I would like to know how I can increase the maximum voltage of inductive spikes.

I’m investigating their use in electrolysis - the bigger the spike the better. To create the spikes, I’m using an IRF840 MOSFET and a 40mH coil as in the circuit shown below.

Pic 1

As I understand it, the peak voltage of the spike will depend on the coil inductance and also di/dt which will I assume will be a function of how fast the FET shuts off.

My coil is fixed with an inductance of 40mH but I have six of them and yet, if I string them together to make an inductance of 240mH, the peak voltage remains roughly the same, maybe because the total resistance of the coils is now about 10ohms compared to 1.5ohms for a single coil.

So the other option would seem to be to encourage the FET to shut off faster and for that, based on another post in this forum, I thought one can use a ‘pull-down’ transistor as in the following diagram.

Pic 2

Are there any other circuit modifications I could make to increase di/dt, including perhaps a different FET for the job? At the moment the peak voltage is a bit over 220V and I'm hoping it can be made over twice that.

  • \$\begingroup\$ What are the characteristics of "signal in"? \$\endgroup\$
    – Andy aka
    Nov 3, 2020 at 9:15
  • \$\begingroup\$ @julianperry: Do you have your output connected to something? The load will have an effect of the voltage of the spike, as well as on the width of the spike. \$\endgroup\$
    – JRE
    Nov 3, 2020 at 11:09
  • \$\begingroup\$ When you string the inductors together, do you increase the duty cycle of the input signal? More inductance means slower rise of current. If you don't increase the duty cycle along with that, peak current will be lower. \$\endgroup\$ Nov 3, 2020 at 11:55
  • \$\begingroup\$ 500V is easy, even 30kV is easy with a TV flyback transformer, but what impedance and rise time do you need and resonant frequency. What kind of electrolysis? otherwise your question is useless. \$\endgroup\$ Nov 3, 2020 at 12:37
  • \$\begingroup\$ The signal input is a CMOS square wave of 10V amplitude. Ther output is connected to the small electrolysis test cell which consists of two stainless steel plates of area each 50cm2 set 1mm apart in distilled water. This is to generate OxyHydrogen. I don't know much about TV flyback transformers and I have yet calculated the resonant frequency partly due to not having a reliable value for the capacitance of the cell. \$\endgroup\$ Nov 3, 2020 at 13:01

2 Answers 2


In fact the voltage at the drain of the MOSFET will begin to "ring" at the resonant frequency of the capacitance and the inductance. The capacitance comes from the MOSFET drain capacitance in parallel with the distributed capacitance of the coil.

The peak will be higher if the capacitance is lower, the on time is very brief and the resulting current peak very high.

In practice, a boost of more than 10:1 or 20:1 is difficult to achieve (and unsupported by most controller chips- the 'on' time becomes very brief), so a transformer (or tapped inductor) makes more sense than trying to use your circuit.

enter image description here

Or, if you want to avoid using that kind of inductive component, you can use two stages to easily get higher ratios.

  • \$\begingroup\$ Thanks, I need to digest all this slowly. \$\endgroup\$ Nov 3, 2020 at 13:07
  • 1
    \$\begingroup\$ You probably know this, but for production of Brown’s gas, the most efficient voltage is only a few volts with an electrolyte (eg. a caustic) added to the water to increase conductivity. 1A at 100V produces the same amount of gas as 1A at 4V, but far more (25x) wasted heat. That’s what my micro hydrogen-oxygen torch uses to create a 1mm or so flame diameter at 2500+ degrees C. \$\endgroup\$ Nov 3, 2020 at 13:18
  • \$\begingroup\$ True but there’s another way to dissociate water which is to treat the water between the plates as a dialectic and to invoke a high voltage between the plates of the order of 5-10kV. This also ionises the water but involves little current flow so keeps the power down. This is the basis of what Meyer did with his now expired patents. But I imagine today there are more efficient and easier ways to achieve that and maybe a fly back transformer could do that? \$\endgroup\$ Nov 3, 2020 at 13:37
  • 1
    \$\begingroup\$ The talk about Zero Point Energy ZPE, is total BS to me. But PD is one of my expertises. It's a nano-sized nuclear energy that causes old grid transformers to explode every year. \$\endgroup\$ Nov 3, 2020 at 14:24
  • 1
    \$\begingroup\$ Sphero you are referring to standard electrolysis but I was referring to the capacitive discharge system that quite a few have been exploring for years. The resonance here is that of a choke and the capacitance of the electrolysis cell itself (also not of the HOH bonds). It seems that interesting things happen at resonance. Anyway, if I come up with some circuit ideas to test this (lock down looming again!) I will post them here for comment/ideas. \$\endgroup\$ Nov 3, 2020 at 16:00

"Bigger" depends on how to define this.

Voltage peak, V x t ,[volt-seconds], or power = V²/R = V²(f,t) / (real{Z(f,t)})

There are a few basic rules to electronics.

  • All insulators = dielectrics
  • this includes the interwinding capacitance of large inductors and the dilectric medium (Water has a Dk=80, Air =1) so bigger chokes is not always better.

Here's a simple experiment.

Use a rasp file as an intermittent switch with a nail slide and a decent air coil with a spark plug and close the electrical circuit with the series loop current using a tap off the larger coil which is open to the spark plug as a step up transformer. 1:10 to 1:100 using a good low ESR battery. Avoid shocks and battery heat. The low C of Air and low L of an air coil will give a very high f resonance yet a very fast rise time and voltage . 2~3kV/mm for a 3mm spark plug = 6~9kV

  • i.e. they all have capacitance and a relative value of Dk compared to air which is close to the vacuum reference. But with a low impedance 12V LED String with reverse diode protection, you can power it from a D Cell temporarily this way. Just remember polarity , you make the low side intermittent to generate a positive spike when current is released from L.

Vp = LdI/dt and dV/dt= I/C which combine give the resonant frequency and impedance ( look up)

  • meaning they have a Capacitance per unit area of the electrodes volume which increases with reduced gap, as does the breakdown voltage in kV/mm. Hydrogen also has a lower explosive limit of 4% with some trigger voltage.

  • the lower the FET RdsOn the larger the Coss during transition , although this has a geometric and packaging variation as well as Voltage rating so examining 50k different varieties of FETs with a spreadsheet from D-K can find the best for T=Ron x Coss time constant not given in datasheets.

  • There are many basic ways to chemically separate atoms in molecules such as electrolysis and it can can be done with DC, AC and HF pulses.

I am by no means an expert on Electrolyser design expert but you can learn what has worked for certain parameters of size and efficiency then match or try to improve it with electrode materials and methods.

With a free registration, you can read how H2 Electrolyzers work. https://www.researchgate.net/publication/341293476_Proton_Exchange_Membrane_Electrolyzer_Modeling_for_Power_Electronics_Control_A_Short_Review#fullTextFileContent

  • unless you are trying something different.

You can use a MOSFET Buck regulator or DC enter image description here

enter image description here

There is a more than a quarter century of books and research papers on hydrogen cells yet this has not stopped Engineers and Chemists to keep improving the efficacy and safety of this next generation fuel cell.

If it was this simple, it would have been perfected long ago.

But if your goal is simpler, it's not that hard, just refine your specs. (not given) and Re-google ehhm, I mean research. ;}

  • \$\begingroup\$ Thanks and lots here for me to explore over time. Yes, I'm exploring OxyHydrogen production efficiency using different electrode formats e.g. conductive plastics and catalysts and also different ways of supplying the energy to split the water. \$\endgroup\$ Nov 3, 2020 at 13:10
  • \$\begingroup\$ Do some solid re-search for methods and observe LEL on H2 when mixed with a molecule of O2 it is easily triggered. and probability increases then stops above HEL. \$\endgroup\$ Nov 3, 2020 at 14:10
  • \$\begingroup\$ I’ve had a read of your material. I’m not working with a PEMEL system but the capacitive discharge system that uses pure potential to split the water with minimal current. This involves treating the water filled cell as a capacitor and achieving a resonance with that and a choke and building up the voltage across the cell till breakdown occurs. Apparently it’s not easy to achieve in a stable way but I’m exploring what might be involved. If I come up with some circuit ideas I will post here for comment and add some more details re the concept. \$\endgroup\$ Nov 3, 2020 at 16:21
  • \$\begingroup\$ Ok, I have revised my approach somewhat in that instead of using back EMF pulses I have put together a circuit to generate HV pulses using the suggested from this forum of a flyback transformer which seems a suitable device for the task. As I don't seem to be able to add images to a comment and, rather than edit my original post ,I will start a new question. \$\endgroup\$ Nov 20, 2020 at 11:27
  • \$\begingroup\$ The equations for PEMEL are relevant even though it is not your application. Water has NO resonant frequency. It is an impure dielectric. "Pure water is poison" if that was all you drank. It must have mineral contaminants. If you are trying to resonate these minerals and detonate them, this is a different method and model with Partial Discharge (PD) being a stoichastic property for breakdown voltage. But pure water is a polar pure dielectric insulator over a wide bandwidth. So you must define your impurities precisely in ppb or ppm for each type to attempt what you are doing using HVac or dc \$\endgroup\$ Nov 21, 2020 at 21:50

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.