A Medium-Wave FET Regenerative Receiver

September 13, 2009

MW Regen Front Panel

Medium-Wave FET Regenerative Receiver

The Synchrodyne project left me with the mechanical framework for a simple broadcast-band receiver. Like most projects, this one went through several iterations. The longer I worked on it the better it got, and the more encouraged I was to keep working on it.

The Synchrodyne used a single 365-pF tuning capacitor and 240-uH inductor. These formed the tank circuit of the first version. The rf and audio stages were essentially duplicated from earlier regens like the Ham Bands Regen described elsewhere. While gain and selectivity were good, even with a 10-turn Regeneration pot it was nearly impossible to keep the detector on the threshold below oscillation. With a 6:1 reduction drive the tuning resolution was too large for DXing.

Redesigned Tank Circuit
Phase II saw the introduction of a three-section, 365-pF per-section variable cap, with all sections connected in parallel. (Mounting this capacitor so it lined up with the vernier drive was a challenge! I recommend you support it in only one place, at the back, and make the mounting adjustable.) The total capacitance is about 120-1095 pF. Some work with the spreadsheet mentioned on the Ham Bands Regen page gave me some clues for bandspreading the receiver, while maintaining relatively low tank-circuit reactances over most of the tuning range.

The photo below shows a vintage signal-generator coil connected to the rf board during development.

Rf board with signal generator coil

See the schematic; briefly, the AM broadcast band is divided in two. The lower half, 520-900 kHz, is tuned with the full 120-1095 pF in parallel with a 470-pF silvered-mica cap. The necessary inductance, 61 uH, originally came from two molded rf chokes (22 and 39 uH) in series. While chokes are intentionally wound for low Q, the selectivity was more than sufficient, though I later replaced them with a coil similar to the one used on the lower band segment. You don't need to get the Q in the inductor; the Q-multiplying effect of the detector, especially when it isn't oscillating, will provide it.

The high end of the band posed some problems. It would be better to cover the band in at least three segments, but that would have complicated switching. I chose to use a 30-uH inductor and a 200-pF silvered-mica capacitor in parallel. The 30-uH coil is wound on an FT50-61 ferrite toroid. Its Q is significantly higher than that of the series-connected rf chokes. Regeneration was hard to manage toward the top of the band, so I placed a 150k resistor in parallel with the inductor. There was no noticeable reduction in gain, but regeneration control was much smoother.

Perhaps the most-decadent part of this receiver is the bandswitching relay, controlled by the front-panel toggle switch. Lack of panel space ruled against using a rotary switch and another chicken-head knob, unfortunately. Running wires between the inductors with their parallel capacitors, and the switch, was sure to cause hand-capacity and microphonic problems. I happened to have a small rf-rated relay on hand, so I used it instead. Had this project not involved using an existing panel and base plate, it would have been possible to use a traditional switch.

Having solved those problems, I connected the receiver to a tuned ferrite-loop antenna made with an 8-inch (200-mm) rod, probably made with -61 ferrite. Sometimes I coupled an indoor random wire to the ferrite rod, sometimes I used it alone. Over a few nights of listening I was able to hear dozens of stations from the Northeast and Midwest U.S., Cuba, and the Bahamas (I live in Southeast Florida). Some highlights from the first week:

Frequency Station City State/Country
 570 CMDC Santa Clara Cuba
 580 WDBO Orlando FL
 600 CMKV Urbano Noris Cuba
 620 WDAE St Petersburg FL
 640 CMBC Guanabacoa Cuba
 650 WSM Nashville TN
 670 CMQ Arroyo Arenas Cuba
 750 WSB Atlanta GA
 790 CMAQ Pinar del Rio Cuba
 810 ZNS-3 Freeport Bahamas
 820 WWBA Largo FL
 820 CMDE Contramuestra Cuba
 820 WWFD Fredrick MD
 870 WWL New Orleans LA
 870 CMDT Sancti Spiritus Cuba
 880 WCBS New York NY
 900 CMKP Cacocum Cuba
 900 WSWN Belle Glade FL
1030 WBZ Boston MA
1120 KMOX St Louis MO
1130 KWKH Shreveport LA
1140 CMBW Habana Cuba
1180 VOA Marathon FL
1240 WINK Fort Myers FL
1250 WHNZ Tampa FL
1510 WLAC Nashville TN
1670 WVVM Warner Robins GA

Here are some RealAudio® sound clips of later loggings:

This was my first experience with long-term listening to AM-voice signals on a regenerative receiver, and I was hooked. (Like many radio hobbyists, I first got interested by listening to distant AM broadcasts as a youth, but using superheterodyne receivers.) Still, there were some lingering problems. First among them was white noise from the detector. I find it interferes with hearing weak signals (see "Noise Suppression of Low Level Signals," by Yardley Beers, elsewhere on this site).

While testing this version I found there was some interaction between the audio-gain control and the detector. MFJ solved this problem in their MFJ-8100 receiver by placing the audio-gain control at the output of the audio amplifier, where headphones are connected. I used an additional FET, a 2N3819, configured as a low-gain preamplifier, to isolate the detector from the audio amplifier. Adding the stages described below removed the need for that extra stage, so it came out again. The added FET helped drive an LM386 amplifier directly, with the audio-gain control connected across its output (through a 0.1-uF capacitor), for headphone-level listening. I wanted more gain than the LM386 audio amp (a noise generator all its own!) could provide. How could this pretty-good receiver be made even better?

Rf board

RF Board Before Q4 Was Installed

In the photo above, the signal runs from right to left. The RCA phono connector is temporary; the final receiver has a back-panel-mounted BNC connector. The voltage regulator is lower left. This photo was taken before Q4 was installed at upper left. (What looks like board-edge curvature is a lens aberration!) (Schematic of the RF board.)

Improved Audio
Fortunately, the solution to both problems lay in a couple of dual-section op amps, NRC NJM2068s (U201 and U202). Three sections make a Sallen-Key low-pass filter; the last section is a preamp for the LM380 power amplifier. The filter-preamplifier has a cut-off frequency of about 8 kHz. Gain of the filter shown is relatively flat to 5 kHz, and down a little more than 6 dB at 8 kHz. For communications-quality speech a lower cut-off frequency would be better. You can accomplish this by substituting 18k- or 20k-ohm resistors for the 15k-ohm resistors I used. As the filter sections have zero gain, you also could switch the first section into or out of the circuit, without affecting audio level. Use higher-value resistors in the first two stages and you'll have a dual-selectivity filter. Again, as the filter sections have zero gain, there is no limit to the number you can use. To make a more-compact amplifier you could even leave out the first op amp, though that will degrade noise reduction.

I tested this filter-preamplifier with National LM358 and Texas Instruments TL072 op-amps, which are pin-compatible with the NJM2068 (and several other op amps), and it worked the same. I installed NJM2068s in the final version only becsuse I had more of them on hand. The 1000- and 2000-pF capacitors in the filter sections should be polystyrene. The breadboarded version shows ceramic caps only because the thin wires on poly caps make them hard to plug into my breadboard.

Audio filter and preamplifier under development

Audio Filter and Preamplifier Under Development

After I started building the audio amp I learned the LM380 (U203) was discontinued by National Semiconductor, but Digi-Key still had some in stock the day I wrote this article, for a reasonable $1.65 each (Mouser isn't carrying the LM380 or the more-popular LM386 anymore). With the LM380 out of production, you may want to use another IC for the power amplifier. Several TDA-series power amps are capable of producing even more power, with similar gain. They were designed for use in car radios, so they perform well on a 12-V supply. The 60-uH choke and .01-uF capacitor on the amplifier output are a low-pass filter to prevent rf from leaking into the receiver from headphone or speaker wires. The choke should have low Q and and low resistance. I used a power-supply-type toroid found at a hamfest; 12 turns on an FT37-43 toroid will work. (Schematic of the audio filter and amplifier.)

You can get reasonable audio driving an LM386 IC directly. It takes a few extra external components to quiet down an LM386, but for headphone or powered-speaker operation it will work. (Schematic of alternate audio amp using LM386.)

Update November 4, 2009: I tried connecting a Sony ICD-UX70 MP3 recorder to the Line Out jack, and noticed it loaded the audio slightly, so I added a FET source-follower. The source resistor was selected for the J201 audio FET I used; if you use another FET, select the source resistor to provide about 0.5 Vdc at the source terminal.

This article continues on the next page.

Back to KR1S QRP Tech Stuff