Coil forms:  many of the projects I see  use a 2" piece of pvc pipe - hard to find and not easy to work with.  I use empty toilet paper roll cores; they're about 1 5/8"in diameter, can be easily worked, and can be prettied up with a little paint or shellac if you like.  Cost: nothing.  Gift wrapping paper sometimes comes on 2 inch diameter cores;  which make for a bit larger diameter but shorter coil.  Keep your eyes open when the xyl is getting ready to wrap for Christmas.
.
For the big diameter forms, I like to use an empty 40 ounce plastic peanut butter jar for kid projects.  The cap can be drilled to mount anything you have in mind, such as a capacitor, switch, or what have you, and serve as your chassis.  Thick walled cores are not generally recommended by the pros.  You can roll up poster board and glue it to make just about any diameter form you might need; rolled around a 40 ounce peanut butter jar gives you a nice 4 inch diameter coil. A couple of guys have found that the thin-walled pvc drain pipe, which is used in septic drain fields, works well and can be used to make hi Q coils.  To hold the wire in place for a "space" wound solenoid coil, you can use the double backed sticky tape on the coil form, and wind the coil with two strands of wire to get the spacing.  When the coil is wound, remove the extra winding.  While I use the peanut butter jars in some projects, I recently found that that is not the best material to use for a low loss form; the answer is found on the bottom of the container - look at the number in that little triangle (the recycling code) - the low loss materials are numbers 2, 4, and 6 while the lossy materials are the odd numbers.  Another way to test for potential core materials is to just pop them in the microwave for a minute or so - the ones that heat up the least are the ones to use.

Detectors:  Radio Shack sells 10 packs of 1N34A type germanium diodes for about $1.20, and they work fine.  I have seen praises of the 1N6263 diode, a hot-carrier type, but the side by side objective comparison against the 1N34 in volume 4 of the Xtal Set Society newsletter indicates that the 1N34A is better.  If you are reading this, you probably already know not to use the 1N914, which is even cheaper, but has the sensitivity of a Marine DI; unless you're into detector biasing and other forms of self discipline.  Actually, with strong locals, the silicon diodes such as the 914 may give you less signal distortion, but should be avoided for dx use.  I know there are other diodes and even transistor junctions that work pretty well, but I use the '34 because it is really hard to beat,  and is readily available at a good price.  Incidentally, not all 1N34s from Radio Shack are alike.  Actually, they are billed as "1N34A type", and may be any of a variety of germanium diodes. It pays to test them for high performance rigs; every now and then you will get one that is a cut above the pack; Radio Shack may not be carrying these packs in the stores much longer, by the way. The 1N914 is a silicon diode, by the way, and it and other silicon diodes are generally considered unacceptable, except as noted above.

Wire:  this is the tough one.  I usually find that for #20-#24 plastic coated wire, Radio Shack is also the best buy.  For  enameled wire, I go to Mouser and buy a 1-lb roll. One pound of #24 is about 700 feet,  costs a little over a penny a foot, and you can do some serious winding with that much;  handles well and makes pretty good Q coils. If you can find a local source of salvageable wire for antennas and grounds, such as someone in a rewind shop or involved in factory instrumentation, exploit them shamelessly.  To put a simple radio together along with materials for an antenna and ground, the cost of the wire can easily be the driving cost.  If you have a local motor rewind shop, they may give you a good deal on enameled copper wire, since they buy in bulk.

Earphones:  I get the high impedance ceramic earphones from Mouser in quantity.  For a one-off project, your local Radio Shack will special order one for about 3 bucks or so, and it should arrive in about a week.  Make sure it is the one they use in the hobby project kits.  If you are into more parts, you might try the low impedance stereo headphones, they run about 16 to 32 ohms, and use a cheap 8:1000 audio transformer to feed them.  I only mention this because I have picked up several of the cheap imported phones in my local dollar stores in the last few months, and they seem to work pretty well for an amplified rig ( I use 'em for copying cw)- probably won't fit the bill for a dedicated audiophile, which I'm not, and you do get to use both ears.  Since the transformer adds more loss, you might want to use a simple  amplifier of some sort to feed them unless you live near pretty strong stations, which I don't.  You will find that the cheaper hi-fi phones are usually not sensitive enough for xtal sets; you get what you pay for, and don't forget to check the sensitivity specs.  If you run across an unknown impedance earplug, and want to see if it is high impedance, or you just want to check a new plug to see if it will work, just stick it in your ear and then touch the leads together.  A good set will give you a nice distinctive click when you do this.If you want to go binaural, you can connect two of the crystal earphones in parallel; you lose a little signal compared to just one, but the ability to listen simultaneously with both ears makes up for it.  You might want to fiddle with the value of the 47k resistor to see if there is a value you like better.  Sticking the earplugs under a set of hearing protection muffs really cuts out the background noise.
    When showing off my crystal radios to small children (mine are 5 and 7)  I notice that they often have a tough time getting the plug of a crystal earphone into their small ears and so don't hear very well.  One solution that I ran across again a couple of days ago was to use the earpiece of a discarded telephone handset.  I had one in the junk box, and took the handset apart to view the guts, and found that it was a piece of cake to extend the wires from the earpiece to the radio, and it worked pretty well.  You can use a saw to separate the earpiece from the rest of the handset, since you are only using the top portion anyway, and you have a nice single headset, which even small kids can hold up to their ear.  Sensitivity isn't as good as a crystal plug, but is still fair, and is a lot cheaper than a fifteen dollar set of magnetic phones.  Of course, you can always use a couple of stages of amplification and a speaker, but then it is just another radio, and you lose the magic.  One of my students plugged in the earphone to an oatbox set I had sitting on the counter at school, heard a couple of stations coming through, and then asked where the radio was. hi.  Incidentally, I think the standard telephone earpiece impedance used to be around 600 ohms.  A word of caution, however; I tore into another cheap handset with all the electronics in it because it was not working, and found that it had a little one and a half inch 8 ohm speaker - guess they did any needed impedance matching and amplification with the electronics.  The telephone handset mentioned above showed 165 ohms dc resistance, and worked fine; even a little better with an audio output transformer from RS.  Put it in my daughter's crystal convertible, and she likes it.
    For the older headsets with excellent sensitivity, I have used Scott's Crystal Radios for a source - good quality and good prices.  He occasionally gets in a couple of sets of surplus sound powered headsets, and these are almost impossible to beat for sensitivity.

Chassis:  The Oatbox radio can be built on just about anything, including an empty oatmeal box.  My personal favorite is the 40 ounce plastic peanut butter jar, but my students and I have used rolled tag board, and empty 2 or 3 liter soda bottles as well.  For projects needing mounted components, such as a polyfilm variabl capacitor, I have used stiff cardboard boxes, as long as the supply lasted.  Then I noticed some scrap 1 x 6 wood in the trash bin at a construction site, and some 3/4 x 5 pallet slats that had been tossed, and with a little sanding and some stain even, they make nice bases when cut into 5" lengths.  For panels, I found some galvanized 12" roll flashing, left over from the last termite cleanup on my house, and this makes a very nice front panel.  With a dollar can of cheap spray paint, it can make your rig look more hand-crafted than homemade.  I intend to use the flashing to make enclosures for a couple of ham QRP rigs I have on my to-do list, and my students have used it for pop-pop boat hulls as well.  If you have to buy it, a roll of flashing, enough for projects into the next millennium, costs just a few dollars, and can be cut up with a sturdy pair of scissors if you don't have tin snips. .  One more thought - you can disturb the Q (efficiency and selectivity) by placing them near chunks of metal - upsets the flux pattern, so make sure the coils are at least a couple of coil diameters away from metal panels.  Another nice, cheap chassis material is wood panelling.  It's only about 3 ply, and about 1/8 " thick, and cuts easily.  Only problem with it is that you don't go screwing stuff into it- solution? Just use a hot glue gun to stick everything to the board.  You can securely glue just about anything to a wood base with the hot glue.  For something a little less permanent, or which might be damaged by heat, just use the double sided sticky tape.

    Here's another neat trick, particularly when making simple transistor circuits.  Make a pseudo printed circuit board.  I cut small strips of brass hobby sheet (or copper if you can find it), about 0.025 " thick, into small strips, and glued them to a wooden base with either barge cement or even white school glue.  You can readily solder to it, and it is easier than trying to fit small parts into a terminal strip.  You may be able to find copper strips with a sticky side, and this works fine too.  Dennis Foster, KK5PY takes another approach to pseudo circuit boards.  He takes a single sided copper covered pc blank, punches out little pads  from the blank, then glues them onto his pc board.  The board is chassis ground, and the pads are used for the other connections; neat.

Need a box?  Here is a small one which is easily cut out of the galvanized flashing or any other thin sheet metal.  Scale it up for larger enclosures.  Ed Ganger sent me a few words to indicate that the cheap galvanized flashing I use may be superior in some cases to aluminum, particularly in a set where you don't want stray rf in the set to find its way back to another part of the circuit.  Seems that aluminum, being less lossy, tends to let the rf wander about, while the  iron chassis will absorb the rf and dissipate it in the form of heat.  Since rf will take off in a straight line until it is reflected or otherwise caught, perhaps those early homebrew breadboard rigs with no enclosure were a pretty good idea (but watch what you touch when using high voltages)

Resistors, fixed capacitors, variable resistors, alligator clips,  and even blister packs of general purpose transistors are usually available at a good price at Radio Shack.  Whenever I want to make a bulk purchase, however, I always check out Dan's  catalog on the web first.  The rule with Dan's is to check early and often.

Variable capacitors of the preferred type (air variable) just aren't available on the cheap, unless you are a diligent flea marketer.  If you want more than a couple, you might do as I do, and look for deals in the catalogues on the polyfilm rf tuning capacitors.  Finding them is still a challenge, however; I don't think they ever really made a 365 pf poly varicap - even the Radio Shack kits only have one between 150 and 260 pf.   Only occasionally will I run across these babies with tuning knobs, so I usually make my own out of plastic soda bottle caps, cutting a hole to fit, and securing it to the capacitor with a washer and screw.  Mouser has screws which can be used for both mounting the capacitor and for securing your homemade tuning knob.  The Xtal Set Society has both single and dual 365 air variables, as well as just about anything else you need.

    About 80% of the designs I see out there on the web tap the detector at the top of the tuning circuit.  However, most of the articles I see on design philosophy indicate that somewhere down on the coil is a better place to tap off the detector; increases selectivity and possibly sensitivity because of less tuned circuit loading and a better impedance match.  I usually find that my typical detector setup, 1N34A germanium diode, and 47 kohm resistor across a high impedance crystal earphone, works best when it is tapped about halfway up the tuning coil.  This is both by subjective (listening) and objective (digital millivoltmeter across the 47k resistor) analysis.   A second tap about 10 turns from the ground end of the tuning coil can be used when you need a little more selectivity (and a little less sensitivity).  Instead of a tap on your tuning coil, you can wind a detector coil over the tuning coil, using about half as many turns as are on the tuning coil.   This often improves selectivity a bit, but is not quite as sensitive as directly coupling the detector to the coil, and is tough to vary, although you can loose-couple it to the tuning coil, and get some variability, I suppose.  If you want to go for the best arrangement, invest in a matching transformer to match the detector/headphone circuit to the top of the tank.  Go to Ben Tongue's pages to see how to do this.  Also see what Ben Tongue and Al Klase recommend about tapping to the top of the tank circuit, where the higher signal voltages (to overcome the detector barrier) are found - it works.

ABOUT WIRE:

     In some of the  literature out there, I have run across several suggestions or outright declarations on the superiority of Litz, or rather stranded, wire over solid.  Having seen Litz wire used in a number of applications, and, drawing on an imperfect memory about the movement of electrons, I decided to pursue this a bit further, at least on paper.  One assumption - 20 gauge is 20 gauge be it stranded or solid, an easy assumption, particularly when you see that the cost of solid and stranded are generally the same.  This means that the volume of metal in a given length of wire is the same, be it solid or stranded and that the cross-section of a given gage is the same for each type.  Why the fuss?  Because the higher the resistance of a wire, the lower the "Q" of an inductor (coil), and thence the selectivity of a circuit.  Since the Q of a coil is the coil Reactance divided by its Resistance, Q can be a hot topic with guys using simple tuning setups.  Resistance also causes signal loss; no surprise here.  Anyway, I dug into my physics books and located a table on the properties of copper wire.  Bottom line; the dc resistance of stranded and solid wire of a given gauge are the same (see assumption above on cross section).  Still, something nagged me about skin effect, or rather the tendency of electrons to run on the outside of wire, which would make surface area of the outside of the wire a greater factor.  A little math further showed that for a given total cross-sectional area, the surface area of stranded wire increased as the number of strands increased.  An old electronics text  discussed skin effect in transmission lines, but stated that the phenomenon  occurred at very high frequencies (high being undefined, but it was clear that the VHF/UHF regions were the areas of concern).  My take?  At the frequencies at which crystal sets are operated, stranded wire may have no advantage re circuit Q.  It is more flexible, and hence easier to work with, and also will not open the circuit if a single strand breaks, but otherwise is no better than solid.  Having said that, Ed Ganger, KB4ORA, finally had enough of my dithering and sent this, which looks pretty definitive, so read on:
"litz wire is stranded, insulated and has a bizarre twist to it that forces the RF current to evenly divide amongst the strands.Each wire spends some length on the outside, and some length inside. The skin effect (by personal knowledge in switching power supplies) can be a significant problem as low as 100KHz. By using multiple windings of smaller gauge wire, the DC resistance goes up a little; the RF resistance goes down a lot."
    As the results from the y2k dx contest came in, we saw a renewed interest in litz wire.  Mike Tuggle used it for his very high performance xtal set - good enough for 2500 mile contacts.  He used basket coils of 200 strand litz.  I used some 420 strand litz for basket coils on my last competition rig, and broke the 100 contact barrier.

    Wire resistance approximately doubles for every three gauges smaller.  Since wire usually comes in steps of 2 units of gauge, the factor is about 1.6 for every two steps.  Have a supply of aluminum wire you'd like to use for an antenna?  Fine, go ahead, but realize that aluminum is about 1.6 time as resistive as copper; in other words, #20 copper has about the same resistance as #18 aluminum.  Want to use that old wire fencing?  Now you're talking about 8 gauge numbers different (about five times as resistive) .  My feeling is that the worst wire in a good location is better than the best wire in ....In the antenna world, location and resonance generally override other considerations.  My summer project, some summer,  is to use some electric fence wire to make a beverage antenna.

COIL WINDING:
    Rather than repeat what is already out there, I'll refer you to Harry Lythall's site; CALCULATING COILS FOR HF  He has some thumb rules on coils and capacitors in tuned circuits for the mathematically challenged as well as the formulas for those who aren't.  Bytemark's site has data on winding toroids, but I haven't found anything definitive about winding ferrite rods; use Harry's thumbrules then cut and try, I guess.
Ed Ganger's page  is shorter, and covers air core coils nicely.  By the way, the conventional wisdom is that the best coil lengths are between one and two times the diameter of the coil.  I did check a couple of ferrite core coils with an inductance meter; removing the ferrite core dropped the inductance of the coil by a factor of about 10; length of the coil and core seem to affect this, so some cut and try is needed.  Still, this will get you more inside the ballpark.

TANK CIRCUIT CALCULATIONS:  When making a resonant circuit  I find that the formula for calculating coil inductance as Ed Ganger shows on his page to be very reliable.  To calculate the expected frequency of a coil and capacitor combination, you can use this equation with confidence:

F = sqrt (25,300/LC)  where F is frequency in MHz, L is inductance in uH (microhenries), and C is capacitance in pF (picofarads = mmf on the old schematics).

In plain English, divide 25 300 by L and C, and then take the square root to get the frequency of the circuit at resonance.
By using the range of values of a variable capacitor in the equation, you can calculate the range of frequencies the circuit will cover.  Generally, since the coil will have a bit of distributed capacitance, your calculated frequencies will be a bit high, but it's no big deal.

One other trick, if you will, when doing tank circuit calculations.  Divide the lowest frequency of interest by the highest frequency of interest, then square the result.  This gives you what the ratio of your highest to lowest capacitance must be to cover the frequencies.  For example, say you are using a capacitance with a maximum value of 20 pF, and it is for a circuit to cover 2 MHz to 4 MHz.  The frequence ratio is 2, the square of that is 4, so your capacitor must go from 20 to  5 pF to cover the frequency range.  If you are using variable inductance instead, the same trick works.  Normally, we try to get a capacitor with a little larger tuning range than the calculated minimum to take care of variations due to construction of coils, stray capacitances, and the like.  My 20 to 400 pF tuning capacitor from the XSS covers the broadcast band in my set with a little room to spare.

I got tired of hooking up two of the 200 pF polyfilm variables to get full BC band coverage, so got out my calculator and figured that 150 turns of #24 enameled wire on a tp core would get me right at 400 uH of inductance.  The resulting coil with a single polyfilm cap gets me just about the whole band, but the coil length to diameter ratio is just over 2:1.  Smaller wire, say #26, will give you a better ratio, but Q goes down.  Okay, so use a bigger diameter core, right?  Well, I like to see what I can get out of the tp cores, since they are readily available, and are fine for student projects.  For my personal rigs... well, let's just say that 3 inches diameter is sometimes small enough.

On the subject of coil wisdom,  you really need to go to Al Klase's  page.  'Nuff said.

BANDSPREAD THAT CRYSTAL SET: HERE'S HOW.

CONSTRUCTION TIPS I HAVE RELEARNED:

    Think big when building something to experiment with.   There are few things more frustrating than trying to fish a soldering iron into a tiny slide switch contact to undo one solder joint and make a new one.  I often build a finished product- neat, compact and nice looking, only to decide later to modify it.  Big mistake.  If your latest wonder rig works fine, but you want to fiddle with it, give it instead to some kid, rig an antenna and ground for them, and start over.  If you want to play around with different coil/capacitor/ tap arrangements, etc., make yourself a real breadboard and use it for experimenting.  It might look plug ugly with some arrangements, but you can always clean it up when you have a design you want to put to bed - on a new chassis.  I think a good breadboard should have a couple or 3 variable capacitors mounted, maybe a couple or so slide switches mounted, and all with leads soldered on and ready to connect to.  Some fahnestock clips placed around the perimeter are nice to make semi-permanent connections, at least one each for antenna and ground, a couple for headphone tip jacks, and what have you.  A board about 6 by 8 inches should give you room to work without having a slew of off-board components.  Use insulated hookup wire for connections.  Mount some of the stuff on a sheet metal or plastic front panel.  If you like to play around with different detectors, changing detector direction etc., make some provision for doing that.  And oh yes, get a bunch of 12 inch stranded wire leads with alligator clips on both ends - makes it easy to shift connections around.  Put alligator clips on your antenna and ground leads too.  If you want to play around with simple transistor circuits too, get one of the Radio Shack Bunchinone project sets, and modify it as you see fit.
    When soldering, tin before you make connections, particularly when working in tight spaces.  Of course, don't forget to make a decent mechanical connection before you solder either.  Trim your leads to the desired length before soldering.  Trying to snip off excess wire after you have made a connection can be, well, trying.  If your chassis is of non-conductive material, and you can poke a hole through it, you will find that the brass paper brads from the stationary store make fine soldering posts.  On my oatbox and peanut radios, I also use the brads to secure fahnstock clips to the chassis, which are then used for antenna or ground connections.
    Another nice little trick for soldering pads came to my attention recently:  brass coated thumb tacks.  They can just be pushed into a wooden chassis and soldered to.  I saw them in the dollar store for a buck for a box of 100 or so.
    For the tough soldering jobs, such as to large pieces, heavy stranded wire (that is hard to get shiny), and even galvanized iron, spread a little rosin soldering flux paste on first.  You'll be amazed at how the solder gets sucked up.

THE SEARCH FOR THE ELUSIVE MULTI-BAND CRYSTAL SET:

     Try as I might, I have never been satisfied when building a set to cover both the BC and the HF bands with one tank coil.  Part of this probably due to a  dead end effect, which reduces the efficiency of the set.  In my case, whenever I try this, my local bandmaster will bleed into the SW bands.  When I tie two coils into the same set, this bleed over works both ways.  One of my problems, I suspect, is that with two coils, I try to cheat in the switching, and so at least one side of an unused coil is connected to ground or something.  I am convinced that this just sets up a sneak circuit which is resonant somewhere, and it feeds the detector circuit somehow.  Since we are not dealing with a highly selective circuit to begin with, failure to completely disconnect an unused component that can be resonant at any frequency at all is a setup for the law of unintended consequences.  The way I see it, there are four ways to attack this problem.  The first is obvious - build only single band rigs.  Number two is to use plug in coils.  Number three is to completely switch out all of the unused coils, including grounds, close inductive coupling, stray capacitive coupling, etc.;  physical separation and judicious orientation will help.  The last way is to reduce the unwanted signal before it reaches the set, or at least the detector.  Resonant antennas, antenna tuners, and wave traps fall into this category.  If you aren't 100 percent successful, you are not necessarily a complete doofus.  Even the latest superhets (and stereos, and TVs) catch a stray signal every so often.  Just try to make your set as efficient on each band as possible, then see if you can keep that efficiency when you try for band flexibility.

DETECTOR STANDS FOR ROCK HOUNDS

    For some people, the thrill of crystal sets is being able to listen to the radio using an actual rock.  In the early days of radio, some pretty high powered talent was engaged in finding the best materials for radio detectors, something most of us seem to take for granted since the advent of the tube, the transistor, and high tech diodes.  The most common materials still in use are galena,  which is lead sulfide, and iron pyrites, otherwise known as fool's gold.  Both of these can be found in the boutique type toy stores and a few other places that sell mineral samples.  Two sites I have on my links page, G3XRT and Gnome, have some info on stands and materials.  Larry Pizzella, who helped me sort out the RS crystal set, has built some stands and has devised a couple of clever ways to mount the crystal.  GO HERE to take a look at one and also at a neat crystal set he built.  Al Klase  has another nice little detector stand on his page.
 

    As I said at the beginning, I am usually trying to kit my sets for a dozen or so kids, so cost and ease of construction are real drivers for me.  If you have any ideas or sources I can use, please drop me a line.