How can I construct my own capacitor?

I want to construct my own capacitor, primarily as an experiment in different plate configurations and orientations, so I can better understand field theory. (In other words, I'm playing, and not looking for an off-the-shelf solution. The building of it is half the point.) I've built small PCB units on the order of a few nanofarads. I'd like to build something larger, on the order of 5-10 uF, which seems to make PCBs impractical. I also want the cap to be unpolarized, so I can run AC across it.

I'm thinking I need a couple rolls of foil, a roll of some insulator, and some way to attach conductors to the foil. But before I just go to Wal-Mart and buy all their aluminum foil, I wanted to ask: what might I be missing? Are there standard techniques for hand-building capacitors? Are there particular materials I should use (or avoid)?

You need to do some basic calculations first. The formula for capacitance is

$$C = \epsilon_R \epsilon_0 \frac{A}{d}$$

Let's say you want to use 1-mil (25.4 µm) waxed paper as a dielectric Note that this will determine the voltage rating of your capacitor. The dielectric strength of waxed paper is about 35-40 MV/m, so this will give you a capacitor that can theoretically handle almost a kilovolt, but be conservative in how you use it!

The relative permittivity of waxed paper is about 3.7, the permittivity of free space is 8.854e-12 F/m. Solve for the area required:

$$A = \frac{C \cdot d}{\epsilon_R \epsilon_0}$$

$$A = \frac{5 \mu F \cdot 25.4 \cdot \mu m}{3.7 \cdot 8.854 \cdot 10^{-12} F/m} = 3.8767 m^2$$

If you get aluminum foil and waxed paper that's about 12" (30 cm) wide, you can probably get an overlap of, say, 25 cm, which means that you'll need a length of about 15.5 m to get the area you need.

If you then roll up your capacitor (using a second layer of waxed paper), the capacitance will be doubled, or about 10 µF. Obviously, this will be physically rather large, over a foot long and several inches in diameter.

• You might get better results using plastic wrap as the dielectric instead of waxed paper. 1 mil is quite thin for waxed paper, but not for plastic wrap. I don't know what the permittivity is, but probably more than waxed paper. – Olin Lathrop May 23 '14 at 15:41
• Household cling film (e.g., "Saran wrap" or "Glad wrap") is typically 0.5 mils thick (12.5 um) and is made from PVC. PVC has a dielectric strength of about 40 MV/m and a dielectric constant of about 4.5. This would give you a capacitor that needs roughly half the area, but it would have half the voltage rating. – Dave Tweed May 23 '14 at 17:06
• @DaveTweed Saran wrap is PVDC, not PVC. $\epsilon_R$ is about 4, and dielectric strength of about 20MV/m. – Spehro Pefhany May 23 '14 at 18:35
• Depending on the brand and product line plastic wrap is often LDPE now. – Phil May 24 '14 at 0:29

Welcome to the world of capacitor building. I have built quite a lot of capacitor from stacked to rolled type. The only thing is that I build them for high voltage AC (4kVAC to 15kVAC). I'm not going into the details of physics or ESR etc....

The best bang for the bucks of all the tried and tested material is low density polyethylene (LDPE), which sounds very technical until you realize that it is basically the same stuff used in resealable freezer bags. If you are still unsure, just head down to any kitchen department and look for the resealable plastic bag, LDPE are quite stretchy. the ones used for trash and dustbin have loads of contaminants from recycled stuffs so try avoiding them. Usually carpet (yes, carpet) rolls comes bagged in long plastic wrap/tube. They are LDPE as well. Try to avoid high density polyethylene (HDPE) as they tend to rip easier. HDPE is also the same stuff normal used in supermarket plastic carrier bag.

If you want to splash some cash then you can try purchasing Poplypropylene (PP) rolls from florist but they are quite expensive. I have no experience on this as I build giant high voltage capacitor at tiny microfarads (0.01 to 0.2uF) and PP would be uneconomical.

The thicker the dielectric, the higher the voltage it will resist and the smaller the uF it will be, but for 240VAC, I think you can get away with thin stuff, I dont have the list of dielectric breakdown voltage on hand right now, try googling for it.

Tesla coil are dangerous but funn at the same time.

If the whole point of the exercise is to experiment with configurations that are mathematically tractable for you, probably the simplest is to have two large and very flat circular metal plates, in air, securely anchored at a fixed distance (use standoffs around the perimeter) that is adjustable and won't short out. Foil-covered discs might also work (you might need to glue down the foil). You can work out what capacitances you should expect from a reasonably-sized apparatus, and adjust your circuit from that. Are you planning to stick probes between the plates to measure fields, voltages, or anything?

From there, you can try various separations, different shapes (square, rectangular, etc.), lateral offsets, putting different dielectrics between the plates, and whatnot. I would suggest sticking to reasonably low voltages. Then you can try rolled-up designs. I'm not sure how the resulting interleaved plates will behave differently than one large pair of plates. Remember the old variable capacitors from radio tuners? Have fun (and stay safe)!

• Try beeswax impregnated paper – you can sell them later for a lot of money to audiofools. – Szymon Bęczkowski May 23 '14 at 18:38

For the order of 5-10uF you'll need electrolytic type capacitors. Check for instance http://www.youtube.com/watch?v=LXf3SUQ8F5g

• I didn't specify, I want an unpolarized cap. My bad. – Stephen Collings May 23 '14 at 14:53

You could go Old School with a Simple Leyden Jar

• Did you notice the part in the OP where he specified 5 to 10 uF? – WhatRoughBeast May 23 '14 at 18:38
• @WhatRoughBeast Nope, guess he'll have to find a BIG jar to do it this way. – Wayfaring Stranger May 24 '14 at 11:34
• According to wikipedia a leyden jar of about 0.5 liter will have 1nF. So you only need to connect 5000 of those to get to 5µF. – SpiderPig May 24 '14 at 22:51