Tests of homemade 200,000 volt capacitors
Copyright 2012 Brian Fraser
Caution: This page is under construction.
This is an "educational toy " Van de Graaff generator from Edmund Scientific that my parents bought for me over 50 years ago. It still runs fine with all the original parts (including the belt). It is advertised as producing 200,000 volts at a few microamperes. It is safe when used as directed in the user manual. Here, however, it is used to test Leyden jars and capacitors. These devices accumulate (store) the output of this machine for 2 to 4 minutes. The accumulated charge can be lethal. Such a setup is not safe for use by children or testosterone crazed males trying to impress their girl friends, or by people who have no experience with high voltage techniques. It can also damage nearby voltage sensitive electronic devices like computers, laptops, TV remotes, ipads, and so forth. Hence, this page is about test results and NOT about how to operate the various setups shown here.
The second picture shows the placement of a spark gap electrode. It is used to prevent overvoltage on the device being tested.
These are two large Leyden jars constructed from a plastic trashcan and a discarded cheese ball jar. Each could only be charged for about 2 minutes. A charge of 4 minutes caused dielectric failure ("punch through") accompanied by a loud firecracker-like explosion. They could stay charged for at least several minutes after the generator was shut off. Note that the foil does not go all the way to the top of the containers. This generous gap is necessary to prevent flashover.
This is a discharge wand with a 2 foot PVC handle. It is made from 1/4" steel rod and two Baoding balls (finger exercise balls) that I got at an oriental food store. The balls were drilled and tapped for the rod. A bolt with two tension washers allows the spherical tips to be adjusted as required. The bolt goes through an eye screw which is epoxied into the handle. During use, one ball is touched to the generator ground and the other is brought near the high voltage terminal. When connected to a Leyden jar, the high voltage terminal discharges with a very loud, thick, bright, long spark.
This shows dielectric breakdown ("punch through") at the bottom of the trash can and at the side of cheese ball container. Ordinary aluminum foil was used in the construction. Breakdown punches a neat pinhole in the plastic and blows back the foil. These holes were later plugged with Silicone I sealant, and normal operation was restored.
This shows the placement of foil (aluminum flashing) on a tube-within-a-tube PVC pipe capacitor. Normally a tube capacitor could be constructed with one tube, with foil on the inside and foil on the outside. But this one was intended to use distilled water as the dielectric. The water will go into the annulus formed when the 1.5 inch pipe is placed inside the 2 inch pipe. Note that thin wall PVC pipe was used in this case.
This shows arc-over tracks from surface corona discharge. The aluminum flashing electrodes had to be cut back about 4 inches (both inside and outside) to prevent flashover in air at 200,000 volts. Corona and arc-over eventually destroy capacitors, and also interfere with charging.
High voltage end of the capacitor tube. The outer foil (aluminum flashing) is at ground potential. The inner electrode connector is a stainless steel scouring pad epoxied to a bamboo stick and is threaded with #16 AWG wire. It connects with the innermost cylindrical electrode of aluminum flashing. The distilled water goes into the annulus between the two pipes. The epoxy coated paper centering ring is used during dry testing and construction. The thick insulation on the high voltage wire is made from three different diameters of vinyl tubing, the smaller ones being pulled through the larger ones. This capacitor is being dry tested for flashover (at 200,000 volts) and charging time characterization. It can store a dangerous amount of energy even with air/PVC as the dielectric.
This shows the scheme for sealing the inner pipe within the outer pipe and the passthrough for the water fill tubes. The brown rings are made from epoxy coated Kraft wrapping paper. The large ones are cemented to the pipe with expoy (the 6 minute kind) and the narrow ones are movable. The water fill tube is pressed against the capacitor tube with tape and then epoxied. The tape is later removed. The helical wrapping of the water tube keeps the inner dielectric pipe centered within the outer pipe. For final sealing, the inner dielectric pipe assembly is slid into the outer dielectric pipe. The inner pipe is then moved right or left a few inches so that a 1/2" wide layer of Silicone I sealant can be applied first to one end and then to the other. With the ends of both pipes flush, the narrow rings are pushed into the annulus with a suitable tool to compress the sealant and fill any gaps. The water fill tubes are stowed by coiling them in the air gap of the annulus.
This shows the completed capacitor. The outer foil (aluminum flashing) and the two copper drain wires are covered with a couple of layers of packaging tape.
The following test results are typical:
PVC Capacitor tube charging test (dry)
Tube: 3’ x 2” double wall coaxial PVC thin wall water pipe (1.5” + 2”)
Conditions: SG = 43 mm; RH 30%; dry capacitor, horizontal
Generator: 200,000 volt @ 5 microamp (nominal) Van de Graaff
Date: July 6, 2001
Time at Spark Gap firing
Difference (seconds)
25:35
42
26:17
39
26:56
39
27:35
39
28:14
38
28:52
37
29:29
36
30:05
Note that the charging time shortens somewhat as the test proceeds. This is probably because the dielectric tends to polarize over time. A single spark does not fully discharge it. This leaves less dielectric that is actually polarizable, and so the charging times decrease. At the end of the test, the capacitor can carry a residual charge even after being discharged with a wand several times. In fact, during early testing, I took this capacitor completely apart, handled all the parts and pieces, let them set overnight on the work bench, and when I reassembled it a day later, I got a mild shock. The lesson: Never trust a "fully discharged" capacitor!
The tests for the capacitor in the vertical position gave the same results as those for the horizontal..
Rough measurements using a 28" length of active plate section gave a calculated annular volume of 175 ml. I then injected 90 ml of "distilled" bottled water (grocery store grade). I found unexpectedly that 90 ml was actually the full capacity. I sealed the tubes and proceeded with another charging test:
PVC Capacitor tube charging test (wet)
Tube: 3’ x 2” double wall coaxial PVC thin wall water pipe (1.5” + 2”)
Conditions: SG = 43 mm; RH 30%; wet capacitor, vertical
Generator: 200,000 volt @ 5 microamp (nominal) Van de Graaff
Date: July 7, 2001
Time at Spark Gap firing
Difference (seconds)
34:03
49
34:52
48
35:40
46
36:26
45
37:11
44
37:55
44
38:39
44
39:23
44 40:07 41 40:48 44 41:32 46 42:18 46 43:04
The results were both encouraging and disappointing. The charging time increased by 8 seconds, indicating higher capacitance, as expected. But the increase was disappointingly small. Still, this was my first experience with a water capacitor. The fact that it has any capacitance after the water was added was encouraging. The device did not leak either, nor flash over, which means that the construction methods are valid, at least for the stated spark gap setting.
In subsequent tests, a Spark Gap setting of 62 mm gave a charging time of 65 seconds, and an SG setting of 70 mm gave about 78 seconds. In the latter case, corona losses at the generator were becoming significant and caused some scatter in the data. A 2.75 inch Spark Gap appears to be roughly the limit of this set up. If the dielectric strength of air is taken as 3 kV/mm, that works out to be about 210 kV. (http://en.wikipedia.org/wiki/Dielectric_strength ) A current of 5 microamps for 70 seconds transfers 350 microcoulombs of charge. Energy stored in a capacitor is U = 1/2 QV . At 200,000 volts that represents about 35 joules (or enough energy to light a 20 watt fluorescent light bulb for almost 2 seconds)
Water, as a liquid dielectric, has the advantage of picosecond relaxation times, which allows for very fast rise times (tens of nanoseconds) in properly constructed high voltage pulse generators (ones that use triggered spark gaps, transmission line techniques, reduction of inductive loop areas, etc.; fast rise times are believed to improve performance in antigravity generators.) Water has a relatively high dielectric constant of about 78.3. A major limitation though is that distilled water tends to be very corrosive. It tries to dissolve just about anything (even air), and becomes somewhat conducting as a result. The residual conductance results in self-discharge, and therefore limits the time available for extracting stored energy after charging. My implementation has no provision for continuous deionization of the water, and this is undoubtedly a limitation. However, the water is in an insulated annulus. (For pulsed switching see: http://event.cwi.nl/icpig05/cd/D:/pdf/18-221.pdf ; http://www.pulsedpwr.com/PDFs/PPLabsInc-HPMPhaseII-PPPS2009Paper.pdf ; http://alexandria.tue.nl/extra2/200712432.pdf )
Other dielectrics could be used of course. Transformer oil (or an ultradry mineral oil) has a dielectric constant of 2 or 3 and is conventionally used in capacitors, and will work at high voltages. Organic conjugated dienes have dielectric constants in the tens of thousands, but saturate quickly when charged with only a couple of volts ( http://www.springerlink.com/content/m117200kq47q1n10/ ; http://www.patentstorm.us/patents/6544651.html ; http://opus.kobv.de/ubp/volltexte/2011/5119/pdf/stoyanov_diss.pdf ). Certain polar organic liquids, with a k in the range of 30-200 can work too. Propylene carbonate, for instance, is especially effective as a capacitor dielectric, as is dimethlyl sulfoxide. Use of these (and others) as a dielectric can give energy densities 2 or 3 times that of a water capacitor, but without the problems associated with water. See US patent 3903460 and 3558908 for more information. Lead magnesium niobate has a k around 10,000 ( http://physics.info/dielectrics/ ), calcium copper titanate, 250,000 ( http://en.wikipedia.org/wiki/Relative_permittivity ; http://www.paper.edu.cn/index.php/default/scholar/downpaper/dangzhimin511435-201001-20[1].pdf ). There are even more exotic materials that have "giant dielectric permittivity" with a k in excess of a billion! ( http://repository.upenn.edu/cgi/viewcontent.cgi?article=1158&context=physics_papers ) See also dessicants)
Barium titanate is another common high k material (k of 1250–10,000). I tried making a high voltage capacitor by using it, paraffin wax, computer printer paper, and four copper foil plates. It was a complete failure. I could not even get it to charge. Apparently, there was some sort of internal leakage, but I did not have a gigohm meter (examples) handy to investigate. At 200,000 volts even a megohm is considered very conducting (do the math). A good insulator would be above 10 teraohms at a minimum.
In retrospect, I should have tried filling the water capacitor with a mineral oil / barium titanate suspension.
A stack of lead plates, paper index cards, and paraffin wax was to be used in a test of Brown's massive cellular gravitator. But because of the previous failure with paper and paraffin, the experiment was postponed indefinitely. However, Brown has suggested that a slightly conducting ("semiconducting") dielectric in this kind of application might have an advantage over a perfectly insulating one.
This is a tubular asymmetric capacitor similar to one described in Brown's patents. It is suspended from two pink nylon strings. The inner tube is filled with white barium titanate and bees wax. The outer one is filled with paraffin. The outside is wrapped with aluminum foil and serves as the negative electrode. The wire down the center is in contact, asymmetrically, with the barium titanate/wax mixture. Upon application of a 100,000 volt DC pulse, the assembly is expected to move in the direction of the ruler.
Links:
http://mark.rehorst.com/Van_de_Graaff/ (construction experiences)
http://distributionbizwiz.wordpress.com/2007/09/05/plastic-best-choice-for-high-voltage-capacitors/ (construction tips)"Experiments Which Show That the Earth Functions As an Electrostatic Machine", C. L. Stong, May, 1957
http://laplace.ucv.cl/Cursos/TrabajoTitulo/ExperimentosBajoCosto/VanDerGraaf/VanDerGraaf02.htmlhttp://www.cn-sphere.com/?gclid=CMPvgPStuq0CFasaQgodlAmTAA (hollow steel spheres)
http://unitednuclear.com/index.php?main_page=index&cPath=90 (spheres, Van de Graaff)
http://www.electricstuff.co.uk/ (lots of ideas and stuff)PropyleneCarbonateCapacitor4.doc (untested)
Interest in machnes that can produce short pulses of millions of volts at 100,000 amps for tens of nanoseconds has grown considerably in the last few years. The related design literature is very specialized, hard to find, and expensive. Some of the following references could be helpful:
"Nanosecond Pulse Techniques", http://ieeexplore.ieee.org/xpl/freeabs_all.jsp?arnumber=149456
J.C. Martin on Pulsed Power, John Christopher Martin, Thomas H. Martin, Arthur Henry Guenther, Magne Kristiansen (1996)
Pulsed Power Systems: Principles and Applications, Hansjoachim Bluhm (2006)
Transient Electronics: Pulsed Circuit Technology, Paul W. Smith (2002)
http://ped.slac.stanford.edu:8080/pem/useful_info/Blumlein.pdf (good tutorial on Blumlein configuration)
"Design and performance analysis of transmission line-based nanosecond pulse multiplier", Rishi Verma, A. Shyam, and Kunal G. Shah http://www.ias.ac.in/sadhana/Pdf2006Oct/597.pdfhttp://blockyourid.com/~gbpprorg/mil/herf/Impulse_Electromagnetic_Interference_Generator.pdf
"Design and construction of double-Blumlein HV pulse power supply", Deepak K Gupta and P I John (2000) http://www.ias.ac.in/sadhana/Pdf2001Oct/pe941.pdf
See also: http://www.ebookpp.com/bl/blumlein-pdf.html (various listings)
