Pulse Power
Tesla coils
High Voltage
Pulse Power
Rail gun
Magnetic Levitation
Nuclear, X-ray
Public Displays
About Me



















































STOP PRESS  This is my old site last updated June 2005.  Enjoy the pics here but it is best to shift direct to the new site. Looks the same but lots more stuff and regularly updated.  The full size pictures  are only available there.

CLICK HERE to go to tesladownunder.com index page


Power topics on this page include:


Can crushing


Can crusher 2


Coin shrinking


Measurement of high current pulses


Measurement of high voltage pulses


Low voltage (but high current)




Power supplies


Low voltage supply using a MO


Wood burn fractals


Big SCR's

Can crushing  (1000 joule) This is done by discharging a capacitor through a coil wrapped around an aluminium drink can.  The high voltage capacitor is necessary to generate very high peak currents to induce currents in the can which magnetically heat and crush it.  Both capacitors come from medical defibrillators.  These normally deliver up to 360 joules into the chest body resistance between large conductive pads of around 50 - 100 ohms (modern machines print this out with each shock). There is a series coil of around 47 mH to limit peak current flow in the medical situation. Both capacitors are Maxwell pulse rated around 36 uF at 5.2 kV and each stores 500 J (total 1000 joules).  I charge it with the supply I made for HeNe lasers.


The can crusher before and during firing (click to enlarge)

The coil is made from 8 turns of 2 mm wire.  Most can crushers only use a few turns.  I have not tried this before as I believed that I did not have enough power for to crush a can.  Interestingly my 800 V supply with SCR switching barely makes a dent in the can although the delivered energy of 1500 joules is greater. A rapid rise time is important.  I trigger it by a momentary switch (the grey and yellow plastic device above) with large metal contacts. Not really cool like a triggered spark gap. On one occasion the can shorted the coil and I could feel the blast wave from the spark. I wear hearing and eye protection and look away (while firing and taking the picture)! 

Larger can crushers will tear the can in half with around 2000 joules. 

Can crusher 2 (and tear in halfer!) Using a 3rd 500 joule capacitor for a total of 1500 joules and with straight and shortened wiring paths this model performs much better.

(click to enlarge)

The can before and after. Note the distorted windings where a short circuit has occurred and vaporized some of the wire.  One wonders how many turns there were effectively.

  Point to play

Video (720 k but worth the download for the few frames with action. Sound is good too).

(click to enlarge)

There is noticeable crushing at 200 J, better at 300 J and the can tears apart at 1500 J.  Also dependent on the turns used. Note that 3 turns at 2400 V 110 uF (300 J) had an effect but 3 turns at 5200 V 35 uF (300 J) had no effect. So experimentation is called for.

Enlarge the picture and you can see the "5% crushed lemons" and "low joule" labeling!

(click to enlarge)

Picture above shows the smaller Red Bull cans crushed at various energies and the effect of crushing a full can.  The full  can is interesting and bears further analysis.  My interpretation is that the thermal and mechanical effects are on such a rapid timeframe that there is no conducting away of heat but the fluid has inertia and incompressibility.  This has the effect of preventing the aluminium from folding in to the centre.  It remains in the area of highest field which is right in contact with the coils. Hence, can disruption into the two halves is enhanced and since there is no infolding the "cut" is cleaner.

The fluid in the can undergoes a major drop in pressure as the two halves of the can start to separate.  Hence the observation above that the upper part of the can is folded in.  It is sucked in by negative pressure rather than by the induced field.

This process is symmetrical and the net result is that the fluid forms a central column that stretches and becomes fountain shaped as in the video. 

  Point to play

Video (700 k) shows effect of crushing a full can.  The can rapidly disappears leaving a long stream of fluid.  Look at it frame by frame if you can.

Some frame grabs at 1/120 sec from a video sequence on tearing apart a full can.

Coin shrinking   This is to be an account of my attempts to shrink a coin in the same way a can is shrunk around the middle.  I am not sure if I can or not.  The power is certainly much lower than is conventionally used by the pro's such as Bert Hickman who use 100kA from Maxwell pulse caps with up to 6 kJ and the best coins.

Things I have learnt or surmise. 
The greatest field is right adjacent to the coil. Minimise it and don't allow the coil to move away.
The force shrinking the coil is the same one acting to push the coil away.  Think of a hammer blow as a fair analogy of the forces involved.
Use the most conductive coin with softest metal. A gold doubloon perhaps?
I may have to use a small coin. I know bigger is better..
There need to be adequate insulation as there is 5kV across the coil.  It will very readily short out.

Version 1.  To start I have used a tiny Malaysian 1 sen copper coloured coin .  This choice was for availability, size and apparent copper alloy composition.  I started with 8 turns of single stranded PVC insulated wire that was too thin.
I covered the taped up coil with a towel and fired. The wire exploded into small pieces and most of it went through the towel.  There was no copper left near the coil. When the smoke cleared the coin was unchanged.  Unfortunately I didn't save the current waveform but it was around 10kA and more heavily damped then the can crushing coil.  In fact it was probably truncated as the coil disintegrated.

Version 2 was heavier with a wood block for support. Still no coin shrinkage.  I was subsequently informed that this particular coin is magnetic and indeed it is.  Possibly with nickel content and generally higher resistance.  The magnetism gives an effect opposite to the induced current and resists the effect of coin shrinking.

 (click to enlarge)

Left photo shows the pieces that I could find of the first coil.  The second coil, (centre photo) used heavier wire and a wooden surround and it still blew the wood apart and broke the wire in one part (right photo).

(click to enlarge)

This is the oscilloscope shot which shows the current reading which is different to the can crushing one.  Current scale is 4250A/div and timebase 20uS/div. Peak current around 13kA. The surprising thing is that the first current reversal is GREATER than the initial pulse. This seems most likely due to inductance (L) reduction by a degree of interturn shorting.  Note that the magnetic forces act to compress axially (lengthways) and expand radially (outwards).  The consequences of this are that this should NOT recur if I prevent interturn shorting or movement.

Version 3 used an aluminium disc in lieu of a coin.  I setup the whole 5 turn coil and coin in a block of ice (made from relatively non- conductive deionised water).

(click to enlarge)

The block of ice was shattered into small pieces but the coil was still enlarged and burnt out.  The "coin" was shrunk by perhaps 0.5 mm with the blue circle being the tracing before and the red after.  So no real success yet.  I am told that a Japanese 1 yen coin is suitable for crushing being light and aluminium.

See the discussion of my coinshrinking on the 4HV forum.

(click to enlarge)

Left photo is of the fibreglass and cloth and 4 inch PVC reinforced ice block housing the coil with inter-turn insulation above. It is shown after firing (centre photo) with a few ice shards that have cracked off in the shockwave. Right photo shows the current/time trace discussed below.
Incidentally I forgot and used plain tap water rather than deionised water (we use rainwater here for water supply). Unlikely to have made a difference.
Well the last 10 turn coil was fired in its block of ice. No exploding shards of ice just the bang of the spark gap. The ice developed cracks through it. After the shot I melted all the ice (similar to 'a watched kettle never boils') to get down to the coil. Unlike the previous attempts, there was no damage at all to the coil, no expansion and no inter-turn shorting. Almost like it didn't fire (apart from the cracks in the ice)

Unfortunately there was no shrinkage of my aluminium disc either.  This is not a 'coin shrinker' yet.
Sadly, I now have to take apart the coil that took so much time to make so I can recover the disc.

So, success in maintaining the integrity of the coil but not in shrinking the coin.

So why didn't it work?  The clue may be in the current time curve above. Current scale is 4250A/div and timebase 20uS/div as before. Peak current is only 2.5kA for the 10 turns. Which is much less than the 13 kA for 5 turns. The ringing is more prolonged and slower suggesting that energy is not being extracted well from the resonant circuit.  I suppose this means a higher Q.

This suggests to me a couple of possibilities:
1  the longer coil of 10 thick turns is not as effective as the 5 thinner ones in concentrating the field near the coin. The higher Q suggests this. 
2  that interturn shorting is increasing the peak current and hence effectiveness of the coin crushing. This means that simply winding magnet wire and expecting it to all explode and have some interturn shorting is important in getting the peak currents, greater than if the coil remained intact and well insulated and not expanded.  Paradoxically I perhaps should have been encouraging my wire to short between turns rather than going to lengths to prevent it.

Measurement of high current pulses (or my multimeter doesn't have a 50,000 amp scale)
Measuring the currents involved with can crushing presents a challenge. The initial pulse is short (25 microseconds and the current oscillates back and forth several times in a resonance that is around 16 kHz.  To understand this look at the oscilloscope shot below.  The upper trace monitors the current with time.  Time is 20 uS per division. The initial current rises, then falls, then reverses as the energy stored in the coil's magnetic field returns to the capacitors and back several times in diminishing amounts. Energy is drained from this process by heating and disrupting the can.

To measure these currents without connecting wires to the high voltage, I have used a small inefficient transformer which does not have a magnetic core called a Rogowski coil.  There is no magnetic core to saturate so the pulse response is good.  It is a transformer so cannot measure DC, however.  The primary of this transformer is only one wire which is the main current carrying cable.

 It will respond to the rate of change of magnetic field or current so has to be "integrated" using, in my case, a passive setup with a resistor of 2.2 kohms in series followed by a parallel capacitor of 0.01 uF.

(click to enlarge)

The left photo above shows the oscilloscope with the actual current waveform upper trace (the Rogowski coil output with passive integration) and the lower one shows the raw Rogowski coil output which shows rate of current change (di/dt).  The middle photo is the winding of the Rogowski coil onto flexible coaxial cable. The coax centre is the return wire.  The right photo shows the Rogowski coil wrapped around the main cable.

So far so good but it needs to be calibrated to work out what current in the main cable gives what output of the Rogowski coil.  This is best done at the frequency concerned.

The calibration procedure setup in summary.  A messy process with lots of cables, meters, high voltage and high frequency power.
High frequency power is supplied from my Royer circuit running 6.5 A, 25 V (160 W) with 2.2 uF polyester caps. Main core is 7+7 turns on an inverter MO core. Gap retained. Secondary is also 7 turns. Resonant frequency is 18 kHz measured on the CRO and frequency meter.

Setup 1:  25 turns around the Rogowski run from the 7 turn secondary, Current is measured in 2 ways. Firstly RMS clamp meter (3% error at 400Hz according to specs) showing 6.4 A AC RMS.  This seems to agree fairly well with:
Second current measurement is DC via a bridge rectifier, 2000 uF and a DC current meter.  This gives about 15% ripple in use. DC meter reads 8.4 A which is essentially closer to a peak reading and is the expected value of a 6.4 A RMS sine wave (increases by  sqroot2)
Hence 6.4 A = 160 A.turns at 18 kHz
Rogowski output is 0.25 V peak = 0.18 V RMS
  ie 1 V Rogowski output = 890 A

Setup 2
Single wire setup with can crusher in place including crushing coil in the circuit.
Current is 13 A DC clamp meter and 12 A with DC bridge method with the latter being more accurate.
Rogowski output is 0.014 V peak = 0.010V RMS
  ie 1 V Rogowski output = 830 A

Average these two readings and 850 A is a fair guess for 1 V output from the Rogowski coil at 18 kHz.

  (click to enlarge)

The oscilloscope shot of an actual firing which shows current at 850 kA/div i.e. peak of 15 kA (15,000 A) and 20 uS/div time base on channel B.  I still haven't found the top half of the can  yet after it went ricocheting around my shed.

Now for more flexibility I have made an 'active' Rogowski integrator using a fast integrated circuit op amp and a few bells and whistles.

(click to enlarge)

Left photo shows me working on calibrating an active integrator for my Rogowski coil using a TL072 op amp to replace the previous passive integrator.  Middle photo shows the frequency vs output voltage for the active integrator. Right photo shows the completed integrator with a peak and hold detector.  The Rogowski coil is the continuation of the white cable around the orange PVC support.   The large black cable carries the high current under test. The thinner black cable is 10 turns through the Rogowski coil for testing at 20 A 50 Hz.  I have two inputs 100 A/V (full scale 1000 A peak) and 5,0000 A/V (full scale 50,000A peak).  Calibration was at 50 Hz but should be OK to 25 kHz from the graph in the last post.  It uses  about +/- 16 V supply which is poorly regulated but current draw is only 3 mA. Output waveform is read on the CRO. 

I have had a further round of calibration at 50 Hz and have been getting current readings of 40,000 A peak.  High frequency output at 18 kHz is about 10% below that expected from the 50 Hz calibration.

I have added a peak and hold circuit with the intention that this can be read from a multimeter and hence be useful for the majority of coilgunners and other using pulse power who don't have a CRO.  I have had to go through a bit of a learning curve to deal with some offset problems which are now largely fixed.  The peak and hold is not really useful (since I have a CRO) but the time constants for a 60us pulse to be held for many seconds are beyond a single IC and I need a voltage follower at least and probably a dual peak and hold with two different time constants as the Merovingian has done.  His op amps have convenient offset pins.

Measurement of high voltage pulses
I have now measured the voltage during discharge along with the current looking for voltage reversal (a bad thing for caps). Voltage reversal occurs when the energy dumped into the coil begins to return to the capacitor which is how an inductance and a capacitance oscillate. The initial 5000 V drops to zero and then reverses polarity and back again in a decaying manner.

 I initially ran my CRO and Rogowski from a 12V 150W inverter to isolate it from the MOT supply which is at a potential to earth.  Later I made a set of dual contacts with a long insulated handle as a DPDT switch to isolate them.  Voltage divider was 27M (330K x 82) and 270K for 100:1 division.

Surprisingly there was no voltage reversal with that setup (see left photo below).  The voltage decayed in a damped sinusoidal waveform to the baseline.  This was not expected and I wondered about stray capacitance with my leads and the relatively small currents through my divider creating excessive damping.

So to see if there was voltage reversal I tried some expendable diodes 5 x 2.5KV 60nS to catch the reverse EMF. If they explode then there was voltage reversal.  Well, they exploded all right. The only piece I ever found of the diodes was the small black fragment embedded in my right wrist (every other part of me was behind a blast shield).

Accordingly, I rebuilt my divider with 10 x 1M uomega resistors to give 10:1 then used my 100Mhz 1.5kV probe and minimized wire length.

(click to enlarge)

Left photo shows the inaccurate voltage curve.  Right photo is more accurate and shows voltage reversal of over 75% for a 2KV (=200J) shot with a can in situ. Scale is 10,000 A/div for current (upper trace) and 1 kV/div for voltage (lower trace) with time base 20 uS/div.

See the discussion about my can crushing and current measurements on the 4HV forum.
Altair 's 4kJ setup 100kV but only run to 400J.
Bert Hickman is renowned for his small change (coinshrinking) and has a few pics of can crushing.
Sam Barros used electrolytic capacitors and SCR switching.  Electrolytics are slow and his 3000J gave similar results to a 300J shot with my setup.
Tristan Stewart has nice pictures of different energies and the can being torn apart with 2100J.  It was the inspiration for my efforts.
Mike Harrison (Mike's Electric Stuff) has some high speed videos of can crushing.

Low voltage (but high current)  One volt can be impressive too.

(click to enlarge)

Check out the 533 A current on the clamp meter from this transformer (above).  The secondary is wound with 12 mm by 3 mm rectangular wire and is rated at 6 V 90 amps. Seen  here heating an iron strip from a transformer lamination with about 1.1 volts across the two bolts = 600 W. At this current it will melt if left on for more than a few seconds for the picture.  Short circuit current is around 830 A plus.  This was another of my junkyard cheapo transformers for around AUD$20.

Spotwelder   I made this as I thought I may need something to make joins for use in a vacuum that may get red hot. Epoxy and plain or silver soldering would not do.  It uses the transformer above and is able to spot weld a variety of mostly thin metal sheets.

 (click to enlarge)

The left photo shows the hinged arm which allows two tungsten electrodes to come together. I also shows various bits of metal joined with this (very 'art nouveau').  The middle photo shows the flash if there is a poor contact. The right photo shows a good contact and 812 amps flowing. About 1-2 seconds later the tungsten rod was red hot.  The 8 mm black cables get hot even with a few shots.

Power supplies 1971

(click to enlarge)

My sentimental favourite power supply was made in 1971 and was developed over the years. It appeared in the local newspapers 8 times as a backdrop to photos about two of four state wide talent quest prizes I won between 1970 and 1973. Currently (no pun intended) it is defunct but previously supplied 1-30 V DC at 1 A with a LM317 regulator, and a range of AC voltages from 3 V to 500 V.

Low voltage supply using a MO.
A  microwave oven transformer can be used for a powerful low voltage supply.

  (click to enlarge)

Left photo shows removing the HV windings the hard way. Usually easiest to hacksaw off the winding flush then use a block of wood and hammer them out. Can't be done easily with this vertical winding style however.  Middle photo shows the 0.83 volts AC (RMS) for a single turn at the rated 240 V AC input. Right photo shows 13 turns being one side of a centre-tapped winding. This gives 11.8 V AC which will be 23.6 V AC which after the diode bridge rectifier will be 23.6 * (sqrt 2) = 33.6 volts.  This will be reduced by the voltage drop of the rectifier to about 32 V DC peak.

(click to enlarge)

Left photo shows the full 26 turns centre-tapped giving 22.86 V which is a little different to the 23.6 above due to slightly different input voltage. Right photo shows the addition of heavy gauge windings of 4 + 4 + 7 windings in series to give 12.72 V AC.

This allows the many AC voltage combinations from 3.3 V to 35 V.  Arranging selection of these is now the consideration, along with a range of DC voltages at an appropriate cost i.e. nearly zero.

Wood burn fractals

   (click to enlarge)

Here is what happens when you put a few kV between two nails in some wet pine. I have adjusted the voltage between 2 to 10 kV from my single MOT with voltage multiplier as used in my laser supply altering it periodically to promote the spark or prevent flames.  If things stopped I added a water spray.  They burns have a curiously fractal nature about them.  Note that the burns are just as likely to burn away from the nails as towards. I didn't encourage things either way.

Big SCR's  I have a number of large silicon controlled rectifiers (SCR's) of the "hockey puck" type. The ratings of this one (Semikron SKT 1200/12E)  are 1200 volts at 1200 amps.  For a 10 mS pulse it can take up to 30,000 amps. They need a large amount of compression (about 1/2 ton) to firmly press the precious metal contact layer onto the silicon to get the best contact.

    (click to enlarge)

Above is my SCR clamped with two heavy copper busbars with insulated bolts and is shown short circuiting a car battery at 210 A. I have to use the quick release locking pliers as there is no switch here and once the SCR goes on it stays on.


This page was last updated August 28, 2005