Stefan's Tesla-Pages

How I built my Tesla capacitors...


Table of contents:

I'll give an overview on some different types of Tesla coil caps and describe how I built my own (glass bottle types, flat plate LDPE/oil and MMC):
000) Intro & safety
00) Proper capacitance value for different types of gap
0) Capacitor failure modes (text from arcotronics)
1) Salt water caps best for beginners
2) MMC (array of small professional pulse caps) best for advanced coilers
3) Big single professional pulse cap only for people with enough money ;-)
4) Stacked glass plate caps (dry) NOT recommended
5) Flat plate plastic foil caps (under oil) NOT recommended
6) Rolled plastic foil caps (under oil) NOT recommended


The capacitor is a critical element in coiling. Critical because it is really punished by the oscillations of the Tesla tank circuit. The tank current can easily reach a value of several hundred amperes. The RF produces intense stress on the dielectric of the cap. For limiting the cap voltage (resonance etc.), a second safety gap is usually set directly across its terminals. But beware: within this safety gap, the maximum current should be limited by high voltage high power resistors to the nominal value of the oszillations (where the inductance of the primary coil limits the current)!

Professional pulse discharge caps with the values needed for Tesla coil circuits usually cost between 200US$ and 1000US$ (depending on the capacitance and voltage value: .025µF $210.00, .05µF $540.00, .1µF $580.00 (quote from Plastic Capacitors, Inc., last two in march'98)). So most coilers try make their own. On this page I try to describe how I built some of my caps. The optimum solution for a quick and dirty coil are salt water glass bottle caps. The newest (spring '99) and best approach for advanced coilers is to connect many small pulse caps in series/parallel and build a so called MMC (MultiMiniCap). This seems to be the ultimate solution as a 20nF unit rated for 10kV in TC use cost below 50$US when build this way - and this is a perfect cap buildt by professionals (at least its components). Selfmade oil/foil caps are definately out of time (high price, big, really a lot of work, messy, not very reliable) and are therefore NOT recommended (though I placed the description of my humble attempt describing all my failures building such a thing here on my website for educational purposes). Some thoughts about professional caps (lifetime etc.) can be found on my BlueThunder page.

Some safety hints: In most cases the main cap in a Tesla coil is discharged as soon as the power is removed, the path being through the transformer winding. If your neon secondary should fail and open up during a run a lethal charge could remain in your caps (also of course if you use some caps in series). If you are going to do high voltage work utilizing capacitors, make and USE a shorting wand with a 10K or more power resistor (and of course one suitable for high voltage!) to safely discharge your caps (tau=10kOhm*100µF=1s, max. current at 10kV will be 1A). Capacitors can still retain a charge after being discharged. This effect is known as dielectric memory (caused by partial discharges). And sometimes capacitors sitting idle can build a charge. The safest thing is to keep all HV caps shorted when not in use! The main tank circuit capacitors are storing terrific amounts of energy. If the cap fails internally that energy could cause the capacitor to explode. I have heard of SW bottle caps blow up and also heard of several commercial pulse caps exploding in Tesla use. Do not use metal cased caps like typically found in microwave ovens. These are meant as power supply filter caps and are not suitable for Tesla use. They use dielectric not suitable for RF and will overheat and likely explode. It's a good idea to use bleeding resistors on every single cap!

back to top of page

Proper Capacitance value for the capacitor:

For static gaps it is recommended to use approx. 1.6 times the resocap size, for synchronous rotary spark gaps between 2.6 and 3.2 times the resocap size.

back to top of page

Failure modes of capacitors:
The following text is from the arcotronics website and describes the failure modes of capacitors:

Plastic dielectric film capacitors can undergo two classic failure modes: opens or shorts. Included in these categories are intermittent opens, shorts or high resistance shorts. In addition to these failures, capacitors may fail due to capacitance drift, instability with temperature, high dissipation factor or low insulation resistance. Failures can be the result of electrical, mechanical or environmental overstress, due to dielectric degradation during operation.

- Dielectric breakdown (Shorts) The classic capacitor failure mechanism is dielectric breakdown. The dielectric in the capacitor is subjected to the full potential to which the device is charged and, high electrical stresses are common. Dielectric breakdowns may develop after many hours of satisfactory operation. There are several causes which could be associated with operational failures. If the device is operating at or below its maximum rated conditions, most dielectric materials gradually deteriorate with time and temperature to the point of eventual failure. Most of the common dielectric materials undergo a slow ageing process by which they become brittle and are more susceptible to cracking. The higher the temperature is, the more the process is accelerated. Chemical or aqueous cleaning may also have an adverse effect on capacitors. Dielectric breakdown may occur as a result of misapplication of high transients (surges). The capacitor may survive many repeated applications of high voltage transients, howewer, this may cause a premature failure.

- Open capacitors Open capacitors usually occur as a result of overstress in application. For instance operation of DC rated capacitors at high AC current levels can cause a localized heating at the end terminations. The localized heating is caused by high RI2 losses. Continued operation of the capacitor can result in increased end termination resistance, additional heating, and possible failure. The open condition is caused by a separation of the end-connection of the capacitor. Both RMS and Peak currents may cause the open condition when overcome. Mounting capacitors by the leads in high vibration environment may also cause an open condition. The lead wire may fatigue and break at the egress area if a severe resonance is reached. The capacitor body must be fastened into place by use of a clamp or a structural adhesive.

- Environmental considerations The following list is a summary of most common environmentally critical factors affecting the life of capacitors. The design engineer must take into consideration his own applications and the effects caused by combinations of various environmental factors.

- Service life Service life of a capacitor must be taken into consideration. The service life decreases when the temperature increases (see page 8). Capacitors Failure Modes

- Capacitance Capacitance will change up and down with temperature due to the dielectric constant and an expansion or shrinking of the dielectric material (see diagram .C/T on page 5). Capacitance changes can be the result of excessive clamping pressure on non-rigid cases.

- Insulation resistance When the capacitor temperature increases the insulation resistance decreases. This is due to increased electron activity. Low insulation resistance can also be the result of moisture tapped in the windings, caused by a prolonged exposure to excessive humidity.

- Dissipation factor tan d The dissipation factor is a complex function involved with the inefficiency of the capacitor. The tg may change up and down with increased temperature (see diagram tan d on page 5).

- Dielectric strength The dielectric strength (dielectric withstanding voltage or "stress" voltage) level decreases as the temperature increases. This is due to chemical activity of the dielectric material which causes a change in the phisical or electrical properties of the capacitor.

- Sealing Hermetically Sealed Capacitors When the temperature increases, the pressure inside the capacitor increases. If the internal pressure is high enough, it can cause a breach in the capacitor, which can then cause leakage of impregnation or filling fluid or moisture susceptibility.

- Epoxy encased / Wrap and fill capacitors The epoxy seals on both epoxy encased and wrap and fill capacitors will withstand short-term exposure to high humidity environments without degradation. Epoxies and plastic tapes will form a pseudo-impervious barrier to humidity and chemicals. These case materials are somewhat porous and through osmosis can cause contaminants to enter the capacitor. The second area of contaminate absortion is the lead-wire / epoxy interface. Since epoxies cannot 100% bond to tinned wires, there can be a path formed, up to the lead wire, into the capacitor section. This can be aggravated by aqueous cleaning of circuit boards.

- Vibration, Acceleration and shock A capacitor can be mechanically destroyed or may malfunction if it is not designed, manufactured, or installed to meet the vibrations, shock or acceleration requirement within a particular application. Movement of the capacitor within the case can cause low insulation resistance, shorts or opens. Fatigue in the leads or mounting brackets can also cause a catastrofic failure.

- Barometric Pressure The altitude at which hermetically sealed capacitors have to be operated controls the voltage rating of the capacitor. As the barometric pressure decreases so does the terminal arcover susceptibility increases. Non-hermetic capacitors can be affected by internal stresses due to pressure changes. This can be in the form of capacitance changes or dielectric arc-overs as well as low insulation resistance. Heat transfer can be also affected by altitude operation. Heat generated in operation cannot be dissipated properly and can result in high RI2 losses and eventual failure.

- Radiation Radiation capabilities of capacitors must be taken into consideration. Electrical degradation in the form of dielectric embrittlement can take place causing shorts or opens.

This text can also be found here (.pdf)

back to top of page

1) Salt water glass bottle caps
The cheap and easiest way is to make a couple of salt water caps. They are nothing else than glass bottles filled with salt water and wrapped with aluminium foil on the outside. Glass is lossy. But here the heated glass is on the outside of the cap and the heat can be transferred to the surrounding air or the salt water (not so if you build a flat cap of stacked glass plates - see below for details and warnings). But there is so much of it that this effect is negligible. This is the ideal cap for the beginner as there is nothing needed than some cylindrical bottles, aluminium foil, tape, and some screws and nuts. Though most of the glass bottles withstand voltages up to 12-15kV, its recommended to place them in a box for safety reasons.

How I built my salt water caps (glass bottle types):

The easiest, fastest and cheapest cap is a salt water cap. Nearly every beginner makes at least one of them. It also was my first set of caps, still recommended for beginners or for a quick and dirty test on the new coil which will require a bigger cap in almost all cases.

I made most of my salt water caps out of Schweppes bottles (1 liter). I wrapped the glass bottles with aluminium foil on the outside (up to 2"-3" below the top) and the bottom. To give them more mechanical stability, I wrapped them also with electrical tape (the thin elastic stuff) from 1/2" above the bottom (don't cover the bottom with the isolating tape) to 1/2" above the upper edge of the foil. After that I filled the bottles with salt water (about 5 table spoons per liter water) up to about 1cm (1/2") below the aluminium foil. This helps to suppress corona discharge on the outside of the bottle. Then I covered the salt water with 1 inch cheap salad oil to prevent corona on the inside of the bottle. The electrodes that stick inside the salt water are threaded rods (M8 = 8mm dia). I screwed them into plastic screw caps from soft drink bottles and fixed them with 2 nuts, one on the inside and one on the outside. The aluminium foil is simply contacted by placing the bottles on a sheet of copper or aluminium.

Here you can see the first bottles I made for my 2"-TC:

Here you can see the bottles I made for the 4"-TC:
salt water caps (bottle type)

There are 8 of the bottles combined in a cardboard box and I have 3 boxes. Box1 is 8.7nF, box2 is 8.1nF and box3 is 12.7nF for a total of 29.6nF. But, as said above, these glass caps are lossy and bulky and shoulud only used as a quick starting point.

Be sure to screw the caps of the capacitor bottles loose so that the gasses produced during excessive runs can vent outside (else the bottle will explode)!

back to top of page

2) MMC:
A nice approach is to use many low voltage (typically 1600VDC or 2000VDC / 700VAC) MKP-caps (PP, film foil, e.g. the WIMA FKP1) and series/parallel them for TC-use, resulting in a so called MMC (MultiMiniCap). Be sure to use ONLY caps build from PP, so called MKP caps. The appropriate voltage derating from AC->TC depends on the frequency, duty cycle and some other values. A typically 1000VDC MKP-cap is rated 350VAC for 50/60Hz. For standard coils, the duty cycle of the RF is low compared to the 50/60Hz sine wave. But there is a phenomenon called 'partial discharge' which is a major problem at high frequency where the resulting corona (and its chemical byproducts) appearing at the edges of the foil eats the dielectric away. So the important voltage rating is the AC-value, not the DC-value. However, most of the 1600VDC caps (MKP with film/foil, of course) seem to survive even when used at the edge of their rating, which is "peak AC in TC use" = "DC rating". It's up to you to decide if you'll spend more money for a "safer MMC" or for replacing parts of it regularily ;-)
When building an MMC, you should series as many single caps as will be needed for your tank circuit voltage. Then connect as many strings in parallel as will be needed to achive the desired capacitance. You should only connect them in series strings and parallel whole strings. Do NOT connect individual parallel capacitors. Terry Fritz wrote on pupman: "I don't think the self healing feature is something to rely on though :-)) However, the self healing feature may be defeated by interconnecting parallel strings of caps!! I would no longer recommend doing that based on this. If a cap shorts internally, it may be able to successfully absorb the pulse. However, if there are say five other caps in parallel with it, it will probably not recover! Of course, I always recommend using drain resistors across each cap to equalize voltage (wrong, would be only useful for DC!) and drain the charge when power is removed (THAT is the real reason...) ."
He also wrote: "One thing I will say about any poly cap. If it gets warm, start to worry!!! Good news is, it takes a LOT to get them warm!"
For additional information, you also can have a look on the MMC-page of the GTL-website (which I'm also the webmaster of) or look at Terry Fritz' website for more theory, but please be sure to come back and read this website completely ;-) Terry made also a small program and collected lots of info to calculate the required number of strings in parallel and number of caps per string depending on the type of cap and the data of your TC. Kurt integrated Terrys program into his EXCEL-sheet

If you want to measure the capacitance of an MMC with bleeders, be sure to add a capacitor of known value in series to the MMC (to block DC current) and calculate the capacitance of the MMC back from that. Many cheap LCR-meters will read false values if the capacitance to measure is bridged with a bleeder resistor!

How I build my MMCs:

Go to my Vitamini-page and my 2"-TC page to read a description of the major steps to calculate the required number of caps. I describe the major aspects of the design there.

I had an interesting discussion with Christoph on the GTL on 09.12.2005 whether the thin wires of a capacitor have an effect on the performance of the TC. A short and rough calculation showed that the power loss is only about 0.5% of the input power.

back to top of page

3) Professional caps
Single professional caps for Tesla coil circuits usually cost between 200US$ and 1000US$ depending on the value. Homemade caps (read MMCs!) give more flexibility for the coiler because there are many single smaller caps needed for the desired capacitance. An additional handicap is that only a few companies in the world (apparently all in the USA) produce capacitors suitable for Tesla coil application. BUT - if you have enough money, no fun with your soldering iron and really know how much capacitance you need, then this type of cap is the best to purchase because it is small and reliable (ideal for TCs you wish to sell...).

back to top of page

4) Glass plate caps (flat)
A simple way to build a high voltage capacitor is to take a couple of thin glass plates (typically d=2mm glass) and stack them with aluminium foil between every plate. The foil comes out alternating on the left side and the right side of the stack. If all foils on each side are connected, the total capacitance of the stack is C=eps*eps0*A/d*(n-1). Of course, there has to be a wide margin around the aluminium foil (>5cm). This sort of cap is good for HV/DC but NOT for RF (read TCs!). There are two major disadvantages with this type of cap: the glass is very lossy which means energy is converted into heat inside the plates instead going into sparks. With this effect in mind, the shape of the cap becomes worse because the heat produced inside can not be transferred to the outside of the stack. This leads to very short runtimes and duty cycles if the cap should survive. I've heard from some coilers who's glass plate stack shattered from the heat (one coiler was injured in the face from the shattered glass flowing through the lab). The second disadvantage is the weight: an old attempt (I never finished it) leaded to 12kg for an 50nF cap.
So one can say that this type of cap is OUT OF TIME (don't build one!!!) because it will be

Let's make a short calculation of the losses in a glass plate cap based on my 4"-system of 8'97 (though I had SW-caps):

With the data above, the power wasted in the cap would be approx. 125W. And that's only the power from dielectric heating. There will be also the heating from the sparks at the edges of the electrodes. With salt water caps this presents no problem for the cap (except the wasted power is missing in the arcs) since the heat is distributed over a big surface (only approx. 1nF per glass bottle!).

back to top of page

5) homemade flat plate plastic foil / oil cap
Homemade highvoltage LDPE/oil-caps will definately extinct in the near future and they will be replaced completely by MMCs. Here are some reasons why the homemade ones are very nasty compared to MMCs:
- big
- heavy
- fire hazard
- smell
- not easy to transport
- VERY messy and a LOT of work
- they cost MUCH more than MMCs (yes, really!)
- you never get such perfect materials as the professionals use in their small pulse caps we combine to MMCs
+ ...(sorry, I can't find anything positive on homemade LDPE/oil-caps compared to MMCs anymore...)

Therefore the next section is OUT OF TIME! Better build an MMC instead!

The best dielectric (besides PP which also has nearly no losses) is LDPE-foil (Low Density PolyEthylene). But then a homemade cap has to be operated under oil to prevent corona discharge which will easily melt the LDPE otherwise. The oil should be a high quality transformer oil because both electrodes are inside the oil which serves as an high quality insulator in this case. Transformer oil can be found at each oil company (yellow pages) for about 80 to 150DM per 20liter here in Germany (that is around 60US$ per 5gallons). The LDPE should be a clear material with no scratches or particles inside. The particles would come from recycling processes and can be everything, even conductive. If the cap is built with flat plates, it has a very low inductivity. Also it can be made 'easily' out of many single caps in one housing to provide a set of different values. Make a vent hole in the container (or scratch the plexi top cover) so that the pressure can be released if something goes wrong. Oil is flammable, so have a fire extinguisher at hand.

How I build my flat plate poly cap:

The next drawing shows a schematic of the flat plate capacitor I tried to build once (click on the image for a better resolution):

The whole thing was housed in a plastic box (blue) filled with transformer oil. The cap consists of 2 packs of 3 series caps each. The series approach is to reduce electrical stress and therefore corona on the edges of the plates. I made the 2 packs for more flexibility in capacitance value. In the next schematic you can see the possible connections of the 6 capacitors:

This arrangement therefore would have allowed three different values of capacitance. C0 should have been around 16nF. The two packs were compressed by 3 compression bars (U-shaped aluminium) on each side of each pack.

Technical data:
Plastic box: 18cm x 32cm, height 24cm
Compression plates: 21cm x 25cm, 4mm thick Polystyrene, that was not thick enough !!!
LDPE-sheets: 18cm x 23cm, 0.283mm (11mils) thick clear material
Aluminium foil plates:   12cm x 16cm (overlapping area), so the insulating margin is 3cm (4cm where the Al foil comes out)   

There were 412 sheets of LDPE and 204 plates (2 LDPE sheets between each pair of plates). The effective thickness of LDPE was 1.7mm (6x 11mils = 67mils) per 10kV (connections A and D for maximum capacitance).

The thickness (LDPE and the plates) of each pack was determined to be 67mm for allowing 10% of the thickness for some oil and wrinkles inside. The formula for flat plate caps C = eps.eps0.A/d.(n-1) (n is the number of plates) applied to my special construction with 3 series caps in each of the two packs leads to a value of 16nF assuming 2.1 as the eps (dielectric constant) of the LDPE and the oil.


In January 1998 I arranged the first of the two packs. I did this under oil to exclude all air from the inside. This way no break-in period should be necessary. I stacked 206 sheets of LDPE and 102 plates to build one pack. I began one afternoon and continued the next afternoon. BIG mistake! The stack from the day before was now about 50% thicker than original. This was mainly because a lot of oil creeped between the sheets again though I placed a heavy screwdriver on top of the stack over night. It was VERY difficult to squeeze the oil out again. The next thing I encountered was that 308 sheets is way to much for one stack. I had serious alignment problems with the slippery sheets. I now recommend not to use more than 200 per pack when building the cap under oil. (Stacking them dry however is no problem.) Mounting the top plate of the pack was a real mess. The oil has eaten my latex gloves away faster than I could replace them (use Vinyl or Nitril gloves!). I clamped the pack as tight as I tried it dry (I marked the screws). Next mistake!
On the next day, the LDPE has swollen so much that it had many wrinkles inside (the sheets looked like ~~~~~~ instead of --------- ). Of course, the clamped stack couldn't expand lengthwise (to high friction), so the wrinkles made it thicker... You know what happened: the compression plates couldn't withstand the pressure buildup and cracked ;-(  The pack stayed in one piece but the sheets separated a little between two sets of compression bars. So I think the capacitance decreased a bit. The current capacitance value is 8.4nF for this first half of my flat plate cap. So the total capacitance will be around 16.8nF. With the predicted 16nF from the formula above we can assume that the effective dielectric constant of the LDPE and the oil appears to be 2.2.

Resumé or how one could try to make it better:
One should take GFK (glass fiber laminated epoxy) instead of PS or acrylic as compression plates. Another idea is to let the LDPE presoak in oil for some days before stacking them together. Also I won't fasten the screws in the first days completely to give the sheets the possibility to expand. If it helps: the oil I use is EltecGK2NA transformer oil from DEA, a naphthenic mineral oil. A HDPE container is advised for storage in the data sheet.
As I encountered with my flat plate LDPE cap, the LDPE soaks up a little oil and gets milky (it was clear before). Other coilers reported this phenomenon also when they repaired their rolled caps. The LDPE had wrinkles especially at the edges.

Another mistake: the plexi top cover of the container also cracked because the gasket swelled in contact with the oil over the weeks.

OUT OF TIME! Better build an MMC instead!

back to top of page

6) rolled cap (LDPE/oil)
An alternative way to make a cap is to roll the (long) plates and place the whole thing in a 6"-tube. To reduce the corona effects on the edges, the voltage per cap has to be limited to about 5-7kV per cap. So for the typical coil, a set of two or three rolled caps (of the same capacitance!) have to be connected in series. This leads to a not so flexible system as the flat plate cap would give, because a rolled cap has usually no center tap (it consists normally of only two plates, that is difficult enough to roll with all the sheets of LDPE).
All the negative remarks made at the beginning of point 5) about LDPE/oil caps apply here, too!
OUT OF TIME! Better build an MMC instead!

back to top of page