pulses are generated. A conventional AFC system alone is not sufficient to allow highly stable operation from the first X-ray pulse after an off period without RF feedback. In an embodiment of the invention, in order to take the same linac system with AFC and provide optimal and stable pulsing from t=0, or what is sometimes called a cold start, a cool down compensation (CC) algorithm has been invented and is used.

(29) With classic AFC circuits, the AFC compensates for thermal effects or long term drift effects that cause a drift in the X-ray pulse performance. If a drift or deviation in performance takes place over many pulses, and if it is due to a correctable frequency mismatch between the RF source and the linac resonance, then the classic feedback techniques used in prior art AFC subsystems can be used to stabilize the system. However, AFC designs and methods typically require many pulses to correct the position of the magnetron tuner and so, if only the AFC is used and if the pulsing is intermittent, that is, shut off for an irregular period of time and then turned back on, the dose and/or energy per X-ray pulse for some number of initial pulses can deviate significantly from the intended dose or energy per pulse.

(30) FIG. 6A depicts the operation of the magnetron in the context of a classic heating and cooling curve as shown in FIG. 6B. The shape of that curve is generally associated with the deposition of heat during pulsing, and then the dissipation of heat once pulsing stops. This is a classic phenomenon known to those of skill in the art, and when heating is occurring due to pulsing, a classic AFC can help maintain frequency match between a magnetron and a linac. However, when pulsing is turned off, then the system will cool, but the AFC cannot adjust the tuner in the same manner because it receives no feedback signals. The conventional feedback signals like forward RF power and reflected RF power do not exist for the AFC during off or non-pulsing periods. When a feedback signal is removed, prior art AFC subsystems have no input information with which to drive a motor or tuner, and may maintain the motor and tuner position that was appropriate for the most recent pulse, or some other position, but that is not appropriate for the next pulse at some indeterminate future point in time. This can result in non-optimum X-ray output or output at unintended values once the system is restarted, because the characteristics of the linac have changed during the off period.

(31) In an embodiment of the invention, cool down compensation logic is provided in the AFC circuit (185 in FIG. 1) that causes the position of the tuner to be changed during non-pulsing periods without the need for feedback, so that the output frequency of the magnetron remains properly matched to the linac resonance characteristics despite the cooling that occurs during an off period. In an embodiment, this technique can achieve stabilities significantly better than 10%, and in some instances approximately 1%. The cool down compensation logic can be a look up table (LUT), an algorithm, or other implementation, and is addressed by the AFC circuit when no pulses are detected for a predetermined period of time, for example, one second.

(32) FIG. 6C is a timing diagram that shows the movement of the tuner during cooling periods shows the changing tuner position that follows, and compensates for, the cooling that occurs when scanning is not active. FIG. 6D is a plot of dose per pulse, and illustrates that, as the result of the cool down compensation, the first pulses after a restart are substantially identical in energy and dose as the pulses that occurred during prolonged scanning with the AFC circuit active. In an embodiment, the AFC circuit with cool down compensation logic drives a stepper motor, and the stepper motor moves the tuner on the magnetron in a conventional manner.

(33) The cool down compensation data used to populate a LUT or define an algorithm, as discussed above, can be developed as follows. For a given RF value, a proper magnetron starting or home tuner position is determined by very low repetition rate operation of the linac system, at the desired energy and dose per pulse. For example, this home tuner position can be determined at 1 to 10 Hz operation, and can be set by an operator who optimizes system performance vs tuner position, either manually or with a classic AFC. The very low repetition rate simulates a scenario where very little heating is occurring. This home tuner position is recorded in system memory. This type of operation simulates the behavior of a linac system with very little stored heat in either the linac or magnetron. In a preferred embodiment, the magnetron tuner is driven by a stepper motor, and the number of steps away from a motor reference point or mechanical stop is used as a proxy for tuner position. As an example, the home position may be determined to correspond to 50 steps away from the reference point or mechanical stop. A tuner position associated with full power and full repetition rate operation may achieved by 100 or 150 steps away from the reference point or mechanical stop, as an example

(34) For the CC technique of the present invention to cause optimal tuning during non-pulsing, a calculation or look-up table is generated with motor positions that correspond to substantial matching between magnetron frequency and linac frequency during the time intervals where no pulsing is commanded. After very long off periods, like minutes, the proper motor position corresponds very nearly to the home position. After long on periods, for example after minutes of operation, the motor position determined by a classic AFC will be many steps away from the home position, in order to tune the magnetron to the linac resonance in a warmed state. For off periods of varying durations, the optimal motor position will be somewhere between these two positions, and must be determined.

(35) A variety of techniques can be used to determine optimal motor and tuner position in the absence of pulsing. For an embodiment of the system, an outline of one procedure that can be used to develop the tuner positions during cool down, and thus the entries for a LUT or other correction technique, is as follows:

(36) 1. The system is operated at full energy and power, so for example, the magnetron output is 2.6 MW, and it is pulsed at 400 pps with pulse durations of about 2 microseconds for the flat-top of the RF pulse. This mode corresponds to the maximum heat deposition to the system in an embodiment. A standard AFC circuit maintains the match between the magnetron frequency and the linac resonance during this pulsing period, with a time constant on the order of 0.5 seconds. Near steady state can be achieved in times on the order of a few minutes for this approach.

(37) 2. The system is turned down to 10 pps abruptly, and the AFC is allowed to tune the magnetron tuner in a standard way, typically by driving a stepper motor. Every 5 seconds, an operator or a data acquisition system records the position of the stepper motor, over a period of several minutes.

(38) 3. A plot of stepper motor position is created, which shows steps on the Y axis, and time on the horizontal axis.

(39) 4. The plot in step 3 above can be fit to an exponential decay with a time constant, or it can be used to create a look-up table. The calculated exponential decay can be used to calculate the proper tuner position as a function of its most recent position; in a preferred embodiment, the farther a tuner has been stepped from its zero heat or home position, the larger its correction per unit time towards the zero heat or home position should be.

(40) In an embodiment, a rate of change in terms of steps per second towards home may be calculated or measured from the data collected in steps 1 to 3 above. The rate of change can be plotted as a function of the steps away from home. In a preferred embodiment, the optimal rate of change is largest for position changes that are large with respect to the home position. With the CC active, for any position of the motor that drives the magnetron tuner, there is an associated rate-of-change in steps/second corresponding to that position. This rate of change can either be calculated by determining how many steps the AFC moved between intervals in the procedure 1 to 3, or it can be derived from a fit to the exponential curve to the position vs time data collected in steps 1 to 3. Alternatively, a look-up table can be created.

(41) In an embodiment, an optimal expression of the steps per second as a function of steps from home can take the form
Y=a*e.sup.(b*x)
where Y is the optimal steps per second, x is the steps from home, and a and b are constants determined by a mathematical fit to experimental data for the linac system in an embodiment. In general, a is a coefficient representing the magnitude of the exponential equation, and b is a coefficient for curve-fitting. For example, in an embodiment where the stepper motor has on the order of 100 steps of range, a can be 0.13, while b can be 0.034. Those skilled in the art will recognize that these coefficients can be different numbers depending on the thermal behavior of the system, or the behavior of the system in response to other parameters. When the CC is used to move the stepper motor that moves the magnetron tuner, and the CC uses the formula, the pulsing performance in an intermittent pulsing scenario was very nearly optimal for every pulse, as shown in FIGS. 6D and 6E.

(42) In an alternative embodiment, the following procedure can be used:

(43) 1. Warm up at 400 pps for 3 minutes, note warmed-up tuner motor step position with respect to home (position as controlled by classic AFC).

(44) 2. Shut off for 5 seconds, turn back on, record tuner motor step position after 2 seconds (position as controlled by classic AFC).

(45) 3. Warm up again at 400 pps to same warmed-up tuner motor step position as #1.

(46) 4. Shut off for 10 seconds, turn back on, record tuner motor step position after 2 seconds (position as controlled by classic AFC).

(47) 5. Warm up again at 400 pps to same warmed-up tuner motor step position as #1.

(48) 6. Shut off for 15 seconds, turn back on, record tuner motor step position after 2 seconds (position as controlled by classic AFC).

(49) Continue this sequence for up to 2 minutes of shut off time. Plot the data as tuner motor position in steps-from-home (y axis) vs cool-down time (x axis).

(50) 7. Fit an exponential to that data.

(51) 8. Take the derivative of that data, and plot it on the same graph, which will now depict both the exponential fits to steps-from-home and steps/second vs cool-down time.

(52) 9. Create a new plot of steps/second vs the steps-from-home.

(53) In an embodiment, this table created from such techniques is used by the AFC with CC during customer-triggered intermittent operation to provide the rate at which the tuner motor should be moved any time the RF is turned off, based upon the last position where the tuner motor was left by the AFC. In this embodiment, the AFC controls the tuner motor position while RF and X-rays are pulsing, and the CC controls the tuner motor position when RF and X-rays are turned off.

(54) In some cases, further improvement resulting in better pulse-to-pulse stability is desirable. In the application of cargo scanning this can facilitate better imaging, and in the application of radiation oncology this can facilitate more stable dose delivery. For example, an RMS stability of the X-ray dose per pulse of nearly 1%, or less than 1%, is desirable. In the case of the linac operating in dual energy on an alternating basis, the dose per pulse at one energy may be less stable than that of the other. To enhance stability of the dose beyond what is achieved by AFC alone, additional techniques can be used. One technique, which uses substantially the hardware structure shown in FIG. 1, takes a digital output from the AFC circuit 185 during X-ray pulsing bouts in order to generate a correction signal for the programming voltage for the modulator 125 that drives the magnetron 140. The programmed voltage for steady state magnetron operation is typically a fixed value, for example, a high voltage power supply programmed voltage of 20 kV will result in a PFN output of approximately 10 kV which results in a pulse to the magnetron of 45 kV (after a 4.5:1 pulse transformer.) This results in an RF output of approximately 3 MW, which in combination with a specific gun current will result in a specific desired energy, such as 6 MeV. Using the AFC circuit's digital output of the AFC stepper motor position, a correction value to the normally fixed programmed voltage can be created, using the reported AFC stepper motor position with respect to home or zero position. In a preferred embodiment, the correction value can be positive or negative, and is added to the steady state programmed voltage that corresponds to one of the desired pulse energies. In a preferred embodiment, this correction value contains a term proportional to the AFC stepper motor position with respect to the home position that corresponds to tuner position with no system heating, and this correction value also contains a fixed offset term. Either term of the correction value can be positive or negative, depending on the direction of desired correction to the pulse-to-pulse X-ray dose over a typical 30 to 60 second scan. The correction value can be added to the programmed voltage to just one of the alternating energy pulses (either the high or the low energy), or to both, or a differently scaled correction value can be generated for each of the energy settings. By adjustment of the correction value, enhanced X-ray stability over the duration of a typical scan is achieved, resulting in an improved image in a security scanning system.

(55) Another method to stabilize the X-ray dose per pulse is by measuring target current on a pulse-to-pulse basis. One approach for accomplishing this is where the target assembly has been electrically isolated from the linac body, and current due to the electron beam striking the target is measured with a simple circuit. Peak target currents of between 25 mA and 250 mA during a pulse are typical values, which allows generation of a measurable signal voltage across a resistor in series with the target current. If the electron energy is constant, the dose for a given X-ray pulse or series of pulses is directly proportional to target current. Therefore, if electron energy is constant, a feedback loop can be implemented using target current as the measured parameter that is a proxy for X-ray dose. The feedback loop stabilizes target current (and thus X-ray dose), on a near pulse-to-pulse basis by adjusting the high voltage power supply program voltage (the voltage that drives the magnetron, as discussed above). Another method to achieve this effect is to measure the integrated target current over an individual pulse, in order to calculate the total charge delivered in an individual pulse. In this alternate method, using integrated target current per pulse, a servo loop that adjusts pulse duration at the target is used to stabilize the integrated target current per pulse with respect to a desired value. In a system with a triode or gridded gun and an independent triode gun driver, a straightforward way to control the duration of a pulse at the target is by controlling the duration of a pulse to the grid as delivered from the gun driver. Certain solid state modulators may also facilitate this form of stabilization. However, this approach can suffer from certain instabilities if a change in energy occurs.

(56) A still further alternative, shown in FIG. 7, offers some advantages because it measures actual dose on a pulse-to-pulse basis rather than the integration-based averaging performed by many other designs, and, by means of a feedback loop, uses the measured dose to stabilize the successive generated doses. The feedback loop can, for example, control either the voltage level at which the magnetron is driven, which adjusts the RF power deliver to the linac beamline; or, alternatively, the pulse duration of the gun driver that drives an electron gun with a controlling grid, where longer pulses correlate to a greater dose per pulse and shorter pulses correlate to lower dose per pulse.

(57) For the purpose of measuring a signal that is proportional to dose, either a scintillator crystal based detector or an ion chamber can be used. A challenge to this approach is the development of a detector with an appropriate response to the X-ray dose, and ideally with a signal level and speed that allows a pulse to pulse measurement of the X ray signal. Scintillator crystals and diode detectors are used for this purpose in the detector arrays used for X-ray imaging of cargothey are designed for detection of the relatively lower level signals that pass through the cargo. A detector using a cadmium tungstate crystal and photodiode can be used, as is well known in the detection of gamma level X-rays. Care must be taken to keep X-ray levels low enough to minimize damage to the crystals, but high enough to provide an adequate signal for the photodiode detectors to measure with respect to electrical noise levels, as is known in the art. Detectors are available by custom order from Berkeley Nucleonics Corporation, of San Rafael Calif. A challenge with an ion chamber is the development of a chamber with a sufficiently fast response time and sufficiently high signal levels. One such ion chamber is the A12, which can be procured from Standard Imaging in Middleton Wis.

(58) The signal from either of these properly designed dose detectors can be used to measure dose per X-ray pulse, and this signal can be used in conjunction with a feedback circuit or computational technique for the purpose of maintaining an X-ray dose per pulse that is more stable on a pulse to pulse basis than may otherwise be achieved. The magnetron voltage is adjusted with a correction depending upon an error signal generated by the difference between the actual measured dose-per-pulse, and the desired dose-per-pulse.

(59) Still referring to FIG. 7, an embodiment of the invention using a pulse-to-pulse dose detector can be better appreciated. It will be apparent that the basic structure is that of FIG. 1, and like elements are indicated with like numerals. As shown in FIG. 7, a pulse-to-pulse dose detector 105, typically either scintillator-based or ion chamber-based, detects a portion of the X-rays emitted from the target, and generates a feedback signal 107 which provides an additional input to the control system 110. The scintillator-based detector is preferred in at least some embodiments. An ion chamber can be designed to intercept and measure most of the X-ray beam by placing it close to the target. However, the ion chamber measures dose or signal in a different way than the scintillator crystal and diode combinations used in the arrays used for cargo scanning. The ion chamber measures a signal proportional to the amount of ionized gas between two conductors, but the scintillator uses a crystal that emits visible light with a relationship to the X-rays incident upon the crystal, with a photodiode and electronics being used to measure the light output of the scintillator crystal. Therefore, it is preferable in some embodiments to stabilize pulse-to-pulse dose from the linac system using a scintillator based detector.

(60) In a typical design, a collimator (not shown) is used to generate the desired X-ray pattern for cargo scanning. In such cargo scanners, the X-ray pattern is commonly a fan beam having an included angle between 50 and 90 degrees, and a width of 2 mm to 3 mm. This line beam of MeV X-rays is used to illuminate the detector array in a scanning system, with this array made of scintillator crystals and diodes and necessary signal processing electronics. In the presently described embodiment of the invention, the pulse-to-pulse detector 105 is placed either within the linac system cabinetry itself, or near it, in a position that is never obscured by the cargo to be scanned. The detector can be placed at the edge of the X-ray fan beam, or just outside of that fan beam, in order to measure a signal that is directly proportional to that delivered in the fan beam for imaging purposes. Other locations are also possible, such as behind the electron gun, as some X-rays are emitted from the target in directions other than the fan beam direction. In an embodiment, the detector is placed in a location that provides sufficient signal for the scintillator-diode detector combination, and provides a signal that is sufficiently proportional to the dose per pulse delivered into the fan beam used for imaging.

(61) Use of an ion chamber for measuring dose in a linac system is well known in the art. However, it is novel to use of a detector that is the same as that used in the imaging array, but here for the purpose of measuring and stabilizing the X-ray dose on a pulse-to-pulse basis. The pulse-to-pulse dose detector signals are measured by the control system, where the measured signal associated with the dose per pulse is compared against a reference value. Standard feedback techniques can be used to generate a control signal to the appropriate linac subsystem for the purpose of stabilizing the dose per pulse. In a preferred embodiment, the voltage level of the HVPS (120) is adjusted on a pulse-to-pulse basis in order to provide an improved stability per pulse. In another embodiment, the duration of the gun pulses is controlled in order to adjust the dose per pulse and improve the stability per pulse; in this embodiment, the duration of the pulses delivered to the grid of the electron gun is adjusted in order to adjust the dose per pulse. In a third embodiment, if a modulator is used to drive the magnetron that allows adjustment of the duration of the pulse duration to the magnetron, such as a modulator available from Scandinova Electronics or e2V, then the duration of this pulse to the magnetron can be used to adjust the dose per pulse. In all cases, benefit is provided by using a pulse-to-pulse dose detector that measures the dose per pulse in a manner that is very similar to, or exactly the same as, the manner used by the imaging array.

(62) From the foregoing, those skilled in the art will recognize that a new and novel linac-based X-ray scanning system has been disclosed, offering significant improvement in pulse-to-pulse stability for interleaved pulses of different X-ray energies, including pulse-to-pulse dose control, all on a rapidly pulsed basis. In addition, in another aspect of the invention, cool down compensation permits substantially improved stability in the dose and energy of the initial pulses after a cold start, or after a restart after an off period of indeterminate duration. Given the teachings herein, those skilled in the art will recognize numerous alternatives and equivalents that do not vary from the invention, and therefore the present invention is not to be limited by the foregoing description, but only by the appended claims.