SKTF -- Fall 1990

The Smith-Kettlewell Technical File

A Quarterly Publication of
The Smith-Kettlewell Eye Research Institute’s
Rehabilitation Engineering Research Center

William Gerrey, Editor

Issue: SKTF -- Fall 1990

Original support provided by:
The Smith-Kettlewell Eye Research Institute
and the National Institute on Disability and Rehabilitation Research

Note: This archive is provided as a historical resource. Details regarding products, suppliers, and other contact information are original and may be outdated.

Questions about this archive can be sent to








Although this is an article presenting variations on designs of crystal radio sets, I tried to put enough theory in here for universal appeal. For example, impedance matching with transformers is discussed, a formula for calculating single-layer air-core inductors is given, and the "Daniell cell" (a voltaic cell made from a lemon) is also described.


[The editor would like to acknowledge the writings and accomplishments of Mr. Joseph Amoros of Richmond, Virginia, who published crystal radio circuits between 1941 and 1953. He sold my father boxes of crystal-set parts when he closed his mail-order business in 1957, and he brought us much joy in experimentation.]

A Bill Gerrish passion is the "crystal set"--the simplest possible radio receiver. It got its name for its crystal detector. Once upon a time, this detector was made by scratching around on a small chip of galena with a rusty wire, hoping to find the happy accident of oxides that would rectify the radio signal and present the amplitude envelope to your headphones.

Actually, various metallic sulfides were used as "crystals." "Galena," lead sulfide or PbS, is the most common ore of lead, and this was often used. "Iron pyrites, iron disulfide or FeS2 (sometimes called "fool's gold") was another ore used. Even Carborundum (a trade name for silicon carbide, SiC) was sometimes employed as the crystal. A blue-steel wire with a sharpened point was the "cat's whisker."

Modern crystal sets use germanium PN-junction diodes for their detectors. These are better diodes than those created from oxides, and they are a lot less fuss.

The purist's crystal radio set is a passive receiver; the power that "runs" it is what you steal from the radio station you are listening to. I leave them connected for years at a time.

With my father's help, I built my first one at age six, and I have hobbied around with them ever since--finding new and better coils now and then, or just putting them in differently shaped containers for the lark of it. The constant faint chatter from earphones hanging on a hook (in my lab upstairs and in the garage) is as friendly as the sound of a familiar clock to me.

Why do I entertain this frivolous pursuit? It's culturally enriching; it causes me to listen to talk shows, music, and yammering advertisers that I would normally leave the room to avoid. I suppose it brings back memories of a kind father revealing his best "shop tricks" to an impatient child who was barely listening.

How Do They Work?

Well, a big long antenna gets in the way of all those waves, and a tuned circuit (a coil and capacitor) resonates to absorb the maximum amount of energy from the station you want--barely attenuating the others. Than comes the heart of the set, the detector.

By itself, a radio signal has no audible component. Stations suitable for this kind of detection have the amplitude of their radio-frequency (RF) signal "modulated" (the power made greater or lesser) by the "intelligence?". (Intelligence on AM radio? Nah!) In order to hear these fluctuations in amplitude, you have to strip away the radio-frequency "carrier" and detect the "envelope" of the amplitude variations.

The diode detector is a rectifier-filter system that feeds the earphones the instantaneous peak value of the RF signal. This peak value is the "amplitude" of the signal, and when you retrieve that information, you get amplitude variations that are the modulating audio.

The rectifier, conducting only in one direction, charges a filter capacitor which then follows the peaks of the RF signal. The filter capacitor is the 0.005uF unit found throughout these circuits, and the diode is the "crystal detector."

Antenna Requirements

The crystal set needs a long antenna and some sort of ground return. A water main is usually the best ground, but a mounting screw of a wall socket, a steam radiator, or a wire fence may be good enough. Avoid using gas pipes as grounds.

There is no substitute for the antenna, however; it must be long. A couple of hundred feet of wire is a luxury. Fifty feet will sometimes do, if it is clear of tall buildings and trees.

My home lab antenna is a continuous overhead central-heating duct which is 50 feet long. (This one is rather funny, too, since it is grounded at the furnace.) My garage antenna is an unused wire in the telephone line. (Be careful of this one. Some telephone installations still use this third yellow wire for the ring signal. first, make sure that there is no DC from the yellow to the red and green, then get someone to ring you and make sure that no high-voltage AC appears between the yellow and the other wires. Finally, always couple to such a thing through a 0.01uF 500-volt capacitor.)

If you can string up a wire antenna, try to keep it from lying on surfaces and bumping into aluminum window frames. You can run a wire antenna inside an attic; hang it from short pieces of string under the rafters (the string acting as insulators). You can use any kind of insulating scheme to keep the antenna from resting against large surfaces. Holding it with plastic strapping of some sort, you can stretch the wire taut between elevated fixtures on a roof so that it doesn't rest on the shingles. On the other hand, unlike transmitting antennas, touching non-metallic objects here and there along the way will not be fatal to an antenna's performance.

That it be long is the important antenna consideration, but beyond this, not having it double back on itself is to be preferred. If you can make it long, but it must go out and comeback, keep the two runs as far apart as possible, and if it can be arranged, avoid running the wire in exact parallel directions.

Luckily, I live where there hasn't been a lightning storm in years. If your situation is otherwise, cultivate the habit of grounding your outdoor antenna when not in use, and if you hear a rumble of thunder, ground it quickly with a clip lead and run like a turkey.

Notes on the Various Components

The crystal set lends itself to parts which have been "scrounged" from far and wide. Don't ever let a dead radio, a 1960-vintage circuit board, or a dead telephone slip through your fingers. From telephones, you can get passably good medium-impedance earphones. From old radios, you can get coils and variable capacitors, output transformers for matching to low-impedance earphones, and cabinets to build your crystal sets into. From old circuit boards with solid-state stuff on them, you can get germanium diodes, or transistors that can be used as diodes. If the family finds the catsup in an orange-juice bottle some day, they will just have to accept the fact that you've used the nice cylindrical plastic catsup squeeze bottle as a coil form.

The Diode Detector

Radio Shack still sells 1N34A germanium diodes--ten for a dollar as cat. No. 276-1123. Mouser also sells germanium diodes--cat. No 333-1N34 and 333-1N60 ($.17 and $.25).

Though not always the case, these diodes are in glass envelopes which break rather easily if you are zealous in bending their leads. Also, their leads are often plated iron, and if you fool around too much in soldering them, the plating will come off, the iron will oxidize, and you will find the process of soldering tiresome, if not impossible. Solder 'em quickly, but deliberately, and treat them gently.

You can use a germanium transistor for its base-emitter diode. Typically, people tie the collector and the base together for one diode connection, and use the emitter for the other connection. If the transistor is burned out, you can often tie the collector and emitter together for one end of the "diode, and use the base lead as the diode's other end.

Silicon diodes and transistors won't work (unless the radio station is breathing down your throat). This is because silicon junctions have a comparatively high forward voltage--typically 0.6V as opposed to 0.3V for germanium junctions. On the other hand, you can apply an external bias voltage to reduce this unwelcome threshold to zero, and a circuit for that is given later.

Another possibility is use of the "Schotky Barrier rectifier"; these begin conducting at 1/3 volt, and they make a loud crystal set. Their disadvantage is that they have a very high capacitance; this is true because they use a large-area metal barrier in the junction. Near 0 volts, their typical capacitance is over 100pF, so use of them means that you have trouble tuning the top of the band. Nevertheless, they are fun to try, and a typical number is the 1N5819.

[You have all heard of the razor-blade detector. You rest a pencil lead (one intended for a mechanical pencil) on the edge of the blade; the oxide inevitably present on the thin edge conducts best in one direction. Like the galena and cat's whisker business, this is not a very good diode, and it's a lot of fuss. Interestingly, it was invented by a prisoner; he probably couldn't get down to Radio Shack.]

Commercially Made Coils

Rather than winding your own coils, you can buy them. The store-bought ones, because there is less market than there used to be, can be quite expensive. Once common place, a ferrite-slug-tuned broadcast coil, the good-ol' Miller 9001, now costs you $11. (The tapped one, Miller 9011, is nearly $13 now.)

Nevertheless, trying circuits with those Miller slug-tuned coils is worth doing, because they have such a high Q. Besides having the advantage of a ferrite core, they are wound with "Litz wire." [To conquer the "skin effect" (the tendency for RF current to use only the outside surface of a wire), Litz wire is made of many thin strands of wire, all insulated from one another. The result is low resistance at radio frequencies.] Moreover, those Miller coils are "scramble wound"; their turns do not lie parallel to one another, thereby greatly reducing the distributed capacitance.

Store-bought coils have a funny mounting arrangement. They have a springy metal cap that fits into a 0.3-inch diameter hole. A small tab 1/4 of an inch away from center keeps the coil from turning in its mounting as you adjust the slug. The mounting plate must be thin metal--an aluminum chassis or a fabricated bracket.

Homemade Air-Core Coils

It is easy to make your own air-core coils by winding magnet wire, or even wire-wrap wire, around a cylindrical form.

"Magnet wire" is copper wire which has been coated with enamel insulation; enamel is used because it is so much thinner than other insulating materials. You strip its insulation with medium-grade sandpaper; you test to see if you have gotten down to bare wire with a continuity tester.

There are various types of enamel--some very tough and some fragile. For radio coils, it doesn't matter which kind you use. (These differences are important to electromagnetics people--designers of motors and transformers.)

Radio Shack has an assortment of "magnet wire"; they give you 40 feet of 22-gauge, 75 feet of 26-gauge, and 200 feet of 30-gauge--Radio Shack No. 2781345. Any electronics supplier or electric motor shop can provide you with magnet wire. Belden Wire Company and Vector Electronics Company sell magnet wire by weight--1/2- and 1-pound spools.

For example, Belden Wire Co. sells "solderable magnet wire" whose enamel is destroyed in the soldering process (as long as the soldering is done at 800 degrees or more). This enamel chafes easily, but it is good enough for these coils; it can be gotten from Newark Electronics. A half pound spool of 22-gauge has the Belden No. 8051, Newark cat. No. 36F1314 (about $20, ouch). A half-pound spool of 24-gauge is Belden No. 8052, Newark cat. No. 36F1315. (about $20).

Though it frightens me to tell you this, Mouser has magnet wire, but only in large quantities. For example, they list an 8-pound 22-gauge spool of magnet wire--cat. No. 501MW22H (4000 feet). I bought one; it only cost me $80, and it is very nice wire made by Phelps Dodge. You can use it as antenna wire as well, but that's a lifetime supply for you and your neighbors.

Typical coil forms for the broadcast band are cores from paper rolls--toilet-tissue cores or kitchen-towel cores, cardboard mailing tubes, or plastic tubes (perhaps bottles with the tops and bottoms cut off). Plastic ones are the best; paper and cardboard take on moisture that causes losses and lowers the Q.

Securing a coil against a panel, especially a wood panel, will cause losses and will add distributed capacitance. Therefore, mount your coils on spacers.

My current favorite plastic forms are the "pill bottles" (vials, 30-dram and 60-dram sizes). These are available from sympathetic pharmacists (you have to convince one that these are for your crystal set).

The 60-dram vial measures 2 inches in diameter--more or less--and 5-1/4 inches long. The 30-dram vial is also 2 inches in diameter, but about 3 inches long (they leave a little room for cotton at the opening). The minor disadvantage of these bottles is the slight taper--smaller at the bottom. Because of the taper, all coils should be wound from the bottom up, so that succeeding turns drift toward the coil, not away from it.

A pill bottle is rather nice for mounting; you just screw the cap to the board and snap the bottle into place so that it stands vertically. This way, spacing the coil away from the board is automatically taken care of.

The ends of windings are anchored by weaving them in and out of a couple of holes. Consider the process of installing a 17-turn antenna winding at one end of a coil form. About 1/2 inch from the end of the form, you poke two holes which are perhaps 1/2 inch apart. You thread the magnet wire through one of the holes--from outside--and pull it through about 8 inches. Then, you poke the wire through the neighboring hole from the inside and pull it taut (giving you 7-1/2 inches of free end now. You now repeat the process; poke the wire down through the first hole and pull it taut, then poke it back out the second hole and pull it taut. This will fix the end firmly in place.

Crumple up the free end and stuff it into the form where it will be out of your way. Next, get a piece of tape ready and stick it to the edge of your workbench where you can grab it handily. Turning the form slowly, wind your 17 turns, every once in a while pressing them together with three or four fingernails so that there are no amateurish spaces between turns. When you are finished (or when you get tired and need to take a break), grab your piece of tape off the edge of the bench and slap it across the coil to hold it in place.

Once the winding has been completed and the tape is holding it, cut the wire to an 8-inch length. Poke two more holes just beyond the winding and weave the free end in and out as before.

Normally, margins of 5/8 inch are left at the ends of the form to permit screw holes to be drilled for mounting. Sometimes, as with the double-tuned circuit using pill bottles, you need to place a winding at the very end of the form; in this case, passing the first half-turn through another pair of holes opposite the ones fixing the end of the wire will keep wraps from spilling off the end of the form.

As an example of anchoring the end-most turn, consider the problem of winding 20 turns of wire around a plastic soda-bottle cap (making a coupling coil for some of these circuits). Very near the closed end of the cap, two pairs of holes are drilled, diametrically opposed. The wire goes in and out of the first pair and is pulled through to a length of 10 inches; then, the end is woven in and out of the opposite pair of holes to firmly "fix the end." A ridge at the open end of the cap means that the last turn can be anchored in one pair of holes; there is no danger of turns falling over the ridge.

You will see that most of the coils in these circuits require a tap one-quarter or one-third of the way up from ground. A tap can be arranged in two ways. The most elegant way is to sand a little bare spot in the wire when you get to that point in the winding, and solder a pigtail of insulated stranded wire to this bare spot. The other way, while it is a little more crude, is to bend the enameled wire double for an inch or so, then twist it several times so that a loop will emerge from the windings at this point in the coil.

Of course, you can have as many taps as you want. This is one way to match the detector most properly to the earphones--getting sufficient voltage to drive the earphones, while preserving the Q by minimizing the load from the earphones.

When it comes to connecting the coil into the circuit, cut the ends to slightly longer than the desired length and sand about 3/4 of an inch bare and make the connections. If the tap is a twisted loop, don't cut it, but just sand the end of the loop bare and make that connection. Test all stripped segments with a continuity tester; you cannot feel whether you have gotten down to copper, or just roughened the enamel.

Traditional dimensions for radio coils is to design them in such a way that the diameter and length are nearly the same. (Having the length equal the diameter makes the so-called "square coil" which is popularly used.)

There is an approximate formula given in the Reference Data for Engineers, copyright 1964 by IT&T, by which you can calculate the inductance of single-layer coils within a few percent. This expression is:

L (inductance in microHenries) equals n squared times r squared, divided by the quantity 9 r plus 10 l.

Lower-case "l" is the length of the coil in inches, "r" is the radius in inches, and "n" is the number of turns.

For the broadcast band, we want the inductance to fall somewhere around 220 microHenries. To save you the algebra, a 2-inch-diameter form will require secondaries of 80 turns of 22-gauge wire, 77 turns of 24-gauge wire, or 70 turns of 26-gauge wire.

[Because the 60-dram and 30-dram vials are tapered, it is good to make the windings 5% too long to compensate for this.]

Notes on Wire Gauging

The copper wire gauging system used in the U.S. is "the American Wire Gauge" (AWG). The wire diameter changes by a factor of 2 every 6 gauges. Thus, since 26-gauge wire is 0.016 inches in diameter, 20-gauge wire is 0.032 inches in diameter, and 32-gauge wire is 0.008 inches. (These are approximate dimensions, and the enamel adds perhaps 0.001 inches.) Since cross-sectional area depends on the square of the radius, this area changes by a factor of 2 every 3 gauge sizes.

Now, if you can remember three wire sizes, you no longer need a wire table. Good-ol' 22-gauge wire (0.0253 inches in diameter) reminds me of my micrometer--25 thousandths per turn, or 40 turns per inch. Wire of 24-gauge (0.0201 inches) gives you 50 turns per inch. Wire of 20-gauge is about 32 thousandths.

Tuning Capacitors

These are the most hard-to-come-by items in our modern age. Big radios don't have mechanical tuning any more, and little radios will soon give it up. The best tuning capacitors are being harbored in antique radios. Built into metal frames, they have a stationary set of plates ("stator") and a movable set ("rotor"). Many of these are multi-section--usually dual.

Unfortunately, most of the dual ones do not give us two usable sections, because the sections are different sizes and values. The smaller one, used to tune the local oscillator in superheterodyne sets, can be ignored and only the large one used. However, some low-budget radio makers of the 1950s used dual variable capacitors whose sections are identical; for double-tuned crystal sets, these are worth "breaking and entering" for.

Because many salvaged capacitors are fitted with dial-cord pulleys instead of knobs, you may end up making a knob for the odd-sized shaft after the pulley has been pried off.

The rotor is usually common to all sections. Since it is usually part of the frame, and since you must operate it with your hand, it is grounded. Connection to the rotor is almost always done via a mounting screw; solder lugs on the frame are rare. Each stator almost always has two solder lugs, one on each side of the stator, and these are electrically identical.

Radio Shack doesn't even have a suitable capacitor. The only one they show of sufficient range is a "compression-type trimmer" that you have to tune with a screwdriver.

The most readily available modern tuning capacitor is a Mouser unit suited for pocket radios. It has two main sections that go from 5 to 266 picofarads, and four 7pF screwdriver-adjustable trimmers which we can ignore for these sets. The Mouser cat. No. is 24TR218 ($2.71). The only trouble is that they don't supply a knob for the accursed thing; you have to make your own and secure it with a 2.6mm metric screw.

This Mouser unit is festooned with solder lugs. Two sets of three long ones are for the trimmers, which we don't need. There are three short lugs, two on one edge and one on the far corner of an adjacent edge. The lug which is off by itself goes to the rotors of both the large sections; the two coming out the same edge are the stators of these sections.

You can mount the Mouser capacitor by sticking its one free edge--or its back--to the chassis or breadboard with double-sided foam tape.

The required range is perhaps 30pF to 365pF. Actually, most go slightly higher than this--400pF or so, and their lower limit can be nice and low--5pF for each section of the Mouser ones. The Mouser units require that you connect the two main sections together, giving you 10pF to 532pF. The important thing is to have enough range to cover the whole broadcast band (given that the coil will have a lot of distributed capacitance).

(For double-tuned sets with matching coils, the Mouser units can be hooked up another way. Two of their 2-gang units are used; the rotors are connected together, the stators of Section A of the two units are connected together, and the stators of Section B are connected together. In this way, you have ganged capacitors; they both have to be at maximum to tune the bottom of the band, and they both have to be minimum to get to the top of the band.)

Caution: When you transport or store open-frame variable capacitors, be sure that the plates are fully meshed. If a capacitor is partly open, the plates of the rotor can be easily bent out of shape.

LC Resonance

As any radio ham knows, the expression for resonance, whether it's a series- or parallel-tuned circuit, is:

f equals 1 over 2 pi times the square root of L times C.

L is the inductance in Henries, C is the capacitance in Farads, and lower-case f is the frequency in Hertz.

Consider the example of 220 microHenries and 400 picoFarads. Expressed in scientific notation, the capacitor is 4 times 10 to the minus 10th power, and the coil is 2.2 times 10 to the minus 4th. Their product is 8.8 times 10 to the minus 14th. The square root of this is 2.97 times 10 to the minus 7th (you have to divide the exponent by 2 when you take the square root). Multiplying this by 2 times pi (2 times 22/7ths), we get 18.7 times 10 to the minus 7th, or 1.87 times 10 to the minus 6th. The reciprocal of this is 0.535 times 10 to the 6th Hz (the exponent changes sign when you bring it up into the numerator). Ten to the 6th Hz is a megacycle, so the result is 0.535 megaHertz.


These are becoming a troublesome item to get, but I've solved it. Crystal sets require high-impedance earphones. By high-impedance, I mean 2,000 ohms or more. Most of the traditional hobby types were electromagnetic; a permanent magnet put a bias on a diaphragm by pulling it in a bit, and then coils around the poles of the magnet would cause the field to fluctuate, causing the diaphragm to move back and forth.

Sometimes seen in junk collections, brands to look for are Trim, Brandes, Telephonics, Cannonball, Scientifics, Baldwin, and Utah. Short of these, you can use the earpieces from broken-down telephone handsets--fashioning a headband as best you can. Trim "Dependables" used to be issued by the Library of Congress for talking book listening. If you know someone who has those, get them somehow, even if you have to rough 'm up a bit.

Military surplus phones, such as those made by Utah, were all medium-impedance--600 ohms. These and telephone earpieces must be coupled to the crystal set via an autotransformer (described herein).

Another type, which affords high impedance, is the so-called "crystal earphone." Older ones used a crystal of "Rochelle Salt" as a Piezo-electric element to drive a foil diaphragm. Old ones of these are most likely bad now; the Rochelle Salt takes on moisture and loses its properties over time. New ones have ceramic elements.

They don't look resistive; they electrically appear as capacitors. For this reason, the diode detector may have to be provided with a load resistor in order to perform well. Finally, if you have some around and you are playing with active circuits, you should know that crystal earphones are destroyed by DC, so other than on your crystal sets, put a resistor across these earphones and capacitively couple them.

Mouser has high-impedance "crystal earphones," Mouser No. 25CR060 ($1.56). In general, pairs of magnetic earphones are connected in series to get their impedance as high as possible. Pairs of crystal earphones are connected in parallel. Earphones of the wrong impedance can be used if you match the impedances with transformers. For example, small loudspeakers can be used if you drive them with an audio-output transformer; the speakers, connected in series, go to the low-impedance secondary of the transformer (a 4- or 8-ohm secondary), and the primary winding goes to the earphone connections at the crystal set. Medium-impedance earphones, such as telephone receivers, can be matched with an autotransformer; if you have an output transformer with a center-tapped primary winding, the medium-impedance earphones, connected in series, go between the center tap and one side of the winding, while the whole winding goes to the earphone connections on the set (the secondary of the transformer is left open).

Impedances are transformed by the square of the turns ratio. For example, when you drive medium-impedance earphones with half the winding as described above, you see four times the earphone impedance across the whole winding.

Anyhow, I made a pair of earphones using two 8-ohm 2-inch loudspeakers connected in series (an earphone impedance of 16 ohms). Then, I used a miniature output transformer--8 ohms to 1.2K--to drive them. (The series-connected speakers are connected across the 8-ohm secondary winding, while the 1.2K primary goes to the earphone connections on the set. If it has one, the centertap on the primary is unused.) Since the secondary is looking at 16 ohms, the impedance appearing across the primary is 2.4K. I glued the speakers to stiff cardboard disks, then covered the backs with disposable drinking cups. The headband was fashioned out of coathanger. They could further be improved by putting foam shrouds to seal around the ears, thus improving acoustical coupling.

I tried my best to make "Walkman" earphones work, but to no avail. Since they are open earphones, and have small diaphragms, they are somewhat inefficient.

Mouser has a fine output transformer for matching to low-impedance headphones (like my PM speakers and drinking cups); it has a center-tapped primary of 20K, and a center-tapped secondary of 4 ohms--Cat. No. 42KM019 ($1.63). A less drastic impedance-matching transformer is the 8-ohms to center-tapped 1.2K--Mouser No. 42KM003.

Vacuum-tube radios have salvageable transformers of two types, center-tapped primaries for 2-tube "push-pull" amplifiers and high-impedance nontapped primaries for "single-ended" circuits. I kind of remember that 10K to 3.2 ohms was common.

Try everything--half the primary, the full primary, and try various output taps if the transformer has them.

Crystal sets don't give you much surplus power, so acoustic coupling to the ears is a critical factor.

Matching the Detector and the Earphones

Since the impedances of earphones you are likely to find are "by guess and by gosh," and the Q they are effecting is largely unknown, matching by experimentation is the best approach. The following effects can be noted to aid in troubleshooting:

If the set distorts on weak stations, either the diode is not seeing enough voltage (the tap is too low), or the impedance of the earphones is too high and the detector is not sufficiently loaded. If the set is nonselective ("broad as a barn"), the impedance of the earphones is loading the coil and greatly reducing its Q, or the tap is too high on the coil.

The serious experimenter will want to build his secondary winding with a selection of taps for the detector. Since the impedance--and the power--go up as the square of the turns ratio, adopt the systematic scheme of doubling the impedance every tap (going in steps of 3dB). This would require taps at the following percentages of the winding (starting at the grounded end): 18%, 25%, 35%, 50%, 70%, and the full winding. Notice that these percentages are related to each other by the square root of 2; the voltage and current either go up by 1.414 or down by 0.707 as you select neighboring taps, while the impedance and power are either doubled or halved as you select neighboring taps.

Solder these taps to successive positions on a 6-position rotary switch. The arm of the switch will then go through the diode, then through the earphones to ground, with the earphones being shunted by 0.005uF (see the circuits to follow).

Crystal Radio Circuits

I give my impression of the performance of any set I have in my collection. I tested them on my little furnace duct antenna, doing so in the daytime.

Some of these have variable coupling between the antenna and the main tuning coil. This creates a sort of combination volume control and selectivity adjustment. In general, the looser the coupling, the more selective the set becomes, although this is at the cost of sound volume. If a station is blustery enough to over-ride all interference, tighten the coupling, turn it up and dance to it.

* * *

I won a science prize once for making the world's smallest receiver. The circuit follows:

Circuit for the World's Smallest Receiver

The antenna goes through the diode to ground. The earphones are in parallel with the diode.

It gets everything all at once. This can be rather fun. You can hear zealous spiritual leaders in shouting matches with "rap musicians." When a famous politician's speech is carried by all the networks, you can hear three or four of him, all accompanied by drum machines.

[It is worth noting that at one time, this was good enough; in the very early days of radio, all the stations were trying to transmit at a wavelength of 300 meters--all more or less at 1 mHz.]

* * *

Perhaps the most common of circuits uses the distributed capacitance of the coil to make a half-baked LC circuit. A long coil is required, perhaps 200 turns of No. 26 wire on a cardboard or plastic form (such as a mailing tube, the core from a roll of kitchen towels, or the 60-dram vial). Then, for mechanical stability, "dope" (coat) the coil with model cement.

By sanding to and fro with medium sandpaper, create a strip of exposed copper across the windings so that a metal "tuning slide" can make contact with the coil. Next, fashion some sort of slider (perhaps a ground-off hacksaw blade anchored at one end) which can be used to make contact with the bared coil windings anywhere along the length of the coil.

What you have created is a coil with adjustable length. Only one end of the coil is connected to anything; the other coil connection is the slider. A circuit for the simplest tuned set is as follows:

Simplest Tunable Crystal Radio

The coil slider is grounded. One end of the coil goes to the antenna; also connected to this end of the coil is one end of the diode. The other end of the diode goes through the earphones to ground. The earphones are shunted by a 0.005uF capacitor.

The disadvantage of this circuit is that it is broad as a barn--for two reasons. First, the tuned circuit is low Q because of the resistance of all that wire used in the coil; tuning the set with a variable capacitor means that a shorter coil could be used. Second, no attempt is made to efficiently match the earphone circuit to the "tuned circuit"; unless very high-impedance earphones are used, the load that they present lowers the Q even more.

* * *

Standard LC-Tuned Crystal Radio Circuit

This was copied from a Knight Kit (Allied Radio) set purchased in the 1950s. The variable capacitor goes from 22pF to 400pF. It uses a Bakelite coil form measuring 1-1/2 inches in diameter and 5 inches long. The coil, wound with 22-gauge magnet wire, has a primary (antenna coupling) winding of 17 turns placed at the cold end of a secondary (main tuning) winding of 120 turns. The detector is taken from a tap which is one-fourth the way up from the cold end-- of the secondary30 turns from the grounded end. The separation between the two windings is 1/4 inch.

The antenna coil, 17 turns of wire at one end of the form, has its outside end going to the antenna, while the end nearest the secondary is grounded, This end of the secondary is also grounded. The other end of the secondary goes to the stator of the variable capacitor; the rotor of the capacitor is grounded. A tap on the secondary which is 30 turns from the ground end goes through the diode, then through a 0.005uF disc ceramic capacitor to ground. The earphones are connected across this ceramic capacitor.

This can be built on a 2-inch coil form; just change the proportions. For example, 80 turns of 24-gauge wire--or 85 turns of 22-gauge wire-- on a 2-inch-diameter form will do; I like the detector tap to be at 25 turns. The antenna primary can stay at 15 or 17 turns; an embellishment is to wind a 25-turn antenna winding with a tap 9 turns away from the antenna end, using two antenna binding posts for selectable coupling.

This set is very selective, but not very loud. I am able to get ten stations, seven of which can be isolated enough to listen to. It can be jazzed up a bit by providing it with variable antenna coupling:

Variable Coupling on the Standard Set

Wind 20-turns of 26-gauge wire on the plastic cap of a plastic soda bottle. Affix this cap to a nonmetallic stick of some sort, and fashion a sliding arrangement that permits the cap to be moved in and out of the grounded end of the secondary winding--from an inch separation to where the coupling coil is encircled by the cold end of the secondary.

Scrap the 17-turn antenna winding on the main coil form. One end of the bottle-cap coil is grounded, while the other end goes to the antenna.

Popular Mechanics Variation

A variation on this basic circuit appeared in the January 1977 issue of Popular Mechanics. The coil was wound on a "salt box" (3-1/4 inches in diameter). The junction between the antenna coil and the tuned secondary is a tap on an overall winding--14 turns below the tap for the antenna coil and 32 turns above the tap for the secondary.

This "tap" is grounded. The antenna goes to the end of the 14-turn segment. The rotor of the capacitor is grounded. The stator goes to the end of the 32-turn segment. The end of the 32-turn coil also goes through the diode, then through the earphones to ground, with the earphones being shunted by 0.005uF.

The editor has not built this Popular Mechanics set, but the closeness of its primary and secondary windings--delineated only by a tap on the continuous coil--will probably make for a loud set with poor selectivity.

* * *

A Double-Tuned Crystal Radio Circuit with Variable Coupling

Two 2-inch-diameter 30-dram pill bottles (vials) are used. The bottoms and caps are drilled to accommodate a wooden dowel or plastic rod. One of the vials is fixed by winding tape around the rod inside the bottle and under its cap; the other is permitted to slide a distance of perhaps 2 inches. When moved toward one another, the vials must be able to touch bottoms.

Matching coils are wound on the vials--85 turns of 22-gauge or 80 turns of 24-gauge wire. These are started as close to the bottoms as you can get, so that they can be brought quite close together. The secondary is tapped 25 turns up from the bottom, or even better, taps are placed at 15 turns, 20 turns, 28 turns, 40 turns, and 56 turns. (If multiple taps are created, connect them to successive positions of a 6-position rotary switch, with position six going to the hot end of the full winding.)

The rotors of two 400pF variable capacitors are grounded. The antenna goes to the far end of the untapped coil; the bottom of this coil goes to the stator of one of the capacitors. The bottom end of the tapped coil is grounded, with its top end going to the stator of the other capacitor. The tap (or the arm of the rotary switch) goes through the diode, then through the earphones to ground. The earphones are shunted by 0.005uF.

Because the coils are matched (except for the load the detector puts on the secondary), you can get away with a ganged 2-section variable, although having separate variable capacitors gives you some freedom to "detune" the set as an aid to interference elimination. Remember, the two sections of a ganged tuning capacitor must be identical.

With this set, I was able to receive ten stations, eight of which could be isolated well enough to listen to.

* * *

A Slug-Tuned Crystal Radio Circuit Using the Tapped Miller Coil

This is quick to build. The Miller 9011 tapped variable inductor is used. (This tap is 30% up from the cold end.) The end of the smaller portion of the winding is grounded. The other end goes to the stator of a 400pF variable capacitor, the rotor of which is grounded. The antenna goes to the tap. The tap also goes through the diode, then through the earphones to ground, with the earphones being shunted by 0.005uF.

The slug is adjusted to "center" the band in the range of the tuning capacitor. Simpler yet, since the coil is variable, the capacitor can be a fixed one--330pF across the ends of the coil. The only disadvantage is that the range of inductance is only about 4 to 1; since the frequency varies as the square root of this, the frequency range can only be 2 to 1 (from 0.6 to 1.2 mHz, for example).

This set is very loud; however, because the antenna coupling is much too high, it's broad as a hippopotamus. Four stations were received, and only three could be isolated enough to listen to.

A profound improvement can be had by fashioning a coil for the antenna. Wind 20 turns around a plastic bottle cap from a soda bottle (which is 1-1/8 inches in diameter). With bee's wax or cement, fix the cap over the end of the Miller coil. On this new primary winding, the end nearest the open end of the bottle cap is grounded; the other end goes to the antenna.

Wow! With this arrangement, I received eleven stations; eight could be isolated enough to be listened to.

Double-Tuned Crystal Radio Circuit with Slug-Tuned Coils

The coils, one Miller 9001 and one Miller 9011 are mounted near each other on a strip of aluminum. They are pointing the same direction and mounted at 1 inch between centers. (One could make this coupling adjustable by using two strips of aluminum in a scissor-like arrangement, but moving these coils detunes them, so you would have to adjust their slugs each time.)

One end of the 9001 (untapped) coil goes to the antenna; its other end goes to the stator of one section of a dual-ganged variable capacitor. The rotors are grounded. The end of the smaller portion of windings on a Miller 9011 coil (tapped at one-third) is grounded; the other end of this coil goes to the stator of the second variable section. The tap on the 9011 goes through the diode, then through the earphones to ground, with the earphones being shunted by 0.005uF.

Back the slugs out 15 turns from maximum (maximum being all the way in). Check to see if the capacitor tunes the whole band; if not, move the slugs in to shift the stations downward, or move them out to shift every thing upward as necessary. Next, find a station near the top of the band and adjust the secondary (the tapped coil) for maximum volume; retune the capacitor and adjust the slug a second time.

This is the best set I've ever had. It's quite loud, and I can receive thirteen stations, eleven of which can be isolated enough to listen to.

Circuit for the "Rocket Radio"

If you happen to find one of these in a junk shop, they're a lot of fun. They use an untapped slug-tuned coil. Pulling up on the nose of the rocket withdraws the slug and raises the frequency. You will notice the conspicuous lack of the filter capacitor; the capacitance of the crystal earphone doubles as the filter.

One end of the coil goes to the antenna. This junction also goes through the diode, then through the earphone (high-impedance crystal earplug) to the other end of the coil. The earphone is shunted by 220K. Depending on the make, the coil might have enough windings so that the distributed capacitance is the "C" of the LC circuit. Some had a capacitor shunting the coil.

The only ground return is you. The main secret behind these radios was the intimate acoustic coupling from the earphone to your tympanum.

* * *

A Hybrid Crystal Radio Circuit with Lots of Adjustments

The secondary winding is an air-core coil--85 turns of 22-gauge on a 2-inch form; it is multiply tapped and the detector is taken from a 6-position switch. The primary is a Miller 9001 untapped inductor. Since the characteristics of the coils are markedly different, you need separate variable capacitors (of identical type).

The Miller coil is mounted on a sliding bracket that allows its penetration into the cold end of the secondary to be adjusted. A knob is made--having a 4-40 threaded hole--and is fitted to the screw of the slug; a "jam nut" ahead of the knob secures it firmly.

The antenna goes through the Miller coil, then through C1 to ground. The bottom end of the secondary is grounded and C2 is across this coil. The taps go to positions of a rotary switch. The arm of this switch feeds the detector.

Adjusting the slug changes the Q of the primary. Sliding the primary in and out changes the selectivity and the volume. Selecting taps optimizes the match to the detector (changing the selectivity and the volume). Juggling the capacitors finds optimal settings to cope with various interference.

The tally isn't in yet on its performance; the instrument is too hard to run. I'm still chasing the last of thirteen stations, and I think I imagine a fourteenth. Watching you operate it can really impress your sister, although, in my high-school days, it seemed to invoke little enthusiasm from girl friends.

* * *

The 15-Minute Set Using RF Chokes

Mouser sells a style of miniature RF choke which has "radial leads"--both leads come out one end. This set uses two of them, Mouser cat. No. ME434-1120-221L (220 microHenries). If you butt them together--head to head--and hold them together with tape, you have the makings of a fine double-tuned crystal set. I "haywired" the whole arrangement on one side of an old open-frame ganged variable capacitor (no board was used). On the other hand, these chokes, together with two of the Mouser 24TR218 variable capacitors, could be put into a fancy little package.

The antenna goes through RFC1 to the stator of C1. The rotors of C1 and C2 are grounded. RFC2 is in parallel with C2. The stator of C2 goes through the diode, then through the earphones to ground. The earphones are shunted by 0.005uF.

I declare, when I counted ten stations on this set (eight of which I could isolate), I jumped right up to "Do the Locomotion" with my soldering iron.

[These chokes are nifty. Good for 50mA, values we are likely to use elsewhere are also available: 2.2mH are ME434-1120-223K, and 10mH are ME434-1120-104K. The "L" appending the low-value 220uH unit stands for "low."]

* * *

Biasing the Detector

Crystal detectors have a serious flaw; a station must be strong enough to overcome a forward voltage drop before it begins to drive the earphones. This can be compensated for with a bias battery. Moreover, someday, germanium diodes may be unavailable; active circuits are taking their place (creating "perfect rectifiers"). A silicon diode, such as the 1N914, can be used if you get rid of its 0.6V forward voltage drop.

"Battery?!" you cry in protest. It's not so bad, in that the current drain is only 15 microamps; a D cell will last its shelf life. This detector will work in any of the sets.

Biased Detector Circuit

The tap on the coil goes to the cathode of a 1N914 diode. To provide the detector with a DC load, the anode goes through 22K, in parallel with 0.005uF, to ground. The anode also goes to the positive end of a 1uF electrolytic (tantalum is best), with the negative end of this cap going through the earphones to ground. The junction of the earphones and the electrolytic goes to the negative terminal of a cell, with the positive of the cell going through 100K to the diode's anode.

The Daniell Cell

Would it help the purist if the cell were organic? How about use of the "Daniell cell?" A lemon, lime, potato, or other piece of food is pierced by a copper wire (untinned copper) and a galvanized nail (presuming this nail to be zinc coated). The zinc electrode is the negative terminal, while the copper is positive. (most wire is tinned, so you have to sand this coating off for the portion in the electrolyte.)

Address List

  • J.W. Miller: 19070 Reyes Ave., PO Box 5825, Rancho Dominguez, CA 90224; Phone: (213) 537-5200.
  • Mouser Electronics: 11433 Woodside Ave., Lakeside, CA 92040; Telephone: (619) 449-2222.
  • Newark Electronics: To find a representative for your area, call: (312) 784-5100.



A low-power tunable medium-wave (from 800kHz to 1200kHz) AM transmitter is described. It draws less than 6mA from 9 or 12 volts, and is equipped with low-level and high-level audio inputs (for use with microphone or talking book machine outputs).


We were made aware of a problem faced by blind competition swimmers. Where, precisely, is the end of the pool? Staying in the lane is no problem, as the lanes are roped-off straight paths. However, the competition swimmer needs to flip over and push off against the end of the pool, and only precise knowledge of when to do that allows the blind swimmer to fairly compete.

One setup in Australia has companions on the pool deck at either end of the lane. Each of these fellows has a long, lightweight pole with a tennis ball at the end with which he taps the swimmer. This is not a bad solution except, by the account I received, these fellows occasionally miss the swimmer. My immediate solution was to recommend that three such people with tennis balls be placed at either end; one of them is bound to find his mark on the swimmer. No one seems to like my idea, so the project described here is being assembled and tried.

I reasoned that having a person on the pool deck making the judgment of distance is a good idea; I am suspicious of detectors that might--or might not--detect the swimmer as he approaches the end. I like the idea of the swimmer being signaled by a thinking human being (armed, perhaps, with a tennis ball for good measure). Next, I reasoned that any equipment the swimmer might have on his person would likely be submerged now and then. Trying to radio the swimmer with short wavelengths when the antenna could be under water, from time to time, led me to dismiss VHF "wireless microphones." I have heard that long wavelengths work under water.

We decided to try medium-wave (the AM broadcast band). After getting hold of the AGC signal from a pocket transistor radio and connecting it to a voltmeter, the radio was sealed in a plastic bag and submerged in a large bucket of water. The AGC signal showed very little decrease in the signal of a local broadcast station, so we went ahead and put together a couple of transmitters.

Construction Details

Each transmitter is built on a piece of perforated board measuring 2-1/4 by 6 inches. With the long side toward you and seen from the bottom view, a ground bus is run along the bottom edge, then turns upward to follow the right-hand short edge. The VCC bus follows the top edge, turning downward 1.8 inches from the right end to create a vertical run that falls just short of the ground bus. (Hairpin loops, or your favorite techniques, are used to hold these buses in place.)

A small area at the right end, boxed in on three sides by bus wire, is reserved for the modulator. To the left of the VCC's vertical bus, the RF choke isolating the modulator is placed, along with a short run of bus wire which supplies the class C amplifier from the modulator.

The class C amplifier is built around the middle of the board. There is plenty of room under the VCC bus, above the transistor, for the tank circuit (the 220uH coil and 420pF compression trimmer).

The oscillator is built at the other end of the board. I was able to squeeze this circuit onto the lower edge of the board, leaving room for a 9-volt battery to lie horizontally along the VCC edge.

Parts layout is not critical. However, attention should be paid to the orientation of the various coils and chokes. As can be seen from the parts list, the two tuning coils are Mouser RF chokes (220uH, 220 microhenries), and the three actual RF chokes are 10mH Mouser chokes of identical style. They all have radial leads; that is, their two leads emerge from one end of the package, and they are made to stand on end. In order to minimize coupling between the circuits, the coils in the two tuned circuits are made to lie horizontally, and are at right angles to one another. In other words: All the RF chokes (all 10mH units) stand up on the board. The class C amplifier's tank coil is lying horizontally, parallel to the long edge of the board. The oscillator tuning coil is lying horizontally, perpendicular to the long edge of the board.

Circuit Operation

Although a Colpitz oscillator was tried with great success, this necessitates a tunable inductor, and those are expensive. A Hartley would have served as well, but this requires a tapped inductor; even though it can be a fixed inductor, tapped ones are hard to come by. So, we chose the Clapp oscillator--which is a modified Colpitz.

It was decided not to modulate the oscillator, but to apply good-ol' high-level collector modulation to a Class C amplifier.

The class C amplifier is highly tamed. The emitter resistor keeps it from self-destruction in the off-resonant condition (although a 2N2222 cannot dissipate enough power to stand this, and a 2N2219 was used instead). The 15K resistor across the tank circuit lowers its Q and minimizes the amplifier's desire to oscillate on its own (although it will do so, especially when tuned to a higher frequency than the oscillator).

The amplifier is resonated by tuning for minimum collector current. Our prototype transmitters use a sensitive LED as the indicator; you tune for minimum brightness. For the editor's purpose, a jack was wired in parallel with the LED so that I could make this adjustment using an audible current meter. (If all you have is a talking meter, have an adjustable afternoon.)

Use of a modulation transformer was deemed too elaborate; instead, the collector supply of the class C amplifier is simply the output of the modulator. The modulator is a 2-stage affair whose output is an emitter follower. Notice that the emitter follower is "in the loop"; the gain of the second stage is determined by the feedback resistor which comes from the emitter back to the inverting input of the op-amp involved.

Because of "motorboating" in the modulator, the first stage was made non-inverting. Building the modulator, in total, as an inverting amplifier eliminated the motorboating.

Making the first stage non-inverting was not without its cost, however, since this requires that a solid bias supply be available. Thus, a quad op-amp (LM324) was used; one section is used as a follower to create a rigidly fixed bias point, two sections are used as amplifiers, and a remaining section is not used. (We may decide to make an audio oscillator out of the remaining stage to "beep" the swimmer.)

From the microphone input, the total gain of the 2-stage modulator is about 900. From the high-level input, the gain is 9.


Power Supply Considerations--The transmitter can be powered by 9 or 12 volts. Because my supply lines on the board are long, two 0.05uF bypass capacitors go from the VCC bus to the ground bus; one is near the oscillator, and the other is near the modulator. Also near the modulator is a 10uF electrolytic bypass from VCC to ground (the negative end is grounded). A 1/2-VCC bias point is created; the VCC line goes through two 47K resistors in series to ground, with the junction of these resistors being bypassed by the parallel combination of 0.05uF and 10uF (the negative of the electrolytic at ground).

Clapp Oscillator Circuit

A 2N2222 is used. The base is biased by a voltage divider; the VCC line goes through 120K, then through the parallel combination of 100K and 0.05uF to ground, with the junction of these resistors going to the base of the 2222. The collector goes through a 10mH (ten millihenry) RF choke, then through a 2.2K resistor to VCC. The collector also goes through 470pF to the emitter, with the emitter going through the parallel combination of 1000pF and 2.2K to ground.

The collector also goes through a 220uH (220 microhenry) RF choke (used as a tuning coil) to the stator of a 400pF variable capacitor (see parts list); the rotor of this cap is grounded. The output of this oscillator is the emitter.

RF Amplifier Circuit

The emitter of the oscillator transistor goes through 100pF to the base of a 2N2219; this base also goes through a 10mH RF choke to ground. The emitter goes through 100 ohms to ground. The collector of the 2219 goes to the stator of another 400pF variable capacitor (see parts list); the rotor of this cap goes to a modulated bus we shall call "Point M." In parallel with this capacitor is another 220uH RF choke (used as a tank coil). Also across the coil and capacitor is a 15K resistor.

Point M is bypassed to VCC by 0.01uF. Point M goes through a 10mH RF choke to the cathode of an LED, with the anode of the LED going to the output of the modulator (which is the emitter of a transistor to follow). This LED is shunted by 0.1uF (disc ceramic). So that you and I can tune the transmitter with an audible ammeter, an open-circuit jack of some sort is connected across the LED--perhaps a 1/8-inch mini phone jack whose mounting is insulated from any chassis.

The antenna, which can just be a long piece of wire, is coupled to the 2219's collector through 100pF (disc ceramic).

Modulator Circuit

An LM324 quad op-amp is used. Pin 4 goes to VCC, while pin 11 is grounded. Barring another use for it, the fourth op-amp in the package is unused; pins 13 and 14 are tied together, while pin 12 is either grounded or taken to the bias point (1/2 VCC).

The third op-amp, pins 10, 9 and 8, is used as a follower; pins 8 and 9 are tied together, while pin 10, the non-inverting input, goes to the 1/2 VCC bias point.

A closed-circuit mini phone jack is used for the microphone. The switch contact is tied to the sleeve, which is also grounded. The tip contact goes to the negative end of a 10uF electrolytic capacitor; the positive of this cap goes to pin 5, a non-inverting input. Pin 5 also goes through a 10K metal-film resistor to pin 8--the buffered bias point. Between pins 6 and 7 (inverting input and output) is a 100K metal-film feedback resistor shunted by 220pF. Pin 6 also goes through a 3K metal-film resistor (value not critical) to pin 8, the buffered bias point.

Pin 3, a non-inverting input, goes to the unbuffered bias point (which already has pin 10 going to it). Pin 2, the inverting input, goes through a 3.3K carbon-composition resistor to pin 7, the output of the mike preamp. Pin 2 also goes through a 10K carbon-composition resistor to the positive side of a 10uF capacitor. The negative of this cap goes to the hot side of an RCA phono jack, the shell of which is grounded; this is the high-level input.

Pin 1, the output of this op-amp, goes to the base of a 2N2222 used as an emitter follower. The collector of this transistor goes to VCC. The emitter goes through a 100K carbon-composition feedback resistor to pin 2, the inverting input; this feedback resistor is also shunted by 220pF. The emitter also goes through 4.7K to ground. The emitter is the output of the modulator; this goes to the anode of the LED mentioned earlier.

Tune-Up Procedure

In principle, the tuning capacitor of the oscillator is adjusted to the desired frequency; then, the RF amplifier tank circuit is tuned to resonance by adjusting its capacitor for minimum collector current (minimum brightness of the LED). There are two complications that arise in this simple procedure: First, before the RF amplifier is resonated, you will hear many false signals as you tune the oscillator. Second, the oscillator frequency will change slightly as the RF amplifier is adjusted.

Start with both capacitors at their maximum settings--all the way closed. With a nearby broadcast receiver tuned to a clear spot on the band, adjust the oscillator's tuning capacitor until you hear a signal. Next, "dip" the class C amplifier by tuning for minimum brightness (or minimum current through your audible meter).

If the signal in the broadcast receiver doesn't get weaker as you dip the amplifier, you've probably got the intended signal. Just to make sure, tune the oscillator to see if you find a stronger one; if not, put the adjustment back where you had it and adjust the frequency to precisely where you want it.

Minor adjustments of the frequency will not affect the class C amplifier very much, but it never hurts to dip the collector current now and then.

Parts List


  • 3--10mH low-current RF chokes, such as Mouser ME434-1120-104K
  • 2--220uH coils, such as Mouser ME434-1120-221L


  • 1--2N2219
  • 2--2N2222
  • 1--LM324
  • 1--Low-Current, or high-output, LED, Radio Shack 276-066 or 276-087, or Industrial Devices, Inc., 4308H1


  • 1--closed-circuit (shorting) mike jack
  • 1--open-circuit jack (insulated)
  • Appropriate Battery Connector
  • Perforated Board (2.2 or 2.3 by 6 inches)

LEDs are a puzzlement to me. The IDI (Industrial Devices) unit lists itself as "extra super bright" and states a brightness of 100 millicandelas. The two Radio Shack ones above are 500 and 2000 candelas, respectively. The candle-power rating has recently been formalized into the "candela," for heaven's sake.



This circuit is a low-power, frequency-modulated oscillator in the FM broadcast band (its frequency range being limited to perhaps the lower third of the band). A high-level output, such as the earphone output from a talking book machine, is enough to modulate it. With a long patch cord between it and the source (the patch cord being its antenna), it can be received from two or three rooms away.

Richard Oehm gave me this little transmitter ten years ago. He doesn't remember where he got it, but it's worth having. It has writing on it that says, "RPI 7021A," which suggests that it must have been a commercial product at one time.

It is mounted on a printed board measuring 1-3/4 by 2-1/2 inches. It has one transistor, a few components and a slug-tuned coil. The layout is probably not critical, since there are so few parts, and the choice of transistor is probably not critical either. A 9-volt battery supplies the unit. The audio input required is a high-level source, about 2V peak-to-peak; a speaker output can drive it if turned up fairly loud. Looking at the audio input, you can see how this might be modified to your liking.


The negative of the 9-volt battery is grounded. The positive goes through an on-off switch to the VCC line; this is bypassed in two locations by 0.1uF discs.

The hot audio input terminal goes through 47 ohms to ground. The hot input also goes through 47K, then through 2.2K to ground--a voltage divider that can obviously be changed. The output of this voltage divider goes to the negative end of a 5uF electrolytic, the positive end of which goes to the base of an NPN transistor (2N2923).

The base of the transistor goes through 100K to VCC, and this base is bypassed to ground by a 500pF disc. The emitter goes through 1K to ground. From collector to emitter is a 5pF disc. The collector goes through the parallel tank circuit to VCC.

This parallel tuned circuit is made up of a 10pF NP0 capacitor and the slug-tuned coil. The coil is 3 turns of heavy-gauge wire on a 3/8-inch diameter slug-tuned form. It is "space wound" with an over-all winding length of 3/8-inch.

Parts List


Ten years! I say, this has been ten years! I started this when I was 33 years old (I'm not saying how old I am now). This is an anniversary issue. Pick it up and smell it. Rub your face in it (unless it's a diskette) and celebrate with me!

What are some of the accomplishments? Dozens of chips have been documented for our use. Lists of books revealed that there actually is a lot for us to read out there, if we look for it. Instruments especially for our use--musical instrument tuners, compasses, beacons, dial markers, weather and medical instruments, soldering aids, and electrical tools of the trade--have been documented so that we can build them.

Consider the soldering aids alone. No fewer than seven inventions have been published here to aid blind technicians: the tubular solder guide by Mike Bhagwandas; a feeder system for it by Jean LeBorgne; a darning-needle solder guide by Frank Jeanmonod; the featherweight, foot-controlled iron by Bernie Vinther; a temperature controller for the Vinther iron by Albert Alden; the little fishpaper dam between closely spaced terminals by Bill Gerrey; and the idea of a wrist support by Dr. Irene Gilbert of University of California Medical Center. These inventions do not include techniques which have come by way of suggestion: wrapping solder around the Vinther iron (suggested by Bernie Vinther), using "resistance soldering" (as suggested by Richard Joy), and so on.

The Smith-Kettlewell Technical File has not merely documented these things, but it has done much to cause their existence. Working with you, dear readers, has fostered growth in our bag of tricks that allow us to accomplish what we wish to. By working with you, let me assure you that my bag of tricks has expanded as much as yours has. Thank you for your inventions and your "hints and kinks."

Along this vein, I would like to salute my idols in this band of readership. Please don't feel insulted if I leave your name out of this list--I know you when I trip over your accomplishments. I started out with idols brought to me through The Braille Technical Press: Gunderson, Swail, Cranmer, Benham, Heavner, ... But there are no fewer notables whom I have met through SKTF: Vispi Mirza, Mike Bhagwandas, Albert Yeo, David Plumlee, Jonny Ott, Luis Suarez, Jaraslav Slezak, Gordon Legge, Bob Norris, Bob Trottman, Boris Meshevtsev (in Frunze, USSR), ... and counting (again, please don't feel left out). To a one, you readers belong to a company of the best people.

Special thanks must go to secretaries who have put these issues together, and in front of whom I have staged occasional "brat attacks": C.J. Combs, Liz Auer, Cathy Maloney, and Ann Lewis.

The Smith-Kettlewell Board has backed me, encouraged me, and seen to it that this creation has not floundered. John Brabyn, our Co-Director of the Rehabilitation Engineering Center, has seen to SKTF's financial support through his own contributions. To the tip of our soldering irons, we thank you.

My career is due to the Rehabilitation Act of 1973, and to the agency that supports my engineering efforts, the National Institute of Disability and Rehabilitation Research, by funding the Rehabilitation Engineering Center which houses my lab.

Ten years? If this were a kid, it would have learned to read itself by now.

Have a glorious new decade, friends and cohorts.