HK1081337B - Power modulator - Google Patents

Description

Electric power modulator

Technical Field

The present invention relates to a power modulator, and more particularly, to a power modulator with a pulse generation module employing a primary winding and a secondary winding.

Background

a. Modulator, general description and definition of terms

The modulator is a component that controls the flow of power. When a lamp is switched on and then switched off again, the current fed to the lamp is regulated, as it were. In its most common form, the modulator delivers a train of high-power electrical pulses to a prescribed load, such as a microwave generator. Most high power radars in the world utilize a modulator to deliver power pulses to a microwave source, which in turn feeds power to an antenna structure in the form of periodic microwave pulse trains. Other possible applications of such a power modulator are listed below.

In decades after the second major war, the basic structure of the power modulator has not changed significantly. Conventional modulators include a power supply that receives power from an AC (alternating current) power line, boosts the voltage, rectifies the power to produce direct current DC power, and delivers the energy to an accumulator, typically formed by an energy capacitor bank. This is necessary because the power input line cannot deliver peak power, so the accumulator delivers peak power in small energy slices and is replenished or charged by the DCPS at a reasonably constant rate with much less average power.

Part of the energy of this accumulator is then transferred to a second, smaller accumulator, usually a Pulse Forming Network (PFN). PFNs are capacitor and inductor networks designed to deliver power to a load in the form of square waves (flat-top waves) that have rise and fall times that are fast compared to the pulse width or duration.

The pulse forming network (artificial transmission line or delay line) is then switched to connect it to the primary side of a pulse transformer, which is usually but not always a step-up transformer. The PEN voltage before switching is V, and the voltage applied to the primary of the pulse transformer is V/2 or slightly less. This is a disadvantage of PFN technology. The turns ratio (step-up ratio) of the pulse transformer must be doubled for PEN and for the present invention.

The PFN is fully discharged within a time T (typically a few to tens of microseconds), which maintains a fairly constant voltage on the primary of the pulse transformer and produces a fairly flat output pulse on the secondary of the transformer. But if pulse flatness of around 0.1 percent is required, the PFN must have a very large number of inductor-capacitor (LC) segments, and this is difficult to adjust. In addition, if any component in the PFN fails, the PFN needs to be readjusted when a new component is installed, because the value and location of all components in the PFN are critical.

After the pulse is delivered, the PFN must be fully charged to voltage V for the next pulse. To maintain pulse-to-pulse repeatability at a few tenths of a percent, large charging voltage "swings" must occur with high accuracy. All PFN capacitors are fully charged and fully discharged in each pulse hundreds or thousands of times per second, which imposes strong strains on the dielectric material of these capacitors, forcing them to be designed with very low stress and hence energy density. This makes the structure of the PFN large compared to the new concept of the invention where the capacitor is not discharged and recharged for every pulse and thus the capacitor can have a much higher energy density.

In summary, the drawbacks of the prior art modulators are:

the voltage on the primary side of the pulse transformer is high, typically 10 kilovolts or more.

The PFN must be fully charged to 10-20 kV for each pulse and fully discharged during the pulse, which places high stress on the capacitor

For the above reasons the energy density of PFN capacitors is low, so that they are rather large compared to the low stress capacitors used in the new concept of the invention.

If the load is short-circuited (as happens frequently with magnetrons for example), the current cannot be interrupted because the high voltage PFN switch (gas-filled thyristor) cannot be turned off before the current drops to zero.

If one element in the PFN fails, the PFN must be retuned to the optimal pulse shape after the element is replaced. This is a cumbersome and dangerous task as it must be done with high voltage applied to the PFN.

If a different pulse width is required, the PFN structure must be replaced and retuned

b. Pulse transformer

The history of so-called fractional-turn pulse transformers began with an invention of nicholas christofilos, which was assigned to the united states government Lawrence Livermore National Laboratory (LLNL) in the early 60's of the 20 th century. Then the laboratory was called the Lawrence Livermore laboratory or LLL. The invention discloses a method using a large number of ring-shaped (toroidal) magnetic cores, each core being driven by a high-voltage pulse generator of several tens of Kilovolts (KV) (using spark-gap switches and pulse forming networks or PFNs) to generate an accelerating potential of several hundreds KV to several Megavolts (MV) to accelerate a charged particle beam. . The basic idea of a so-called Linear Magnetic Induction (LMI) accelerator is shown in the following fig. 1 and 2.

Figure 1 shows a set of torroidal cores arranged so that their central apertures surround a line along which a particle beam is accelerated.

FIG. 2 shows an LMI structure with more detail added; a High Voltage (HV) driver system (one for each core) is shown and indicates the beam path.

A key feature of this type of accelerator is that it has an outer surface at ground potential, as with all LINAC accelerators. The voltages driving the various cores all appear to add in series down the central axis, but not anywhere else. This means that the accelerator does not radiate energy to the outside world and is easy to install in the laboratory since it does not require isolation from the environment, LLL established an 800kV LMI (ASTRON accelerator) in the 60's of the 20 th century and was used for electron beam acceleration in fusion tests. In the laboratory in the 70's of the 20 th century a larger LMI machine (FXR, for transient X-rays) was built and electron beam pulses were accelerated to the X-ray conversion target. FXR is used for frozen frame radiography of explosions.

The principle of operation of the LMI accelerator can be explained by means of fig. 3, which fig. 3 shows a cross-sectional view of the machine in a plane containing the beam axis.

Some rules are required for discussing the behavior of the multi-core structure shown in fig. 3. First, a right-handed rule is required. The rule of thumb states that if you hold a conductor with the right hand and point in the positive direction of the current with the thumb, your fingers will hold the conductor in the direction of the lines of magnetic flux around it. Applying this rule to fig. 3, the flux induced in the torroidal core will circulate as shown. With "dot" indicating the flux vector is directed toward the reader, and "X" is used to indicate the flux vector is away from the reader.

Applying this rule to the beam flowing along the axis to the right of the structure in fig. 3, we find it to be true that the flux generated by the beam circulates in the opposite direction to the flux induced by the primary current. If we consider it to be a transformer and the beam is a short circuit on the secondary winding, the current in the secondary will flow in a direction that cancels the flux induced in the primary, causing no net flux to be induced in the core and thus representing a short circuit to the primary power supply. The absence of a change in flux in the core means that there is no voltage across the primary winding and by definition this is a short circuit. Thus, a beam of positively charged particles (protons) will accelerate through the structure to the right, while a beam of negatively charged particles (electrons) will accelerate to the left.

Another rule of electricity now applies, namely that the voltage induced in the conductor surrounding the magnetic flux is equal to the rate of change of this magnetic flux. The path ABCD of the magnetic flux around all five cores was studied. The voltage induced in the wire flowing through the path is equal to the rate of change of the magnetic flux common to all five cores. But each core is driven by a primary voltage V so each core has a rate of change of flux equal to V. Thus, the voltage induced along path ABCD should be 5V. This structure functions as a step-up transformer. Another rule is that the ratio of secondary voltage to primary voltage in the transformer is equal to the ratio of the number of secondary turns to the number of primary turns, so that the LMI accelerator effective turns ratio of figure 3 is equal to 5, although path ABCD represents only a single turn. Thus the primary must be 1/5 turns, so the LMI accelerator can be seen as a transformer with fractional turns primary

c. Other related arts

FIG. 4 is a sketch of the pulse transformer connection disclosed in U.S. Pat. No. 5,905,646 to Crewson et al, 5, 18, 1999. For simplicity, two pulse generation modules are shown. As can be seen, each module drives a single turn primary (1) around one of the two cores. Each module contains a capacitor charged to a voltage V and has a clamping diode or reverse diode D to prevent the switch from undergoing a destructive "reverse spike" of voltage caused when the switch is opened.

The U.S. Pat. No. 5,905,646 given above emphasizes the idea that each module drives a separate turn in the primary structure. This is achieved by ensuring that all module switches will conduct the same current. This limitation leaves the modulator with a potentially destructive failure mode. To understand this destructive failure, assume that the two switches in fig. 4 are not accurately conducting simultaneously. The nuisance suddenly occurs if the upper switch is turned on a fraction of a microsecond earlier than the lower switch (or vice versa). If this happens, the upper core produces magnetic flux in the direction shown (flowing into the page at "X" and out of the page at the dot symbol). This flux induces current flow into the secondary and the load as shown. There is no magnetic flux in the lower core because its modular switch is not yet conducting, but the current flowing in the secondary winding will induce a magnetic flux in the lower core in the direction opposite to that indicated. This flux will induce a current in the lower module connection shown and this current will cause the diode D in the lower module to conduct.

Now, when the switch in the lower module does turn on, the applied voltage reverse biases the lower diode (which is conducting) and this forces the diode to turn off. A conducting diode is usually damaged when it is turned off within a few nanoseconds under high current conduction. When the diode breaks down, it becomes a short circuit. The short circuit then draws an almost unlimited current through the switch of the module below, damaging the switch.

Disclosure of Invention

It is an object of the present invention to obviate the above-mentioned disadvantages of prior art modulators, as in U.S. patent No. 5,905,646. It is a further object of the invention to provide a power modulator whose primary winding connections do not necessarily require the same number of pulse generating modules and primary windings. It is a further object of the present invention to provide a power modulator which eliminates the aforementioned drawbacks of prior art modulators as follows:

the voltage on the primary side of its pulse transformer is low, typically 1kV or lower;

there is no PFN, avoiding all the disadvantages of PFN, since the modulator switches are semiconductors such as IGBTs or mosfets, which can be turned off when current flows in, terminating the pulse.

The energy storage capacitors do not discharge more than a few percentage points during the pulse, so that their energy density can be much higher than that of PFN capacitors;

if a short circuit occurs on the load, the short circuit can be detected by observing a sudden drop in the load voltage, generating a signal that releases a fast comparator that removes the low voltage strobe from the semiconductor switches, thus ending the pulse (prior art modulators employ overcurrent detectors for this purpose, which are much slower in operation and let a much higher current flow before opening); and

if a different pulse width is required, this can be provided by simply changing the timing of the solid state switch trigger, which is done at low voltage and can be done from a computer control station, allowing simple electronic adjustment of the pulse width.

These advantages result in much smaller size and longer service life of such solid state modulator systems compared to the older technology PFN/thyristors.

It is a further object of the present invention to provide a power modulator that can turn different pulse generating modules on or off at different times, which can be useful for eliminating overshoot or ringing at the start of a pulse.

According to an aspect of the present invention, there is provided a power modulator comprising: at least two pulse generating modules; and including at least one secondary winding surrounding all of the plurality of magnetic cores; wherein each of the at least two pulse generating modules comprises: an energy storage capacitor; a switching device which can be switched on or off in an electronically controlled manner; and at least one diode or series-connected diode assembly connected in parallel with an output conductor of the pulse generation module, and wherein each pulse generation module further comprises a set of primary windings, each primary winding of the set of primary windings surrounding a corresponding one of the plurality of magnetic cores, and each primary winding of the set of primary windings being connected in parallel with at least one of the pulse generation modules, thereby providing a power modulator in which each of the at least two pulse generation modules drives all of the plurality of magnetic cores.

For simplicity of illustration, the discussion is limited to the case of two pulse generation modules. This is in no way meant to be limiting to the invention, which may instead operate with any number of pulse generating modules.

The addition of two single turns as shown in fig. 5 completely eliminates the aforementioned overvoltage fault mode and at the same time eliminates the limitation of equal number of pulse generating modules and primary windings. The prior art modulator built according to the us 5,906,646 patent is limited by having one pulse generating module per primary winding and at least one primary winding per core segment. This limitation can be removed with the present invention so that any number of modules can be used. The prior art requires that each core segment be driven by the same number of modules. But with the present invention any number of modules can be used and still provide the same drive signal to each core. This is a strong economic advantage in favour of the present invention.

When conductor (11) in fig. 5 is added, then whichever module switch turns on first controls the circuit until the other switches close. If the upper switch is turned on before the lower switch, the upper module drives the flux in both cores, not just the upper core. This will prevent the underlying diode from being pulled into a conducting state because the diode will be reverse biased. The effect is almost the same if all primaries are connected in parallel, i.e. the "early" switch will apply a positive voltage to the diodes in all "late" modules.

In fact all primaries may be connected in parallel for further simplification. This is not obvious, but examining fig. 4 will see that this is functional. If in fact all switches are closed at the same moment, there is no voltage between points P and R in the figure. If points Q and T are connected together, there is no voltage between these points either. If there is no voltage between P and R, it is also possible to connect these points together without causing additional current to flow, so that the circuit will work the same way as if all primaries were connected together.

Both of the above connections are in fact used to equalize the module currents, which cannot be achieved with the independent connection disclosed in the us 5,906,646 patent. This is because the invention ensures for the first time that the load impedance presented to all pulse generating modules is identical. Which is not guaranteed in the prior art. As for the two primary connections given above, the connection shown in fig. 5 is preferred over the idea of simply connecting all the primaries together in parallel, because any diode failure in any module when all primaries are connected in parallel will pull all the current from all the modules into the failed diode, which can be very disruptive to the switch. The connection of fig. 5 eliminates this possibility.

Drawings

FIG. 1 shows a core arrangement for an LMI accelerator;

FIG. 2 shows an LMI accelerator with a PFN and a beam path;

FIG. 3 illustrates the structure of a five-center LMI accelerator;

FIG. 4 is a schematic diagram of a half turn primary connection with two pulse generating modules according to the prior art;

FIG. 5 is a schematic diagram of a half turn primary connection with two pulse generation modules in accordance with the present invention;

FIG. 6 is a detailed diagram of one pulse generation module in one embodiment;

FIG. 7 illustrates some typical high voltage pulse transformers;

figure 8 is a three-dimensional view of a pulse transformer having two cores side-by-side.

Detailed Description

How to design the power modulator of the present invention and some preferred embodiments of the present invention are described in detail below by way of examples. These preferred embodiments and this design example are merely illustrative and are in no way to be construed as limiting of the present invention.

The invention provides a power modulator. The power modulator includes: a set of pulse generating modules, wherein each pulse generating module has a primary winding connection around a set of magnetic cores and has a secondary winding around all these magnetic cores. The pulse generating modules further comprise an energy storage capacitor, switching means and a diode or series connected diode assembly. The diode or the series connected diode assembly is connected between the switching device and a conductor returning current to the capacitor. The switching means may be any switch known to those skilled in the art of electronics, such as an IGBT solid state switch. As mentioned above, the pulse generating modules/module groups in the modulator have one connection of the primary winding and in fig. 5 the connection is shown for the specific case of a power modulator with two pulse generating modules and two cores. The power modulator with these features is the power modulator cited in each of the following embodiments.

As highlighted previously, the connection shown in fig. 5 completely eliminates the destructive failure mode that often occurs in the prior art. Fig. 5 shows a connection with two pulse generating modules and two cores. This is merely illustrative and the number of modules may be any number and the number of cores may be any number. All that is required is to have a power modulator with more than two pulse generating modules connect all the modules by means of a primary winding around each core, so that each core is surrounded by one primary winding from the respective module. Thus, this is a simple way to add any number of modules to any number of cores. As an example, a power modulator with N pulse generating modules and M cores can be considered. If one is to build a power modulator with only a single-turn primary winding and the connection of the invention, the total number of single-turn primary windings should be 2 mxn, where 2M single-turn primary coils for each pulse generating module should enclose M cores (one primary turn per core "leg (leg) and two" legs "per core as shown in fig. 7), and N pulse generating modules. It is important to note that in a preferred embodiment these pulse generating modules are not connected together by external connections such as wires. The only connection existing between the modules is an inductive connection caused by the magnetic flux in the core. It is of course possible to interconnect the modules by connecting all primary windings in parallel, although this may lead to the aforementioned failure, i.e. when all primary windings are connected in parallel, any diode failure in any module will pull all current from all modules into the failed module, which may be very disruptive to the switch.

Another embodiment of the invention includes a power modulator with a single pulse generation module having a primary winding that surrounds all of the cores. In this way a single module drives all the cores.

It should be noted that the primary winding discussed above may be a single turn or a multi-turn primary winding. In the latter case there is the condition that the voltage per turn in all primary windings is the same, i.e. if the voltage V in the capacitors belonging to different pulse generating modules is different, the difference should be reflected in the number of turns N of the primary windings, so that V/N is equal for all primary windings. If this condition is not met, the primary windings may contend with each other and draw excessive current from each pulse generation module. If the voltage across all capacitors is the same, the number of turns in each multi-turn must be equal to satisfy the voltage per turn condition.

Power modulators typically have a wide variety of applications, such as: radar systems, lasers, cancer therapy, microwave heating, sterilization of materials in the process, particle acceleration (LINAC) driving, plasma heating for nuclear fusion, semiconductor cleaning, surface treatment, electron beam pumping of gas lasers, curing of inks in the printing industry, driving of piezoelectric or magneto transducers in sonar, ultrasonic medical imaging, driving of antenna structures for monopulse broadband radars, driving high currents and voltages in aerospace vehicles, lighting, simulation of nuclear weapons action, direct driving of electron beam sources for material modification, driving of klystrons for microwave generation for radars, magnetrons, gyrotrons or crossed-field amplifiers, etc. Of course, a power modulator such as the present invention may be used in any location where electrical pulses are required.

Example Modulator design

The following is an example of how a power modulator may be designed according to the principles of the present invention. This example is merely illustrative and all numbers and specific components are included for pedagogical purposes only and are not to be construed as limiting the invention.

a. Selection of number of pulse generating modules

An effective way to understand the concept of the new modulator is to make a preliminary design exercise for such a modulator. Assume that a 120kV pulse is to be achieved with a peak power of 70 amps on a 5 microsecond (μ s) pulse width and with a Pulse Repetition Frequency (PRF) of 800 pulses per second (800 Hz). It is also assumed that a 1600 volt, 2200 amp rated IGBT solid state switch is available.

The first step is to calculate the peak power output because it strongly influences the number of switches needed. The peak power is 120kV by 70A, or 8.4 megawatts (8.4 MW). The safety margin of the switches is taken into account so that they do not operate at their maximum rated condition. Experience has shown that a peak rating of 75% allows a long life safety margin. The operating value of each switch should therefore not exceed 0.75 × 1600 ═ 1200 volts and 2200 ÷ 0.75 ÷ 1650 amperes. Thus, each switch can produce 1200 × 1650 ═ 1.98 MW. To deliver 8.4MW, 4.24 switches are needed at 8.4/1.98, so five switches are used to keep the design safe.

In prior art modulators built according to U.S. Pat. No. 5,905,646, there is a limitation that each primary winding has a pulse generating module and each core segment has at least one primary winding. With the present invention, as noted in the foregoing description, this limitation is eliminated and any number of modules may be used. For example, if a pulse transformer with two cores is employed to obtain a primary winding of 1/2 turns, there are four "legs" or four core segments in such a transformer. The prior art requires that each such "leg" or core segment be driven by the same number of modules. Four modules cannot be used because this violates the safety margin chosen in this example. Thus, in the prior art, eight (8) modules must be used so that each core "leg" is driven by two modules. But with the present invention it is possible to use five (5) modules and still provide the same drive signal to each of the four core "legs" of the transformer.

b. Capacitor size and pulse flatness

The number of pulse generating modules is decided to be 5 in this example (at least in the preliminary design). This data may be slightly changed in further iterations involving the design of heat transfer and cooling of the switches, capacitors, and other components, but the level of detail of such a design is beyond the scope of the present invention. Turning now to the capacitors used in each module. If a capacitor of farad C is charged to a voltage V and then connected to a load to draw a current I in T seconds, the voltage of the capacitor will "droop" or "slide down" as follows:

ΔV＝ΔQ/C＝IT/C

where Q is the sign of the charge (in coulombs) and the delta sign means the "change amount". This equation reads as "voltage change equals charge change divided by capacitance C". This is immediately derived from the definition of capacitance, which is the amount of charge in the capacitor per applied voltage:

C＝Q/V

now assume by way of example that the pulses are required to have a flat top that does not deviate more than 0.5% from the mean value of the voltage of the pulses. In this example, the starting voltage is 1200 volts, so that the voltage at the end of the pulse should not be lower than 0.5% of this voltage, or 0.995 × 1200 ═ 1194 volts. This deviation value of V is 6 volts. The current for each module has been determined to be 1650 amps (1650A) and the pulse duration T is 5 mus, so that the capacitance is:

C＝Q/V＝ΔQ/ΔV＝IT/ΔV＝1650×5×10^-6/6

＝1375×10^-6＝1375μF

this is a very large capacitance. To reduce the need for such large capacitors, a pulse-flattening circuit may be selected that reduces the capacitance by about one-tenth. If such a circuit is used, the capacitance can be reduced to 137.5 μ F instead of 1375 μ F.

c. Module circuit

The number of switches (5) and the size of the module capacitors (1200 volts, 138 muf) have been determined. For cooling reasons, the capacitor can be divided into two or three separate containers in order to increase the surface area to volume ratio, since this improves the ability to remove waste heat. The final capacitor value may thus be 138/2-69 μ F or 138/3-46 μ F. The components are now assembled into a pulse generating module in the form shown in figure 6.

In fig. 6, a ground connection is shown attached to the negative terminal of the capacitor. This is a flexible option-the ground connection can also be arranged at the emitter terminal of the switch. This results in the trigger generation circuit being ground based, which eliminates the small pulse transformer T1 used to isolate the gate connection from ground in the circuit shown.

A reverse diode is also shown connected across the output of the module. The diode is reverse biased when the switch is conducting, but when the switch is off (causing non-conduction), the output current continues to flow due to the inductance of the pulse transformer. At which point the reverse diode becomes conductive and clamps the output voltage to near zero to protect the switch from damaging overvoltage conditions that would damage the switch if the reverse diode were not present. In the prior art, the timing difference between two adjacent pulse generating modules causes the reverse diode to turn on prematurely. The conducting diode is then suddenly switched off when the switch connected to the diode is switched on. This transient often destroys the diode, shorting it out. This in turn destroys the switch by allowing excessive current to flow into it.

The optional pulse flattening circuitry in fig. 6 is shown connected in series with the low voltage end of the pulse transformer secondary winding(s). The damping circuit is shown connected in series with the module output in fig. 6. As previously explained, prior art modulators require one such damping circuit per module, but the novel connection concept of the present invention allows modules to be turned on at different times, a feature that allows only one damping circuit to be used. The module connected to the single damping circuit is first turned on, and when the output pulse reaches a peak and the voltage has stabilized, the other modules are then turned on to carry the pulse load. In which case all modules are disconnected at the same time. This saves cost and reduces complexity.

d. Consideration of pulse transformer

Fig. 7 shows a typical high voltage pulse transformer used in a pulse modulator. The core is marked (15) and the high voltage output connection (shield ring) is marked (16). The geometry is proportional to the modulator output voltage and power. Typically, such transformers are immersed in tanks filled with mineral-based or silicon-based insulating oil to improve cooling and high voltage insulation.

The magnetic core shown in fig. 7 is a single core assembly. In a two-core transformer, two identical magnetic cores are arranged side by side and are both surrounded by a secondary winding, while each magnetic core carries a set of separate primary windings. This provides half turn primary performance as explained previously. Three or more cores may be similarly arranged to provide 1/3 or 1/4 turns of primary winding performance, respectively. For example, a SLAC (Stanford Linear Accelerator center) modulator uses many more cores than four cores and has few (one to three) turns on the secondary. The pulse rise and fall times guide these choices depending on what choice core number and number of secondary turns as will now be explained. The design example will be returned to aid in the description of this.

The gap between the two cores can be clearly seen in figure 8. Comparing fig. 7 and 8, it is seen that there is a volume of space between the secondary winding (the winding around all the cores) and the primary winding. When the transformer is in operation, there is almost no magnetic field in the transformer, according to ampere's law, except for the volume between the windings, including the core material itself. This magnetic field stores magnetic energy and it is equal to exhibiting induction in series with the load. This induction is in series rather than in parallel because the current does not reach the load unless it first flows through the windings and creates a stray magnetic field. This magnetic field represents a "tax" on the performance of the transformer that must be paid in order to obtain any output power. The inductance created by this magnetic field volume is referred to as the leakage inductance of the transformer.

As is well known, the coil inductance is proportional to the square of the number of coil turns. This leakage inductance is proportional to the square of the number of secondary turns, as measured from the secondary or output side of the transformer. Thus, any event that can reduce the number of secondary turns will significantly affect the leakage inductance. The smaller the inductance, the faster the output pulse can rise to its maximum power level. By arranging two cores side by side, each with its own set of primary coils, using a fractional turn primary will halve the number of secondary turns for a given input and output voltage. This will (almost) reduce the leakage inductance to one quarter of the original and will allow the pulse to rise almost four times faster than if a single core were used.

This argument is not exact because the leakage inductance is proportional to the volume contained between the primary and secondary components and the square of the secondary turns, and this volume increases when more cores are added to the transformer. For this reason, the reduction in leakage inductance when the core is changed from one to two is closer to 1/3 than 1/4. But this effect is still very useful in obtaining faster rise and fall times for the pulse transformer.

Again, looking at fig. 2 and 3, it is seen that in the case of two cores there are four vertical magnetic material "legs" that must be surrounded by the primary coil. This has been mentioned in the previous discussion, and the reason for doing so is evident in both figures. For a single core, there are two such vertical "legs". In summary, there are 2N "legs" for N cores that need to be surrounded by the primary coil. Basically, a multi-core transformer is equivalent to several transformers connected in parallel.

To accomplish this design example, the transformer needs to be sized. It is assumed that the magnetic material has a saturation magnetic flux density equal to Bmax tesla. Typical values for this flux density are around 1 tesla or 10,000 gauss with older magnetic unit systems. A set of coils is wound around each core to carry a DC current for resetting the cores. The coils are connected to the DC power supply through a series reset choke (reset choke) because the reset coils develop a pulsed voltage when the transformer is driven by the pulse generating module set and a high series impedance is required to shield the DC power supply from the pulse and to ensure that the reset current is not affected by the pulse. The reset choke keeps the reset current flowing unchanged regardless of the pulse action in the transformer.

The DC reset current induces a DC magnetic flux density equal to-Bmax in each core, the negative sign indicating that the magnetic flux is in the opposite direction to the magnetic flux induced when the transformer is pulsed. This allows these cores to be made using half the material needed without a reset, since the flux density can now swing from-Bmax to + Bmax during the pulse so that the total flux variation range is 2 Bmax. Each core has a cross-sectional area of AC square meters (see fig. 2). While Lenz's law states that if a single turn primary winding is used, the cross-sectional area of each core is calculated as follows for a given module voltage V and pulse duration T:

Ac＝V×T/2Bmax

substituting the values V1200V, T5 μ s, Bmax 1 tesla, results in Ac 0.003 m or 30 cm. So that the cross-sectional area of each of the two cores is about 5.5cm wide and 5.5cm deep, which is a typical transformer built. It is usually designed for a wider pulse width of 10 to 12us, so that the core is larger than the core, again according to the above.

It is then desirable to have a 1.2kV input and 120kV output so that the voltage gain is 100: 1, which requires 50 turns per cage of the two parallel connected secondary "cages" shown in figures 2 and 3. This reflects the fact that the primary is actually 1/2 turns. The voltage gain is equal to the turns ratio, which in this case is 50/0.5 — 100.

As mentioned before, there are four vertical core "legs" surrounded by primary windings, since in practice two transformers are made in parallel connection. Each of these "legs" requires five turns around it because five modules are used. Each module is then connected to a set of four single turn primary windings, one turn for each of the four "legs". The total number of primary turns required is thus 5 modules times 4 leg turns per module, or 20 turns of primary winding. This gives five turns per "leg", which is true.

It should be mentioned that in order to keep the leakage inductance low, a good "current plate" approximation needs to be used around each core "leg" to minimize the magnetic field between the wires. Five small diameter wires spaced along a core leg for 120kV, which may be 15cm in length, are not a good approximation of a uniform current plate, so five single turns can be made: for example, ten or even twenty individual wires are wound around the core legs, and these single turns are then divided into five bundles of two or three or four single turns each, each bundle being connected in parallel to equate to one wide single turn.

In this design example, an 800Hz PRF is desired, and each pulse delivers an output energy of Vsec × Isec × Tpulse of 1200kV × 70A × 5us of 42 joules per pulse. At a repetition rate of 800Hz this represents an average power of 800 × 42 ═ 33,600 watts or 33.6 KW. The duty ratio is the ratio of the pulse duration and the time between pulses, in this case 800 × 5 μ S — 0.004 or 0.4%. The RMS (root mean square) value of the current is equal to the peak current times the square root of the duty cycle, in this case 70 amps by 0.063, or 4.4 amps of total secondary current. The secondary is formed of two windings connected in parallel as in figures 7 and 8 so that each of these windings carries an RMS current of 2.2 amps. The wire cross-sectional dimensions are determined from a standard wire table carrying the current.

Similarly, the peak current is 1650 amps for each primary winding, so that the RMS current is 104 amps. Also, the primary conductor is sized from a standard table carrying the current, taking into account the heat transport/cooling means that can cool the conductor to a safe level.

This completes the connection of the modulator, except for sizing the reset winding and reset current. The dimensioning of the reset winding and the reset current relates to the magnetic permeability of the magnetic material, which is the ratio of the magnetic flux density B in tesla and the magnetic potential H in ampere-turns per meter and is a property that is tabulated in the catalogue of magnetic materials. When determining this value, the selection of the number of reset turns and reset current involves simply applying ampere's law and therefore does not need to be further explained. This is a well known technique.

The embodiments and design examples given above are illustrative only, and other embodiments within the scope and spirit of the invention will occur to those skilled in the art. The invention is therefore defined in the appended claims.

Claims (8)

1. A power modulator, comprising:

at least two pulse generating modules; and comprises

At least one secondary winding surrounding all of the plurality of magnetic cores;

wherein each of the at least two pulse generating modules comprises:

an energy storage capacitor;

a switching device which can be switched on or off in an electronically controlled manner; and

at least one diode connected in parallel with an output conductor of the pulse generating module or a series-connected diode assembly; and

wherein each pulse generating module further comprises a set of primary windings, each primary winding of said set of primary windings surrounding a corresponding one of said plurality of magnetic cores, and each said primary winding of said set of primary windings and at least one module of said pulse generating module are connected in parallel, thereby providing a power modulator in which each of said at least two pulse generating modules drives all of said plurality of magnetic cores.

2. A power modulator as in claim 1, wherein said at least one diode is part of a series-connected diode assembly.

3. A power modulator as in claim 1, wherein the set of primary windings is a set of single turn primary windings.

4. A power modulator as in claim 1, wherein the set of primary windings is a set of multi-turn primary windings.

5. A power modulator as in claim 1, wherein the number of pulse generating modules and the number of magnetic cores are the same.

6. A power modulator as in claim 1, wherein the number of pulse generating modules and the number of magnetic cores are different.

7. A power modulator as in claim 1, wherein each pulse generating module can be turned on or off at different times by manual or automatic means.

8. A power modulator as in claim 1, wherein the switching device is an IGBT solid state switch.