A 40 meter SSB QRP transceiver

How low is "QRP"? Some people define it as anything up to 100 Watt, but the more commonly accepted limit is 10 Watt input to the last stage, or 5 Watt output. There are many contests in which one has to stay within 5 Watt output to be considered a QRP station. This was one of the reasons to design the transceiver described on this page for 5 Watt power output. Another reason was that this power is a good compromise between being heard, and not needing too much energy, a very important consideration in backpack portable operation, which was the main purpose of this radio when I designed it back in 1987.
 

History

Following that, in 1982 I built my first radio that included some design of my own. It was based on another QST article for a CW transceiver, which I modified for DSB. This radio too was a direct conversion design, and worked quite well. Then, in 1985, I built a more elaborate version of that same thing, with higher power. Finally, in 1987 I turned my back to direct conversion, and designed the radio described here, from scratch. This is still the crowning of my QRP career. It works very well to this day, has never had any failure and always attracts crowds when I show it off in public.

Some photos illustrating this history are on my homo ludens radiactivus page.

My previous QRP transceivers used to carry just the year as a model number, so I had the QRP 82, QRP 85, and some others. But the performance of this new QRP transceiver was so much in another league than all the previous ones, that I gave it a new model designation: PQD. This stands for the Spanish phrase primer QRP decente, meaning the first decent QRP!
 

The Design

This block diagram shows the basic layout of the radio. You can click the image to obtain a high resolution version of it, which is a lot better for printing. If your browser refuses the open the high resolution image, right-click instead of left-click, choose to save the file, and then open it with a good image viewing program, such as IrfanView. 


Let's start the functional description at the antenna (lower left corner). The receive signal enters through the TX low pass filter and is routed straight to the receiver via passive circuitry. There it passes a band pass filter tuned to the 40m band, and goes without any preamplification to a MOSFET mixer. The reasons for both using a MOSFET-based mixer and not using a preamplifier are routed in the need to get the best possible ratio between receive capabilities and power consumption. MOSFET mixers have a high impedance, so their power consumption can be low; and they need almost no driving power, so there is no need for powerful buffer stages after the oscillator; and they provide a substantial conversion gain, obviating the need for a preamplifier! The downsides are more noise and less dynamic range than some other mixers. But on 40 meters, band noise is so high that the mixer noise is of no consequence; and not using a preamplifier, the incoming signals are usually not strong enough to cause overload. Certainly when out in the wild, there are no strong nearby stations! In large cities with lots of hams, front end overload can occur, but has not been a problem for home use of this radio. In any case, for the power consumption limits I set myself for this radio, the MOSFET mixer offerred the best possible performance.

The VFO runs at a rather low frequency of roughly 2.5 to 2.8 MHz.. This produces a very high frequency stability. The oscillator uses a JFET, and a single buffer stage, also employing a JFET. As a result, the entire VFO/buffer circuit has a very low power consumption, which as an added bonus reduces any heating that could cause frequency shifts. The VFO uses high stability polystyrene capacitors and an air-wound coil. The output of the buffer drives the MOSFET mixer. The output of the mixer goes through a crystal filter, which has a center frequency near 9.8MHz and a bandwidth of 2.4kHz. Then comes a gain-controlled IF amplifier in an MC1350 IC, which is the main gain contributor in this radio, but also the main power consumer! The output of the IF amp goes to a MOSFET product detector, which gets its BFO signal from a crystal oscillator and produces audio output, which after passing the volume control is amplified by an LM386 IC and delivered to an internal, relatively large loudspeaker, or to external phones.

A sample of the IF signal is further amplified by a JFET and a bipolar transistor to a rather high level, rectified, and applied through a gate circuit to the gain control input of the IF amplifier chip. This IF-derived automatic gain control eliminates the typical bumps and pops so common of audio-derived AGC, which is used in so many simple transceivers!

For transmission, the audio signal coming from a tiny electret microphone is directly applied to an IC balanced mixer, which also gets a signal from the BFO. The resulting double sideband output is applied to the crystal filter,  and the SSB signal that comes out of it is amplified by the MC1350, then applied to a second balanced mixer, also implemented in an IC. This second mixer gets the VFO signal, thus producing a SSB output on the transmit frequency, at a power level of roughly 2 to 3 mW. This signal is filtered by a band pass circuit,  and then amplified to the 5 Watt output level  by just two tuned high gain stages. After low pass filtering, the RF power reaches the antenna connector.

A sample of the RF voltage in the output stage is rectified and used for automatic transmit level control. This eliminates the need for a microphone gain control and maintains proper modulation over a reasonable range of voice intensities and mike distances, while also affording a moderate amount of speech compression! Together with a clever shaping of the TX audio response, this gives the little radio a transmit punch that makes many old-timers shake their head in disbelief: "This can't be just 5 Watts!"

TX-RX switching is done with electronic switches, using no relay. A CMOS gate IC and three small transistors take care of this task.
 

Specifications

Frequency coverage:                7.0 to 7.3 MHz
Operation mode:                       LSB
Frequency stability:                  100 Hz in 30 minutes, worst case
Power supply voltage:               12 V nominal, while 10.5 to 14 V is OK

Current consumption:   RX:      30 mA at no output, 33 mA average, 90 mA at max output
                                       TX:      270 mA at no output, 500 mA average, 1 A at max output

Meter readings:                        RX signal - TX power - Battery voltage
Sensitivity for 10dB S/N:          1 µV
USB rejection:                           50 dB
Image rejection:                        60 dB
IF rejection                                60 dB
Audio output power:                  0.6 W
AGC range:                               65 dB
TX output power:                       5 W
Mike sensitivity:                       50 mV
Carrier suppression:                 50 dB
USB suppression:                      50 dB
Harmonics suppression:            48 dB
SWR protection:                        Infinite
Audio processing:                      8 dB compression, high frequency emphasis
Dimensions:                               208 x 128 x 61 mm, including all projections
Weight:                                       0.61 kg

As you can see, the specs are quite attractive for the use intended! And practical use of this radio over many years confirmed its great usability and fun!
 

Circuit design details

Here is the schematic diagram. You can click on it to download a high resolution version for printing. As with all high resolution images on this web site, if your browser refuses to open it, right-click, save, and open it with an image viewer.

If you live in the USA, you may be surprised to see resistors drawn as rectangles instead of zigzag lines. This is the European standard, and I hope you won't be bothered by it. At the time I designed this radio, I drew all my schematics in European norm, and even if the schematic shown here was re-drawn by computer in 2002, I tried to make it look as close as possible to the hand drawn original of 1987! I did not use a scan of the original, simply because a hand-drawn and scanned image has an immensely larger filesize and lower quality than a digitally generated image!

Note that all inductors have their inductance values shown. 3µ2, for example, means 3.2 µH. When there are taps, their position is shown in percentage of turns from one side. All values of adjustable inductors are nominal values, and the inductors should be adjustable over 30% or so to each side, unless noted otherwise.


Now, let's start in the lower right corner, where the VFO sits.

The VFO is a Colpitts design with Clapp reminiscences. It uses a 450pF (maximum) variable capacitor taken from an old AM pocket radio. As these variable capacitors have a nonlinear capacitance to position ratio, I used a series capacitor to almost fully linearize the response. The oscillator coil is wound on a former with a small adjustable ferrite core, using enough turns to barely introduce the core into the coil for the desired inductance. This reduces any instability caused by the core. The VFO output is amplified by a tuned JFET buffer, which has two taps on its tank: One delivers several Volt of RF to the MOSFET RX mixer, while the other delivers a small RF voltage to the IC TX mixer. The oscillator and buffer are powered from a three-terminal regulator, which also powers the crystal oscillator, while everything else runs directly off the 12V supply.

Now let's jump to the lower left side. From the antenna connector, the signal gets through a low pass filter section and is coupled into a double tuned circuit, which passes the 7 to 7.3 MHz  range and little else. Two diodes across the first tuned circuit will clamp the high level  RF reaching here during TX. The output of the filter gets to one gate of the mixer MOSFET, which gets the VFO signal on its other gate. The bipolar transistor at the output of the filter conducts during TX, thus avoiding the TX signal leaking through to the following stages. During RX this transistor is off.

The mixer  has its drain permanently powered from 12V, while the source is grounded during RX and put to 12V during TX. In this way the mixer is switched off during TX. While receiving, the IF signal appearing on the drain is filtered by a tuned circuit, and  link-coupled into the crystal filter. As you may know, crystal filters require well defined input and output load impedances to work well. The turns ratio of the transformer, together with the 10k resistor across the primary, play a vital role in this. The circuit's resonance is broad enough to properly terminate the filter on all frequencies where this is important.

This signal is also applied to the TX mixer's output, but since this one is off during RX, it does not significantly load that line.

The crystal filter's output is impedance-matched with a PI network to the IF amplifier. As this IC has a well defined input impedance, it can be used to load the filter, achieving best efficiency. The differential output of the IC is converted to a single-ended one in a tuned circuit, the output of which goes to the product detector. The second gate of this detector MOSFET gets a signal from a simple crystal beat frequency oscillator, which can be set to the exact frequency by a small trimmer. The output of the detector is low-pass filtered to remove any RF signals, and the resulting audio is fed to the volume control and then to an LM386 amplifier IC, configured for its minimal gain, which is ample in this case, thanks to the high conversion gain of the MOSFETs. The 386's output goes to a headphone connector, which passes it through to the internal speaker when no headphone is connected. This speaker is a relatively large one - as large as I could fit in the small box of this radio. I learned early in my portable operation career that the added weight of large speakers pays for itself: They are a lot more efficient than small speakers, so less audio power is needed, which reduces battery drain! 50 more grams in a speaker can save several hundred grams in additional batteries!

The IF amplifier output is also passed through a trimpot and applied to a two-stage tuned amplifier, followed by a detector. The resulting AGC signal is current-amplified by a transistor powered from the 12V RX bus, which serves several purposes: It switches off this AGC source during TX, makes the proper fast attack and slow release of the AGC in conjunction with the 100nF capacitor at the base, and affords quick discharge of this capacitor during TX, so that a new RX period will not have AGC hangover from previous reception.

When the operator presses the PTT button, a 4069 CMOS IC will get this signal, and switch on the 12V TX bus, which was actively held down during RX. It will also switch off the 12V RX bus. Several things happen at this point: The RX MOSFET mixer goes into high impedance, the product detector shuts off, the RX AGC line is interrupted, the RF input to the receiver is clamped to ground, while power is applied to the electret microphone, the two TX mixers and the the bias circuit of the TX driver. This sets the scene for transmission.

Electret microphones are very small, lightweight, inexpensive, and contain an internal FET that delivers a pretty high output level. This allowed me to completely skip any further audio amplification, and instead apply the microphone signal directly to the rather sensitive AN612 doubly balanced mixer IC. The relatively small coupling capacitor produces a low frequency rolloff, which is surprisingly effective in making the signal sound larger than it is! 90% of a voice's power is concentrated in the low frequency, while 90% of the intelligibility is in the high frequencies. Thus, it's a really good thing to emphasize the high range, and avoid wasting too much of our precious power on the useless lows! This radio invariably draws reports of "extremely crisp, clear, highly intelligible audio"!

It's important not to apply excessive carrier voltage to the mixer IC, because it will simply block, and deliver no output! The capacitive voltage divider at the crystal oscillator cares for this task.

The mixer's output is applied to the crystal filter. At this time, the RX mixer MOSFET is in high impedance, guaranteed by having low voltage on its gate 1 and 12V on both drain and source! So it will not conduct, allowing the 10k resistor across its tuned circuit to keep performing its task of providing a well defined source impedance for the filter, into which the TX mixer injects its signal through the rather high 2.2k resistor. The crystal filter selects the lower sideband, as the crystal oscillator sits a little bit above the filter's passband.

The MC1350 works in TX just like it does in RX, only at a higher signal level. Its output couples into the second AN612, which gets a low level carrier signal from the  VFO buffer and generates outputs on 7 and 12 MHz, of which the following double tuned bandpass filter selects the 7 MHz one.

The collectors of both the driver and power amplifier transistors are permanently powered from the 12V line. The bias of the driver is controlled by the switching circuit. When enabled, the driver transistor is biased into class A, and its emitter current is used to bias the power amplifier into class B operation! This is highly power-efficient, and at the same time allows to switch the entire power amplifier chain with just a few mA! This is a scheme I invented, and I was mighty proud of it at the time. Years later I have seen it in use by other designers, who probably came up with it independently of myself. It's just such a logical thing to do!

A small note about the 1N5401 bias diode: This is a 3 Ampere diode, but don't be tempted into replacing it by a smaller one! It has to conduct much less than one Ampere, but it is still necessary to use the big diode, because smaller ones have a slightly larger voltage drop, and will over-bias the power transistor, making it run at low efficiency or even burning it out by thermal runaway! Don't even replace this diode by one rated at higher voltage, such as the 1N5408, because this one also has a higher voltage drop!

Otherwise these two stages are pretty straightforward. The band pass filter is impedance matched to the driver transistor's base, while L-C-C networks do the interstage matching and the power amplifier collector matching. The transistors are intended for 27 MHz operation, and at 7 MHz they are quite "hot", delivering high gain but also loving to self-oscillate. Good construction practices are essential for achieving stability when putting this much gain into just two RF stages which are tuned at all ends!

A simple RF rectifier samples to collector RF voltage of the power transistor, and produces an ALC voltage from it, fed into the IF amplifier. This achieves automatic transmit level control, and the time constant was selected such as to provide a moderate level of speech compression. As a byproduct, this circuit limits the output during high SWR conditions. The trimpot is set to obtain 5W output power.

The power supply is routed through a fuse and clamped by a reverse-connected diode. In case of applying the wrong polarity, the fuse will open, avoiding further damage. Then comes the power switch, linked with the volume control. A separate toggle switch allows to power two little lamps that light the dial and the meter. During normal battery operation these lamps are left off, since they drain several times as much current as the entire radio in RX operation!

The meter is normally connected to the AGC/ALC line, but can be switched over to the 12V bus. In both positions there are trimpots to properly calibrate meter response.

Mechanical construction

A frame made from 1mm thick aluminum sheet runs around the printed circuit board, forming the front and rear panels, and the internal sides. Two almost symmetrical U-shaped pieces of the same material form the top and bottom covers while also doubling the sides. This allows very easy access to both the component and solder sides of the board. The speaker is glued to the top cover, which is properly drilled to let the sound pass.

The printed circuit board measures 20 x 10 cm, and is double sided. The top side is left mostly intact, to serve as ground plane, except only in the VFO area, where the top side was etched away, in order to avoid unstable parasitic capacitances which could affect frequency stability. All ground connections are soldered on the top side, which is made easy by the relatively low population density of the board. Some parts, like the IF transformers, have to be mounted a bit off the board to allow soldering under them. All holes which are not ground connections had their top side copper cleared away around them, using a 3mm drill bit.


A plated-through board would of course avoid the need for top side soldering, but I didn't have the technology back then to make one! The board was made using my home methods of the time: I designed it with a graphite pencil on plain paper, then clamped a transparent foil onto the design, applied rub-on pads, then made the traces using self-adhesive black tape of different widths. This original was then photographically copied onto the board, using Kontakt-Chemie Positiv-20 spray, and then the board was etched.
The result is what you can see here! The etching quality was quite good, even if the layout does not look as clean and computerish as a modern software-generated one. But for RF use, these softly curved tracks are actually better than the sharp bends most PCB software makes!

I made a high resolution scan of the original PCB layout. It will probably not be usable as a mask to directly make a PCB, but should be useful as a base on which to do a digital edit for generating a workable mask.

I did never make a component location diagram, but instead I made a high resolution photo of the radio's inside. You can use this photo to find out which part goes where, if you copy this design!

The unsoldered pins are those ground connections that are soldered on the top side. Note the three jumpers: The red one is a 12V line, while the other two are oscillator signal connections, made with small shielded wire. There are no jumpers on the top side.

The 100 Ohm resistor was originally intended to serve as winding core for the decoupling choke of the power amplifier stage, but later I found it better to wind this choke with a larger diameter, and so I mounted the resistor on the bottom side.

The spacing between the PCB and the bottom plate is barely 5 mm.


Here's another view of the radio's inside, exposing the otherwise hidden components along the rear panel, among which the TX power amplifier chain is prominent.

At this time you may be asking where I found all those special components, like the crystal filter. Well, it's quite simple: The early 1980's brought a rush of Citizen's Band activity, and by 1987 the rush was ending. There was a surplus of CB radios, so if a CB radio failed, the likelihood for the owner to send it in for repair was small. During the early design stages of this radio I asked around, and was given an old Royce 639 transceiver, which had a burned PLL chip and thus could not be repaired cheaply. PLL chips were not easily available here. I took that Royce apart, and used the crystal filter, BFO crystal, signal meter, coil forms, and many other parts from it!


Here is yet another view, that allows you to peek at the back side of the front panel. The volume control potentiometer I found was not intended for PCB mounting. So I simply wired it to the board, just like the connectors and switches.

Note the wiring of the illumination system. Actually, there is a resistor in series with the two tiny bulbs, and this resistor is not in the schematic... I could not find 6V bulbs, so I had to use 5V ones!

You may have noticed that I used RCA connectors for both the 12V supply and the antenna, which may look like a dangerous situation. But it is not! If you mix up the cables and connect 12V to the antenna input, nothing will happens, since it is DC-blocked. RCA connectors behave quite well at RF, are much lighter and smaller than common coax connectors, and also can easily carry the 1 Ampere of supply current needed by this radio. The only true disadvantage is that a male RCA carrying 12V power poses a short-circuit hazard. But I had standardized on RCA connectors for all low current 12V needs in my shack in times when I still didn't think much about safety, so I didn't want change this standard. In fact, to this day I still use RCAs for all small 12V connections, and simply use care to avoid shorting them.

The 3.5mm connector in the middle of the back panel is not on the schematic. It is simply an output of the 12V TX line. I used that output to switch an external linear amplifier for QRO operation at home. The first amplifier I built was a broadband push-pull design, using two CB transistors and delivering about 30 Watt. Then I built the big one, which used two 6146B tubes in parallel, powered from slightly over a kilovolt, running at a 300 Watt input level! By the way, that one was my first tube project.

But what will probably most catch your attention is the tuning mechanism. I made a detail scan of that area of this photo, to better show it - because it's really worthwhile to see! :-)


The gray piece is a bracket specially crafted from 1mm aluminum sheet, that holds both the variable capacitor and the tuning axle. And the tuning axle is, well, nothing else than a discarded Parker ballpen refill! The thick part of it has the right diameter to carry the large tuning knob, while its tip is nicely thin so that it will have precise guiding and low friction, while also allowing a great reduction ratio between the tuning knob and the variable capacitor. It is hollow, saving some weight, and made of high quality steel. And it was free!

The ballpen refill was simply cut to the proper length, inserted in the bracket, and two washers were soldered to it, in order to keep it in place. The original pulley of the old AM radio was kept on the variable capacitor, a cord was strung around to complete the reduction drive of about 9:1.   The capacitor tunes from 7 to 7.3 MHz over about 120 degrees, so this gives about 100kHz per revolution of the tuning knob. This is quite fast for older gents, but I never had trouble precisely tuning in a station. It is millimeter work, in any case, but perfectly possible to do quickly.

The dial was made of stiff paper, and calibrated by measuring the frequency on a frequency counter. The markings were made with a graphite pencil, which is a lot more permanent than ink pens of any kind! The dial was glued to the pulley, and another piece of paper was glued in as a backplane, to keep people from looking into the radio through the dial opening! The small bulb was glued to the front panel in such a position as to evenly illuminate the portion of the dial visible through the front panel cutout. A small piece of clipped resistor lead serves as a pointer. On the dial, 10kHz markings are only about 1 mm apart, but still the dial is precise enough to get within 2kHz or so of any given frequency - close enough to find your buddy or a net!

Using this radio

Over the first months I made a lot of interesting contacts. This photo was made for the 1988 publication in RadioAfición, showing a colorful sampling of QSL cards earned with this radio: A contact with a DXpedition on the highly wanted San Felix Island, the card of a special event station that followed Pope John Paul II during his 1987 visit to Chile, and some less special QSL cards, including my own instead of a signature... The other items in the photo are all that's needed for a complete backpack HF SSB station: The radio, the tiny microphone, a battery box with 10 NiCd cells of 1.2 Ah capacity (with modern cells that box would be a lot smaller!), a small solar panel to charge the battery, and the antenna:  This antenna is a full size dipole made from thin, flexible AC power wire, fed by RG174 miniature coax cable. This cable saves a lot of weight and causes less than 1dB loss on 40m. Three pieces of 3mm nylon rope, of three different lengths, are included in the roll. This allows to quickly raise the dipole, using any tree, post, mast, roof, or whatever as a support.

Forget about loaded whip antennas for QRP use! Such antennas have a low efficiency, and an inefficient antenna fed by low power is an excellent recipe to not be heard! But with the full size dipole in a decent location, this radio generates surprising signal reports! I often got reports of 5-9 +20 from places 500 km away, and usually these reports came paired with comments about the beautiful, crisp and clear audio!

Is it for you?

I know of roughly 10 people who built this radio after I published it in RadioAfición. Some of them also gutted a Royce 639, like I did, to use the parts. Others modified the details, for example by using a crystal ladder filter made from widely available 10 MHz microprocessor crystals. Whatever you decide, exact copy or rough copy, you should first find out if you can get the parts you need. Miniature plastic variable capacitors are to be found in many hams' junk boxes, but are not easy to buy new nowadays! Also I don't know if the AN612 ICs can still be found. So, shop around before deciding!