Stefan's Tesla-Pages

High energy page
Exploding wires, can crushers, coin shrinkers, ball lightning

The more energy, the better   ;-)

WARNING: big caps (>50J) are absolutely deadly!
Defibrillators have approx. 500J, we are dealing
with energies an order of two magnitudes higher (up to 50kJ) here!

50kJ equals dropping a VW Golf (a german car with 1400kg) from 3.5m, better not to stand under the weight!


Want to know how to produce YOUR OWN indoor LIGHTNING ???
Check out the rest of my web site !!!
Read how I created 56" long artificial lightning bolts with my Tesla coils.
A good place to start is here.
My website covers all from theory (cookbook) to systems I buildt
(great pics!), Tesla coil links and a lightning page.

The first setup:

Having some fun with 2.5kJ (75µF @ 8kV, MP-Caps):

For a private show on 23.05.09 in the family of my girlfried, I build a first version of the capacitorbank. Testing was done in my cellar vaporizing sone centimeters of thin aluminium strip. On the falily event, I vaporized some iron wire.

The next appearance of this capacitor bank was at the GTL Teslathon 2009 (11th July 2009, Zollernalb Sternwarte), where I  vaporized some copper and magnesium strips, erased some CDs and produced my first ball lightning.

I've measured the capacitance before and got 75.1uF. After probably 15 discharges (approx. 3 of them with ringing due to a too long or too thick wire), I now measure 73.7uF, that's only 98% of the original value! Perhaps the temperature also plays a role here, I don't know. I'll proceed to monitor the capacitance over time after each event to see if there is a continious decrease.

Statistics on capacitance decrease:  
  original after 3 ringings and
12 vaporizations
  uF uF  
C1   13.6  
C2   13.6  
C3   14.7  
C4   14.6  
C5   14.9  
total 75.1 73.7  

Up to now, I operate the capacitor bank manually with the help of some sticks on a long string (Q1, Q4, Q5) which I just pull out to operate the HV switches. The only electrical switch is Q6 & Q7 which are operated electrically by K10 which itself is activated by pressing a rubberball on a long thin hose. However, the circuit plan already shows the electrified version (lower part of the schematic):

Click for a high resolution image!

This image shows the first part with the 6kVAC neon siogn xfmr (lower left corner), bridge rectifier (8pcs. red brick diodes, 6kV@1A each, directly right from xfmr), current limiting resistors (4pcs. green tubular resistors 3.5kOhm each, top left corner), HV-probe (Fluke 80k-40 for up to 40kVDC, red thing in the middle) and panelmeter (grey box labeled "Vorsicht") and finally a selfmade manually operated switch which disconnects both HV wires (plus and minus) by gravity (to protect the diodes when the capacitors are rapidly discharged):

Here you can see the 5 capacitores I currently use in this "small" capacitor bank: 3pcs. Atesys caps (black, 16µF@10kV nominal each) and 2pcs. Cornell Dublier caps (grey, 15µF@10kV nominal each) for a total of 75µF@8kV (charge to 80% voltage):

On this image you can see the wooden blast box which will direct the residues of the (sometimes only partly) vaporized material upwards (away from the audience). Between the red and blue screws you can see 35cm of 0.4mm diameter iron wire, which is about 3 times the length I really can use with the energy I have here:

The lower box contains the dump resistors and saftey switch. Five fat carbon-ceramic tubular resistors can take the whole energy multiple times without getting to hot. Their resistance (3x 6kOhm = 30kOhm) limits the peak current to max. 0.27A (still more than 2kW peak power) so that a simple gravity operated switch with two ball electrodes can be used. In fact, one ball is just an acorn nut ;-) The second switch in the lower box is the saftey switch which makes a hard short on the caps. It's the last one to be pulled. I don't even trust it, so I put a "Jesus-stick" (also called "chicken-stick") across the terminals before touching anything. That Jesus-stick is not a grounding stick, because both terminals of my capacitors are floating. It is just a long isolated handle with a big fat 2kOhm carbon resistor at its end and some soft braided flat wide wire to make good contact.  

Here you can see the whole setup (purple box is empty, just for adjusting height and isolating from carpet). You can see the Jesus-stick (grey handly poking out from the right side of the capacitor box) and an analog voltmeter on top of the capacitor box:  

Calculation of maximum peak current with the help of an RLC-circuit simulator (

L=5000nH (guess)
R=1Ohm (guess)
=> Imax = 7kA

Spark gap resistance might be around 0.02Ohm in best case (one gap with large diameter, here I have 6 gaps in series with not so large diameter)

This is the data of the copper strip I vaporized at the Teslathon 2009 (sorry, still to be translated):

To be continued...

The BIG one:

Having more fun with 26kJ (24pcs. 137µF@5kV each, MP-Caps):

We aquired some pretty big capacitors (many thanks again to Max!):

The 5kV-caps will be connected in an array of 6x4 caps, that means four caps in series at 16kV (that is again 80% of their maximum rating) and six of these packages in parallel to achieve 206µF@16kV.

The first thing we did was to perform a measurement how well they will hold the charge when charged to 5kV:

Result: All performed well, two of them extremely well:

The capacitance values are ok and do not spread too much:

Though we had some damage due to the transport (dumb driver did not fix them at all on the pallet, so they slipped off), all caps held the charge well at nominal voltage:

Here you can see why it is really a good idea to install some bleeding resistors:

As said above, the caps will be arranged in a series/parallel array. To achieve a good distribution of voltage over the 4 caps in series, I selected them by capacitance:

Some facts (sorry, to be translated later):

Russischer Pulskondensator  2-5-140 Y4
Nominalspannung: 5kV
Nominalkapazität: 140uF
Eigeninduktivität: <600nH
Pulsstrom: 2,5-3kA
Ladungsrepetition: 10-12/min
Temperatur: -50ºC bis +50ºC
Log. Dekrement (CLR) Lambda=2-2,5

(siehe auch

also eingesetzt:
Lambda = beta*T

= R/(2*L) * T
= R/(2*L) * 2*Pi/Omega

= R/L * Pi/Omega
= R/L * Pi/{Wurzel(Omega0^2-beta^2)}
Lambda = R/L * Pi/{Wurzel([1/(L*C)]-[R/(2*L)]^2)}

Und das dann nach R auflösen... - ich war faul und hab das in EXCEL eingetippt und R variiert:
R = 0.040 Ohm liefert Lambda = 2
R = 0.049 Ohm liefert Lambda = 2.5

Damit nun Imax ausrechnen:
Imax = U/R =5000V/0.04Ohm = 125kA kann nicht sein...

=> L berücksichtigen und einen RLC-Schwingkreis Simulator nehmen.: liefert 51kA, also immer noch viel zu viel...
( liefert hier auch 51kA)

Wenn man den RLC-Simulator rückwärts anwendet und auf 3kA einstellt, kommt man auf R=1.65Ohm, also einen 41-fach höheren Wert als im Datenblatt angegeben???

Der Kondensator ist speziell für Impulsanwendungen gebaut und soll in einem Temperaturbereich von 1-35ºC betrieben werden.

Schwankungen der Kapazität betragen bei 20ºC zwischen -10% und +20%. Der Frequenzverlustwinkel tan phi beträgt 4,5*10e-3.

Der Kondensator besitzt zwei Hochspannungsdurchführungen und einen verschweißten Einlass für Isolieröl. Im Inneren befinden sich parallel verschaltete kapazitive Sektionen, die seriell einzeln mit einer Sicherung bestückt sind.

Das Isolieröl im Kondensator enthält kein PCB, was durch eine Ölanalyse bestätigt wurde.

Die mindestens garantierte Pulszahl beträgt 150.000 bei einem Signifikanzniveau von 10%. Es ist davon auszugehen, dass die Kondensatoren dieser Bauart ca. 10 Jahre lang mit jeweils 30 Schüssen pro Tag belastet wurden. Daraus ergibt sich eine bisherige Pulszahl von ca. 110.000, so dass sich ein durchschnittlicher Erwartungswert von mindestens 40.000 weiteren Entladungen ergibt.

Here a simulation of what might be achievable:

Spark gap resistance might be around 20mOhm in best case (one gap with large electrode diameter), see experiment of Bert Hickman simulated here:

To be continued...

Project for the future:

The Coin Shrinker (MP-Caps):

For the coin shrinker I'll use some pulse caps. I'll arrange them in modules of 10kVDC rated voltage. They usually will be charged to only 80% of their voltage rating for lifetime reasons (15th power law: charging to only 80% of maximum rating will increase lifetime by factor 28 (see my Blue Thunder page for details on derating). Unfortunately, most of the caps will die from overcurrent, not overvoltage). Since I recently acquired some 25kV caps, I'll be able to combine them with two 10kV-modules at a charging voltage of 16kV if I ever feel that 8kV is not enough for any reason. That way I'll bo compatible to the 26kJ capacitor bank I described in the section above ;-)

Up to now, I have the following pulse caps:
A1 18 caps
40µF @
MP-caps from Bosch
(single bushing,
housing is live, too)
5 modules
of 4 caps in series
=> 5x 10µF@10kV
2kV across each cap 1.6kJ  50µF o=10.5cm
A2 2 caps 40µF @
MP-caps from Siemens
(single bushing,
housing is live, too)
will be used with row "A1"   - - o=7.5cm
H=24cm incl. central bottom screw
A3 (1 cap) 40µF @
MP-cap from ITT
(single bushing,
housing is live, too)
replacement cap for row "A1" or "A2"   --- --- o=10.5cm
  4 caps 60µF @
MP-caps from Bosch
(single bushing,
housing is live, too)
1 module
of 4 caps in series
=> 1x 15µF@10kV
  0.6kJ  15µF  o=10.5cm
H=19cm incl. central bottom screw
B 12 caps 10µF @ 
MP-caps from Bosch
(9x single bushing,
housing is live, too.
3x double bushing)
4 modules of 3 caps in series
=> 4x 3.3µF@8(11.25)kV
2.83kV across each cap 0.42kJ 13µF o=10.5cm
H=20cm incl. 11cm x 11cm bottom plate

(double bushing?)

  1 cap 10µF @ 
MP-caps from Bosch
(double bushing)
    --- --- ---
C 4 caps 250µF @ 
Aerovox YD252EW250R24A
pulse discharge cap
one module of 4 caps in series
=> 62.5µF@8(10)kV
2kV across each cap 2kJ 62.5µF T=10cm
D 3 caps
(+1 cap
80µF @
red Maxwell pulse caps (type 34160) 1 module of 3 caps in series
(1 cap for spare)
=> 27µF@8(9)kV
2.7kV across each cap 0.85kJ 27µF o=18cm
(incl. top screw)
D' 1 cap
80µF @
SIEMENS B25352-S3806-A009 (just like the red Maxwells above)          o=17.5cm
(16.5cm incl. top screw)
E 2 caps 32µF @
still need two more for 16kV  perhaps to be combined with row G for 16kV 4kV across each cap 0.51kJ 16µF T=12.5cm


E' 1 cap 28µF @
CSI 5F653TN perhaps to be combined with row E       T=10cm
F 1 cap 20µF @
MP-cap from Bosch
(two bushings)
     --- --- o=10.5cm
G 2 caps 32µF @
MP-caps from ITT
(two bushings)
still need the third one for 8kV
 perhaps to be combined with row E for 16kV    0.34kJ 16µF o=12cm
H 3 caps
(+1 cap)
6µF @
MP-caps from Bosch
(two bushings)
1 module of 3 caps in series
=> 2µF@8(10)kV 
  0.06kJ 2µF o=10.5cm
I 2 caps 15µF @
   2 caps in parallel
=> 2x 15µF@8(10)kV 
   0.96kJ 30µF  
J 10 caps (+1 cap) 4µF @
MP-caps (?, probably plastic film!) from Siemens, single bushing 5 modules of 2 caps in series
=> 5x 2µF@8(12)kV 
4kV across each cap 0.32kJ 10µF o=10cm
H=22cm plus central screw (L=15mm)
K 10 caps (+4 caps) 100µF @
MP-caps from Bosch
(two bushings)
1 module of 10caps in series
=> 10µF@8(10)kV 
0.8kV across each cap 0.32kJ 10µF  
  1 cap spare

+4 caps

10µF @
MP-cap from Bosch
(single bushing)
2 module of 2 caps in series
=> 10µF@8(10)kV  
    20µF o=10.5cm
75 (+1defective +4spare) 80µF @
red Maxwell pulse caps (type 34152),  evtl. for long time charged, ok for higher voltage in short time use (evtl. tested at 120% voltage) 15 modules
of 5 caps in series
=> 15x 16µF@10kV
2kV across each cap 7.7kJ 240µF approx. 3.2kg each
            sum of 10kV-modules above 15.4kJ 480.5µF
@ 8kV
L 4 caps 1.4µF @
General Electric      0.72kJ  5.6µF  
M 1 cap 1µF @
CSI, pulse rated for
2 million shots up to
60% voltage reversal!
this cap will be the one nearest to the experiment (exploding wire or various coils) to deliver the fast current this cap has a center tap, so it could be used as a 4µF cap for half the voltage if the current will be limited (not for pulse discharge!)  0.13kJ  1µF  
N 3 caps 1µF @
Atesys this cap will be the one nearest to the experiment (exploding wire or various coils) to deliver the fast current this cap has a center tap, so it could be used as a 4µF cap for half the voltage if the current will be limited (not for pulse discharge!) 0.38kJ  3µF o=20cm
incl. screws
O 2 more caps 1.2µF @ 25kVDC CSI, pulse rated for
2 million shots up to
60% voltage reversal!
P 2 more caps 1µF @ 20kVDC Atesys
            sum of 20(25)kV-caps above: 0.2kJ
9.6µF @ 8kV
9.6µF @ 16kV
(9.6µF @ 20kV)
            total @ 16kV
(some caps will not fit into this arrangement!)
7.3kJ 57µF @ 16kV  

One possible arrangement for 8kV will be:

All the 10kV-modules listed above in parallel @ 8kV for a sum of 480µF @ 8kV =  15.4kJ.
Row L (5.6µF @ 25kV) but only charged to 8kV = 0.18kJ.
Row M: use the center tap of  cap and connect cap as 4µF @ 12.5kV but only charged to 8kV = 0.13kJ.
Total value will be: 7.24kJ @ 8kV w/o red Maxwells
(advantage: only identical caps in series!)

One possible arrangement for 16kV will be: w/o red Maxwells

Row A (50µF @ 10kV) in parallel with row B (10µF @ 10kV) for 60µF @ 10kV, this connected in series to row C (62.5µF @ 10kV) for a total of 30.6µF @ 16kV = 3.9kJ.
Row D (27µF @ 10kV) in parallel with row H (2µF @ 10kV) for 29µF @ 10kV connected in series to row I (30µF @ 10kV) for a total of 14.8µF @ 16kV = 1.9kJ.
Row J (10µF @ 10kV) in series with row K (10µF @ 10kV) for a total of 5µF @ 16kV = 0.64kJ.
Row L for a total of 5.6µF @ 16kV = 0.72kJ.
Row M for a total of 1µF @ 16kV = 0.13kJ (pulse rated).
Total value will be: 7.3kJ @ 16kV
(disadvantage: different types of caps in series!)

Rectangular Aerovox caps: The Aerovox caps are an older technology that is constructed from metalized kraft paper, foil and polypropylene. (Bert Hickman got the specs from Aerovox some years back for some 25uF @ 2500volt caps YD252EW025D21A. Aerovox indicated that the maximum surge current spec for these was 1,000 amps. Hopefully mine are rated for significantly higher peak current.)

Cylindrical red Maxwell caps: Here is the data I got from Maxwell about their pulse capacitors #34160 (red cylindrical housing, one has the serial number 72722): They do not have a complete specification for this 1976-vintage capacitor in their files today. The following data was given from their database and a test report they found on file:

CSI pulse cap data sheet:

I have no idea about the ESL and ESR of all the other MP caps up to now.

On the General Atomics Energy Division website  ( one can find a paper which describes the derating when higher voltage reversals are applied (

There is also a paper on safety (, where the following paragraph is taken from:
"One particularly violent failure mode that must be considered with large capacitor banks that are designed to operate in air, as opposed to under oil, is the ignition of a mixture of dielectric fluid and air after the capacitor has ruptured. In this scenario, when the capacitor case ruptures, it sprays the dielectric fluid into the air and then the arc caused by the internal fault ignites the mixture." So please be careful and take appropriate measures for that case when operating your capacitor bank!

Arrangement of the caps:

The best effect will be achieved with the highest voltage (and therefore highest current) by connecting all caps in series. This is ok for the electrolytics (see section on can crusher below). But the pulse caps from Aerovox and Bosch are different types (read ESL, ESR). It's not a good idea to connect caps from different vendors in series for HV or high current applications. Differences in surge impedance may result in accidental overvolting and failure of some caps followed by cascading failure of the rest. Also, lower voltages will be easier to cope with (think of corona). As described above, I'll build some modules rated 10kV and operated at 8kV. Each module will contain only identical caps.

With a 1" diameter 10 turn #10 AWG work coil, the peak current will be in the 50-75 kA range (according to Bert Hickman). It's hard to say what the lifetime of the caps will be since this will really be pushing them...

Charging circuit:

6kV neon sign transformer (rated 80mA, will be boosted to approx. 200mA) with bridge rectifier (2 stacked 6kV@100mA - diodes each leg) for 8kV charging voltage, small variac to adjust voltage.

Releasing the energy:

A) Firing the bank of MP-caps will be done via my EG&G GP-41B-25 triggered spark gap. It usually should be operated between 12kV and 36kV, can cope with a charge of up to 0.5 coulomb and peak currents of up to 50kA. Triggering will be achieved by a simple ignition coil and light dimmer circuit via two 0.5nF/30kV capacitors like Ross Overstreet suggested on his website. 

B) Should the EG&G gap ever die, I can build a robust gap out of two 60mm spheres (approx 7.5mm for 25kV according to the table in he HV-book of Jim Lux). The gap setting can be tested (VOM with HV-probe) with a small capacitance and big charging resistor.

C) NEW: On the teslamania-website of Bert Hickman I found a nice solution of a solenoid spark gap with moving contact. The contacts are brought close together so that the voltage is sufficient for breakdown. But they do not make physical contact. This way, no welding together can occure.
I intend to use such a gap here to cope with the smaller voltage and higher energy I have now available (not so good for the EG&G gap). Perhaps I'll use some graphite electrodes on a pneumatic cylinder driven from a small compressor. Setup will be: compressor, switching pressure gauge (to switch compressor off via a contactor when a certain pressure is reached), pressure regulator (with integrated particle filter), 5/2-way valve, pneumatic cylinder (10bar). The valve will be pneumatically activated by a 3/2-way valve which itself is activated pnematically when the operator turns a pneumatic switch. This pneumatic switch is actually a manually operated 3-way-valve. It assures that the tubing vented normally so that no accidential activation can occure. This way, the operator only has contact to two insulating tubings with a pneumatic switch at the end, no electric conductive wires involved.

D) Another idea is to use a pressure dependent switch (low pressure) to activate the 3/2-way valve mentioned above. In this case, the operator only has contact to an insulating tubing with a rubber bulb at the end. One squeeze and the switch will be activated.

Experiments with the bank of MP-caps:

I'll do a series of experiments with different energy levels and different number of turns and different wire diameters to find an optimum in shrinking performance. I'll start with a 10 turn coil of #10 AWG.