K7ITM wrote:
Huh? You wrote, "if tight winding results in a lower Q/other
effects, why space the windings for air-core, crystal radio coils,
period?" Do you not want a higher Q? Generally, people try for the
highest unloaded Q they can get, under some set of constraints.

Sorry, I made a typo. Rather, why do people tight-wrap coils, period?
Just ease of construction?

Close spacing lowers the Q mainly because the current in the wire is no
longer radially symmetrical, if you look at a cross-section of the
(round) wire. That raises the RF resistance of the wire. For decent
(low-loss) form material, it's mainly the RF resistance of the wire
that determines the loss and therefore the Q. Generally, highest Q for
a given diameter and length is obtained by spacing the wire about two
wire diameters, center to center, at least for high frequency work. If
you want to use Litz wire, there's an optimum stranding...more, finer
strands are not necessarily better as you get to either lower or higher
frequencies. You should be able to find info on that, if you do some
searching.
SNIP
Cheers,
Tom

Thanks for the information! Is there an advantage to wrapping cylinder
coils as opposed to spider or torroid, other than ease of construction?
I'm debating which one I'd likely get the best result with... I'll
probably make both to try it out, but I'd like to know which one would
'probably' work best.

Thanks again,
Dave

dave.harper wrote:
-ex- wrote:



From what I've read here and elsewhere, I'm debating either making a
spider coil or a tight wound 4" cylinder ( about 1" length). Is there
any significant advantage to either?

They're going to be very similar in performance. If you do the solenoid
(cylinder), go with one wire spacing between turns instead of tight
wound. Thats been pretty well proven to give a bit better Q. And 16-18
ga wire is also in the 'best' range for both the spider-web and
solenoid. Next step up would be some serious litz wire.


I can see with a powered ferrite core how the spacing would make less
of a difference... but if tight winding results in a lower Q/other
effects, why space the windings for air-core, crystal radio coils,
period?

I'm not 100% sure I understand the question....On a
rook/basketweave/spider coil there's inherent spacing already. And a
solenoid coil will also do better that way...on larger coils. The way I
understand it is that there are several factors at work - primarily
interwinding capacitance and overall l/d ratio. In practice, as you
change one you also change the other. You can compensate one for the
other somewhat with a different diameter coil, different gauge wire, etc
but the general concensus is that 4"/16-18 wire/~1 wire diameter spacing
is pretty close to the best you can squeeze out of that class of coil.

HTH.

-Bill

I s'pose Reg, the local expert on proximity effect, etc., should pop in
here and 'splain it all. Seems, though, like it's wrapped up in
practicalities. For low-frequency work you typically want a lot of
inductance, so you use fine wire so you can get a lot of turns in a
relatively small volume. The wire diameter is small enough that, at
low frequencies, the skin depth is large compared with the wire size.
I believe you will then find that the proximity effect won't have as
much influence on the Q as in the case where the skin depth is a small
fraction of the wire diameter. So for a 50Hz/60Hz power transformer,
you won't find the turns spaced apart any more than needed for
insulation.

You can do a Google search for conductor proximity effect and find a
bunch of references. The stuff at
http://www.national.com/nationaledge...c_article.html has some
nice pix to show the effect in a bit different environment than we're
talking about here.

The Q you actually obtain may depend on so many other things than just
the shape of the windings that it's not possible to tell you the "best"
geometry. But I can tell you that if you make a large coil of good
design, you should be able to get to a high enough unloaded Q that
doing better with a different geometry about the same size will give
you only small returns on the performance in the circuit. That is, if
you do manage to make a solenoid coil say 5 inches long and 5 inches
diameter, maybe getting the Q up near 500 if you're careful, then
operating it at a loaded Q of 100 (for a 10kHz bandwidth at 1MHz), the
loss in the coil compared with an INFINITE unloaded Q is so small as to
be nearly unnoticable. If my mental arithmetic is right, it would be
about a 1dB difference, just barely audible. And of course, you won't
get anything like that much improvement in Q with a different shape.
Plus--the standard solenoid shape is easy to construct! (There ARE
reasons for wanting higher unloaded Q, if you want to operate at a
higher loaded Q and if you want to build a multiple-resonator tuner,
but my impression is you are not there yet!)

Cheers,
Tom

dave.harper wrote:


K7ITM wrote:

Huh? You wrote, "if tight winding results in a lower Q/other
effects, why space the windings for air-core, crystal radio coils,
period?" Do you not want a higher Q? Generally, people try for the
highest unloaded Q they can get, under some set of constraints.


Sorry, I made a typo. Rather, why do people tight-wrap coils, period?
Just ease of construction?


Close spacing lowers the Q mainly because the current in the wire is no
longer radially symmetrical, if you look at a cross-section of the
(round) wire. That raises the RF resistance of the wire. For decent
(low-loss) form material, it's mainly the RF resistance of the wire
that determines the loss and therefore the Q. Generally, highest Q for
a given diameter and length is obtained by spacing the wire about two
wire diameters, center to center, at least for high frequency work. If
you want to use Litz wire, there's an optimum stranding...more, finer
strands are not necessarily better as you get to either lower or higher
frequencies. You should be able to find info on that, if you do some
searching.

SNIP

Cheers,
Tom


Thanks for the information! Is there an advantage to wrapping cylinder
coils as opposed to spider or torroid, other than ease of construction?
I'm debating which one I'd likely get the best result with... I'll
probably make both to try it out, but I'd like to know which one would
'probably' work best.

Thanks again,
Dave


Me again...I should have read the later threads before my earlier reply.

Tom is correct about the Q using litz. Some guys have tried 48 ga litz
and said it nosedived in performance from the more-common 46 ga litz.
I've seen that explained with a critique of skin depth in that the rf
resistance of 48 is considerably higher at those freqs. Strand count
seems to still be in the 'more-is-better' range at BCB. 660-strand is
commonly used in DX sets...although I haven't graduated to that level of
expenditure myself

As to which to try....in a single-tuned set you won't notice the
difference. If you have a strong local BCB station the toroid will do a
very effective job of decreasing direct pickup by the coil. You might
still want a trap inline, though. Guess which type of coil makes the
best trap in this scenario!

My own tests gave a slight nod to the spider web coil. Not enough to be
noticeable in reception but enough for "spec-talk". The spider-web is
also less prone to proximity effects and even direct pickup because it
is 'directional'.

My own dx set which is admittedly tailored for my particular environment
uses a toroid on the first tuned stage, a toroid inline trap, then a
spiderweb on the detector stage with a loose coupled trap made with a
loopstick ferrite. A little of each, huh? I can receive stations
within 80-100 kc of the 5kw local that is 1/4 mile away on 1370.

My best recommendation would be to seriously consider a double-tuned
set. Its a whole different world than a single-tuned one.

-Bill

K7ITM wrote:

There is such a thing as TOO HIGH a loaded Q. Let's say you start off
with a coil with unloaded Q of 500, and couple lightly to it with your
circuit (antenna and detector), so the loaded Q is 250. That means the
bandwidth at 3dB points, if you tune a station at 1MHz, is 4000Hz. If
you've tuned to the center of the station, your demodulated bandwidth
will be only 2kHz. Since the rolloff is gradual with a single-tuned
circuit, voice should be OK, but you'll be missing out on a lot of the
highs. (Mind you, it's not easy at all to get an unloaded Q of 500 at
1MHz!)

Ok, I'm back with a couple more questions, since y'all are so
informative... I know it's not a true xtal set if I add an amp, but
wouldn't the best way to minimize the drop between the unloaded and
loaded Q be to add an amp? With a transistor/op amp, it seems that you
could tailor the load imposed on the tank circuit so that it's minimal,
while it seems a headphone load would be pretty significant (depending
on the headphone)...?

Also, I've seen some schematics with transistor-based and Op amp-base
amplifiers. Generally speaking, are there advantages/disadvantages to
either transistors or op-amps as the first-stage RF amplifier?

Thanks again for all the information!
Dave

dave.harper wrote:

K7ITM wrote:


There is such a thing as TOO HIGH a loaded Q. Let's say you start off
with a coil with unloaded Q of 500, and couple lightly to it with your
circuit (antenna and detector), so the loaded Q is 250. That means the
bandwidth at 3dB points, if you tune a station at 1MHz, is 4000Hz. If
you've tuned to the center of the station, your demodulated bandwidth
will be only 2kHz. Since the rolloff is gradual with a single-tuned
circuit, voice should be OK, but you'll be missing out on a lot of the
highs. (Mind you, it's not easy at all to get an unloaded Q of 500 at
1MHz!)


Ok, I'm back with a couple more questions, since y'all are so
informative... I know it's not a true xtal set if I add an amp, but
wouldn't the best way to minimize the drop between the unloaded and
loaded Q be to add an amp? With a transistor/op amp, it seems that you
could tailor the load imposed on the tank circuit so that it's minimal,
while it seems a headphone load would be pretty significant (depending
on the headphone)...?

In my set I simply use an audio matching xfmr. I'm using sound-powered
phones which are in the 200 ohm impedance range, and around 50 ohms DC.
A little xfmr like the Calrad 45-700 is a good choice. Some of the
guys use a switchable matchbox using a tapped xfmr like the Bogen. You
can see some of these at http://www.crystalradio.net/

As for an amp...mine plays great into my computer's sound card! I tried
a little one chip amp but it only had about 20db of gain which isn't
really enough to do a lot of good.

-Bill

Yes, by using an amplifier, you can lower the loading caused by the
detector. There is an optimum load impedance for the detector output
in terms of best output power for a given input signal, and for small
signals it's a pretty high impedance (resistance), because the diode's
dynamic impedance is quite high for very small signals. You can find
info on this at the Agilent web site (at least till Agilent sells off
their semiconductor business...). Look for ap notes and data sheets
covering zero-bias detector diodes. It gets a bit technical. But the
optimum load is, as a rule, rather high resistance. An FET-input
amplifier chosen for low input voltage noise is probably ideal.
HOWEVER, the crystal radio purists would probably complain that it's
not a crystal radio then. As Bill says, a matching transformer can
help you out a lot. I've done some work using zero-bias Schottky
detector diodes driving DC amplifiers to look for small signals, and
can detect signals down in the few tens of microvolts---but the output
is in the vicinity of a microvolt at picoamp currents.

The other thing that lowers the loaded Q of the coil is coupling to the
antenna. Remember, the antenna looks like some impedance. A resonant
antenna looks like a resistance, and an antenna coupler or tuner will
make a non-resonant antenna look resistive also. And that resistance,
coupled to the tank coil in your crystal radio, will lower the Q. If
you couple too lightly, you won't get all the signal you can, and if
you couple too heavily, you will lower the Q so much that you won't get
the desired selectivity. It's a balancing act. In fact, in a
multiple-resonator tuner, the bandpass shape is adjusted by changing
the coupling from one resonator to the next, which changes the loaded Q
of each resonator. When you have but one resonator, you just change
the bandwidth (and signal level) as you change the coupling and loaded
Q. Coupling that's too light mostly just changes the signal level,
with minimal change in bandwidth. Coupling that's too tight mostly
changes the bandwidth, with minimal change in signal level.

Hope these thoughts help some...

Cheers,
Tom

In theory, one could also use a synthetic inductance, aka 'gyrator". I
took a gyrator based audio oscillator that used 741's and on LTSpice
rebuilt it using 1000 Mhz GBW op-amps. Using an FFT of a transient
analysis I had a nice narrow adjustable center frequency peak of about
10 Mhz, But I never did have a chance to wire it up, as I have a baby
to take care of.

The Eternal Squire