Originally published: 2010
A Marx Generator is a type of voltage multiplier circuit constructed using capacitors, resistors and spark gaps. They produce high voltage, high current DC pulses (non-continuous). They consist of multiple similar stacked stages where each stage effectively doubles its input voltage. The simple design and basic components allow one to that make this electrical circuit quickly and relatively cheaply.
The bipolar Marx generator I built in this project took a 15kV input and produced an output of approximately 220kV. Each discharge occurred with an excitingly load bang every couple of seconds! Unfortunately, before I could take some pretty pictures or a video of the device running I managed to get carried away and blew the device by raising the input voltage above 20kv which was too much for the maximum rating of the ceramic capacitors I had used! Looks like I will have to build a bigger and better version 2…
You might have noticed that I have mentioned the device I built as a “bipolar Marx generator”. The bipolar part indicates that I have built two Marx generators and connected them in an arrangement that supplies opposite polarity to their first stage inputs. This results in a positive and negative high voltage output relative to ground for each Marx generator, however, between the outputs of the two Marx generators there exists a voltage difference twice that of the magnitude of just one of the Marx generators alone.
Looking at just a couple of stages in single Marx generator, a DC input voltage charges the capacitors through the resistors until the voltage across the capacitor exceeds the breakdown voltage of the spark gap and they discharge resulting in a spark. It can be seen that the reason this functions as a voltage multiplier is that the capacitors charge in parallel through the resistors, but they discharge in series once the spark gap voltage break down is exceeded. This can be cascaded in multiple stages to keep doubling the voltage to a desired level. This is an ideal case where you could theoretically keep doubling indefinitely, however, this is both impractical and not possible due to losses and inefficiencies with the components used. In theory my bipolar Marx generator should be able to produce 300kV with a total of 20 stages and an input voltage of 15kV (20x15kv=300kV). However, it only produced sparks up to about 22cm which implies an output of approximately 220kV.
Another important and interesting point to be aware of is that the voltage across the resistors at each stage within the Marx generator is only equal to (or less than) that of the input of the first stage. Also remember that the capacitors discharge in series. This means that the capacitors and the resistors in each stage only need to have a maximum voltage rating equal to the input of the first stage (i.e. the components in each stage do not need to progressively be rated for higher and higher voltages as you add more stages). This is good news for us as it means we don’t have to buy expensive and rare ultra high voltage components! It is however, never sensible to load components at their absolute maximum ratings especially when dealing with high voltages and pulsed discharging of capacitors – a margin of around 50% should be incorporated (i.e. if your input voltage was 10kV, you should try to use components rated for at least 15kv). Otherwise you do what I did and end up killing your lovely new Marx generator!
Making the device
I decided that a nice round number of 10 stages in each leg of the bipolar Marx generator would be sufficient, giving me a maximum of a x20 multiplier. The bipolar Marx generator arrangement that I have used is shown in the circuit diagram below. Note that this diagram shows only five stages per leg to keep the diagram compact for easy viewing.
The next thing was to source 20 high voltage capacitors. My input voltage was supplied by a variable high voltage DC power supply that I built using a flyback transformer set to about 15kV. I found a cheap set of 20kV rated ceramic capacitors on eBay for about £30. If you try to source quality high voltage polypropylene/polystyrene capacitors with similar voltage ratings you will find that it will cost an awful lot more!
For the resistors you will also find that buying quality 20kV rated resistors will cost you a small fortune, that’s if you can even find any at all. There is a nice trick you can do and that is to buy multiple lower rated resistors (standard 1kV rated resistors, which can handle a fair bit more than their trusty data sheets tell you) and solder them in series. Each resistor will see a relative fraction of the total applied voltage across the bulk resistor string, e.g. 10kV across three resistors in series will subject each resistor to 1/3 of the 10kV. If you use power resistors 5W+, the carbon/metal-film tracks will be thicker and wider spaced with the insulation coating keeping a reasonably high voltage at bay. The distance across each resistor (from leg to leg) also needs to be considered – it needs to be long enough that it doesn’t arc over the whole resistor (a 1mm air gap per kV is a sufficient guideline to the breakdown voltage of air). The resistance value needs to be reasonably high, this is to reduce the bleed/leakage current during charging and discharging of the capacitors which will increase each stages voltage doubling efficiency. The resistance value can’t be too high as this will increase charging times between each full discharge. A value between 500kΩ and 2MΩ is a reasonable guide but not absolutely critical – I settled with two 5W 500kΩ resistors soldered in series for a total of ~1MΩ for each resistor in each stage.
I decided to mount the array of components onto strips of polypropylene sheet which I obtained from a couple of cheap cutting boards from the supermarket. For the spark gaps I wanted some nice looking wires/electrodes. For this I decided to get some 3mm dia. steel ball bearings and solder one to the end of a length of thick solid copper wire. Each wire was then bent the same distance from the ball. I then marked and drilled the necessary holes in the polypropylene strips to allow me to mount the components similar to through hole components with all the connections soldered together on the rear of the strip. Keeping the holes a tight fit ensured the spark gap electrodes and components remained rigid. The air gap between the electrodes was estimated to the equivalent breakdown voltage of 15kV which was about 15mm.
Below is the finished bipolar Marx generator mounted on a polypropylene base.