A Simple but Very Sensitive Regen Receiver for 40 Meters

by Rick Andersen, 2004

(revised and expanded 1/15/06)

[I have revised this article as of January 15th, 2006, showing an alternative to the toroid-core inductors of the original design. The revised version uses #26 or #28 enamelled copper 'magnet wire' (such as sold in 3-packs by Radio Shack) wound on McDonalds' plastic straws as coil forms, then covered with molten candle wax to hold the turns in place. See details below.]

I absolutely love regenerative detectors. They amaze me to no end. For those of you who don't know the basics: 99% of all commercially-made radios these days are based on the Superheterodyne design. It has proved itself worthy of universal praise and reliability many times over since it replaced earlier designs in the 1930's. It made Television practical. It will probably always be used as the core of any communications receiver ever built in the future.

But there were earlier beasties that worked well, too, at least for the era in which they lived. The...THE...most sensitive design was the Regenerative circuit. The idea, invented by Major Edwin Armstrong in the 1910's, was to 1) tune in a feeble radio station, 2) amplify it at RF [he used a vacuum tube; we use a transistor today], and, here's the punch-line: 3) feed a small fraction of the amplified signal back to the input, in phase with the incoming antenna signal. A snowball effect occurred, where the signal was reinforced by a boosted version of itself, over and over again -- the precise amount of positive feedback usually held in a delicate balance, right at the edge of the point where the tendency would be to break into a squealing oscillation (like the inevitable feedback howl that your high school principal would experience at the podium microphone when he was getting ready to bawl the student body out about some prank somebody pulled). It is at this threshold that the regenerative detector has unmatched sensitivity (well, the Superregen, covered elsewhere, is actually better, but not as selective).

It was found that this just-before-oscillation setting of the feedback or Regen control was best for listening to AM radio stations; advancing the regeneration any further would cause the station's carrier frequency to be heard as an audio beat note -- a loud squeal -- so your Grandpop would tune in a station with one hand, then set the regeneration to just below oscillation with the other hand. For CW (Morse Code transmissions) or SSB (Single Sideband transmissions) he would leave the regeneration in its oscillating region; the only way to hear the "beep, beep" of a Morse Code signal would be to purposely generate that audio beat note, by "beating" or heterodyning [subtracting the frequencies of] the on-and-off, inaudible CW carrier with an almost-the-same-frequency "fake" carrier generated inside the radio itself... which is what you get when the feedback is past that 'avalanche' point mentioned up above.

So why do we use Superhets nowadays, and why are Regens ancient relics?

Well, because Superhets are free of almost every problem that plagued the earlier types of radios. But we Radio Geeks don't care about the Good Life, do we? We would rather build something that takes a brain surgeon's steady hand and patience to operate, than be hypnotized by the wiles of Madison Avenue! Besides, We built it ourselves!

So be warned: Regens are not "user-friendly" in the convenience-store, modern-day, spoiled-brat, lazy-ass sense. (Yep, I'm soap-boxing again!) Instead, they're fussy little circuits that need re-tweaking and fidgeting as you tune up and down the band.

They're notorious for accidentally re-radiating (transmitting) their own internal RF oscillations back out into the world (when you use a single regen detector stage, directly connected to an antenna. You can solve that problem by putting an RF Amplifier stage ahead of the main detector, which then serves to isolate the antenna from the oscillating detector. The gain of the new stage need not be high, since the Regen stage already has fantastic sensitivity, and you want to avoid overloading it with too strong a signal).

They share, along with another kind of receiver design called the Direct Conversion receiver, an annoying susceptibility to front-end AM overload (you hear your normal stations normally, but at the same time some super-powered station like Radio Moscow or Radio Havana Cuba can be heard simultaneously, all over the dial, even though you're not tuned anywhere near its actual frequency) but, again, this can be improved by adding another stage with another tuned circuit, or else using a resonant bandpass antenna tuner to provide some extra rejection of out of band signals. (The Direct Conversion crowd does it by using what is called a Doubly-Balanced Mixer in the front end; I felt I'd better mention that before they start sending me nasty emails about my ignorance of the new and improved direct conversion designs out there.)

Without the front-end isolation mentioned above, regen receivers will also rock back and forth slightly in tuning frequency as the wind blows your antenna wire back and forth outside -- very annoying if you're listening to a single sideband voice signal or CW.

If you don't build 'em with at least some metal ground plane or shielding in the box, you'll notice that body capacitance can be a problem: the tuning shifts when you reach for the tuning knob! (A similar effect with nearby metallic objects led to the development of metal detectors; one man's problem is another man's solution!)

And, most annoying to me, but not mentioned too often in other articles about the regen, is that the tuning changes when you advance the regeneration into oscillation. So if you build your own Shortwave radio using a regenerative detector, and attempt to calibrate its tuning dial (I paste a piece of semicircular paper on the front panel and make pencil markings), you will find that the tuning is 'off' when you're listening to CW or sideband, from when you calibrated the dial for AM, just before the point of oscillation.

Whew! What a pain in the butt! Why would anyone want to mess with one of these beasts? Shoot it and put it out of its misery. Then bury it!

Well, not a true Electronics Geek / Radio Head. We love this little circuit because it's almost magic. It can dredge the weakest signals almost out of thin air due to the ingenious application of controlled positive feedback. We just need to be able to control it, that's all! If you're a Generation-Xer, but have been bitten by the Bug (why else would you still be reading this page, you sadist!?), know that your Grandpappy loved, cherished, and relied on his Homebrew "Genny"; your Dad (my age) at least built a few of these as "toy" kits sometime during the 1950's - 1970's, and a spate of articles in QST and other Amateur Radio publications appeared in the 1990's, especially by Charles Kitchin, that revived the old Regen for a new generation of tinkerers, with some new bells and whistles added to smooth out some of the bugs that Regens were known to have.

I guess I've waxed eloquent enough.

My 40 Meter Regen Receiver

Here's a schematic of my version of the trusty old Regen radio. The heart of it is Q2, where I use a common JFET (Field Effect Transistor), an MPF102, in a modified Armstrong "tickler coil" design where I supply feedback from the "bottom"-- the Source of the JFET -- rather than the usual top -- the Drain of the FET. Works just as well either way.

The tuned circuit is set up for the 40 meter Ham band -- approx. 7.1 - 7.35 MHz in this radio.

I happened to have some nice little old-fashioned mesh-plate tuning capacitors for this project. They're getting very rare these days; you may end up using a cheap plastic tuning cap from a small transistor radio, etc. In mine, the "real" tuning cap is the 365 pF, but since it tunes way too "fast" through too large a frequency range, I've done what they often did "way back when": renamed the main cap as the "Bandset" tuning cap, made it an internal trimmer cap rather than an external panel-mounted one, and paralleled it with a smaller-valued "Bandspread" capacitor that I've renamed the "Main Tuning" cap, so that it will cover the small slice of spectrum occupied by the 40 meter band. [Please note: I only had a 100 pF cap on hand, so I put a 15 pF ceramic in series with it, so that the total combination "looks like" a 15 pF variable bandspread circuit.]

So I find the 40 meter band, initially, when first calibrating the receiver, and then leave that internal 365 pF trimmer alone. Now I can mark up the dial using the 15 pF "main tuning" (but actually a bandspread) cap on the front panel. And how do I "find" the 40 meter band initially, if I don't have a frequency counter or marker? I turn on a store-bought shortwave radio nearby, set the Regen control into the oscillating region, and then tune the Bandset cap until I hear a whooshing/quieting in the nearby shortwave radio set to, say, 7 MHz. (The 15 pF Main Tuning cap should be fully meshed at this point; all other settings will be >7 MHz which will put you inside the 40 meter band. You can figure out the rest.)

The feedback "tickler" winding on the red-core toroid coil connects to a 1K resistor, a switch, and a 10 K ohm potentiometer. I wanted, not simply a Regen knob, but also a switch position called "CW/SSB" which would force the detector into its oscillating mode regardless of where the Regen knob was set when I was listening to Amplitude Modulated Shortwave stations such as the BBC, in the non-oscillating mode.

NOTE: A couple of readers have built my regen and written to say that they had a hard time getting the amount of feedback just right for smooth regeneration control in this receiver. I have not yet been able to explain the variations in performance, besides advising them to carefully adjust the spacing of the turns on the toroids, and then melt some candle wax onto the coils to hold the turns in place. All I can say right now is, if the Amidon T-50-2 toroids don't seem to work for you, there's no reason you can't wind your coils on air-core forms, such as plastic pill bottles, etc. Just be ready to adjust the number of turns to tune the 40m band, if you decide to use inductors different from those in my design.


As an alternative to the Amidon T-50-2 toroids used in the original design,

1)Get a plastic McDonalds straw and cut it into several 2-inch lengths. You will need 3 lengths or segments for the 3 coils of the complete receiver with Active Antenna stage included. Or just 2 straw segments if you're leaving the Active Antenna circuit out and are using an external antenna.

2)For each 2" segment, poke two small holes in the straw, each 1/2 inch from the ends of the straw. If your original length was 2 inches per straw segment, the holes will be 1" apart; this is just about right for the 50 turns of #26 or #28 wire you will need to wind as the Main Winding on each coil form.

3)Wind 50 turns, as mentioned above. The wire lead at each end of the 50t coil goes through one of the holes and out the end of the straw segment, from inside the straw. This gives some mechanical stability to the wire which would otherwise tend to spring out and off the straw.

4)In the case of the coil that makes up the L-C Tuning circuit in the actual Regen Detector stage, FET Q2, you need to add two more windings. First, wind a new and separate coil of 10 turns next to one end of the Main coil of 50t. This 10t winding is the input coil that the schematic shows as being connected to the Drain of Q1 and the +Vcc (+9v) power source. (If you have a different color of wire in #26 or #28, use it here; it makes things easier to see when you're wiring the coil(s) into the circuit. For example, I used green wire for the 50t Main coil and red wire for the other windings.)

5)At the other end of the Main coil, wind a new coil having 7 turns. This is the Positive Feedback or "Tickler" winding that enables Regeneration to occur. Note: You may have to reverse the connections if the receiver won't break into oscillation/Regeneration when you attempt to use it. One polarity puts the feedback in phase with the original input signal, and causes Regeneration through Positive Feedback. The other way gives Degeneration through Negative Feedback. In short, the high amplification ("gain") of a Regen receiver depends on the polarity being correct for positive feedback/Regeneration to occur. If you hear squeals and whistles at some setting of the Regen potentiometer, you have the phasing right. If the phasing is wrong, you probably won't hear anything, or signals will be weak at best.
One more thing: Eliminate the 1K resistor shown in the schematic, below the coil and above the Regen potentiometer. At least in my updated prototype, it was no longer needed and in fact prevented oscillation/Regeneration until I removed it and wired the dangling lead from the 7t feedback winding, directly to the pot.

6) Drip melted candle wax over the coil to hold the windings in place. In the case of the single, 50t coils, we want to squeeze or spread out the wire turns slightly, listening for a peak in loudness which indicates resonance with the frequencies we're trying to tune in. So the coils must be soldered in-circuit, a signal must be tuned in (or at least listen for a noise peak), and the spacing of the 50 turns adjusted slightly; when you've found the optimum spacing (max loudness/sensitivity), freeze the turns in place by dripping hot wax onto the coil and blowing on it to cool it down quickly, hardening the wax.
(Don't take the receiver to the beach on a hot Summer's day; the coils may melt!)

Back to the circuit: The audio is demodulated in the same FET, Q2, and we tap off the Source resistors, filter the RF part out with a .01 uF cap, block the DC component with the .47uF coupling cap, and send the weak audio signal out to a jack that we will connect to a Radio Shack amplified speaker, an amplifier of our own design, or maybe some further filtering before being audio amplified.

Adding some Audio Filtering

Now I know I advertised this project as a simple 2-transistor receiver. But part of the fun of "rolling your own" is in trying to improve on it from there. So now I'm using more than just two transistors' worth of circuitry. But the results are usually worth it.

For example, I built a 700 Hz bandpass filter, that I can switch in or out of the audio path, to give some narrow-band audio filtering for CW reception. Preceding that stage is a 2 KHz lowpass filter which is for AM/SSB speech signals, which tend to sound too "tinny" without some extra filtering (i.e., the 2 KHz speech filter is always in circuit, but the 700 HZ CW filter can be switched in and out). Here is the circuit (built with plain old 2N3904 NPN transistors, rather than the more real-estate-efficient op amps):

Regens are not Single-Signal receivers!

The idea of filtering the audio brings up another point: Simple receivers like Regens and Direct Conversion radios are unable to eliminate the opposite sideband, a natural consequence of the heterodyne mixing process. With AM, we never notice; an AM signal consists of a carrier and mirror-image sidebands above and below the carrier. When demodulated, they are essentially "folded" on top of one another, and you hear one full AM signal.

SSB signals deliberately suppress the carrier and one of the sidebands as redundant and wasteful. Your oscillating detector supplies the missing carrier inside the receiver, and allows proper mixing down to baseband (demodulation). However, since SSB has eliminated a sideband at the transmitter, there is now extra space available in the frequency spectrum -- so we can pack twice as many SSB stations into a given number of KHz as we can AM stations.

The point? In the demodulation process, your simple Regen will pick up the SSB signal you want -- plus another station now occupying the "other side" where the other sideband would have been, but with an "inverted spectrum", like a scrambled police transmission. So the superhet user hears just the station he or she has tuned to, while the Regen or DC receiver user hears the desired station, plus an annoying chirping gibberish voice, if a station happens to be transmitting at that spot on the dial. And unless you build a special "phasing" -type receiver for single-signal reception, there ain't a thing you can do about it. You're stuck with hearing "backwards saxophone grunts" and "Donald Duck quacks", simultaneously with your desired station, if the band is crowded -- something Grandpa didn't have to deal with in his day, when there weren't as many sideband stations on the air.

So, going back several paragraphs, you'll find that some extra filtering -- cutting the highs off above 2-3 KHz -- will really help the fatigue that sets in after a couple of hours of listening to high-pitched chirps from guys several kilohertz away... that would not be heard by even a mediocre superhet receiver.

Back to our Regen circuit:

Q1, another MPF102 JFET, operates as a common-gate preamplifier, serving to isolate Q2's internally-oscillating signal from being radiated out the antenna. I purposely chose to put 100 ohm resistors in the source circuit and at the antenna input, to approximately simulate a 50 ohm load as seen by the antenna / transmission line. A series LC (toroid wound with about 23 turns of #28 copper magnet wire on an Amidon T 50-2 red core) forms a fixed, low-Q bandpass filter centered on the 40 meter band. The 100 pF cap and toroidal coil are made to resonate in the band by adding or subtracting turns on the toroid, until the received signal is peaked at its strongest.

This extra tuned circuit, absent from the Gennys of the 1930's, helps reject out-of-band signals and to combat AM overload from out-of-band signals. If there's an unusually strong station within the 40 meter band (and there often are, since that band is shared with high-powered European shortwave broadcasters which blanket the band from late afternoon through the evening hours), you may be out of luck; try attenuating the signal coming in from the antenna by putting some resistive attenuation or a series capacitor in the 1000 pF down to 1 pF range.

Adding an Active Antenna

Another add-on that I use with this regen is an Active Antenna. This frees me up from having to connect an external wire antenna to the radio; the Active Antenna receives its signals via a 24" whip antenna which feeds into the top of another LC tuned circuit and is then fed into an emitter follower circuit to keep loading to a minumum. (In the schematic below, you will note that I have a 10K resistor across that LC tank as "optional"; I found that there was some parasitic oscillation without it. You may or may not need it.)

The 2nd transistor in the Active antenna, a PNP, feeds its output current (with signal riding on it) into the 100 ohm resistor at the antenna jack of the original Regen's front-end RF amplifier. This circuit represents my seat-of-the-pants attempt at impedance matching; someone else may decide I used 'overkill' and may want to modify accordingly). A recent modification, shown on the updated schematic above, is the optional .01uF ceramic capacitor across the 47 ohm resistor in Q2's emitter circuit. I noticed that the apparent strengths of the signals in the 40 meter band are relatively weak during the day; adding this bypass cap increases the RF gain (and sensitivity) noticeably. On the other hand, at night, when the big European stations come booming through, you may find that there's a little too much overload and 'AM bleed-through', where a strong station comes in all over the dial, or where you can't clarify the sound of the SSB stations because their signals are drowning out the 'local oscillator' (i.e., oscillating detector) and the mixing process is not 'clean'. In this case you might want to cut the gain down a little. Although I didn't do it in my prototype receiver, you might want to install an SPST switch to allow you to increase or decrease the gain of the circuit-- you might label the switch positions "hi-gain/lo-gain" or "day/night".

One more comment on the general performance of any Regen receiver: Some people confuse the Regen's tendency to overload with poor selectivity. Nonsense! The Regen detector has excellent selectivity, as long as you're near the oscillation point on the feedback control. I've built AM broadcast versions of this circuit that were so selective that, with the Regen control set right on the edge of oscillation, you could hear what seemed to be hundreds of carriers whooshing by, when tuning through the broadcast band late at night; When any one station was tuned in, I could increase the regen far enough to severely limit the sidebands containing the music the station was playing - without the detector breaking into oscillation - so that the music sounded like it was being funnelled through a sharp audio bandpass filter..... which it was, effectively.

If you end up building only one radio circuit from my small collection on these web pages, my recommendation would be to go for the Regenerative (for AM/SW/CW/SSB) or the Superregenerative (for FM/VHF aircraft). They will amaze you with their performance from such a minimal parts count!