INTRODUCTION

      Welcome to Crystal Set Design 102.  It is assumed that you've already completed the 101 level course elsewhere, and know, at least, something about electronics in general and crystal radios in particular.  If not, hit the books. The stuff in your public library will get you started.   The early chapters of the "Radio Amateurs Handbook" are especially concise and approachable.
       I've been a little disappointed at the lack of meaningful crystal set technical information on the web and in current literature.  This is my attempt to at least partially remedy this situation.  This work is the result of about ten years part-time investigation of crystal radios from an engineering perspective.  Serious development of passive receivers pretty much came to an end with the introduction of reliable vacuum tubes around 1920.  A lot of crystal sets, both commercial and home brew, have been designed in the interim, but most are mediocre performers.  So the mission turned out to be one of rediscovering the the secrets of the age when spark was king.

GETTING STARTED

    My approach is to build crystal radios out of quality vintage radio parts.  They are available in great profusion at amateur radio "hamfests" and antique-radio meets (and in my garage).  If you don't have access to these sources, I'd suggest you start at Radio Shack:  Buy their Crystal AM Radio Kit, 28-177, for $6.99.  This is actually not a bad crystal set, and it contains a coil, a variable capacitor, a germanium diode, and, most importantly, a reasonably sensitive high-impedance ear phone.  You're also going to need an antenna.  This generally means wire up in the air.  The attic may be the next logical alternative.  Apartment dwellers may be in trouble unless you're near an AM radio station, or can arrange a "stealth" antenna of some sort..  Again, if you don't (yet) have a junk box, get Radio Shack's Outdoor Antenna Kit, 278-758, for ten bucks.  Wire has always been expensive, and hard to find retail, so keep your eyes open for bargains.  Another RS item, that's almost indispensable, is a set of mini-alligator jumper cables, 278-1156, 10 for $3.99.  These are how you make temporary "breadboard" hookups while experimenting with new circuits.   You'll also want basic electronic hand tools and a soldering iron.
 
CIRCUITS
 

The simplest radio you can build is just a diode detector and a headset.  With a reasonable antenna and ground you will hear the strongest stations, albeit all a once.  This is not much of a radio. but it will give you some indication that you have enough signal strength to continue experimenting.

 
 
 

The primary problem with the above set is that it offers no selectivity.  We'll solve this problem by building a tunable filter.  This will generally consist of a coil and a capacitor forming a tuned circuit.  Either the capacitor or inductor, or both need to be variable so circuit can be tuned to different stations.  The classic values are 250 uH and 365 pF to tune the broadcast band.  This is the basic crystal set schematic you'll find almost everywhere.  It works better than just a diode, but has some serious short comings that are easily remedied.

 
THE IMPORTANCE OF IMPEDANCE MATCHING

    In a crystal set, all the audio power that arrives at your ear drum came from the distant transmitter.  If the transmitter is hundreds of miles away, the amount of power captured by even a good antenna is reckoned in nanowatts .  At each point in the set we must strive to transfer at least a reasonable percentage of the available power to the next stage.  Perfect impedance matching is not necessary,  don't fret over a 2 to 1 mismatch, but let's eliminate as many of the 10 to 1 and 100 to 1 mismatches as we can.

SINGLE TUNED SETS
 

In the case of the simple set in the last example, lets do two things:  Tap the antenna input "down" on the coil.  The impedances of the antenna and tuned circuit will vary with frequency, so it's a good idea to provide multiple taps and a selector switch or movable jumper.  As a starting point tap the coil at about 5%, 10%, 20%, and 50% of the total number of turns from the ground end of the coil.  Secondly, connect the detector to the 50% tap.  This does two things, both of them beneficial:  It provides a better match to the detector when it's connected to the usual sort of crystal set headphone that has an impedance of about 10K ohms, resulting in a louder signal.  It's also reduces the loading on the tuned circuit, increasing it's Q and consequently it's selectivity.

 
    These improvements result in a better than average crystal radio.  With 50-75 feet of wire up in the air, you should hear daytime 50KW stations out to 40-50 miles, and night-time skywave stations will come in form hundreds of miles away.  This is essentially the circuit I arrived at for my Cub Scouts a few years back.  (See The Den Two Crystal Radio in Crystal Set Projects published by The Xtal Set Society)  You'll also notice it's almost exactly the same circuit used in the Radio Shack set
 

Another effective way to match the antenna to the tuned circuit is to use a variable capacitor as shown in the drawing.  A cap in the 300-500 pF range is appropriate.  Tune in a station then change the coupling and retune and see if there's an improvement.  Some sets have the coupling capacitor in the ground lead instead of the antenna lead.  Electrically it's the same thing, but sometimes it's more convenient for mounting and grounding and eliminating hand capacitance effects.

 
 

With the previous circuit, connected to a good sized antenna, you'll find you can set the tuning capacitor to it's minimum value, or even remove it completely and still get good performance.  To understand how this "series tuned" receiver works, I've inserted the equivalent circuit of the antenna in the drawing.  Any Macroni antenna less than a quarter wavelength long appears to be a capacitor in series with a small resistor, known as the radiation resistance of the antenna, and an RF voltage source. This is almost always the case as a quarter wavelength, even at the top of the broadcast band, is 187 feet.  Our aim is to make as much RF current as possible flow from our antenna.  When the series value of the antenna and tuning capacitors and the inductor  are tuned to the frequency of interest, the inductive and capacitive reactance's cancel out, leaving only the DC resistance of the inductor.  This results in maximum current flow in the tuned circuit, and maximum voltage applied to the detector.  A receiver of this sort will usually require a variable inductor to cover the entire broadcast band. In such cases, keeping the detector connected to the optimal point on the coil presents challenges.

 
 

One classic variation on the series tuned receiver, known as the "two-slider tuner", dispenses with the tuning capacitor entirely.  Instead a sliding contact on the coil varies the inductance in the antenna circuit.  A second slider connects the detector at the best point on the coil.  Total inductance should probably be the better part of a millihenry.

 
 

Yet another, historically common circuit, connects the detector in series with the variable inductor.  This is not such a great idea from the standpoint of impedance matching, but it is simple.  Series tuned sets all have difficulties with short antennas, because their capacity is low.

 

Another way to implement the series tuned set, that improves the match to the detector, is to connect the detector and headset across a second coil in the series circuit.  For the broadcast band, this coil has a fixed value of approximately 80 uH.  The variable capacitor is optional, especially if the larger inductor is continuously variable.

                Single-tuned crystal sets, whether series or parallel tuned,  leave a lot to be desired in terms of selectivity.  Yes, the nose selectivity can be improved by increasing the Q of the tuned circuit, but the skirt selectivity remains hopelessly broad.  The obvious solution is to use more than one tuned circuit.  The point of diminishing returns is between 2 and 3 circuits for a crystal set due to cumulative losses and tuning difficulties.

THE TWO-CIRCUIT RECEIVER
 

The classic solution is the "two circuit" tuner.  The antenna circuit is series tuned by a variable capacitor and an inductor, while the detector circuit is connected to a parallel LC circuit.  The amount of mutual inductance, or coupling, between the primary and secondary circuit is generally made variable.  This allows light coupling to be used to obtain the sharpest tuning, while increasing the coupling increases sensitivity at the expense of selectivity.  This is essentially the same architecture used to great effect in  the communication receivers of the wireless era.  If you're seeking better performance for your crystal set this increased complexity is well worthwhile.

 
     There are several way to implement variable coupling in the two-circuit set.  The navy style "loose couplers" used a secondary coil that telescoped inside the primary.  Other possible schemes include variable taps on the low end of the secondary coil, link coupling between the two coils, or the use of a small variable inductance common to both circuits.
 

Variable coupling schemes for two-circuit tuners.

ADDITIONAL DETAILS

     The primary inductance will want to have a maximum value of about 500 uH to reach down to 530 KHz.  However, a lesser value in needed to  reach the 1600 KHz end of the band.  I've been building coils with five or six evenly spaced taps.  This lets you tune around for an optimum match to the antenna.
     Install a DPDT switch to allow the primary circuit to be operated in a parallel-tuned mode:  The input end of the primary capicator is grounded, and the antenna attached to the top of the coil.  This allows effective operation with short antennas.
     If a non-fixed detector is used in a double-tuned circuit, it's a good idea to include a buzzer to generate a local signal to adjust the detector.  The circuit is a low-voltage mechanical buzzer, a battery, a push-button switch, and a one-or-two-turn link to the secondary coil.  A low voltage relay with it's normally-closed contact wired in series with the coil is a good substitute for a buzzer.