Reference Data for
Hawg Capacitor
Design and Experiment
THE HAWG CAPACITOR
Hawg projects are homegrown Hi Tek Gadgets made from common materials. The first Hawg project, popularized by the illustrious Kip, is an 11 dB gain 1250 MHz antenna made from hardware cloth, or here in Georgia, "Hawg Fence Wire" In keeping with Hawg tradition, here is an investigation of high voltage capacitor banks suitable for use in small Tesla Coils. The high voltage dielectric of Hawg Capacitors was originally plastic soda pop bottles.
A capacitor is two conductors separated by an insulator (dielectric). In the Hawg capacitor, a plastic bottle separates two conductive salt solutions. Salt water capacitors are common among novice Tesla Coilers, although most experienced hands graduate to surplus capacitors or more sophisticated homegrown capacitors. The main problem with plastic soda bottles is they cannot withstand the electrical stress, and the dielectric will soon puncture, short-circuit, leak, etc. Later experiments turned to glass bottles, cider juggs, mineral water bottles and the like to improve operating life.
Packing Identical Circles in a Larger Circle
THE HAWG CAPACITOR EXPERIMENTS
TEST FIXTURE A: the Test Article is connected across a 450VA, 15 kV neon sign transformer (NST) in parallel with a 2cm fixed spark gap. PVC pipe caps support the victim.
TEST ARTICLE : a 2 liter coke bottle with saturated saline inside, set inside a cut off gallon of fermented gatorade and salt. A dead ant was allowed to remain in the electrolyte. #14 Cu wire was immersed in the interior and exterior electrolyte.
TEST 1 : When the PSWC was clip leaded across the gap, the arc became a very loud and satisfying roar. The internal electrolyte level was about 2 in above the external level, and streamers were observed over the exterior of the dielectric in a band between the inner and outer electrolyte levels. Especially bright discharges were seen on the dielectric close to the outer electrode wire, and the test was terminated.
TEST 2 : Extra dielectric was placed as a spacer between the outer electrode and the bottle dielectric, and high voltage was reapplied. Corona and surface streamers were seen in the band previously mentioned. After about one minute, the PSWC shorted out very suddenly. A small dielectric puncture was found several inch below the surface, under a fragment of the bottle label.
TEST 3 : Another PSWC was prepared with a well-stripped 2l coke bottle. This failed after about 3 minutes at the inner (upper) electrolyte level by arcing thru to the outer electrode wire.
Conclusions : Everybody is right. Polyethylene Terpthalate is a great dielectric. Soda pop bottle PSWCs are vulnerable to dielectric rupture, puncture and leak. Foil is not the way to go.
Recommendations : Try again using a metal outer vessel for more symmetric current distribution, probably with a spacer covering the bottom to prevent dielectric stress concentration. Equalize inner and outer electrolyte levels. Use an oil layer above the electrolyte. Cleanliness is next to Godliness and Saint Tesla, get bottles 'squeaky clean' before use. Keep electrodes clear of dielectric to allow current density to diffuse thru electrolyte. Symmetric current diffusion and smooth, clean dielectric seem to be essential. No failures were attributed to the dead ant.
TEST ARTICLE : Outer electrode and vessel was a cheap K-mart stainless steel stock pot, the bottom of which was lined with disk cut from plastic ceiling light grille. The soda bottle was thoroughly stripped and all label adhesive removed with methanol. The inner electrolyte started out as carbonic acid, magnesium sulfate and sodium chloride. The outer electrolyte was magnesium sulfate and sodium chloride in a solution of water and fermented Gatorade. The inner electrode was 10 inch of 1/2 in copper pipe suspended in the bottle neck by a copper tee. The dead ant from last night was retired from duty. Electrolyte levels were equalized within a mm, and no oil layer was added.
TEST 4 : The test article was connected in parallel with a 450 VA, 15,000 Volt NST bridged by a 2 cm static arc gap. During the test, intermittent streamers were observed on the upper dielectric, which mostly appeared internal. Some condensation was also observed on the internal surface. After several minutes the streamers subsided although condensation remained. The plastic bottle dielectric failed at about t+30 min, with a small rupture at the electrolyte surface. All components had about a 10 C temperature rise. A narrow band of plastic at the liquid surface had shrunk during the run, making a smooth, symmetrical groove or constriction all the way aroud the bottle at the liquid line. The groove was about one cm in width and depth. Crystalline deposits were also observed around the entire circumference.
This was the most reliable and longest lived plastic bottle capacitor tested.
RECOMMENDATIONS : Filter contaminants from electrolyte, use a mineral oil film on both electrolyte surfaces to reduce corona. This may aggravate heat accumulation at the top of the liquid level.
TEST ARTICLE : Same as test 4, but with 5mm mineral oil film on electrolyte surface inside and outside dielectric.
TEST 5 : The test article with oil layer was tested for 30 seconds. A slight foam of gas bubbles was generated around the dielectric at or near the surface. A milky cloud began spreading in the oil layer or interface which appeared to be composed of tiny gas bubbles. Two hours after the test, the milky layer had dissipated.
TEST FIXTURE B: The Test Article is connected in a standard TC circuit. The excitation is a 450VA, 15 kV NST in parallel with a 2cm fixed spark gap. A 40 in dia 10 turn solenoidal coil and a Flexigap complete the primary circuit. PVC pipe caps support the victim.
TEST 6 : Three test article with oil layer were each run in turn until failure. Input voltage and Flexigap settings were varied. In the first case the failure was near the electrolyte/oil interface, in the next two test articles failed beneath the surface. Failure typically induced by running more than 8 gap sections and/or more than 70 vac input to NST.
TEST 7 : A 2 quart polyethylene food storage bottle was run until failure, about 3 minutes under conditions above.
Selected Email Excerpts During THE HAWG CAPACITOR Experiments
The Dielectric :
The operating life of a capacitor is roughly inversely proportional to the 5th power of the applied voltage, and is halved for every 10 C temperature rise. Dielectric in thin layers will stand more voltage per mil than in thicker layers. Plastic soda pop bottles are usually made of Polyethylene Terpthalate ( PET or Mylar ) with a wall thickness of 0.010 +- 0.0005 inches.
"Reference Data for Radio Engineers, 4th Edition" (1957) list of Material Properties :
Polyethylene DE-3401 with 0.1% antioxidant additive :
Freq K TAN K 60 Hz 2.26 <.0002 1 kHz 2.26 <.0002 1 MHz 2.26 <.0002 100 MHz 2.26 <.0002 3 GHz 2.26 0.00031 25 GHz 2.26 0.0006
Dielectric Strength (0.033") is 1200 V/mil at 25C.
Softening point is 95 - 105 C
Moisture absorption 0.03%Factors that promote failure of the dielectric are : concentration of E-field at sharp corners, corona discharge (bewaare of corona in trapped bubbles) and heating. PET bottles are blown into molds which 'orients' the molecules and improves the dielectric strength. Heating will cause the plastic to shrink and molecularly disorient, leading to dielectric rupture.
The Conductors :
A common technique using bottles is to wrap the outside with foil and put a condusctive liquid inside. With a foil wrapping, my 2 liter bottle measured 2.4 nF, I fully expected to get one or two nF more with electrolyte replacing the outer foil, but to my surprise the capacitance increased to 8.5 nF !! By putting liquid inside and out, a smooth, close contact is made, increasing capacitance, reducing field concentrations and corona. Corona is BAD NEWS for capacitors.
I dont know a lot about electrolyte conductivity, but ion concentration, ion mobility and valence will be important. However I suspect in this case the major portion of resistance is contact resistance. I normally use table salt (sodium chloride) and/or Epsom salt (magnesium sulfate). Jim Lux suggests copper sulfate for compatibility with copper electrodes. Watch out for generation of dangerous chemicals, for instance sodium chloride may decompose into chlorine. Floating a layer of oil over the top of the liquid, both inside and out, reduces corona discharge at the edges. ******** NOTE see "Conductances of Selected Electrolytes" above ********
.. thin layers have higher dielectric strength (per unit thickness) than thick layers. A phenomenon that has surprised many. Ruby mica has such a high dielectric strength partly because it is composed of many very uniform very thin layers stuck together.
I would consider one hour as the minimum acceptable life of a disposable dielectric capacitor. I would still like to compare notes with any other folks who attempt to use plastic soda bottle SWCs, the more experience i can gather from others, the further/quicker I can progress. Rules of thumb: dielectric life goes as 1/e^5 with voltage and 1/2 with each 10 degree temperature rise. Any comment ?
From: Kip Turner[SMTP:kip@mindspring.com]
Your extrapolation is interesting but may not work. You have to look at the BIG picture. Capacitors are usually tested at twice working voltage to weed out obvious defects. In going to very high voltages, you may excite other failure modes, such as arcing other than through the dielectric. Also, using your 6x example, you will get 36 times the normal internal heating, and most dielectrics are poor thermal conductors.
Yes, but I control the construction to eliminate other modes. Your point about heat is well taken.
I have made PET (mylar film) capacitors for AC operation at 1 KVA per cubic inch. This produces about 10 watts of heat per cubic inch and the units would increase in temperature at a rate of several degrees C per second. The units were rated for 30 seconds on-time per 10 minutes and required a internal cooling arrangement to even survive that.
I consider min acceptable life = 1 hour. Active cooling not desirable, mabe passive cooling.
Since you are operating at a higher frequency and very high fields, I would expect a much more rapid rise to the failure point (150-200 C). It is possible that, even with a cooling liquid for electrodes, the interior of the PET is overheating. The plastics are not very good thermal conductors.
My observations above point to localized heating at the liquid surface. Oil layer may/not help.
You might want to look it up, and calculate what happens to the internal temperature, assuming that the heat input is uniform and about 1 % of the reactive power in the capacitor. You should be able to integrate that and estimate the hot spot maximum temperature. You should keep PET well below 150 C, to avoid damage that weakens it and drastically increases the loss. You might want to check the "bottle" to be sure it does have the lower loss. It may not be as good as the film...which you can buy easily as "drafting mylar". Mylar (PET) is non-linear with voltage. It's resistance has a significant field intensity related effect, with the resistivity dropping at high voltages. This can help, in some cases, but you are above the fields which give long life.
Remember that there are several other causes for failure. Not just the voltage. You may reach gradients which produce field emission, especially at discontinuities.
Hey, I don't care if you pass along my "OPINIONS". Or you can take the credit ( or blame ) if you want to !
Kip
I'm probably running 2 to 3 kV per mil, but maybe this is still possible. So far I have blowed up two PSWCs, the first lasted 10 sec the second one survived 3 min. If I can cut the E-field in half again, that would give a life of 1.5 hours. According to you, if I take a PET plastic dielectric capacitor with a rated life of 100,000 hours to 3x its design voltage, it should survive 400 hours. At six times over voltage, it should survive 10 hours. Corona problems are still my main concern, although none of the Tesla List responses specifically mentioned corona in their PSWC attempts.
Since the Tesla list is generally interested in high voltage caps and especially home grown ones, I'm forwarding your experience to the Tesla list server unless someone objects.
Will
ps : Mike Foster tells me the PET in soda bottles is molecularly oriented as a byproduct of the blowing process. My CRC handbook does not list the real part of the vector permittivity nor the dielectric strength. Yes, I know about teflon, but unfortunately have no source of cheap Teflon soda bottles ;-)
From: Kip Turner[SMTP:kip@mindspring.com]
As mentioned by one of your respondents, Mylar dielectric film is grain oriented PET. I don't know if the soda pop bottles are or not. The DF (Dissipation Factor) is about 1 % at low frequencies(the dielectric stuff). If heated above about 150 C, it will un-orient and the DF (DF= watts/VA... like power factor) will increase to 3 or 4 %. I never checked it at RF frequencies. Obviously, it's rather lossy for RF use. Teflon as a DF of about 0.01%...a factor of 100 lower.
I highly recommend the oil floating on the electrolyte, to reduce corona which is very damaging to the dielectric. You may have a higher than average field at the edge of the electrolyte. Life on a dielectric varies approximately as 1/(E^5) where E is the field intensity in the dielectric. Working fields, depending on life requirements, are usually in the range of 600 to 1,000 V/mil for foil construction. Self healing types, such as electrolytic and metallized dielectric construction operate at somewhat higher fields, though a lot of the space saving is the lack of foil (typically about 6 micron thick).
Hope this helps you to estimate life. Remember...corona in a capacitor is a NO-NO !
Kip
Well, if 6u PET holds 600v then 9 mil PET should hold 900 kV, no? I should be able to get a pop bottle to stand up to a 15 kV Tesla coil, even with 15 kV of LF riding ont he 15 kV of 60 Hz. I notice the CRC handbook lists the dielectric constant of PET the same value for all three frequencies in their table. Thats gotta be a good sign for low dissipation. I dont have a microwave oven for dielectric loss testing right now.
Will
That does not confirm my experiences with PET. I have been able to get 6 micron PET coated with 0.025 micron aluminum to operate at 600 VDC, but that is a "self healing thickness of aluminum. ( It will evaporate around a fault ). Very small pieces of the material might withstand somewhat higher potentials, probably due to favorable variations in thickness and other random events.
I think he missed the decimal point somewhere. Suggest checking that hi-pot.
For the moment, at least, I'm a non-believer.
Kip
Last night I did a few experiments with a soda pop bottle PSWC (Plastic Salt Water Capacitor) . The wall of a 2l plastic coke bottle measured 0.010 +- 0.0005 inch wall thickness. With a foil wrapping the bottle had measured 2.4 nF, I fully expected to get one or two nF more with electrolyte replacing the outer foil, but to my surprise the capacitance increased to 8.5 nF !! Mike you were right about my skill at the subtle art of bottle-foiling !
From: Payne, Will E [SMTP:will.e.payne@lmco.com]
My recent query on use of plastic soda pop bottles for SW caps reaped two sets of responses. Some folks warned me about their bad experiences with soda pop bottle SW caps, which perforate, arc thru and dribble SW on the floor. However, Mike Foster is a fount of useful knowledge on sucessful use of PET dielectric bottles at very high DC voltages.
From: Michael Foster[SMTP:michael.foster@mailexcite.com]
If your 2-liter Coke bottle only had a capacitance of a little over 2nf, then the bottle walls must be thicker than the mineral water bottles, or you are not as good as I am at applying the foil. Pop bottles have to withstand 6 atmospheres of pressure because of the carbonation, so they might be thicker than mineral water bottles. Why don't you try immersing the pop bottle in a larger container of salt water and see the both made of the same stuff, polyethylene terphthalate. PET has a dielectric constant of about 3.2. Both the dielectric constant and the dielectric strength vary a bit with the average molecular weight of the polymer and with the stress of molding, or how far the plastic is stretched when it is blown into the mold. Similar effects are taken into consideration when rating commercial metallized polyester caps as the film used is the same material as the pop bottles. In this case, the film is "biaxially oriented" meaning that it is stretched in two directions while it is hot. This increases both the mechanical and dielectric strength, which are really the same thing. I find that it is important to have the salt water at near saturation to work really well. In fact, one of my favorite tech- niques is to saturate the solution with both salt and Epsom salts (magnesium sulfate.)
Just to see what would happen, I made a salt water cap which was two cells separated by PET film only .00048 inch thick (12 microns.) And no, that's not an extra zero. Anyway, I charged this up to 25kv repeatedly with no trouble at all.
The reason I bring up all this stuff about the stress and orientation of PET, is that if you heat up a spot on this plastic, it tends shrink there and get a little thicker. I just thought that with the addition of water cooling on both sides of the polyester, this might constitute a "self-healing" dielectric.
I guess I'm just going to finally break down and build a coil myself to test it out. And by the way, I will cut apart some 2-liter Coke bottles and measure the wall thickness with a micrometer, which is how I know the thickness of the mineral water bottles. They are all amazingly consistent.
Michael S. Foster
Altair Home
TCrefdata.html, © October 1998 Altair.
EMail altair@altair.org
