15-Band World Receiver          

In the year 2010, thirteen years before writing this article, I happened to have a lazy day, and decided to play a little with electronics. At first it wasn't clear what exactly I would play with, but rummaging through my junk box I happened to re-discover a bunch of old MPF201 transistors. No, that's no typo! They are not the much better known MPF102 JFETs, but the pill-shaped MPF201 dual-gate MOSFETs.  When I was a university student, several decades ago, I had gotten a chance to buy all the ones a store in my town had, for very little money. The store owner was glad to get rid of them, since nobody else wanted those strange things. I bought them because it's always good to have a bunch of RF dual-gate MOSFETs at hand, but had never really used any. So, playtime began!

Dual-gate MOSFETs work pretty well as variable-gain amplifiers, for example in IF amplifiers of radios. So, to see what these things are capable of, I grabbed a small scrap of unetched circuit board, some nondescript 10.7 MHz IF transformers, and built a two-stage IF amplifier, in dead-bug style. It worked quite well. And the rest is history, as they say!

During three years I kept adding parts, ran out of space on that little board, added more boards, added even more parts, and after a lot of play time the thing grew into a fully fledged, microprocessor-controlled, synthesized, pretty good receiver, that covers all short wave and medium wave AM  broadcast bands!  Complete with 100 memories, notch filter, and whatnot.

The idea was to make a modern classic, by using an eclectic mix of modern and antique technology, building it into a chassis, and installing that chassis inside a beautiful lacquered wooden cabinet, maybe even one in cathedral style. But when the radio was technically ready, working perfectly, I got lazy, and never built that nice cabinet! So the radio lingered in my closet for years, in the growing section reserved for half-finished projects, until I decided to use it as it is, without a cabinet, and placed it on my bedside stand, where it now serves duty to lull me into sleep every evening, with the classical and irreplaceable sound of shortwave broadcasts.

I delayed writing and publishing a web page about it until now, in 2023. Given that apparently I will never get to building a cabinet for it, I decided to publish it as it is, naked, bare bones, as it has been for 10 years now.



Here is the complete schematic diagram, of course much too small to be legible, but you can click it to get the full size version.

The antenna signal first goes through a 6-pole bandpass filter. There are just 5 filters, despite this being a 15 band radio. Each filter is used for a relatively broad frequency range, and most of these ranges include cover several broadcast bands. The filters are diode-switched.

The signal then goes through a 6-pole IF trap, and then via an impedance matching transformer to an SA602A doubly balanced mixer. I was a bit hesitant at first to use this mixer IC, due to its fame of having a relatively small dynamic range when compared to diode or JFET mixers. But it turned out to work plenty well enough for this task, and the simplicity it provides is appealing. The only downside really is that I had to add a dedicated voltage regulator IC for it, to provide the mixer with a supply voltage it likes.

The IF signal delivered by the mixer goes through an 8-pole crystal filter, made up from two cascaded 4-pole filters, each consisting of two 3-legged crystals. The filter provides about 12kHz bandwidth, which I find optimal for AM shortwave broadcast reception, because it will pass the carrier of a signal even when the receiver is tuned 5 kHz high or low, which in many cases allows rejecting interference from a neighboring station by simply tuning to the far side of it.

After the IF filter comes the circuit with which the whole project began: The two-stage IF amplifier using two MPF201 dual-gate MOSFETs. It's a very simple circuit, and it uses a very old, time-honored technique that comes from the age of the very first tube radios: A bias battery!  Dual-gate MOSFETs need an AGC signal that gets a little below their source voltage, to provide a wide gain adjustment range. This is most commonly done by lifting the source voltage a little above ground, using various bias schemes, but I found that the MPF201 likes high supply voltages, and since my main supply is only 12V, I decided to simply leave the sources at ground to let the MOSFETs get the full 12V, and place a little coin cell in series with the AGC line, so that the bias voltage can go 1.5V below ground! Since the gates are insulated, no current is ever drawn from this cell, and it will last as long as its shelf life allows. I used a silver oxide cell, and expect it to last for several decades, probably longer than I will last.

The detector is a conventional one, using a real, true, old germanium diode. This is another element that defines this receiver as a modern classic! To load the detector with a high impedance, an operational amplifier is used as a buffer, and another operational amplifier (actually the second section of the dual op amp IC) acts as integrator to provide the proper AGC voltage.
 
The audio chain looks pretty long, but really isn't that complex. It first features a 5kHz low-pass filter, which suits the audio bandwidth used by most AM stations (although some are wider). This is an 8-pole filter too, giving a pretty sharp cutoff, and it's built around a quad op-amp. And after it comes a special feature of my radio: A fixed-tuned, sharp and deep 5kHz notch filter, which totally eliminates the nasty high-pitched carrier beat that usually plagues shortwave broadcast reception!  Since almost all broadcast stations use very accurate frequency control, these beat tones are almost always very precisely at 5kHz, and so they can be rejected by such a fixed filter. This notch filter is built around a dual op-amp.

The end of the audio chain is a GL386 power amplifier chip. This is an equivalent of the very well known LM386, but I found the GL version, made by Goldstar/LG in Korea, to have lower distortion than the original LM versions found in my junk box! It's just a fractional watt amplifier, but when using a decent speaker this is plenty enough power to give room-filling volume.

The local oscillator is an AD9951, a high-performance Direct Digital Synthesizer, not to be confused with the much more humble AD9851. It's the most expensive part of this radio, and since it comes in a 48-pin SMD package that has a pin pitch of 0.5mm, and has a metal patch on its belly that needs to be soldered to a small heatsink, it's not easy to mount for an aging homebrewer. It took me three attempts to get all pins soldered down, without shorts!
 
I used the internal reference oscillator of the AD9951, with an external 20MHz crystal, and used the internal PLL frequency multiplier of the chip to produce a 400MHz clock signal, which drives the DDS proper to synthesize a 14-bit resolution sine wave in the range of  roughly 11 to 37 MHz. The radio uses high-side oscillator injection on all bands, but this could be easily changed in the firmware, if so desired.

There is no trimmer to fine-tune the crystal frequency. Instead the software in the controller allows setting the required compensation for any frequency shift of the crystal. By the way, the software also allows to shift the IF frequency, in case the crystal filter isn't spot-on, or shifts over time.

The output of the DDS chip goes through a balancing transformer and a 5-pole low-pass filter, and then on to the SA602A mixer.

The DDS chip is powered through three separate, dedicated voltage regulators! One provides the 3.3V required for its input/output circuitry, while the other two are 1.8V regulators, one for the digital and one for the analog side. This radio is really a celebration of 3-terminal regulators! There are a total of six of them.

The controller is based on a very inexpensive 28-pin PIC microcontroller. It does everything: Reading the encoder used for all user input, driving the display, controlling the DDS, switching the bandpass filters, storing the memory frequencies, sensing the AGC voltage to display a signal strength indication, and it even senses the supply voltage, to quickly store the operating conditions (frequency, memory channel, tuning mode) into non-volatile memory when the user switches off the radio, while the voltage is falling!

It also supports low-voltage in-circuit programming, and a DB-9 connector is included to reprogram the PIC without having to remove it. This allows correcting for aging-induced frequency shift of the master oscillator crystal or the IF filter, changing the band limits when the broadcast frequency ranges are changed, making improvements, adding features, fixing bugs - but I haven't found any bugs in 10 years!

The display is an extra large, backlit alphanumeric LCD, that has 2 lines of 16 characters each. It's conected by means of a DB-15 connector, so that it can be easily disconnected when taking out the chassis from the cabinet - if I ever build a cabinet, that is!

All of the radio's frequency-related operations, like tuning, band switching, memory programming, recalling and deleting, are done with a single optical step encoder that also has a pushbutton function. I chose the more expensive optical encoder because I have seen too much trouble with mechanical encoders, even after just a short time of use. The optical ones instead work flawlessly, virtually forever. That's important in a radio that needs lots of tuning up and down in the bands, hunting for interesting stations.

Oh, let's not forget to mention that the PIC also has its dedicated 5V regulator! Its input is buffered by a large electrolytic capacitor and a diode, to keep the supply to the PIC going for a moment while it's already coming down for the rest of the radio, so that the PIC has time to write the operational data into its non-volatile memory.

The last part of the radio is the power supply. It's a totally standard, run-of-the-mill transformer-based, regulated 12V power supply, using yet another 3-terminal regulator!



Here you can see the general layout under the chassis. On the left is a PCB material scrap to which I soldered the perfboard carrying the controller circuit, an SMD carrier board that carries the frequency synthesizer, and also its lowpass filter. The TO-220 soldered to it and screwed to the chassis is the 3.3V regulator. It needs heatsinking because it's dropping 8.7V, and the DDS chip it powers consumes a considerable current.

The little board scrap in the middle carries the power supply parts: Rectifier diodes, capacitors, and the 12V regulator, which is also bolted to the chassis.

The right top unit is the bandpass filter board, full of homemade polyethylene coil formers. The three IF cans on its left are the IF trap.

The right center board is the receiver chain. At its left is the impedance matching transformer (toroidal), the mixer IC, the 6V regulator powering it, then comes the IF filter with its input and output tanks, then the two IF amplifier stages, the detector, and the op amp circuit that provides the AGC and buffers the audio signal. The bias battery is on the lower right of that board.

And lastly, the board in the lower right of the chassis contains all the audio circuitry.

The left knob is attached to the optical encoder, while the right one is the volume control potentiometer, with on/off switch. Totally classical! The antenna connector is on the upper right, the PIC reprogramming connector is on the upper left, and the power cable, a safe, three-wire cable, in the middle.



Here is a closer view of the filter, mixer/IF, and audio boards. Long life to the glorious dead-bug construction style, which is fast, practical, and often even better than PCB construction, if done on a continuous ground plane! It's just not particularly service-friendly. But then, this is a good radio, and should never need service! :-)

Many of the discrete parts are recycled ones. They came out of old TVs and VCRs.

Note that I simply bolted the board scraps directly to the chassis. Simplicity rules.





The PIC circuit was built on a scrap of perfboard, of the style that has a copper pattern laid out just like a solderless protoboard. I find this sort of board most practical to build such circuits.

The DDS instead had to be built on an SMD carrier board, due to the minuscule size and horribly tight pin spacing of the IC. I mounted the 1.8V regulators, the crystal, and most of the passive parts right on that same board. Some are legged parts, while others are SMDs. The balancing transformer and lowpass filter were built directly onto the base board.

The DDS IC needs to have the thermal pad on its underside soldered to a heat sink. It doesn't need to be large, but the IC is intended to be soldered to a copper groundplane on a PCB, connected to the other side of the board by many small thermal vias. I can't do that without designing and ordering a custom PCB. So I drilled a hole into the carrier board, then soldered down the pins of the IC, and then soldered an U-shaped piece of copper sheet to the IC's thermal pad, through that hole, and soldered that copper sheet to the groundplane of the board. It doesn't look tidy, so I'm not showing it, but it works!
 
At the left edge of the image you can see a part of the optical encoder. It's quite small, but good. I have already used it a whole lot, without any signs of wear so far.

Here is a close-up of the DDS.

The pins of this IC are 0.3mm wide, and have 0.2mm clearance between them. I exhausted all my extensive vocabulary of power expressions, in three and a half languages, while soldering them!

It's a real shame that so many nice ICs aren't available in homebrew-friendly case styles.




The top of the chassis looks pretty boring, when compared to an antique tube radio. The only part mounted on the top is the power transformer, and the connectors for the cabinet-mounted devices: A DB-15 for the display, and an RCA phono connector for the speaker.

The volume control potentiometer came with a very long shaft. The encoder instead came with a very short one. If I ever build the nice wooden cathedral-style cabinet, I will have to cut off the excess length of the potentiometer's shaft, and make an extension for the encoder's shaft. So is life. It's never fair! Some get it all.

Note that the exact location of the two controls on the chassis' front side is intended to look nice and balanced on that wooden cabinet I have in mind...  The display will sit in between the two controls, slightly higher, and the speaker will go atop it. A cathedral cabinet lends itself very well to such an arrangement.


And here is a quick view of the rear side.

I used an SO-239 antenna connector just because all my HF ham radios use them. So I can easily use the same antenna cables.

You might wonder about the nice chassis?  I made it from a single piece of 1mm thick aluminium sheet. My technique is to draw the flat design in CAD software, duly considering the material thickness when designing the overlaps. Then I print the design on paper, stick the paper to the aluminium sheet using double-sided tape, then I cut the sheet using metal shears. Then I flatten the piece in a vise, to fix the deformation caused by cutting. Then I use a hobby knife to cut through the paper and into the aluminium, along all bend lines. Then I drill all the holes through the paper and metal. Then I remove the paper, deburr the edges, and bend the aluminium sheet, which is easy and precise thanks to the knife cuts.

The four sides of the chassis are held together by sheet metal screws.



The firmware for the PIC compiles in PICBasic Pro, version 2.60. It can be edited with any text editor, and then compiled. I included a lot of comments, so you should be able to understand the program. It's long, but not complicated. A lot of the length goes just into the bandswitching routines. There are many instances of the same small actions, for each band, with the proper values for each. It's not particularly elegant programming, but it works fine.

The source code includes two encoder reading subroutines. One is for encoders like the one I used, which advance the 2-bit coding by a single step per click, going through a full cycle in 4 clicks. The other is for encoders that go through a full cycle of steps in each click. I started developing this program using a mechanical encoder that used the latter system. If you copy my radio but use such a full-cycle-per-click encoder, just remove the comment characters from the first encoder routine, and comment-out the second one, that I left active.

The constant "steps4k" can be tweaked to correct for frequency error of the reference crystal.  The "frecif" value can be tweaked if you need to shift the intermediate frequency.
 
My firmware uses 5kHz frequency steps in all shortwave bands, and 10kHz steps in the medium wave band, consistent with the standard in my part of the world. If you want to copy my radio and use it in a country where the medium wave band uses 9kHz stepping, you will need to modify the program accordingly.

Also if you want to add any bands that I didn't consider, or change any band limits, you need to edit the program.

So there are a good number of reasons why you might need to edit and recompile the program. But in case you can live with the values I used, I'm including the compiled hex file too. You can directly burn this into your PIC, using any PIC programming tool. This hex file also contains the correct configuration of the PIC. If you recompile the program, you might want to program the PIC only with the recompiled program data, and burn the config word of this hex file into it, because PICBasic Pro will most likely set the config word to something different and incorrect, unless you edit the configuration files for the PIC16F886 that come with the compiler.



The whole radio is operated just with two knobs. The left knob controls the volume, and turns the radio on and off. The right knob is the do-it-all: It can be rotated in steps, and pushed.

The radio has three main operating modes: Tuning, Band Switching, and Memory. Short pushes on the knob cycle through the three modes.

In tuning mode, the display shows the band and the frequency in the upper line. The second line is only used for the signal strength indicator, which uses a single character, that grows and shrinks and changes shape according to the signal level. 21 different signal levels can be shown.

When rotating the encoder knob, the frequency changes in 5kHz steps in the shortwave bands, and 10kHz in the medium wave band. When the limit of a band is reached, the radio automatically moves into the next band. If moving up in frequency, it jumps to the lowest frequency of the next higher band, and when moving down it jumps to the highest frequency of the next lower band. When it reaches the end of the highest band, it wraps over to the beginning of the lowest band, and vice versa. This allows tuning continuously through all AM broadcast spectrum, without having to care about the bands.

A short push on the knob takes us into band switching mode:

The band indicator gets underlined, and rotating the knob will cycle through the bands. While cycling up, we will land on the lowest frequency of the next higher band, and while cycling down we land on the highest frequency of the next lower band. Of course when reaching the highest or lowest band the radio covers, it wraps around.
Pushing the knob after having changed band takes us back into tuning mode.

Pushing the knob two times, starting from tuning mode, takes us into memory mode. It's just like tuning mode, except that the memory channel number is also shown.
Rotating the knob cycles through the 100 memory channels.


On the medium wave band, MW is shown instead of the wavelength, because this band stretches over a very broad range, from about 182 to 560 meters.

 
When listening to a station in tuning mode, and wanting to store the frequency in a memory, one simple gives a long push to the knob. Then the display will show the currently selected memory channel, and its content. Now one can rotate the knob to scan through memory channels and their contents, until either finding a blank one, or finding an unimportant one that one wants to overwrite.


Blank channels show up like this.
Meanwhile the radio continues receiving on the VFO frequency. A second long push writes the VFO frequency into the selected memory channel. The display will then show the newly programmed frequency. But if you decide that after all you don't want to store the frequency, a short push aborts the mission.
 
To erase a memory channel, simply select it, in memory mode, and then give a long push to the knob. The display confirms erasure by briefly saying "Blank".

This single-knob user interface could be refined and embellished a lot, but as it is, it works quite well, so I never further developed it. And please excuse the visible adhesive tape over the display, in these photos! I taped a protective plastic sheet over the display, to keep it safe from damage during development, and that plastic screen is still in place, 13 years later...



The radio, with the current firmware, covers the following bands:

MW:  530-1650 kHz
120m: 2300-2495 kHz
90m: 3200-3400 kHz
75m: 3900-4050 kHz
60m: 4400-5100 kHz
49m: 5800-6300 kHz
41m: 7200-7500 kHz
31m: 9250-10000 kHz
25m: 11500-12160 kHz
22m: 13570-13870 kHz
19m: 15000-15825 kHz
16m: 17480-17900 kHz
15m: 18900-19020 kHz
13m: 21450-21850 kHz
11m: 25670-26100 kHz

To the best of my knowledge, this is the best compromise to get all broadcast stations that transmit on legally assigned frequencies, without getting too many non-broadcast signals. After all, this is a broadcast receiver, not a communications receiver! So it's undesirable to tune into SSB signals, Morse code, RTTY and other digital modes, and so on. It's impossible to completely avoid such unwanted transmissions, because some of the broadcast bands partially overlap amateur and marine bands, and also because pirates transmit wherever they please, without any regard for any band plans at all.

I use this radio with the same multiband antenna I use for my ham radio activities. It's a multiband antenna, consisting of four dipoles, for the 80, 40, 20 and 10 meter bands, with capacitive loads to also cover the 17, 15, 12 and 6m bands. The four dipoles share a common balun and feedline. This antenna obviously was built for the ham bands, and is very much out of tune on most broadcast bands, but since broadcasters transmit with high power, it works fine. Any reasonably long antenna should work with this radio. But a short telescopic antenna won't be very good. The sensitivity of this radio is well matched to a relatively large outdoor antenna, and is not enough for using a tiny antenna. Anyway indoor antennas are pretty useless for shortwave these days, because of the extremely high levels of interference created by switching power supplies in a myriad of electronic equipment, from computers and TVs down to LED lamps and cellphone chargers. We really need to get our antenna as far away as possible from these noise sources. For the same reason, a dipole antenna, with balun and coaxial feedline, is a lot better then a simple long wire used against ground, because in the latter case the whole long wire and the grounding wire harness all pick up signals, and that includes wires that are very close to strong noise sources. In the case of a dipole antenna with balun and coax cable instead, only the antenna receives signals, and the antenna is hopefully farther away from the noise sources.
 


Some of my dear readers might ask what sense there is in painstakingly building an AM shortwave world receiver, in this age of the internet, smartphones, and instantaneous, perfect, almost free worldwide communication in audio, video, text and data?  Well, it's pure nostalgia, of course! I grew up in the glorious times when lots of countries ran huge shortwave broadcast stations, transmitting worldwide in many languages all around the clock. Such as the British Broadcasting Service, Deutsche Welle, Voice of America, Radio Moscow, Österreichischer Rundfunk, Radio Exterior de España, Radiodiffusione Italiana  all'estero, Radio Habana, Radio France Internationale, Radio Berlin International, The Voice of the Andes, and many more. There were some years too in which several Arab countries tried to outdo each other, and certainly the whole rest of the world, in setting up and operating absolutely huge transmitters. One megawatt of transmitter power, into a 15dB gain antenna array, resulting in roughly 30 megawatts of EIRP, was not unheard of. I couldn't understand Arabian languages, but the music they transmitted, often for long times, was certainly nice. It was particularly funny to contrast the views expressed by the official stations of countries on opposing sides of the iron curtain. My country, Chile, was a dictatorship in those years, and it was great fun to listen to the news published by certain politically motivated stations, about streams of blood flooding the streets, terrible hardship of the people, and so on, while living here and knowing better how things really were. Sure, not everything was right and ideal here, we all knew that, and some people did suffer, but it certainly wasn't like those radio stations told the world, and I took those news programs as comical acts. I miss them!

Unfortunately we have lost almost all of this juicy, funny, rich shortwave broadcast scenery. Most of those big stations are gone. A few try to cling to life, but have a hard time. The only really big international broadcaster these days is China Radio International, which has many simultaneous signals, in various languages, with high power, although most is in Chinese, and again I'm limited to just listening to the music they transmit. They also transmit Chinese lessons, but I'm too old to learn Chinese, as much as I would like to do so. Radio Habana is still on, but the transmitter serving my area has had a serious modulator problem for years, so the transmissions are impossible to understand, and strangely the people running that station seem to totally lack any interest or ability in repairing that transmitter... At the same time, their famous opponent, Radio Martí from Florida, USA, is going strong, keeping some of the old days' political propaganda style alive. The world divided into good and bad, right and wrong, black and white, top and down, so childish and simple that it is endearing... I also often hear Radio Romania International, with nice programs about their country, and sometimes various much smaller stations, like Radio Cultura from Mexico, and a guy from the USA who bought one of the large transmitting stations and is broadcasting a mix of paid-for religious and colorful philosophical programs, and in the remaining time retransmits internet radio stations.
 
My main language is German, and I'm a bit sad that there is absolutely no German-language broadcasting on shortwave now. At least I haven't heard any in ages. Back in the 1980's the Deutsche Welle was a big player, and I often listened to them. Gone with the wind...
 
So, some of the old glory of shortwave broadcasting remains, but it's a lot less than there once was. Still I enjoy my evening bedtime shortwave listening sessions. Listening to anything, even in Chinese, mixed with the mysterious noises of nature, atmosphere and the cosmos, has a very special charm.
 
And on medium wave, here in Chile in the evenings I hear a huge lot of Argentinian stations, and also quite a few Chilean ones. Rarely any others, but occasionally I have heard Falklands Radio, at the very low end of the MW band, or some Brazilian station. On the MW band I get some Tango programs, and even some occasional Radio Drama! But MW listening in the evenings is always affected by deep fading, and since several stations share each frequency, I hear one station for a while, then it will fade out and another will rise and replace it. It has its own charm!

Back in the 1980's, WWV and WWVH were very weak here, compared to the huge signals of many big broadcast stations. Nowadays very often WWV and WWVH are stronger than the broadcasters! I don't think that WWV and WWVH increased their transmit power so much, so this change seems to show that in average SW broadcast signals are far weaker nowadays than they were 40 years ago. The megawatt stations are all gone.



I have over two dozen antique tube radios in my collection, most of which can receive short waves. As a ham, I also own several HF transceivers. But what receives short wave broadcast stations best is my homebuilt 15-Band World Receiver! The combination of rock-steady frequency stability, optimal bandwidth, high selectivity, 5 kHz notch filter, 100 memories, and tuning range limited to the broadcast bands,  beats each of my other receivers, in performance and convenience.
 
I don't expect any of my readers to build an exact copy of my radio. The problems would start with procuring the exact same parts! But if some of you find usable ideas or tips in my design, such as the 5 kHz notch filter, or how to control the AD9951 DDS chip, then it made sense publishing this page! 


Back to homo ludens electronicus.