This is my version of a one-transistor radio that drives a speaker to room-filling volume and has very good selectivity and sensitivity without an outdoor antenna.
The project came about after seeing Tom Polk's website "The Macrohenrydyne" and his statement about it being the "Most Powerful One-Transistor Radio". This looked like a challenge, so I set about to design a better, or more powerful version of his circuit. The following are the results of that quest:
Before I begin, I want to thank Tom Polk for his expertise on the circuit and Charles Wenzel for not only advice, but for generously providing some of the critical parts of the project. Both of these gentlemen were willing to share their knowledge and time without asking for any compensation. Also, I want to point out that the project present here, is in no way representative of a better or more improved circuit than was previously published by the folks mentioned above. The following project merely documents my attempt to improve the design. My final circuit is, in my opinion, at least as good as my predecessors. However, without actually doing a side-by-side comparison, it is impossible to tell which design is better. I have determined, after all my experiments, that it is really hard to get any more performance out of this type of circuit. Regardless, I had fun with this project, and the results are simply amazing! I only wish I had the parts and this circuit when I was a kid. Back then, I thought a two transistor "boy's" radio was super. This set easily outperforms these!
The final circuit represented
below, is a combination reflex and regenerative receiver. To
briefly describe the operation: The signal is tuned in with the primary
coil on the ferrite rod and variable capacitor combination. This
signal is loosely coupled to the base of the transistor where it (RF)
is amplified. At the same time, part of this signal is fed back
into the ferrite rod where it is amplified several more times,
comprising a traditional regenerative circuit. From
the diagram, It is not apparent how regeneration takes place.
This is accomplished by mounting T2 in close approximation to the
ferrite antenna bar to provide the correct amount of feedback.
The amplified (RF) signal is coupled through T2 to the detector diode
where the audio is recovered. Note that the diode also passes a
DC bias current, supplied by the 1 meg resistor, to the base of the
transistor. The .01 mfd capacitor removes any remaining RF so
that audio is also presented to the base where it is simutainously
amplified, along with the RF, by the transistor. This is the
reflex action. Finally, the amplified audio appears across T1 and
is coupled and impedance matched to the 8 ohm speaker. The reflex
action and regeneration is controlled, at the same time, by the pot in
the emitter circuit.
My first attempt at making improvements, was a double-tuned circuit where the collector, as well as the ferrite bar antenna, were tuned together. It looked good on paper, but in real life, it proved impossible to make the two circuits track. It appears that with a regenerative circuit, the selectivity is so sharp, that tracking can only be accomplished with the absolute best-matched components. Even with a precision dual-section variable capacitor, a matching variable inductor and trimmer capacitors, I could only manage to make it track over about a third of the dial. However, the part that did track, worked amazingly well. If you wanted to make a simple, single station radio, this would be practical. You could just use two trimmer capacitors. Maybe I will post that circuit someday...
Closeup of the "Pacent"
After going back to a single-tuned circuit, one thing became apparent, the so-called "Straight-Line" variable capacitors of later vintage were, in fact, not that straight. They tended to crowd the upper portion of the dial. The solution was to use an old stock 1920's style, straight-line capacitor. The one I chose is an all-brass Pacent that is about 365 picofarad. With the Pacent, the center of the dial came out to 1100 kHz. as compared to the more modern capacitor, which placed it at 950 kHz. This provided a dial, which was very evenly distributed over its entire length.
In reference to Tom Polk's circuit, he used a Contra-Wound ferrite antenna with a switch that divided the broadcast band into two segments. This resulted in spreading out the band in two sections over the length of a slide-rule dial. After experimenting with the Pacent capacitor and using a 14 to 1 tuning wheel, I decided to use a single space-wound ferrite bar and spread the whole band over one section of a similar slide-rule dial. This still gave a resolution of about 1/16 inch per kHz. which is plenty for Dxing. The ferrite bar, I used, is one that was salvaged from an old Zenith Model 1000 TransOceanic Radio. This is a flat bar, 8 1/2 inches long. On this, I space-wound 58 turns of 15/44 Litz wire for the main, tuned circuit, which came out to 235 microhenries. This was followed by 2 turns for the coupling winding. Here, 4 turns gave much more sensitivity, but a reduction in selectivity. 2 turns is a good compromise. With this ferrite bar and variable capacitor combination, the low end of the band started at 520 kHz. I set the high end to 1710 kHz. with the trimmer capacitor that is in parallel with the main variable.
Another difference between Tom's circuit and mine, is the method of controlling regeneration. He added a feedback winding to the ferrite bar and with a combination of a variable capacitor and variable resistor, he was able to obtain smooth regeneration. After trying his circuit, which does work very well, I noticed that there was a noticeable loss in power to the output transformer through the variable resistor. Also, this design makes it necessary to have an additional control on the front panel. I thought if I could eliminate the variable resistor, I could increase the speaker volume somewhat. One method I tried, was controlling the gain with a variable resistor across the coupling winding on the ferrite antenna. This, in effect, shunts the signal that is applied to the base of the transistor. Although this works, it seemed to reduce the "Q" of the ferrite bar antenna circuit when it was set at low resistance, thus reducing the selectivity. This method is used in both Robert Bazian and Charles Wenzel's designs. Another method is to place a variable resistor in the emitter circuit to control the gain. This was a method used in older designs, one of which is presented on Charles Wenzel's website. With this method, there was negligible loss of power to the output transformer since the value of the resistor is very low. Also, since the resistor is in the emitter circuit, there is no loading of the ferrite antenna circuit and no loss of selectivity. I used a 100 ohm, ten-turn pot to provide a very smooth gain control. I mounted it next to the circuit board and used a fiberglass shaft to the knob so that there would be no hand capacity effects. The ten-turn pot is not necessary, I just happen to have it. A single-turn pot will work nicely. I placed a 180 ohm fixed resistor across the pot to limit the minimum gain to the point where a high-power local station could be attenuated to a good listening level. With this setup, the one control varies both regeneration and volume (reflex action).
Closeup of the
Most of the circuit was built on a 2" x 2" piece of perf
board. T2 (black rectangle) is
an ISDN type of transformer that passes RF in the broadcast band of
and rejects audio frequencies. I used a
FIL-MAG part # 66Z3088A that I had in my junk box.
Both Charles Wenzel and Tom Polk also used a commercial
transformer part # PE-64934. Charles
generously gave me a couple of these to try in my set.
In comparison, I was able to get more volume
at a given gain setting, with my transformer. This
is probably because of the 2:1:1 winding ratio that I
series for the primary. Regardless,
this increase in gain can be made up with a lower resistance setting of
gain pot. This
transformer is physically larger than what Charles and Tom used.
As I mentioned earlier, the regeneration is determined by the placement
of T2 in reference to the ferrite bar. In order to optimize the
placement, I mounted the circuit board to a 2 inch square piece
of Garolite, then with double-sided tape, I placed the board in the
correct location and stuck it to the base. I chose a location
the feedback just under oscillation while having the gain control set
to maximum. If you are not able
to obtain proper regeneration, reverse the connections at points "C"
and "D". This will reverse the phase of the signal in
reference to feedback. The glass part, to the right of the
transformer, is the .22 mfd. capacitor.
I tried a number of different transistors, including darlingtons, in this circuit. Several were generously supplied for comparison by Charles. Although many worked, either with or without a change in bias, the MPSA18, which was in the original design, always seemed to edge out the others in gain and stability. I used socket pins on the circuit board for the transistor and diode to make change-out quick.
Some of the circuits presented by Robert Bazian and Charles Wenzel used a 1N5711 Schottky diode for the detector. I have tried this diode, and although it does increase the volume somewhat, it distorts the audio noticeably. So, I used a 1N34 for the better fidelity.
audio output transformer (bottom left) is a common 70 volt line
transformer as used
in PA systems. I used the 4 watt tap on
the secondary to
drive the speaker. I brought the speaker audio out with a 1/4
inch phone jack (below the transformer).
the supply voltage indicated is 18 volts, the set performs almost as
supplies as low as 8 volts. So, for a smaller version, a 9 volt
battery is sufficient. Battery drain is very low, in fact, when I
accidentally left the power switch on for over a week with the speaker
disconnected, the batteries were still in good shape when I found
it. Plus, they are just common carbon-zinc, not alkaline.
The only other changes in the circuit were the values of the
decoupling capacitors, which also improved the
The dial scale can be made by first measuring the available
area then, in a CAD drawing program. draw and print out a graduated
scale with 1/10 or so increments, numbered sequentially. This will be your calibration scale.
Temporarily attach it to the dial then, with a signal generator, mark
every 10 kHz. across the whole dial. You can now remove it and
transfer the numbering to the drawing, fill in the 1 kHz. markers then
delete the calibration scale. After printing it out, attach it to
the front with either paper glue or double-sided tape. My scale
was drawn in Corel Draw and turned out to have 1/16 inch spacing
between each kHz.
I used 1/4 inch Garolite for the base and front panel.
The dial pointer slide and ferrite bar holders (white material) are
high density polyethylene.
Finally, to operate the set, turn the gain control fully
clockwise (least resistance), tune in a station then reduce the gain
until the distortion is eliminated. Further reducing the gain
set the volume. Since the ferrite bar antenna is very
directional, you can rotate the
set to bring in DX and also to null out strong local stations then
receive distance ones just 10 kHz. away!
You can improve the performance by just adding a ground
connection to the common ground on the diagram. Also, an outside
antenna could be added by coupling one turn to the ferrite bar.
You might have to readjust the trimmer to keep the dial tracking since
this will load the tuned circuit slightly. However, I prefer to
use the set without any outside connections.
This was a fun project and I have enjoyed several hours
logging many stations with this set. I plan on building more of
these marvelous radios. The next one will be a pocket version.
Click here to return to the MODERN HOMEBREW page
© 2006 - 2008 ~ Michael R. Starcher ~ All Rights Reserved