In a previous paper, a high-voltage nanosecond pulse generator circuit was considered, built using eight series-connected avalanche transistor blocks. This circuit design ensures high stability of the output pulse parameters, as well as a relatively high efficiency. This article discusses an alternative approach to constructing a nanosecond pulse generator, characterized by a significantly simpler circuit implementation and readily available components. The proposed solution is characterized by good parameter repeatability and ease of manufacture. Its limitations include slightly lower efficiency compared to avalanche-transistor circuits, as well as the need for selection and tuning of magnetic cores.

High-voltage nanosecond pulses are widely used in modern technology and experimental physics. They are used to trigger discharge devices and high-speed switches, pump semiconductor lasers, in radar and pulse diagnostic systems, to study free energy systems, and in high-temporal-resolution measurement technology. Short pulse durations with high amplitudes allow for high peak power with relatively low average power, which is critical for a number of applications, including probing and precision measurement systems.

The circuit design of the pulse generator under consideration is based on the work of Belkin and Shulzhenko [1], which proposed effective methods for generating high-power nanosecond pulses using semiconductor components. This implementation utilizes modern components and introduces a number of circuit design improvements. This enables the generation of high-voltage pulses with nanosecond rise times using readily available components with relatively simple device setup.

The developed pulse generator supports input signals of arbitrary shape and duty cycle with amplitudes in the 5–15 V range. The input signal parameters have virtually no effect on the output pulse shape only the repetition rate is retained. Thus, the device functions as a pulse normalizer. The output pulses are 1–3 kV in amplitude (depending on the load resistance), with a duration of approximately 5–6 ns, and a rise and fall time of 2.5–3 ns. The corresponding voltage change rate reaches values ​​of approximately 400–450 V/ns and higher, making the device suitable for a wide range of high-speed and high-voltage applications.

This paper considers two implementation options for the pulse generator. The first option features a simplified output stage and higher efficiency. Its schematic diagram is shown in Figure 1.

The circuit is divided into two functional blocks, highlighted by dashed-dotted frames. The left block is a low-voltage control pulse generator. Its purpose is to generate signals with a fixed duration to control switch Q1, which is part of the second (right) block of the circuit. For the switch to operate correctly, it is critical that the duration of the control pulse be strictly fixed and independent of the input signal parameters.

The left block in the diagram is powered by 12 volts supplied to the XDC connector, and is implemented using two drivers integrated into a single IXDN602 microcircuit package. A protective circuit is provided at the input of the first channel (pin 2 of the microcircuit), including resistor R1, diode D2, and suppressor D1. This allows signals with a wide amplitude range (approximately 5–15 V) to be fed to the XIN input without risk of damaging the microcircuit. The presence of a small hysteresis in the input stage contributes to the generation of pulses with steeper edges.

The signal from the output of the first amplifier is fed to the input of the second amplifier through the RC-forming circuit C1–R2–R3. This circuit limits the pulse duration to a strictly specified value, controlled by resistor R3. Additionally, positive feedback is used through capacitor C2, which accelerates the switching process and generates steeper pulse edges. This technique was previously used in a similar circuit for generating short signals.

As a result, a signal is generated at the output of the second driver (pin 5 of the microcircuit).A control pulse is generated for the power stage. An important feature of this unit is the invariance of the output pulse duration to the shape and duration of the input signal: regardless of the input parameters, a pulse of a strictly specified duration with fast edges is generated at the output. The external generator determines only the repetition rate, which can vary over a wide range from a few hertz to approximately 1 MHz. Thanks to these properties, this generator can be used as a universal unit in other high-speed pulse devices.

The right block of the diagram includes the power switch Q1 with the protection circuit (R4, D3, D4), as well as the pulse generator circuit consisting of the saturable inductor (transformer) TI1, capacitor C6, and diffuse diode DHV. When switch Q1 is opened, a voltage of approximately 100 V is applied to the inductor, causing it to become magnetized to saturation. The time to reach saturation is determined by the magnetic circuit parameters and the applied voltage, and can be estimated using known relationships for saturable inductors.

After the switch is closed, the main stage of pulse generation begins. Since the current through the inductor cannot change suddenly, it continues to flow through the DHV diode, gradually decreasing. Simultaneously, the voltage across the diode increases, and upon reaching a critical level, it quickly turns off. This process is accompanied by a sharp voltage surge, which forms the front of the output pulse. The diode's diffuse capacitance is discharged through resistor R5 and a load connected to the XOU output.

An important element is the power supply to the right block, supplied through the XPW connector. To achieve maximum pulse amplitude and improve the efficiency of the entire device, it is necessary to properly bypass the Q1-TI1 circuit at high frequency. This is accomplished by capacitors C4 and C5, which should be located as close to this circuit as possible.

Thus, the pulse generation process can be divided into three sequential stages. In the first stage, a control pulse with specified parameters is generated. In the second stage, energy is accumulated in the magnetic element by magnetizing it to saturation. In the third stage, when the diffuse diode is rapidly turned off, an output nanosecond pulse is generated. The characteristic time intervals for these stages are 100 ns, 10 ns, and 5 ns, respectively.

Figure 2 shows a second implementation option for this circuit design, distinguished by the use of a transformer in the output stage instead of a choke (element TI1). This solution ensures galvanic isolation of the load and also helps improve the stability of the generated high-voltage pulse parameters. The disadvantages of this option include a more complex magnetic element design and a slight reduction in efficiency due to additional losses in the transformer. Otherwise, the operating principle of the device is completely consistent with the previously discussed option and does not undergo significant changes.

Figures 3–5 show output pulse waveforms for different loads: 1 kOhm (Figs. 3, 4) and 75 Ohm (Fig. 5). It can be seen thatThe pulse character depends significantly on the load resistance, which is due to changes in discharge conditions and output stage matching.

It should be noted that the given pulse timing parameters (duration, as well as rise and fall times) are limited by the capabilities of the measuring path used. Measurements were performed using a RIGOL DS1202 oscilloscope in combination with a high-voltage probe with a division ratio of 1:100, designed for voltages up to 2 kV. The oscilloscope's rated rise time is approximately 1.7 ns, and the probe used has a comparable value. Thus, the total rise time of the measuring path is on the order of a few nanoseconds, which limits the system's bandwidth and prevents the accurate reproduction of faster transient processes.

Consequently, the actual parameters of the generated pulse may be better than those observed on the oscillograms. In particular, the pulse duration and edge steepness could potentially be significantly shorter, down to a few nanoseconds. However, the limited bandwidth of the measuring path does not allow for a reliable recording of these values, and the oscillograms provided should be considered estimates.

One of the important components of the circuit is the power switch Q1. This position requires a high-voltage transistor capable of handling high pulse currents and offering high response speed. In practice, both MOSFETs and IGBTs can be used, but they have strict requirements for minimum output capacitance and short switching times. The following transistor types are most suitable (in descending order of preference): K40H1203, G30N60A4D, IHW20N120R2, P10NK60ZFP. The choice of a specific type has a significant impact on the device's efficiency and the shape of the generated pulse.

Note: If the pulse repetition rate is up to 5 kHz, the transistor does not require a heatsink. At higher frequencies, the transistor will require a heatsink with an area roughly proportional to this frequency.

An equally important element is the DHV diode, which largely determines the output pulse parameters. For proper operation of the circuit, a diode with a large stored charge and a developed PN structure is required, ensuring a rapid recovery of the blocking properties. Experience shows that the best results are achieved using Soviet-made diffusion diodes: Д242А, Д245, 2Д243, 2Д201В, КД203А, Д112-25Х-16. For the first circuit variant, the Д242А [2] diode is optimal, providing maximum amplitude and pulse rise/fall speed. The other types demonstrate slightly worse, but acceptable, characteristics. In the second circuit variant, all of the listed diodes can be effectively used (in descending order of preference).

In theory, there is an analog of the Д242А the 1N1621 diode but the author has not tested it.

10A10 rectifier diodes can be used as a modern replacement; however, to achieve comparable parameters, it is recommended to connect them in parallel (usually at least two). However, such diodes show satisfactory results primarily in the second circuit variant. The SF58 diode can also be used for this option, but its characteristics are significantly inferior to the above solutions. In some cases, individual selection of samples from a batch is required due to the significant variation in parameters critical for generating nanosecond pulses.

The key element of the circuit, which largely determines its timing and energy characteristics, is the magnetic element TI1 a saturable inductor or transformer, depending on the implementation. For the first version of the circuit, a ferromagnetic ring inductor with the smallest possible cross-section and high-speed material properties is used. This choice is driven by the requirements for saturation time: the smaller the effective cross-section of the magnetic core, the faster saturation is achieved, all other things being equal. Additionally, the applied voltage has a significant impact increasing it leads to a proportional acceleration of the magnetization process. nanocrystalline toroid measuring 4.3×7.2×3 mm (size 0703), possessing high magnetic permeability and a short magnetization reversal time. The inductor winding contains 4–5 turns and should be made of wire with minimal high-frequency losses preferably Litz wire. The wire cross-section is selected so that the turns fill the inner volume of the ring as tightly as possible, minimizing parasitic inductance and resistance.

In the second circuit variant, TI1 is implemented as a transformer. In this case, a two-winding design is used with 3 turns in the primary and 5 in the secondary. The key requirements remain the same: minimizing saturation time and parasitic parameters. The primary winding, through which the fastest and most powerful pulse currents flow, is recommended to be made of Litz wire, while the secondary can be made of regular wire. A nanocrystalline toroid measuring 10x14x4.5 mm (size 1405) is used as the magnetic core, providing the necessary compromise between saturation speed, energy transfer rate, and mechanical feasibility of the design.

Capacitor C6, which operates under significant pulse currents and high reactive power, deserves special attention. Essentially, a low-loss pulse capacitor (with a low dissipation factor) and a high permissible RMS current is required. Film capacitors (such as polypropylene) are preferred, as they offer stable parameters over a wide range of frequencies and temperatures. Ceramic capacitors can be used as a compromise solution; however, it should be noted that they can become noticeably hot due to dielectric losses, which reduces the overall efficiency of the device and degrades its stability. The capacitor's rated working voltage must be at least 3 kV, taking into account possible surge voltages and the circuit's operating conditions.

List of other circuit components:

• U1 - two amplifiers in one housing IXDN602;
• D1 - suppressor P6KE9.1A;
• D2, D3 - diodes UF4007;
• D4 - suppressor P6KE15CA;
• C1-C5 - film capacitors: C1-C3 for 50V, C4-C5 for 680V;
• R1-R5 - any resistors with a power rating of 0.125 watts or more.

Pay attention to the placement of capacitors C4 and C5 on the printed circuit board. They should be placed as close as possible to the Q1-TI1 circuit, ensuring minimal parasitic inductance in the connections. In effect, these capacitors act as a high-frequency shunt, closing the power circuit on the AC component of the current. The efficiency of energy transfer to the load, as well as the shape and steepness of the output pulse edges, directly depend on their correct placement.

The device's power supply deserves special mention. The low-voltage control pulse generator is powered by a standard 12 V DC source. The power section of the circuit requires a separate source with an output voltage of 100–140 V. Experience shows that further increasing this voltage does not result in a noticeable increase in the output pulse amplitude, but it does result in increased switch heating and a decrease in the overall efficiency of the device. The DC-AC Converter Booster Module 12V to 110V 200V 220V 280V 150W module can be used as a power source for the power section. On the board, install the first jumper corresponding to the 110V output voltage. In this case, the entire device can be powered from a single 12V adapter, and the power source additionally provides galvanic isolation of the output stage from the primary power supply.

The setup procedure is the same for both circuit variants. Initially, variable resistor R3 should be set to its minimum resistance. The oscilloscope is connected in parallel with load resistor R5. An external master oscillator is then connected to the input of the driver, and power is applied to the circuit.