Interesting topics about electronics, product development and more.

High-power, high-voltage pulse power: The basics

by Timo Stehmann4. June 2013 20:01

A short history

In some applications pulse power supplies need to handle very high currents (>10kA) and voltages (>30kV). For many decades the only devices that could come close to the required specifications were vacuum tubes. These vacuum tubes are highly specialised and were graced with their own name: Thyratrons (see image below). These are amazing devices, capable of handling very high pulse currents at mind-blowing speeds (<100ns). Since no silicon is involved, these devices are very robust but they have a limited life-time. In the early days the only alternative to thyratrons were spark gaps. Spark gaps, even today, are very fast (<50ns) and are able to handle very high voltages. Unfortunately, the life-time of spark gaps are even more limited than thyratrons. However, if you want to generate very fast pulses the quick and dirty way, then spark gaps are the way to go. In our modern commercial world thyratrons have been become out-dated for many reasons. There is the limited life-time for one and then there are the commercial restrictions. Thyratrons were/are used for nuclear weapons as well and using them in commercial devices can be troublesome. What are the alternatives?



All solid-state pulser (ASSP)

Over the years there have been significant advances in solid-state switches. Most notably the MOSFET and the IGBT. The first solid-state switches made use of thyristors. Thyristors are also very robust and can handle very high pulse currents. Turning them off is the main problem for pulse power supplies that need to generate more than 1,000 pulses per second. As a result pulse power supplies that use thyristors tend to have complicated driver circuits. Driver circuits are used to turn switches (e.g. thyratrons, thyristors, etc.) on and off. In order to provide a reliable commercial solution for high-power, high-voltage pulse power applications the emphasis has shifted to more modern solid-state devices. Most notably the IGBT. The IGBT has proven as a better candidate compared to the MOSFET. The MOSFET in a saturated on-state is similar to a resistor and this can cause a lot of dissipative losses. The IGBT, however, has a more or less constant on-state voltage which has lower losses in pulse applications compared to the MOSFET. But there is a problem: IGBTs have limited voltage ratings and tend to be slower than MOSFETs. What can be done if the IGBT is too slow and cannot handle high voltages?


High-power IGBT module

Pulse transformers

Unfortunately, for high-voltage applications (>5kV) a pulse transformer is required. This can be a challenging problem and there are many patented designs. The transformer needs to be fast and must be able to handle very high voltage differences. In engineering terms this translates into: very low leakage inductance and electric insulation. These two factors work against each other and better electric insulation in most cases increases the leakage inductance of a transformer (i.e. makes it slower). However, there are some tricks and in some designs both can be achieved: fast and high-voltage.


Magnetic pulse compression

The second problem is speed. IGBTs in general cannot generate pulses faster than 1 to 5 micro-seconds. Anybody who paid attention during lectures during university will remember that higher frequency content (i.e. faster pulses) can only be generated by non-linear devices. In this case an interesting technique called magnetic pulse compression (MPC) is used. MPC makes use of the saturation of magnetic materials. This is a highly non-linear process and inductors are used as switches.

Have a look at the circuit below. This is a 2-stage MPC. The capacitor C1 is charged and C2 and C3 are discharged. As soon as the switch is closed the voltage over C1 is "seen" over the inductor L1. At this moment, however, L1 is not conducting. Remember, L1 and L2 are inductors using magnetic materials which will saturate as soon as the current through them become larger than a certain threshold current. For very small currents both L1 and L2 will not be saturated and the magnetic material will have a very high permeability, i.e. the inductance will be high. After the switch is closed a small current will start to flow through L1 into C2. This current will start to increase until the inductor L1 is saturated. At this stage the inductance of L1 will suddenly drop to a very small value and suddenly the current through L1 can increase quickly. A resonant transfer will occur from C1 to C2. If C1 is equal to C2 (ignoring losses) all the charge in C1 will be transferred to C2. At the end of the transfer C1 should have zero voltage and C2 will have the same voltage as the initial charging voltage of C1.

Now the magic happens. It is possible design L2 to saturate as soon as C2 is fully charged. This will result in a second transfer from C2 to C3. If a good design is used, the second transfer will be between 4 to 10 times faster than the first transfer. There you have it: a faster pulse than you started with!


It is possible to build your own magnetic pulse compressor using ferrites. In commercial pulse power supplies (pulsers) more exotic magnetic materials are used. These materials, when saturated, have a permeability close to vacuum (or air for that matter). In addition they have very low losses and saturate at higher magnetic fields. These factors all help to increase efficiency and improve the compression factor (i.e. the amount by which pulses are made shorter). In a next post I will describe how to build a simple pulse compressor.


Power electronics | Power supplies | Pulse power

Permalink | Comments (0)

Pingbacks and trackbacks (1)+

Add comment

  Country flag

  • Comment
  • Preview