A Direct Conversion Receiver for 80 Meters
July 15, 2009 by Rick Andersen, KE3IJ
with "subharmonic" VFO in the "Polyakov" style
[Revised February 23, 2010]
NOTE: I have revised this article to include some pictures of my prototype, originally ugly-construction-only, now installed in a Mouser Electronics plastic enclosure.
I have also done away with the 600 Hz CW filter; the inductor picked up too much 60 Hz AC hum for my liking, even after I took measures to kill the hum.
Instead of the previous 100mH inductor and .68uF series capacitor, there is now a single 10uF electrolytic cap connecting the output of the diode detector to the input of the audio amp transistor Q2. Rather than rewrite the article, I decided to
cross out the revised sections, followed by yellow-highlighted comments on the revisions.
You may have noticed that almost all my receiver projects on these webpages, up to now, have been based on some variation of a Regenerative Detector, usually a Colpitts derivative. Most QRP designs over the last 30 years, on the other hand, have favored the Direct Conversion approach, which essentially means that we inject our incoming RF signal into one port of a mixer (usually a double-balanced diode ring) while injecting a tunable VFO into the other port, with the VFO being at just about the same frequency as the signal we want to detect. In such an arrangement, the beat- or difference- or INTERMEDIATE-frequency (I.F.) is reduced to (or close to) 0 Hz -- i.e., the beat frequency is at baseband and is the same as the audio intelligence, now demodulated. So, although we can describe the Direct Conversion circuit as a sort of "superheterodyne" whose local oscillator matches the RF input frequency and therefore has an I.F. of zero, it's more common to interpret it as a "non-superhet" without any I.F. at all -- we convert the incoming signals directly to baseband (audio); hence the name "Direct Conversion" (also called a Synchrodyne detector). This is not too different from a Regen detector, in self-oscillation, very close to the incoming signal frequency, producing an audible beat note. (Another name for a Regen is an Autodyne detector.) In fact, it can be said that the main difference between the two is that a Direct Conversion receiver has a separate Variable Frequency Oscillator (VFO) and RF input which are then combined in a mixer; in the Regenerative receiver, the mixer/detector/oscillator are all rolled up in one stage, into which the RF is coupled. And the Regen has legendary sensitivity, but it's also more sensitive to outside influences on the oscillating detector.
While the Direct Conversion (DC) receiver is thus conceptually very simple, it also comes with its own baggage (otherwise known as "Gotchas!") that can potentially make or break a homebrew receiver, causing an unnecessary amount of hair-pulling during the debugging phase.
I've tried designing a few DC receivers over the years. In the breadboard stage on the bench, they would often appear to work great... until I put them into enclosures or hooked them up to an electrically-unbalanced antenna like an end-fed wire. That's when the trouble usually began.
The two gotchas that most often caused me to go running back to my first-love (Regen receivers) were microphonics and common-mode hum, which can dampen any enthusiasm you might have for DC receivers pretty quickly. Not to say that Regens don't have idiosyncracies of their own that must be dealt with or lived with. Choose your poison. But as I often say, that's the price we minimalists have to pay for our bare-bones toy radios. But we're not fatalists; we do try to come up with tricks to minimize the gotchas ;-)
"Microphonics" simply means that, since most or all of the approximately 100 dB gain of the receiver is in the audio stages [in contrast to a superhet, where most of the gain occurs at I.F., e.g., 455 KHz in an AM table radio], the receiver becomes very sensitive to mechanical vibration, like a microphone. So when you tap the case or touch a knob, you hear a thump with a bit of bell-like resonance to it. Any wiggling of the box that causes a micro-wiggle of the turns of the VFO's coil also gets amplified; turning a speaker amplifier up too loud may also induce acoustic vibration of the enclosure and cause audio feedback that can howl like a guitar amp if not controlled.
"Common-mode hum" is a much more difficult buggar to deal with. It's a nasty-sounding, raspy 120Hz hum that appears at various points across the tuning dial, usually at the frequency you're most interested in. The mechanism that produces it has to do with the receiver's own oscillator signal radiating back into the DC power leads and being rectified in the mixer -- or something like that (look it up in a textbook on DC Receivers; I never get the explanation quite right!).
It is exacerbated by connecting your DC radio to an unbalanced antenna system, particularly end-fed long wires, Zepps, etc., whose improper installation or inadequate RF grounding might also lead to RF in the shack while transmitting. A related symptom is that your receiver snaps, crackles, and pops with every electrical transient in the house wiring, like a refrigerator compressor turning on, lights being turned on and off, etc. Sometimes you may also find a tendency for the receiver's VFO to be "pulled" excessively as you rock your antenna tuning and loading capacitors back and forth through resonance (I use a resonant antenna tuner, rather than a PI network. I like the extra rejection of out-of-band signals that a resonant tuner gives me -- essentially, it's a bandpass filter in addition to being an impedance-matcher).
So what's a DC enthusiast to do? Well, the gurus' advice to us is to:
1) switch to a balanced antenna like a center-fed dipole (I use an 80 meter full wave horizontal loop nowadays; no more trouble with electrical pops and clicks originating in household appliances)
2) try to run your QRP equipment on batteries rather than AC-powered DC supplies or "wall-warts" (I use batteries almost exclusively, yet I could still hear common-mode hum at various spots on the dial. Also, I've found that most DC power supplies that are considered "well-regulated" are just a disaster when used to power the DC receivers I've built. With all that audio gain, what little 120Hz ripple residue there is, riding on that "pure" DC output, gets highly amplified by the receiver's audio stages.)
3) put "common-mode choke" toroids in the DC power leads to present a high impedance "open circuit" path to the common-mode hum energy (didn't work very well when I tried it)
This is where I can begin my story that led to the new DC receiver described in this article, which I've named the "DC-80."
Some years ago I was sitting at the bench, trying to design a DC receiver for 80 meters, a band I tend to favor (relatively easy to design simple equipment for this band). For my "VFO" I was using a bench RF signal generator in the usual way: tuning it between 3.5 and 4.0 MHz, which worked just fine except for the microphonics problem that I outlined above.
At some point, whether by accident or intentionally -- I can't remember -- I switched the signal generator to a lower frequency range, turned the tuning dial, and was somewhat surprised to hear some very crisp, clear SSB banter in the phone portion of... what?... the 160m band?... because the dial was set at 1.9 MHz or so.
Then I realized that the front end preselect filter I had installed between my antenna wire and my diode product detector, was tuned for the 80 meter band.... were these 160m signals blasting through the 80m filter THAT loud and clear?...
Within a few minutes of listening to the SSB'ers yammering, I was able to identify them as operating on 75 meters, not 160! I pondered this for a minute, then thought "Oh! Maybe I'm mixing a harmonic of the 1.9 MHz generator signal, not the fundamental itself, with my incoming 75m SSB stations... The second harmonic (a.k.a. "first overtone") of 1.9 MHz is 3.8 MHz, and 3.8 MHz is in the 75 meter phone (voice) band! This is kind of neat!"
Well, what was also neat was that I couldn't hear any microphonics at all, with my VFO being at precisely 1/2 normal frequency. And, being set at "160 meters" frequencies rather than directly at "80 meters", the tuning knob was less "touchy", less prone to drift, too. Very cool.
And it got late that night and I hit the sack, and that's where I left my little discovery, until recently.
The Polyakov Detector
In June of 2008, I had an email conversation with Jerry, K9UT, who had built my AGC-80 Regen and wanted me to think about a 40 meter version of that project. He also mentioned that he had been reading about DC receivers whose VFOs were set to operate at 1/2 the normal frequency, and the benefits of that configuration. As I was working on my AGC-80/40 at the time, I filed his remarks away in the back of my mind and forgot about them for the time being.
Over the past few months of 2009, I've been reading here and there about a neat little circuit called the Polyakov detector, which is the circuit that I think Jerry was referring to. Also called the "Russian mixer", it is described by RA3AAE (and/or others), in its simplest form, as two antiparallel diodes (i.e., two diodes soldered together in parallel, but with their cathode bands in opposite directions). Rather than needing any broadband transformers driving them, the simplest version of the circuit has them driven by incoming RF, on one end, through a 100pF capacitor. The other end is sometimes broken apart so that a 500 ohm pot [for nulling out AM 'blanketing'] can be inserted between the two diodes' right ends; the tap of the pot goes to another 100pF cap which is the VFO injection cap (VFO energy enters the diodes through this cap); simultaneously, audio energy (the beat frequency between incoming RF and VFO) is sent out from the same point and into an R-C low pass filter which removes the RF energy and mellows the audio somewhat. From there the audio is amplified, etc. (Other, more elaborate variants of the circuit do involve broadband transformers with multiple windings.)
Further Internet surfing informed me that this circuit concept has been around for a while, and the part that makes the Polyakov detector unique is that the VFO driving it is supposed to be running at half the desired RF input frequency... the Internet sources call this a "subharmonic" mixer. And the hams and QRP tinkerers are talking excitedly about the Polyakov circuit because it apparently solves the microphonics, frequency-pulling, and "touchiness" problems of the DC receiver, to some extent, because the VFO is running at only half the usual frequency; hence, it is more stable!
While I was in that netherworld between reading about it, and getting up the gumption to go down to the shack and actually try building one, I pondered the actual mechanism...
Was it really the '2nd harmonic' of the VFO at 1.9 MHz that was mixing with the incoming signal at 3.8 MHz, which was the 'magic' behind this circuit's performance? Why does 2nd harmonic mixing automatically "isolate" the VFO fundamental frequency from the RF signal in such a way as to suppress microphonics, etc.? Anyway, does the presence of a 2nd harmonic mean that the VFO signal is "dirty"? Because I would expect the 2nd (and subsequent) harmonics to be several dB down in amplitude, from the fundamental, and that the injection level wouldn't necessarily be adequate for that nice, loud, crisp detection I heard during my lab bench experiment.
If my VFO is too "clean", will the circuit fail to detect?
Or does even the "cleanest" VFO output sine wave get mutilated by the mixer diodes -- i.e., Does this "2nd harmonic" really originate in the mixing process of the diodes themselves, via their nonlinearity? The explanations of the Polyakov circuit that I saw attempted to correlate the dual-directionality of the two diodes with the "doubling" effect, sort of like a full wave rectifier.
Then I pondered another question I've always thought about, but have never seen a clear answer to: If I mix two frequencies in a diode mixer, unbalanced, I know I will get four freq's out: f1, f2, f1+f2, and f1-f2. The question is, do all these new frequencies also automatically mix with one another, generating higher-order mixer products (I believe they do)... so where does this stop? It would seem that the original mix would "snowball" into a near-infinite number of new frequencies being generated in the mixer...but each new frequency having less amplitude, so that the majority of the snowball would be down in the mud, for all practical purposes.
...which brought me to another thought: Hold on a minute; maybe this isn't about "harmonics" at all; instead, maybe this is about pure heterodyne mixing! After all, if I mix a 4 MHz input with a 2 MHz VFO, both being pure sine waves with no harmonics contaminating them, then our sum and difference frequencies are going to be 4+2 = 6 MHz, and 4-2 = what?...2 MHz!
In other words, the beat frequency between any frequency and its half-frequency, is another copy of that same "half-frequency" (i.e., its "phase conjugate"). So maybe a Polyakov mixer is not really a "subharmonic" mixer after all; maybe it is simply exhibiting a regular ol' heterodyne process where the "beat" happens to match the lower of the 2 input frequencies. In other words, a "superhet" whose local oscillator is half the input frequency, and whose "I.F." is therefore equal to the "local oscillator" frequency. And when THOSE TWO frequencies beat in the diode, 2 MHz minus 2 MHz = 0....Lo and behold, the output is baseband -- in the audio range!
To test this idea, I built the simple RA3AAE circuit. Then I modified it in at least 3 ways, which led to the DC-80 receiver described later in this article. The test was: Will any other, NON-Polyakov mixer perform in the same way, with f and f/2 applied to their input ports?
The Polyakov circuit
It didn't seem to matter where the 500 ohm pot was set; the AM bleedthrough was still there, although centering the pot lowered the overall volume of all signals, desired and undesired. I'm unimpressed so far. I did try a different null pot value (1k), put it on the other end of the diodes where the RF comes in, and various other things. It worked, but not as advertised, or so it seemed to me.
A Single Diode
I took out one of the diodes. Guess what? As long as there was a resistor to ground for continuity through the diode, the circuit worked as a "normal", single-diode AM detector...even though the VFO was running at half-frequency. This leads me to believe that my "heterodyne" idea is correct, and that it's not necessary to have two diodes connected antiparallel, to get the "doubling" effect [which is a misnomer, in my opinion].
AM bleedthrough, however, was objectionable, which is why decently-designed Direct Conversion receivers usually use a double-balanced, 4-diode ring mixer, along with a diplexer (impedance-matcher which properly terminates the RF and suppresses it from getting into the audio amp, which can cause unwanted "AM blanketing" detection in the audio transistors' base-emitter junctions).
A 2-Diode Product Detector
Next, I tried an old product detector configuration I'd seen in the ARRL Handbooks: Solder two diodes in series, but in opposing polarities; put RF in the left side, take audio from the right side, and drive the junction in the middle (through a cap) with VFO energy. Well, you need terminating resistors on each diode to make this one work right. Once again, the 2-diode product detector worked well as a 3.8 Mhz detector whose VFO was running at 1.9 MHz. But still some AM bleedthrough.
A New 4-Diode Detector
Finally, I tried something that I haven't seen anyone do before: Strap another 2-diode product detector in parallel with the first, but with the relative polarities of the second set of diodes opposite to those of the first set. So now we have 4 mixer diodes, not chasing each other as in a ring mixer, but as 2 "series" product detectors, of opposite polarity, in parallel (see the schematic of the DC-80, below). It can also be described as 2 Polyakov detectors connected together with the VFO injection at the junction between them. (The 500 ohm balance pot is eliminated.)
I call this a 4DPD [4-diode product detector] mixer.
And it works very well. With VFO at half the RF input frequency. So when I call this a DC receiver with a "subharmonic VFO in the Polyakov style" up in the title of this article, I'm aware that it's not an actual Polyakov receiver; instead, it's based on the spirit of the Polyakov mixer, especially with its half-frequency VFO. A run-of-the-mill doubly-balanced ring diode mixer, it ain't. But it performs nicely just the same.
So, no disrespect intended toward our Russian friends, but I think the benefits of this "new and improved" Polyakov Direct Conversion approach stem from the unique heterodyne products of a 2:1 frequency ratio between RF input and VFO, not on the bidirectional transfer curves of two antiparallel diodes. At least that's my take on it, and I'll gladly retract this opinion if the facts warrant it in the future.
I learned a couple of other things in designing the new DC-80 receiver: Terminate the antenna preselector bandpass filter with a 51 ohm resistor, at the input of the diode mixer (see schematic). This seems to reduce AM bleedthrough and makes little difference in overall volume level (I thought the reduction from RA3AAE's 500 ohms down to 50 ohms would kill all my audio output. It doesn't. There's something to be said about designing around a stated impedance level, like 50 ohms.)
Also, I learned that the first audio transistor ought to be configured as a common-base circuit, whose input impedance is also low (close to 50 ohms). Not only does this seem to help with the rejection of AM bleedthrough, but it makes a surprising difference in the quiescent noise level of the audio amp, with the antenna disconnected! Some of the other projects on my webpages would probably be better performers if I had known this before this particular project taught it to me. I deliberately reconfigured Q2 as a common-emitter amp with the signal going into the base in the "normal" way we think of transistor amps being wired.....
...as a common emitter amp, the receiver was quite noticeably noisy (a constant hiss) with the antenna disconnected. When I went back to the common-base setup, with audio going into the emitter, the amp quieted down to a whisper without antenna, yet was amply loud and crisp when SSB or CW signals were coming through, with antenna connected.
I'm also happy to report that there are neither microphonics, nor any common-mode hum, discernable in this receiver -- although I have not yet enclosed it in a box, which may rain on my parade, based on previous experience. The only problem I encountered was some genuine, non-raspy 60 Hz AC hum getting into the 100 millihenry inductor that I use for the CW filter.
I mounted it upside down so that its top was glued to the copper ground plane floor of the circuit, and I tightly twisted the short length of hookup wire coming off that inductor and connecting to the CW filter toggle switch. There was a big reduction in hum once I twisted those switch wires together. There remains a slight hum when the filter is switched to the "CW" position.
I eventually eliminated the inductor, .68uF cap, and CW filter switch and replaced them with a single 10uF electrolytic cap connecting detector output to audio amp input (emitter of Q2). IGNORE the CW Filter section of the schematic diagram, below.
PLEASE ALSO NOTE: Two other changes I made to the schematic below: 1) Change the 470pF ceramic cap across Q3's base resistor to a .0027uF. This compensates for the lack of 600 Hz CW filtering which I removed. 2) I removed the Volume Control pot shown in the schematic, for lack of real estate in the small box I finally ended up with. I connected the audio output cap directly to the audio output jack. A cable plugged into the jack connects to one of the infamous beige-colored Radio Shack Amplified Speakers, which is my preferred method of dealing with audio amp requirements.
Schematic Diagram of the DC-80 Receiver
Let's get into the nitty-gritty of the new Direct Conversion receiver. The schematic, below, is relatively simple, but to make things easier I have broken it up into 5 sections, outlined by the thin blue blocks superimposed over the schematic itself.
Referring to the schematic above, let's go through a description of each section:
Front End Preselector
If I can get away with avoiding large-turn coils up front, I do it. The 80-meter band front-end "preselector" filter described here was also shown as an alternate filter in my AGC-80 series of articles. It's essentially a "Pi" lowpass filter combined with series input and output capacitors, whose combination gives a sharp bandpass response. I designed the filter using an old version of MicroCap circuit analysis software which shows me a plot of the frequency response as I vary the parts values. I'm pretty sure this filter is my own invention; I haven't seen it used anywhere else. It works for me.
This configuration allows me to use fairly large caps -- .001uF and .0022uF (which are the same as saying 1000pF and 2200pF) -- along with a fairly small inductor -- 1.3 microhenrys -- which is easy to wind. Just take about 8 inches of #28 enamelled copper magnet wire (Radio Shack sells a bag of 3 sizes) and wind 10 to 12 turns on an Amidon T50-2 red-gray toroid core....which Electronix Express sells as part number 152750-2. Adjust the spacing of the turns on the coil for peak reception. Drip some melted candle wax over the windings to hold them in place, once you've squeezed or spread them for the loudest volume on an incoming signal.
The caps I use, by the way, are the green mylar caps that Electronix Express sells (I call them "chiclet" caps, because they remind me of the litte green cubes of spearmint gum that go by that name). I believe plain old ceramic caps would work, too, but I have always used the mylar "chiclets" in my designs.
The 80m filter is intended to be used with a 50 Ohm antenna impedance, and it is terminated in a 51 Ohm resistor in the radio circuit. If you want to scale the filter up or down (40m or 160m bands), you just need to calculate the capacitances and inductance from the standard formulas, and knowing the reactance values of the components, which are constant for a 50 ohm system, keeping the reactance ratios the same for all bands. The reactances are: 42 Ohms for the .001uF series caps, and 19 Ohms for the shunt caps to ground. The toroidal coil's inductive reactance is 31 Ohms.
The standard formulas are as follows:
Inductance L = Reactance XL divided by (2pi x Frequency) L=XL/(2pi F)
Capacitance C = 1 divided by (2pi x Frequency x Reactance Xc) C=1/(2pi F Xc)
So, for example, if we wanted a front-end filter for 40 meters, using 7.2 MHz mid-band for F,
1) the series caps at Xc = 42 Ohms would be calculated as 1/(2pi x 7.2E6 x 42) which comes out as 5.263E-10 Farads on the calculator. Multiplying that by 1E12 converts the answer to picoFarads.... = 526pF. Since that isn't a standard value, I might try a 470pF or 560pF cap instead, and modify the turns on the coil later, if necessary.
2) For the shunt caps at Xc = 19 Ohms, the same formula gives me a capacitance of 1163pF, so I'll substitute a 1000pF = .001uF cap for that.
3) For the toroid coil, with XL = 31 Ohms, the equation would be 31/(2pi x 7.2E6) or .000000685 Henrys, which, when multiplied by by 1E6 to convert to microHenrys, gives 0.685uH for our inductance.
But how do I know how many turns this equates to on the toroid? Well, lacking an exact formula, I can reason that, since I doubled my frequency to get from the 80 meter to the 40 meter band, and my original inductance of 1.3uH has just about been cut in half, at 0.685uH, then maybe I ought to put about half the number of turns on the toroid....I'll try 6 turns, and adjust the spacing for a peak later on when I'm testing out the receiver.
Yes, this is "designing by the seat of my pants" but it's OK; it works well enough; don't sweat it.
Notice that the coil is already less than 1uH at 40 meters. This tells me that my 'special' pi-bandpass design won't work well for any other high bands; for 20 meters I would need to change the filter design to something more conventional. But it should be OK for 160, 80, and 40.
Since my design has such easily-obtainable parts and low coil-turns, and has 50 Ohm impedance levels throughout, this also means you can cascade two of these filters in series, if you want even more out-of-band rejection. Again, simple, non-critical, and intuitive.
On another subject relating to the receiver front end:
The RA3AAE-style receiver schematic showed a transistor RF amplifier up front, before the diode mixer. I had a front-end RF amp when I first built the RA3AAE circuit, along with its 500 ohm impedances in the mixer. Bottom line, the increased gain up front invited noticeable AM bleedthrough, no matter what I tried to do to minimize it. When I remembered an ARRL publication stating that, on 80 meters at least, a well-designed DC receiver doesn't need an RF amp, I took mine out, feeding the output of my antenna preselect filter right into the diode mixer, with a 51 Ohm termination resistor to ground at that point. The result was that I was still able to receive loud signals and the AM bleedthrough was cut way down, almost to the point of disappearing, although you can still hear it occasionally when tuning to a blank spot on the dial.
4-Diode Product Detector
I've already described my experiments with the Polyakov configuration and other variations, so I'll just stick to a description of the detector stage here.
The bandpass preselect filter, just described, sees a matching termination in the 51 Ohm resistor that shunts the input of the 4DPD diodes to ground. This helps kill the "AM broadcast blanketing" that plagues simple receivers. The resistor also provides a DC path to ground between the "pumping" action of the VFO input cap, and the diodes.
The diodes, which are plain vanilla 1N4148 (=1N914) switching diodes -- nothing fancy, nor did I make any attempt to "match" them -- can be seen either as a left-hand Polyakov antiparallel pair, joined to a right-hand Polyakov pair, with the VFO injecting its signal in the middle of the two; or, one can view the arrangement as an "upper" 2-diode product detector, joined with a "lower" 2-diode product detector of opposite polarity, in parallel, with the VFO driving both their midpoint junctions simultaneously. When the polarity of the VFO signal is positive, one pair of diodes conducts. When the VFO goes negative, the other pair conducts. Same thing as a Polyakov, but done a little differently. However you want to view this 4-diode arrangement, IT WORKS. In fact, it sounded a bit louder to me when I strapped the 2nd pair of diodes into the circuit, as if it were more efficient.
The VFO signal is coupled into the center of the detector through a 100pF ceramic cap.
The output is taken off the right side of the diode array, with a 0.1uF cap to ground, which bypasses the RF output of the mixer and leaves only our baseband audio. I tried putting another 51 Ohm resistor in series with that cap to ground, in an attempt to imitate one of the "diplexer" circuits I found online in a "DC 'Popcorn' Receiver" schematic, but there was noticeable AM bleedthrough(!) When I removed the resistor and put the bypass cap straight to ground, the AM stuff disappeared. Go figure.
For those of you whose 'common sense' tells you that 0.1uF is much too large a cap to use -- that it will make the audio sound too mellow... -- Surprise! -- it's not like that when you're living in a 50 Ohm world! That 'common sense' only works when your "background" impedance is up around 1K to 10K. The signal is still quite "tinny" at the output of the 4DPD, and needs some more mellow-out filtering later on, in the audio amp.
(I bring this up because I once came across a pair of pundits critiquing one of my radio circuits on some blog somewhere. They applied the standard "f=1/(2piRC)" roll-off formula to my component values and concluded that my audio must be as mellow as jello because my bypass cap appears to be way too big. Not in a low-impedance circuit, though. Hasn't it ever occurred to you to wonder why the coupling cap feeding the 8-Ohm speaker in an audio amp has to be way up there around 470uF or more? You need to dump more charge through a smaller resistance in order to get the same power.)
But I digress.
We leave the mixer/detector output for now and travel down to the VFO.
As any good Ham knows, the VFO can make or break a receiver. So let me say right here: I wanted a quick and dirty, functional VFO; simple and cheap. So I used only a single stage, with no buffer. I basically took a Clapp Oscillator design and used the perfect "coil" for that oscillator, at 1750 - 2000 KHz..... one of those little oscillator cans with the red screwdriver-adjustable cores, which tunes the Local Oscillator (L.O.) in an AM broadcast radio. Voila! No coils to wind with dozens of turns at 160 meters. Because -- remember -- we are running the VFO Polyakov-style, at 1/2 the RF input frequency.
To make things simple for me and for you, I also borrowed the variable tuning capacitor from cheap transistor AM radio design -- one of those little plastic-cased tuning caps with the 3 leads coming out the side. Unlike the metal-plate 365pF variables of the days of yore, these plastic "polycaps" have a total capacitance of around 140 - 170 pF for the RF section, 60pF or so for the L.O. section [intended for a superhet receiver]. I soldered the two cap sections in parallel (the outer leads) while grounding their bottoms through the center lead. Then I also have a 100pF cap in parallel with the whole thing, and series connected 220pF's as the feedback caps in the base-emitter circuit of the 2N3904 transistor, Q1. These values were obtained by experiment and could be improved upon. In fact, the whole VFO was hastily designed and could be improved upon. It's a bit drifty in frequency, but it works! Without taking any "dBm" measurements, it provided a very adequate injection level from the first moment I turned the radio on, as evidenced by the crisp, loud signals I'm able to hear, and by the relative lack of "spurs", "tweets", and unwanted intermodulation products.
So there! If you don't like it, make it better! I'm admitting that I scrimped where I probably should have been a lot more careful.
2/23/2010 Revision: Added a Fine Tune "Clarify" Control
The right-most knob on the front panel of the encased DC-80, which you'll find farther down the page, is the Fine Tune pot. It operates like the Bandspread tuner in old shortwave radios, and like the "Clarifier" knob on some old CBs. In a nutshell, the 140 pF plastic tuning cap tunes the whole 500 KHz -- 3.5 to 4.0 Mhz -- of the 80 meter band. With no gear reduction, one half-twist of the knob covers the whole band in one fell swoop, and the tuning rate is too fast for comfortable operation.
I've added a parallel "fine tune" which is a 10k potentiometer controlling a reverse-biased rectifier diode, used as a varactor tuning diode. The 0-9v at its wiper is isolated from the VFO through a 1 Megohm resistor, and lightly coupled to the top of the plastic Main Tuning cap via a 10pF ceramic cap.
To operate, start with the Fine Tune knob set in the center of its range. Tune in a station with the Main Tuning cap, and tune for a 'zero-beat' with CW signals, or, tune for a deep growl on SSB voice signals. Now, turn the Fine Tune left or right until the "growling" SSB voice "clarifies" to a normal pitch. If tuning CW, tune to the desired beat note.
Since one of the distinctive traits of a simple Direct Conversion receiver (or Regenerative receiver, for that matter) is that it lacks single signal capability (i.e., you are able to hear the signal on both sides of "zero-beat"), you will be able to tune SSB voice signals either "correctly" or "backwards"--- if you're on the "left" side of zero-beat, the voices will sound garbled and impossible to understand [because of audio spectrum inversion of the sidebands]. Turn the Fine Tuning control clockwise through zero-beat, and you will end up on the "right" side and will be able to hear intelligible speech, although it will vary from a deep/low sound to a high "chipmunk" sound; somewhere in the middle of that range you will hear a normal-sounding voice. The Fine Tune pot helps you clarify that voice with a slower tuning rate.
Switchable CW Filter
Back up to the output of the mixer/detector:
I use one of the small, cylindrical blue inductors from Electronix Express in a series bandpass filter, as my very simple CW filter [the inductor also blocks any RF that the 0.1uF bypass cap may have missed]. The choke (as they call it) has a value of 100 milliHenrys, with a DC resistance of around 100 Ohms (they must have a lot of turns of very fine wire inside). That 100 Ohms knocks the Q down, making a rather broad filter, but it's adequate for such a simple filter. The other component making up this filter is a mylar cap whose value I have listed on the schematic as 0.68uF; this value, along with the 100mH choke, gives a resonant peak at around 600 Hz. If you prefer a 750 Hz peak as many QRPers do, change the cap to 0.47uF.
A toggle switch is wired across the 100 mH choke. When the switch contacts are OPEN, the choke is in-circuit and you're in 600 HZ "CW" filtering position. When the switch is CLOSED, the choke is short-circuited and the audio signal travels straight from the detector output to the 0.68uF cap, which doubles as a blocking/coupling cap for the input of the audio amp, Q2. With the choke out of circuit, the audio is unfiltered and quite "bright" [tinny-sounding], to be mellowed out just slightly in the audio preamp circuit, described in the next section.
One "gotcha" that I experienced with the 100mH choke inductor was that it picked up a healthy amount of 60-cycle AC hum from the soldering iron. I mounted it upside-down, gluing it to the copper ground plane on which I built the prototype (see picture near the end of article). The hum got worse when I wired the CW Filter switch across its leads. I then twisted the two switch wires together, which killed most (not all) of the hum.
Your mileage may vary, so keep the physical distance short between the choke and the panel-mounted CW Filter switch, and twist the two wires leading to the switch, tightly, down their length.
Quiet Audio Preamp
Well, a lot quieter than some I've built, anyway.
As usual, I built this receiver without an audio power amp. Instead, I route the output to a Radio Shack Amplified Speaker (steadily rising in price; now about $14.99 I think; I'd like to believe that all my articles extolling the virtues of this little amp has caused business to boom for Tandy Corp.; at any rate, I remember these little beige boxes costing around $11.99 back when I started this website) and lately I've been plugging a computer speaker into the aux output of the Radio Shack amp for better sound.
As I've described elsewhere in this article, I discovered that this receiver's quiescent noise, heard when there is no antenna connected, was drastically reduced when I compared Q2 wired in a common-base configuration, with Q2 wired in a common-emitter configuration. I'm not sure of the correct explanation, other than to say that in the common-base circuit, the input impedance is low, giving a good match to the 50 ohm level designed into the receiver front end.
The same 0.68uF cap [or 0.47uF alternate] that serves as the capacitive element in the L-C CW Filter, also serves to couple the weak audio from the detector into the emitter circuit of Q2, a 2N3904 like I always use.
I now use a 10uF electrolytic here.
The cap serves to block DC at the same time; there is approximately 1 volt DC at the top of the 1K emitter resistor, and we don't want that bias to affect the diodes in our mixer. So we "cork it up" with that cap, which serves to "couple" AC (audio), while "blocking" DC (Q2's emitter voltage). [I used to tell my electronics students that a cap passing AC while simultaneously blocking DC is like waving at someone through a glass window: They can see an image and movement (change; AC) but if you try to blow a stream of air at them, the glass blocks the flow (flow; DC). I'm sure that that explanation caused less brain damage than a friend of mine did when he used to play mind games with his young boy by shaking his head up and down when he meant "No", and side to side when he meant "Yes". The kid is probably doing prison time by now.]
A 220k resistor supplies positive current to Q2's base, while a 10uF electrolytic cap bypasses any audio on the base, to ground. Doing this forces the collector to output an amplified version of the audio variations coming in at the emitter. The amplified audio is picked off Q2's collector and 4.7k collector load resistor, coupled to the next stage through another 10uF electrolytic, and sent into the base of Q3 for further audio amplification. Q3 is grounded at its emitter (a common-emitter configuration) while the base is supplied with DC bias through another 220k resistor. Both stages connect their base resistor to the collector; this is known as "collector biasing" and is a form of negative feedback that automatically sets the transistors' bias at the optimum operating point in their bias curves.
Notice that there's a
470pF .0027uF cap across Q3's base bias resistor. This is a bit of "mellowing" for the tinny-sounding audio I mentioned earlier; a simple low-pass filter to take some of the edge off the highs... so that the piercing whistle of W1AW's code practice signal doesn't drive you batty even when tuned 8 KHz away!
But note that when the CW Filter switch is "off" -- shorted, actually -- the only filtering that the audio sees is that of the 0.1uF cap at the detector output, and this
470pF .0027uF cap across Q3's base bias resistor. So the guys on SSB and AM phone sound sharp and right there in the room with you... and static crashes are that much more noticeable, whereas in the CW position, static is greatly reduced by the action of the 600 Hz bandpass filter, and the extremely high-pitched CW tones are mostly filtered out.
At the collector of Q3 we couple the audio out to a 10K potentiometer (Volume Control), which is a variable resistance voltage divider whose wiper connects to the audio output jack. High-impedance crystal or magnetic headphones may be connected here, although I prefer to connect my amplified speaker here instead. You may ask why I put in a volume control, if the Radio Shack amp already has one at its input? The reason is that there is a lot of amplification (audio gain) from Q2/Q3, and the Radio Shack amp was actually being overloaded, during the peak of the SSB yakking hours in the evening, even with the amp's volume control turned almost all the way down. This made me happy, because as a seat-of-my-pants designer, I knew that enough audio to overload the Radio Shack amp was certainly enough to drive most headphones to at least adequate volume. (Since I don't have a pair of decent 2000 ohm headphones, I have to make educated guesses about how the receiver will sound to someone who does have them ;-) I left the Volume Control out of the revised version because it would not fit in the small enclosure from Mouser (although they do sell larger versions of the box). I made the decision to simply send the strongly-amplified audio to the rear audio jack, and to let the volume control on the Radio Shack Amplified Speaker suffice as my gain control. If you decide to build the DC-80 in a bigger box, you may decide to put it back into the circuit.
A Picture of the Prototype
It's ugly! -- it's scary! -- it's coming to a theater near you -- it's not in an enclosure yet! -- but here it is. I've marked the major components with labels or with black magic marker on the copper-clad board. If you build it and it starts giving you trouble in an enclosed box like some of my earlier DC receivers did, you have the right to box my ears and pommel me. But I wanted to get the new design up as soon as I could. If you look closely at the picture of my prototype, you can see that there's really not much to it, that it should work as well as it does. This 1/2 frequency VFO thing really seems to have breathed new life into the lowly Direct Conversion Receiver. Build one of your own, and if you don't think it performs, look at it as a learning experience.
2/23/2010 UPDATE: The DC-80 in an Enclosure
Mouser Electronics to the rescue! I found a cute little "bone-white" plastic box in the Mouser catalog, part number 616-79160-510-039, which is a small 4" x 6" x 1.5" high plastic box with top and bottom pieces, plus removable front and rear plastic panels that fit in grooves between top and bottom sections. The top and bottom are held together with long screws that go in through the bottom. The enclosure comes with 4 press-on adhesive rubber feet. Only $7.70 + shipping at the time of this update in February 2010.
Here are some photos of the DC-80, rebuilt on a small copper ground plane and mounted in the Mouser box:
PARTIAL Parts List
I'm not gonna give you part numbers for every resistor and cap in the circuit. The resistors are 5%, 1/4 watt types and most of the caps are the green mylars. The big cylindrical caps with polarity markings are electrolytics and should have a DC working voltage over 10v. The VFO injection cap happens to be a 100pF ceramic. You can go to the Electronix Express website and browse their online catalog for yourself. But the 'unique' or unusual parts are listed here, for your convenience and abatement of confusion.
I've decided to give Electronix Express a big plug in this article, since I've dealt with them before and found that they almost always have the kinds of components I need for QRP radio projects -- like the inductors and variable capacitors. Sometimes a few of their parts are back-ordered and take a while to arrive, but for the most part, my orders arrive at my door within a week. (I used to teach Electronics at ITT-Tech, and Electronix Express -- under their alternate name, RSR Electronics -- was our source for electronic parts and kits used in our Lab courses.)
Antenna connector - BNC female, panel mount. You'll notice in the picture that my prototype doesn't use one, but uses alligator clips instead. Well, when I put it in a box, I'll use a BNC connector, OK?
L1 = 1.3uH coil. Wind 12 turns #28 magnet wire on a T50-2 toroid core (Electronix Express # 152750-2) and adjust turns spacing for loudest signals in the middle of the band (windings cover about 1/3 of the core).
Caps = green mylar types at Electronix Express
Diodes = 1N4148 / 1N914 silicon switching diode
Q1 is 2N3904 NPN silicon, like I use in most of my projects.
Transformer is the red core Local Oscillator "can" that is used in most transistor radios. Electronix Express # 16IFR, or Mouser/Xicon 42IF100-RC.
CALIBRATION: Set a nearby receiver to 2.000 MHz, and, with the DC-80's tuning cap turned fully clockwise (top of band; least capacitance), use a screwdriver to adjust the red core until a loud "whoop!" is heard in the 2 MHz receiver. This sets the top of your DC-80 at 4.000 MHz (top of the 75M band), with the VFO running at half the RF input frequency. Re-calibrate if the bottom of the 80m CW band falls out of range, if needed (i.e., the receiver may not tune the entire 3.5 - 4 MHz band, depending on parts tolerances, etc.)
Tuning capacitor is Electronix Express p/n 14VCRF10-280P, the typical AM transistor radio plastic tuning cap with 3 leads. In this receiver we connect the two end leads together (connects to L.O. transformer) while grounding the middle lead.
Inductor is a 100 millihenry cylindrical choke, Electronix Express # 150100M.
CW Filter switch is SPST toggle or slide switch, whichever floats your boat.
Q2 & Q3 are 2N3904 NPN silicons, like I use in most of my projects.
Volume Control = 10k Ohm potentiometer with ON/OFF switch, Electronix Express # 18PMS10K
Power & Miscellaneous
9 volt battery
100uF, 16vDC electrolytic capacitor (bypasses entire receiver, preventing motorboating oscillation)
The receiver needs a "Fine Tuning" control. This could be a small-value variable cap strapped across the Main Tuning cap in parallel with it. The single, miniature plastic variable cap knob has no gear reduction or vernier action (obviously) and it's hard to properly "clarify" SSB voice signals. There's also some body capacitance effect with the circuit unenclosed. Tuning is pulled off frequency when you've zeroed in on a station and then taken your hand away.
Added a Fine Tune "Clarify" Pot; see above.
There is still some discernable AM bleedthrough, but not much at all. It's much easier to mentally screen it out with this receiver, compared to, say, a Pixie or other simple Direct Conversion receiver which lacks a double-balanced diode ring mixer up front.
The receiver could also use some form of AGC, like I use in the AGC-80 series. But, as I said, this prototype was meant to prove the 1/2 VFO frequency concept of the Polyakov-style mixers; these other comments are refinements for another day.
If you decide to put one of these simple receivers together, I'd like to know how yours works out. Email me at firstname.lastname@example.org