Another super simple medium wave radio.
This time a Franklin Audion (synchronous detector).
Well I assembled and tested the noisy regen today:
Performance wise with the short antenna it was a little better than the first one. With a long antenna not as good. Selectivity was better (likely due to the higher input impedance as a result of lower collector currents).
Exciting? No. The reflex circuit was much better (but then I used a proper coil). Would I recommend it? No. Is it all the coil problem? No, I don't think so. To get an idea about how bad this receiver is, without the high gain audio amplifier you would not hear anything. I suspect a Ge diode followed by the same audio amplifier would work better!
Here is the layout:
Alan Yates (http://www.vk2zay.net/article/128) has stripped down the Franklin Audion circuit to its basics and taken the audio off the emitters rather than the tuned circuit. This seems to make more sense to me:
(Note that the 47n output filter capacitor should be 4.7n!)
So I stripped my circuit down and was able to get the simulation to honour the expected frequency locking and demodulation:
The signals (in clockwise order) show the receiver:
The green is the input signal, the red is the oscillator and the yellow the output audio.
Note how the oscillation can be quenched in case 2.
Now that I have a working simulation I can test component values. Or in this case the component value the 100k emitter resistor. You may have noted I increased the value from 10k R to 100k R in this version. The results indicate that the same operating point can be found with the 10k resistor as the 100k resistor. The main difference is that the oscillation amplitude can be much higher (the full available collector voltage or about 0.7v pp) and consequently the available mixer gain (i.e. higher audio output). This is reflected in actual use. The oscillator frequency also shifts with higher oscillation amplitudes.
Using the higher (i.e. 100k R) emitter resistance will tame the receiver. Audio gain is best found elsewhere.
Added the schematic to help show how the oscillator works:
I have used NPN transistors (as this is what most people are used to) and split the emitter resistor to show the two transistor amplifiers.
Here is the updates strip-board design:
If you are interested, the input impedance of the Franklin oscillator in this circuit is about 1.2k (approximately that of the coil impedance tap) and the output is about 10k (approximately the same as the low pass filter).
Although I have had success with the lumped inductors, I know the Q is quite low.
But how good or bad are they? That depends on the wire diameter and core material.
An Internet search suggests that they roughly match the Vishay IRF-24:
Now a Q of 60 is far less than 300 to 500 you can get with an air coil.
For MW (i.e. 1 MHz) a Q of 60 translates to an unloaded bandwidth of 17 kHz.
Which is why they work in this application although not great when the coil is loaded.
The alternates are:
I have designed and built MW radios using the first four options.
Using the data from an Amidon datasheet for their mix 61, the maximum Q is about 270 (between 1 MHz and 2 MHz). This is the best ferrite material for MW frequencies.
Options not tried:
Option 2 seems to be the best for a PCB mount but the toroids are a few weeks away (just orded them on ebay).
Using a 4 foot wire aerial the radio worked but not very loud. Regeneration was smooth and sound quality was good. The circuit Q is pretty poor and regeneration only improves the Q to a limited extent. I was not that impressed! Here it is:
Hooked up a 14 m aerial (no earth) and the volume was quite loud. Clearly, the "detector" is not that sensitive (need a strong signal to perform). There is a design which has AGC, now I know why:
I was thinking it may be better to take the audio off the emitters rather than from the coil tap. I found an example at http://www.vk2zay.net/article/128:
Often radio circuits have coil taps and/or coil couplings. These be simplified as lumped inductors.
For example a 50% tapped 200 uH inductor is the same as two 100 uH inductors in series with regard to LC oscillation frequency and impedance transformation. The main down side with lumped inductors is that for similar coil construction the total resistance will be higher, that is the Q will be lower (i.e. a 200 uH inductor needs only 41% more turns than a 100 uH inductor).
So my tapped 320 uH coils consists of a 100 uH and 220 uH inductors in series. The simulation suggests a frequency range (using a BB212 varicap diode not the 1SV149 diode) of 1.27 MHz (for an 8 volt control voltage) and 564 kHz (for 1 volt control voltage. Note the series resonance response and "roll-up"!:
This is half way between the 25% and 50% taps shown the the schematic:
Modelling a coupled coil has no series resonance response and a normal "roll-down":
Changing the coupling produces a normal response:
I was rummaging through my junk box looking for some inductors and came accross an old super-regenative tube radio I build in the mid-nineties. I used the lumped inductors then and this receiver worked okay:
Here is the schematic:
I have not used one before but my reading suggests that they work pretty well as a "single" in receiver projects (refer to "B") but for (high powered) oscillators a "twin serial" set up (refer to "C") is recommended:
In each case value of "R" is not critical but 1M ohm is common. Usually a capacitor at "Vc" is added to minimise unwanted RF in the control voltage (Vc) circuity as shown in the following:
I purchased a couple of these some time ago off ebay. There is the datasheet (http://www.qsl.net/df7tv/datasheets/1SV149.pdf). At 1 volt the capacitance is about 485 pf and at 8 volts the capacitance is about 25 pf.
Using a 320uH inductor and a 1000 pf series capacitor the tuning range should be about 500 kHz to 1.8 MHz.
The Franklin oscillator is a little strange and not immediately obvious how it works:
The following helps a lot:
(source: www.radiosparks.com/images_d/Franklin Oscillator.png)
So positive feedback at parallel resonance (0 degrees phase shift).
The next image is a transistor version:
Here is a rationalised (long tail pair) FET version:
Still it takes a leap of faith to get to here:
But at low voltages most of the biasing components can go!