Building a radio with homemade condensers (that’s the old word for capacitor I’ll be using herein) is an art as old as radio itself, but one that had largely passed away by the mid 1920’s. Constructing a condenser for your spark transmitter was a right of passage in the days before radio licenses. The old spark sets often used tin foil and various materials for dielectric: glass, mica, oil or oil-soaked paper, etc. Air could be used, but it was both mechanically difficult and the relatively low dielectric constant and breakdown voltage was a challenge for the amateur who needed a large capacity at at least a few kilovolts for a small set. Oil dielectric was a messy but viable option, especially for its self-healing properties in case of an arc (and is still used in high-voltage capacitors today). Power loss due to corona discharge from the edges of a the metal plates in air were a problem with the RF voltages involved, and this was often mitigated by rounding corners of the plates and even immersing the whole sandwich of materials in a high-dielectric-constant binder such as shellac or paraffin. Using a solid binder also tended to improve the physical construction. Discarded window glass and 4×5″ photographic plates were popular materials; it helped to be friends with a glazier or photographer.
For typical example of homemade glass-condenser construction for a 1 kW amateur transmitter, see the July 1914 issue of Wireless Age, p. 847, available over at American Radio History. The only materials are shellac, glass, and tin, so quite a bit of of money would be saved compared to purchasing a commercial mica condenser, like the ones here:
WWI came and went, and vacuum tubes started to replace the whirling lightning, ozone, and interference-spewing monsters of yesteryear. Now a 10-watt CW transmitter could effectively replace a 250W spark set in terms of communications effectiveness while not causing the neighborhood’s lights blink. Another big benefit of this was that the typical RF voltages in an amateur’s transmitter dropped accordingly, meaning components could be smaller and use less material. The same construction practices hung on for a while in the tube era, but the lower voltage requirements along with the nascent broadcasting industry created a mass market for smaller and more affordable condensers. By the end of the 1920’s almost nobody was making their own, at least not out of economic necessity. By the time the 1930’s rolled around it was easier to salvage parts from obsolete broadcast receivers than it was to build from scratch.
Building a mica condenser
By 1924, a homemade capacitor was really nothing more than a sandwich of tin foil, dielectric sheets, some insulating material to clamp it all together…if in a higher voltage transmitter it probably had a paraffin wax binder as well. It’s important to note that component tolerances are usually huge in most of these early 20’s circuits. My advice to anyone attempting this is to find a published value for your dielectric and work with the standard equasions or just use the published values in a reputable period source–if you are off by a factor of two it really won’t matter, and if you are any good at arithmetic and can use a ruler you will do better than that. This is not a world of calibrated dials and three significant figures of accuracy. It always amuses me that people look long and hard for the magical 250pF grid condenser when 2000pF will (and did) work just as well.
Ancient wisdom: it is best to have an odd number of plates. Apparently this helps keep the stray field at bay, particularly if the outer plates can be at ground potential with the things around it. These things are big and they have big fields that can couple into stuff just like those big inductors you know all about.
The good news is that you can find raw mica sheets eBay. The old books tell you to use “ruby mica” or “Muscovite mica” if you can obtain it. This is relatively clear mica that is mined from central and western Asia and is supposed to be superior to the local North-American stuff which was also available back then. I’ve had good luck with some of the eastern-European sellers who are selling packs of old Soviet stock in packs of 50 or so. This usually in the <0.1mm thickness that was probably intended to be punched for things like TO-3 transistor insulators. I used multiple sheets to build up the desired thickness for the voltages I was anticipating. This also has the advantage that any impurities in the mica (scratch, pinhole) that might be a puncture point at high voltage are extremely unlikely to overlap from one random sheet to the next (so shuffle them if you get a pack). Over-engineering with a 4x safety factor or more is well worth it. I figure that if you are going to the trouble you should make something worth keeping.
The plates of the condenser were commonly made from tin foil–real “Sn” tin foil. This was a household item in the 1924 kitchen (the room where most radios were born), but has been replaced by aluminum foil today. I used a surplus roll of real tin for this project, but copper foil would be a period-appropriate upgrade and is more widely available. The “Bakelite” I used for the clamp was a modern 3/16 thickness XX-grade paper phenolic (XX Garolite from McMaster-Carr). This is the same basic recipe as the various radio panel materials from the 1920’s, but one for which I had the MSDS and thermal properties. Personally, I would not put some unknown 80-year old plastic-ish material in an oven, even at the relatively low temperatures I worked with.
The old 4×5 photgraphic plates were the go-to material for small glass condensers in the 1910’s and 20’s. The September QST transmitter article that I was working from mentions that photographic plate glass is usually just a hair under 1/16″ thick. I discovered that the picture frames at the dollar store are the right thickness, and you can get even get them in something very close to in 4″ x 5″ or the nearest Chinese equivalent (but look for bubbles). This is better than ruining real 4×5 photographic plates by scraping off the chemicals (these are collectors items today). Since I was building a replica, I had to cut my glass to fit the dimensions of the plans (i.e., 4×5 photo plates cut in half). I hate cutting glass, and glass this thin is a little harder to cut because it can crush under the pressure of the cutting wheel. With a little practice I found that I could score it and get a clean break by just dropping over the edge of a tile from an inch or so. Fortunately, any rough or chipped edges are not going to show in the final product. It is good that electrons don’t care about aesthetics.
You can build a glass condenser without a wax binder, but then you have the problem of squeezing the thing together so that stays in one place. When doing this there is a risk of stressing the glass or the compression loosening over time, temperature and humidity. Moisture ingress might be another concern. There was a further complication in my transmitter….as crazy as this sounds, my glass condenser was going to be the tank condenser and would determine the oscillating frequency of the set. Any mechanical expansion, shifting due to vibration, etc., would make the thing unusable. The wax keeps things in one place.
So, how do you make a sandwich with hot glass, wax and tin? I found a way that kinda works. First I heated the glass in the oven (about 170F–lowest setting on mine). The thermal mass of the glass allows “greasing” it with the paraffin wax as shown above. Next I placed the tin conductor on one of the sheets:
Now here’s the trick: let the one with tin on it cool and put the others back in the warming oven to heat back up. When you take them out you can drop the drop cold one with the tin on it face down on the next hot piece of waxed glass. You now have a glass-tin-glass sandwich with paraffin holding it together. Put it back in the oven to warm, and then apply paraffin to the top of the sandwich and put a new sheet of tin on it (glass tin glass tin, from the bottom up), being careful to make sure the tab comes off the opposite direction from the first sheet. Take this off to cool to cool and then you can drop it face down on the next piece of waxed glass. Keep repeating, being careful to make sure that the metal tabs alternate sides. It is not a fast process, but all this avoids having to pick up a hot piece of glass with liquid paraffin and risk accidentally upsetting the placement of the metal because it can move while hot. When finished with the desired number of glass plates you have something that looks like this with glass on the top and bottom:
At this point I put it back in the oven with a sharpening stone on top just to liquefy one more time and gently squeeze any excess wax and air bubbles out. I then turned off the oven with the weight still on it and let it cool in place.
In my replica transmitter they specified mounting the glass condenser to a shellacked piece of dry wood–Bakelite would have been a better RF material and period appropriate, but this was meant to be as economical as possible at the time so things that were “free” were preferred where possible. Finally it was to be wrapped with friction tap. This is the stuff with which you wrap hockey sticks and the grips of baseball bats. The tape protects the owner from sharp edges and corners as well as providing physical support for the glass on the wood base (shellac isn’t the greatest glue). I found that coating the outer layer of the tape with shellac took the stickiness out of it and keeps it from collecting dust and dirt. It should be good for 100 years.
Glass condenser from the Setptember 1924 QST (p. 56) and the and the one built by the Idiot from those instructions.
Here’s what the mica and glass condensers look like when mounted in the finished transmitter. Despite my doubts about that wood-and-glass condenser in the tank circuit, the frequency stays in one place and the losses don’t seem to be excessive. The circuit has other quirks, but it’s not the fault of the condensers!