August 15, 2009
KR1S Synchrodyne MW Receiver
In 1972, G.W. Short described a "Modern Homodyne Receiver" in the UK publication, Radio Constructor. I first became aware of this receiver when someone mentioned it on the Yahoo regeenrx group. His post led me to one by Mike Tuggle, which led me to the original article "Homodyne," by G.W. Short). I've created a PDF version of the article.
Circuit Theory
The homodyne or synchrodyne is a type of regenerative receiver in which the signal is split into two paths. In Short's version, both paths are amplified, but in one path the sidebands are clipped, and the resulting amplified carrier frequency is fed back to the input. When an amplitude-modulated carrier is tuned in, the receiver oscillator locks to the carrier. The two paths are then combined in a diode mixer. Because the carrier is augmented by a stronger, locally-generated signal, signal-fading effects are reduced. Apparent selectivity is enhanced because adjacent-channel sidebands are somewhat clipped as well, though their carriers heterodyne with the oscillator and are still audible. An audio-frequency low-pass filter reduces their amplitude.
So far, so good. The problem with Short's design is that he used a now-obsolete PNP transistor as the main rf amplifier. There were good reasons to choose this transistor in 1972. It offered extremely high gain at medium- and short-wave frequencies and, being a PNP device it made the circuit simpler.
Simplified schematic diagram of G.W. Short Homodyne detector.
The PNP bipolar transistor (TR2 in Short's schematic) is direct-coupled to the FET (TR1) drain. Were TR2 an NPN transistor, the positive bias on its base would drive it into saturation. Unfortunately, the AF239 transistors are expensive now, but fans of this receiver design still use them. It seemed to me the design could be modified to use an inexpensive, currently-produced NPN transistor instead. Which one?
Like most decisions of this type, the driving force was "What devices do I have?" The AF239 has a transition frequency of 700 MHz. Its gain at medium and high frequencies is quite high, just the ticket for this receiver. On the standard broadcast band, it seems likely that a workhorse 2N3904, with a transition frequency of 300 MHz, would probably be good enough. (Click the device numbers to download the datasheets.)
| Device | hfe | fT (MHz) | NF (dB) | Price |
|---|---|---|---|---|
| AF239 | 50 | 700 | 5 | $8.99* |
| MPSH10 | 60 | 650 | not spec. | $0.16 |
| 2N3904 | 40-100 | 300 | 5 | $0.07 |
| 2N5179 | 25-300 | 900 | 4.5 | $1.10 |
* On eBay, $3.00 plus a whopping $5.99 for shipping (from a U.S. dealer!).
The AF239 compared with currently available transistors that might replace it.
A look through my parts drawers turned up some MPSH10 transistors, with transition frequency of 650 MHz. They cost 16 cents, versus 7 cents for the 2N3904, so I decided to use the MPSH10.
NOTE: After posting this article I was informed of the Fairchild MPSH81 PNP complement, Mouser 512-MPSH81, with fT of 600 MHz. This transistor should directly replace the AF239 without modifications.
Now I was hooked! Could I modify the G.W. Short design to use an MFSH10 transistor instead? Obviously, the polarity of TR2's emitter and collector and associated circuits would have to be reversed. TR2's base would have to be isolated from TR1's drain. That was easy: Just use a coupling capacitor. TR1's drain load resistor doubled as TR2's base resistor. That wouldn't be possible with a coupling capacitor. Short notes in his article that TR1's drain current is about 3 mA, and the load resistor is 1000 ohms. A voltage divider of 1800 and 1000 ohms would have worked, but I went with 18k and 10k ohms to reduce system current consumpton. Short operated his radio from a 9-V battery. I prefer to use 12-V batteries. I didn't have a 9-volt regulator so I used a 78L08 8-V regulator, and calculated the bias voltage divider based on that voltage, to provide 3-Vdc base bias for the MPSH10. (View the schematic).
To clip sideband modulation, Short used back-to-back OA90 diodes, also obsolete. At low currents, it has a forward voltage drop of 180-250 mV. Another trip to the parts drawers turned up the diodes listed in the following table. I measured the forward drop of a small sample.
Test set-up for measuring diode forward-voltage drop at ~1 mA.
| Sample | 1N34 | 1N60 | 1N4148 | 1N5711 |
|---|---|---|---|---|
| 1 | 377 | 316 | 760 | 396 |
| 2 | 347 | 306 | 458 | 400 |
| 3 | -- | 328 | 409 | 400 |
Forward=voltage drop in millivolts of a small sample of diodes
I have only two, used 1N34s, and they didn't match well anyway. The 2N5711 Schottky diodes were the most consistent, but their forward voltage drop was about 100 mV higher than the 1N60s. BGMicro has 1N60s for 59 cents. The 1N5711s are less expensive and more consistent, but the lower forward voltage drop of the 1N60s got them into the receiver. Two more diodes act as the demodulator. which I followed. The second pair of diodes, D3 and D4 in the partial schematic above, are actually a mixer, where detected audio is re-mixed with the amplified carrier created by limiting diodes D1 and D2 and amplified in TR2. My modified circuit functions the same way. Note that D3 and D4 are connected like a half-wave voltage-doubling rectifier. I noticed that Mike Tuttle used a single-diode demodulator in his version; I think he'd see more audio output if he used Short's method,
The rf stage is built on a 3x2-inch board with a partition between the 2N3819 and MPSH10.
The ferrite rod antenna I had came from a Zenith portable radio. Its main coil measured about 725 uH. I unwound about 20 turns to reduce it to 230 uH, so I could use it with a 365-pF variable capacitor.
It was tempting to use an MVAM109 varactor, but that vernier dial was begging to be used. After I'd drilled the front panel to mount it, I discovered its shaft coupling was undersized, and the shaft of my variable capacitor wouldn't fit into it. A vernier that won't work with standard 1/4-inch shafts is useless, so I grabbed the coupler with Vise Grips and drilled it out. That was the hardest part of this project, which went amazingly well. Sometimes even the simplest projects can be frustratingly difficult to complete.
The only other changes I made to Short's design were to dispense with is RC low-pass filter and substitute a FET audio preamp, like the one in my Utility Audio Amplifier.
But does it work?
Yes, the modifications work just fine. The first time I powered up the rf board I didn't have an audio amplifier connected. The oscillator started right away and I could see the frequency changing as I tuned it, by watching my oscilloscope. My utility audio amplifier let me hear more AM broadcast stations than I have been able to receive on an indoor antenna connected to any other receiver I've tried. A standard 3/4-turn Regeneration control provides enough resolution, though like any regen receiver it gets touchier at the high end of the band. Adjusting the trim pot got the oscillator working with the regeneration pot mid-range over most of the band. By the way, I used a 1000-ohm trim pot because I didn't have a 500-ohm pot.
Tuning this receiver is pretty much like using any regen, except for weaker signals you want the oscillator to be running. Short relied on the RC low-pass filter to reduce adjacent-channel heterodynes; I plan to use an external audio filter (Autek QF-1A) that includes excellent notch filters. Properly adjusted, though, the receiver will clip the sidebands of the station you're listening to. When you're trying to dig out a weak station you may need a filter. The gain offered by the amplifier, and the generally excellent selectivity more than compensate for this deficiency. When the signal is strong enough you can back off the Regeneration control and eliminate heterodynes. Then it operates like a traditional regenerative receiver.
Continue to page 2 for my impressions of this receiver.