High Speed Air-gap Flash

A bullet hitting a solid brass rod.

If you just want to buy an assembled and tested airgap flash, I have made an updated versions that can be purchased at Information Unlimited.

In my quest to capture amazing high speed photographs I notice that when photographing shooting bullets the bullets were blurred. I found that standard xenon tube, which standard flashes use, is very bright for the energy put into it because of glowing xenon gas. The book Electronic Flash Strobe by Harold Edgerton explains all the calculations, but in practice this means all the flashes from Nikon, Canon and others that use xenon flash tubes have a minimum duration of 1/40,000th of a second. That’s fast enough for most things, but not for a shooting bullet travels around 1000 feet/second. In 1/40,000th of a second that bullet can travel about 1/3rd of an inch leading to blurry photographs of bullets.

To solve this I had to make a faster flash. I’m certainly not the first to do this. I think that was Harold Edgerton. He actually created a company called EG&G to sell a product called the 549 Microflash, but that company has been dissolved and the product discontinued. Sometimes you can still find these flash units on ebay, but the ones I saw were selling for $8K+. There is also a company called Prism Science Works making a modern version of these for researchers, but you’ll need really deep pockets to afford one of those. I saw directions on how to build one in the August 1974 issue of Scientific America and emailed Alan who had already built a few. After this research I realized I could build a sub-microsecond flash for just a few hundred dollars. A sub-microsecond flash means the flash duration is less than 1/1,000,000th of a second or about 25 times faster than a xenon flash.

Here is an image showing how an air-gap flash compares to a standard flash when photographing a 1000 feet/sec pellet.

I am a strong believer in sharing knowledge so I’ll explain how I made my air-gap flash, but I am knowledgeable with high voltage safety procedures. You should not build this because this flash will kill you. It really will. This flash requires charging a 35,000 volt capacitor that will easily kill a person for a single mistake. I have a safety checklist that I use ever time I plug in this flash and it still scares me. If it didn’t scare me then I shouldn’t be using it because this thing is dangerous. The below information is for educational purposes only. Do not build one! If you go against my advice and do build one, I am not responsible for any injury, death, or any other problems it causes.

Here is a list of the main parts I used to build my air-gap flash:

Here is an overall diagram of how all the parts fit together.

This is the schematic for the custom circuit board I made. Basically it doubles the 120V AC and uses an external signal to trigger the 45 kV transformer.

And here is a picture of my custom circuit board.

Here is a picture of the air-gap flash.

This shows what is plugged into the air-gap flash. 120V to run the 45K trigger transformer, 14V to charge the big capacitor, and a 3.5mm cable to use for triggering.

Here is a side view. Notice that the reflector has a thin sheet of Plexiglas over it to make cleanup easier. The metal reflector is also grounded in case the plexiglass tubes shatter, and they do shatter sometimes. I have found they last longer if I use latex gloves so no oils from my hands get on the glass tubes.

This is a close up of the glass tubes where the spark happens. The outer glass tube is just for protection and to reduce noise. The inner tube has the wire 45 kV trigger transformer inside. When that transformer is triggered the air ionizes and that cases the big capacitor to discharge in a 1.5 inch spark that creates the flash of light. To seal the inner tube I melt one end with a torch until it is sealed.

A top down view that shows how everything is connected.

This rod is very important. It is how I discharge the big capacitor after every use so I don’t die when I’m working on the flash.

Those extra wires and circles are there so I had an easy way to discharge the capacitor.

Note about using it with the Camera Axe. The Camera Axe only allows a trigger voltage of up to 20V. This has a trigger voltage of 240V so I needed to to put a high voltage protection circuit between this and the Camera Axe. There are many examples of these circuits on the web.

Here are a few results:

Glass rods being hit by a bullet.

A balloon being hit by a bullet.

Thanks Alan for your help and letting me use some of your amazing photos.


  1. Paul said,

    May 19, 2013 @ 1:45 am

    Hey Maurice,

    Brilliant work mate. So your 35kv cap is only charged to 25kv? I have a 50kv 0.01uf cap, do you think this will function in your design with 1/3rd capacitance? what if I extend the V multiplier
    to provide 50kv @ 0.01uf? What differences should I expect in light output?

  2. Maurice Ribble said,

    May 19, 2013 @ 6:07 am

    Correct, the 35kv cap is only charged to 25kv. If you 50kv cap is rated for pulse discharges (most caps are not) it will work fine at 25 kv and generate the same light.

  3. Heron said,

    May 30, 2013 @ 12:58 pm

    Hey Maurice,

    Very good work. I want ask you about the marx generator, Bart ask you about that, and you said that wouldn’t work is there is not enough energy to generate enough light. A discharge time of capacitors is proportional to capacitance, means that we need a very low capacitance to have a very fast discharge, but, reducing a capacitance we also reduce discharge energy, and the light. The discharge energy is calculated E= (C.V²)/2, so, to compensate a low capacitance and reach a good energy level to have enough light, we need to increase the voltage. Because this, high speed flashes uses a high voltage discharge. Marx generator charge the capacitors in parallel and discharge in series, if we use a marx generator with 10 capacitors 10KV 0,04uF, a discharge will be equivalent to a capacitor 100KV 0,004uF, resulting in almost 8 times faster discharge and 10% more energy, comparing with your capacitor 35KV 0,03uF, also, we can use a lower voltage supply and cheap capacitors. Seems to me that a marx generator is the best option. I really not an expert and maybe I misunderstood something, can you help me? I used this online calculator (http://www.digikey.com/Web%20Export/Supplier%20Content/tt-electronics-welwyn-985/docs/tt-electronics-capacitor-discharge-calculator.xls?redirected=1). Thank you.

  4. Gavin Melville said,

    October 5, 2013 @ 3:05 pm

    I’ve got most of the bits together for an air gap flash, and I had had trouble finding an inner tube. Currently I’m trying the tube off a 1500 watt 8 in long halogen light bulb, about 5/16 in diameter, quartz and I’ve cut the ends off. This is the big version of the work light bulbs so common in hardware stores. I’m having trouble getting it hot enough to close the end off. From poking around on the net you need a hydrogen flame to melt quartz – has anyone had any success closing quartz?

    The other issue I’ve got is that I’m testing at about 1 joule, 35 kv and the arc is not hugging the tube, it seems to be slightly above the surface. Is the solution to make the discharge spiral slightly ie try to go through the tube?


  5. Maurice Ribble said,

    October 5, 2013 @ 4:05 pm

    I’ve found these tubes work better as an inner tube than what I used here. When you melt the ends like I recommended here the inner tube tends to shatter after some use. These tubes work much better. http://www.lcmlab.com/13x100mm_Pyrex_Glass_Test_Tubes_p/206-0002w.htm

  6. Gavin Melville said,

    October 6, 2013 @ 12:24 am

    Thanks Maurice,

    I’d been thinking of Pyrex test tubes, but wondered if I was running out of room inside the bigger tube. I’ll try this first. Any thoughts on making the arc hug the tube?


  7. Maurice Ribble said,

    October 6, 2013 @ 4:57 am

    I have not had any issues with having the arc hug the tube. One thing that seems different is you are running at 35kv, but I only run at 20kv.

RSS feed for comments on this post