Crystal radio antenna impedance matching

[Elerion]

Experimenting with crystal radio, I've reach a point where I'm not able to understand the explanations about antenna and tune coils coupling, and impedance matching.

Most radios use a step up coil scheme, where the antenna coil has fewer turns than the tune coil (where the resonance is achieved, using a variable capacitor).
But most loop antenna based radios do the opposite.

Just picking two references:
------
"The impedance of the signal coming from the antenna is probably in the 50-300 Ω range. Crystal radios are usually listened to with high impedance headphones, usually around 2 kΩ. Having a 100 Ω source driving a 2 kΩ load is inefficient, which ultimately means lower volume...
[a] transformer steps up the voltage, making it higher than what comes out of the antenna directly."
------

If stepping up voltage, that means the antenna "sees" the diode side as a lower impedance than it really is.
------
"The basic circuit consisting of the diode, load resistor and the smoothing capacitor will never present a good match to 50Ω
If the detector diode is in its on state, the circuit will appear to be less than 50Ω.
To overcome this issue it is normal practice to use an impedance transformer to step up its impedance."
------

This contradicts the previous statement. Is the diode side a high or a low impedance?

 

[cowboybob]

Low (although varies depending on the diode).

Some notes for your perusal.

 

[Elerion]

Great website, I already read throughout.

"In a very high impedance tank circuit, a standard 1N34A diode just won't cut the mustard. The diode will load the circuit"

Then why use a step-up voltage transformer?
Like this one? (near the top of the page, under "Resonant LC circuit")
**broken link removed**

 

[cowboybob]

... Then why use a step-up voltage transformer?...

The circuit you reference is, while looking like a "step-up" transformer, is rather an improved resonant tank with a higher "Q" than a single coil arrangement. This produces (as noted in the text) improved tuning precision and RF signal strength.

Don't confuse a "coupled resonant tank" circuit (note the use of VC1):

with a "step-up (or a step-down, for that matter) transformer" circuit

 

[Elerion]

Sure they are different, but doesn't the "coupled resonant tank" raise the voltage on the higher turns winding?

The first reference in my first post argued:
"steps up the voltage, making it higher than what comes out of the antenna directly."
I've read this kind if statement many times from several websites.

But at the same time, it is (many times) recommeded to tap the diode at hald way the tune coil (instead of at the same point as the capacitor) to not lower Q so much.
This does make sense, as tapping at less turns, voltage is lower, but current is higher, so taking less current from the coil/capacitor circuit.
Am I right?

 

[JimB]

There are two things to be considered with this tuned circuit:
Sensitivity and selectivity.

Basic crystal sets are not known for being good with either property.

Connecting the antenna to the top of the tuned circuit will have two effects:
It will effectively add resistance across the tuned circuit, thus damping the circuit and lowering the "Q" and hence the selectivity (the ability to discriminate between different frequencies) will be reduced.
It will add reactance, either inductive or capacitive, depending on the relative length of the antenna compared with the frequency to which the circuit is tuned.

As others have commented, using a link winding with fewer turns than the tuned winding will result in the antenna having less effect, both resistive and reactive, on the tuned circuit.
Another effect is that there will be some voltage magnification depending on the Q of the tuned circuit. This will improve the sensitivity.

Where to connect the diode?
Conventional wisdom and experience tells us that the diode detector is likely to have a fairly low impedance, so connecting it to the top of the tuned circuit is likely to result in loading the circuit, lowering the Q and so reducing the selectivity.
In which case it can be advantageous to connect the diode to a tap on the coil.
If we wanted to be really clever, we could connect the diode to its own link winding and adjust the coupling to the tuned circuit as we desired.

There is not one short simple answer to your question.

JimB

 

[Elerion]

I understand better now.
Still I'm not sure about diode impedance.
What exactly loads the tank circuit? DC, audio AC, RF AC?

After the diode, there's a filter capacitor, to remove RF component from the audio signal. This makes a short circuit for RF, and thus, a very low impedance to ground. As the diode, while conducting, is a low impedance, the RF signal loads heavily the tank circuit, no matter what yo do.
The audio AC signal goes to a high impedance headphone, so it is a mild loading. Am I right?
The DC component is not negligible, because high impedance headphones are low impedance at DC. This can be eliminated by using an RC network in series.

So, when people say that the detector loads the tank circuit, are they talking about DC or AC? what is most offensive for Q?

 

[cowboybob]

Look up the difference between resistance, reactance and impedance.

 

[Elerion]

Why ?

Resistance doesn't depend on frequency.
Impedance depends on frequency. It is the sum of DC resistance and reactance.
Reactance is the capacitive or inductive portion of impedance.

Is there anything more I should study?

 

[cowboybob]

I'm suggesting that you need to research how AC and DC figure into circuits that exhibit one or more of the three.

In other words, do the math and you'll get your answer(s).

 

[Elerion]

I've been thinking about this, and doing simulations, but I'm not still sure. Really, I must be missing something very evident, but not able to see what it is.
AFAIK any current taken from the LC tank, in any form or frequency, degrades its Q.

1) the resistor makes the DC resistance high enought to be similar to earphone/output transformer AC impedance.
2) the AC impedance at 1-10 kHz needs to be high enough. Using crystal earphone or impedance matching transformer, this is easy.
3) THIS is where I'm having trouble.
C2 must be high enought to filter out the RF carrier, but much lower than to filter demodulated AF signal.
A 1nF meets this requirement. But this capacitor has less than 100-200 ohms at RF frequencies.
Doesn't this load the LC tank circuit?

 

[MikeMl]

The secret to crystal sets is D1. In a working crystal set, the I/V characteristics of D1 near zero bias is unlike any model built-into LTSpice.
A cat-whisker, point-contact detector is nothing like a Silicon diode. Spending time optimizing impedance matching, and modeling the Q of the tuner is not productive unless the detector model is realistic.

Try this search:


Found this.

 

[cowboybob]

... Doesn't this load the LC tank circuit?

Simply put, yes, of course it does.

But to what degree (as MikeMi notes) is almost entirely dependent on the electrical characteristics of the particular diode used (obviously, other elements of the circuit play a part as well).

And to no small degree (back in the old days), was an impetus for the development of amplifiers ...

 

[audioguru]

A crystal radio has poor sensitivity and poor selectivity. Also it is missing the automatic gain control like in a real radio so a strong local station is too loud and a weak distant station can barely be heard.
The detector diode conducts strongly on a strong local station reducing the Q and causing poor selectivity and increased interference but since the audio level is so high then the adjacent stations and interference can barely be heard.
The detector diode conducts weakly on a weak distant station increasing the Q and causing interference to be low.

 

[JimB]

I had some time to play this afternoon, so I knocked up a little circuit as shown here:

The idea was to test the effects of different antenna connection methods on the tuned circuit of a crystal radio.

The tuned circuit is an 80 turn coil with a 330pF capacitor in parallel, this resonated around 710 kHz with the X10 scope probe connected across the tuned circuit.
There are two means of injecting a test signal:
An 8 turn link winding which is inductively coupled to the bottom end of the tuned circuit. The degree of coupling is adjustable by varying the spacing between the windings.
or
Via a resistor which is connected to the top of the tuned circuit. There are two resistors in series, a 1k and a 10k Ohm, to examine the effects of different antenna loading when connected directly to the tuned circuit.

This circuit was connected to a spectrum analyser which has a built-in tracking generator which was used as the signal source.

Looking first at using the link winding with loose coupling to the tuned winding:

And then moving the link winding to give tight coupling:

I saw response curves like this:

We see that the voltage across the tuned circuit is the same (in this case) for loose and tight coupling, but the selectivity is much better with the loose coupling. A radio built with the coils loosely coupled would be better able to discriminate between signals on different frequencies.

I then tried feeding the signal into the top of the 10k resistor:

Connecting to the 10k resistor gave less voltage across the tuned circuit and a very flat response, ie poor selectivity.

So what happened when I connected the signal to the 1k resistor :

Yuk!
There is next to no selectivity when connected to the 1k resistor.

Adjusting the analyser to sweep across a wider frequency range:

Here we see a comparison of the relatively sharp peak in response when connected to the top of the 10k resistor, and the very flat response when connected to the tuned circuit via the 1k resistor.

So, going back to what was said earlier in the thread, I hope that I have demonstrated why it is not good to just connect an antenna on to the top of the tuned circuit in a crystal radio.
It is because it works very badly.

If I get some more enthusiasm in the next day or so, I will do some more experiments on the detector side of the crystal radio.

JimB

 

[Elerion]

Thank you everyone, particularly JimB for such a detailed comment on the matter.

I also played a little bit. I built this simple rf amplifier

The sensibility has been improved a lot, but also the selectivity.
Anyway, I don't get the "regenerative" effect that its author claims.
AFAIK a regenerative receiver should boost selectivity big deal.
If I set the pot to low value, I get a lot more gain, but the selectivity degrades a little bit, not improve. Even if I get to the point where the noise gets annoying, and it seems that I approach oscillations. (it makes sense, as lowering the pot lowers the input impedance, doesn't it?)
If I set the pot to > 500 ohm, then the signal is nice and clear, but the volume drops a lot, to be just a little bit over the diode on itself.

The best selectivity improvement I got by tapping the amplifier to a low impedance coil tap.
Rememeber I'm using a loop antenna, that does the job of tank inductor too.

The next experiment will be infinite impedance detector using JFET.
I hope to improve selectivity and sensitivity.
Any bets?

 

[unclejed613]

The secret to crystal sets is D1. In a working crystal set, the I/V characteristics of D1 near zero bias is unlike any model built-into LTSpice.
A cat-whisker, point-contact detector is nothing like a Silicon diode. Spending time optimizing impedance matching, and modeling the Q of the tuner is not productive unless the detector model is realistic.

Try this search:


Found this.

for germanium diodes Eg should be around 0.3 to 0.4, but they have it set to 0.69 in these models (for a forward voltage of 0.69). there's a chart on wikipedia with a Vf of 0.27 for germanium.

one interesting thing is that some mounted crystals used to be carborundum (silicon carbide SiC), and there are silicon carbide diodes for switching power supplies that are available. it might be interesting to try some of these. Vf for SiC diodes is about 2V, so a bias supply might be needed to bring the diode near conduction. if choosing a SiC diode for use in a crystal radio, take care that the junction capacitance is as low as possible.

 

[unclejed613]

adding a circuit for clarity. as you can tell, the bias supply brings the diode close to conduction, and the diode begins to demodulate the signal at somewhere between 5-10 microvolts. without the bias supply the threshold is 680mV. the choke (L3) blocks the RF, and so maintains a relatively high impedance on the RF side.


the LTSpice circuit is an adaptation of the 1917 carborundum receiver, and uses a bias battery for the detector. modern crystal radio enthusiasts seem to think using a battery is "cheating", but biasing the detector to near conduction was in use in the equipment of the day.

 

[Colin]

A lot of the things you are saying, are wrong.

You should read the following article to let you understand how the crystal set works:
The Crystal Set

 

[audioguru]

Colin, why do your 2-transistors and 3-transistors radios have no AM detector diode?