This note is based on the results of our previous studies, dedicated to the operation of bipolar transistors in the avalanche mode. This mode allows to significantly improve the parameters of high-voltage pulses - in particular, to reduce the front time while maintaining a significant amplitude. In this article, we will consider one of the practical options for circuit implementation, capable of generating pulses with an amplitude of up to several kilovolts at a slew rate of about 30 V/ns and higher - which is comparable to the voltage characteristics in the spark gap.

For comparison: such parameters are typical, as a rule, for expensive high-speed semiconductor switches. However, in the proposed circuit they are achieved using affordable bipolar transistors operating in the avalanche mode. Moreover, the amplitude of the generated pulses can significantly exceed the maximum breakdown voltage of most MOSFETs and IGBTs, which makes this approach especially attractive for creating short high-voltage pulses in a variety of technical applications.

Pulses with such characteristics are used in a wide range of areas - from pulse electronics and scientific experiments to non-standard areas of research. They are of particular interest to enthusiasts researching alternative energy sources, where such high-speed and high-voltage pulse effects are required.

The already standard transistor avalanche mode circuit was used to implement the device, but with minor modifications. The storage capacitor C2 is mandatory here, and the high-voltage pulse transformer THV is used as a load (Fig. 1a). For some transistors, mainly composite ones, the base resistor Rb is not needed, since it is already installed in the transistor itself (Fig. 1b). The operating principle of this rather simple circuit is described in detail here.

Almost any source capable of providing an adjustable DC voltage in the range from 70 to 350 V at the output can serve as a power source for the device (see Fig. 1 c, d). Current protection is not required in this circuit, since current is limited by resistors R1 and R2. The filter capacitor C1 should be placed as close to the circuit as possible to minimize interference and voltage drops.

We use several types of pulse transformers as a high-voltage transformer (THV). One of the key parameters of the circuit, the output pulse rise time, depends on it, as well as on the VT1 transistor. The first option is a pulse transformer wound on a ferrite ring, described in detail in this work.

This completes the preparation, and immediately after switching on the device, the first effects that cannot be easily explained begin to appear.

This mode occurs automatically at the moment of discharge of the avalanche on the primary winding of the THV transformer and its ferromagnetic core. The oscillations that arise in this case are clearly visible on the following oscillograms: their frequency lies in the range from 6 to 17 MHz - which is quite remarkable, considering that the maximum operating frequency of the ferrites used is usually only hundreds of kilohertz. For more information on the kacher mode of transistor operation, see here.

The first version of the device uses a BD235 transistor in combination with a ferrite core. The oscillograms obtained in this mode are shown in Figures 2 and 3. The high-voltage probe of the oscilloscope is connected to the secondary winding of the THV transformer. Figure 2 clearly shows the characteristic oscillations with a frequency of about 10 MHz - which is quite surprising, considering that the maximum operating frequency of the BD235 transistor is only 3 MHz, and the frequency limitations of the ferrite core have already been mentioned above.

With a ferrite core, the pulse rise time reaches 30 V/ns. For higher values, you need to make nsome changes, which we will discuss further.

One of the most remarkable properties of this circuit is the ability to change the pulse amplitude, while almost not changing its speed characteristics. By increasing the capacitance of C2, we increase the pulse amplitude, and the observed proportionality is approximately as follows: \[ U_{pulse} \sim (C_2)^{1/2} \tag{1}\] So, by increasing this capacitance from 470 pF to 7200 pF, we get the pulse shown in Figure 4. Note that its amplitude has increased from approximately 460 V to 1600 V, although the overall pulse width has also increased.

When the capacitance C2 changes, the frequency of the oscillations also changes, but to a lesser extent. The proportionality here is approximately as follows: \[ f_{k} \sim 1 / (C_2)^{1/4} \tag{2}\] which corresponds to the formula for resonance of the second kind.

It is interesting that this frequency also almost does not depend on the inductance of the THV transformer windings. That is, there is no need to talk about LC resonance here. Resonance at the wavelength of the wire is also not suitable for explaining this effect, since the length of the winding does not coincide at all with the duration of the pulse propagation along it, moreover, its frequency, although weakly, still depends on the capacitance C2.

The use of the transistor TIP110 increases the pulse amplitude by approximately two times, however, the supply voltage Up also needs to be increased by approximately the same amount. Oscillograms reflecting the operation of the circuit with this transistor are shown in Figures 5 and 6. Increasing the capacitance C2 here also leads to an increase in the pulse amplitude (Fig. 7). Note only that the base resistor Rb does not need to be used for this transistor, which is reflected in the circuit 1b.

Figure 6 additionally connects the blue beam of the oscilloscope to the primary winding of the THV. The avalanche strike on the core and the subsequent antiphase oscillations in the primary and secondary windings of this transformer are clearly visible.

The slew rate and pulse amplitude can be increased by using a more expensive and less common core made of amorphous material instead of ferrite. Next we will use the following core: nanocrystalline ring, 32*20*10 mm in size, and a composite transistor TIP120 (TIP122), which also does not require a resistor Rb.

Thanks to this core, the rate of increase of thepulse increases by 1.5 times compared to a ferrite core and is 45 V/ns or more. In addition, all other things being equal, using such a material for the core allows for a larger pulse amplitude with lower consumption of the entire circuit. This means that the efficiency of the circuit becomes higher.

Even better results are obtained from nanocrystalline with smaller dimensions of 21*16*10 mm. Under the same conditions, the pulse amplitude increases by about 30%, and the slew rate becomes 50 V/ns and higher. Based on the trend, we can say that the smaller the core size, the better the results will be. However, we must not forget that reducing its size will also entail a smaller diameter of the winding wires, which can affect the overall efficiency of the circuit. In addition, with a relatively small core window width, it will be increasingly difficult to comply with high-voltage standards. Apparently, it is necessary to look for a minimum, but still optimal core size for a specific task.

Oscillograms with a small core are shown in Figures 11, 12. As before, by increasing the capacitance of C2, the pulse amplitude can be increased even more (Fig. 13). The author did not check a further, quite logical increase in amplitude, since the high-voltage probe of the oscilloscope was designed for a maximum voltage of 2 kV.

As already noted, the ferrite core is described in detail in this note. Next we will move on to discussing nanocrystalline cores and THV transformers made on their basis.

The winding pattern is approximately the same everywhere: the primary winding consists of 2 turns and about 16-20 turns - the secondary winding of the THV transformer. The wire for the primary winding should be thicker and necessarily - in double insulation, for galvanic isolation of the primary and secondary circuits. You can take almost any wire for the secondary winding, in varnish or polyethylene insulation. The windings are wound using the «bulk» method (Fig. 14, 15).

It is necessary to note an interesting property of the current flowing through these windings. Even despite the high voltage, even weak insulation does not break down. Apparently, this is due to the displacement current, abundantly present in such pulses, and slightly different laws of its interaction with the conductor and insulation.

It is best to select transistor VT1 according to maximum output characteristics by installing it in the circuit. The following brands work well in it: BD235, BD237, TIP110, TIP120, TIP41C, BD911.

Resistors R1 and R2 should be taken for a power of at least 0.5 W, more is better. Capacitors C1 and C2 should be rated for a voltage of at least 500 V. It is highly undesirable to use ceramic capacitors in this circuit, all the others will do.

The circuit can be powered from a VVT, observing all the necessary precautions when working with it (Fig. 1c). But the best solution would be to use an adjustable DC voltage source rated for a range of at least 70-350 V (Fig. 1d).

The method is very simple. After installing the transistor in the circuit, it is necessary to gradually increase the voltage of the power source until stable pulses appear on the secondary winding of the transformer. This will be the optimal operating mode of this circuit. If it is necessary to further increase the amplitude of the pulse voltage, increase the capacitance of the capacitor C2 to the desired value. This additional capacitance can be approximately determined using formula 1.

If even higher pulse rise and fall rates are required, a more complex converter circuit can be used, discussed here.