When I was a little kid I asked my Dad, (who was a TV repairman), "If a 6 transistor radio works good, wouldn't a radio with 100 transistors work better?"
He simply answered, "Why use a 100 when 6 will do?". That was an important lesion for someone who was to go onto designing QRP radios. Of course, if you count the transistors used in highly intergrated IC's like microprocessors, DSP, and DDS chips often used in today's radios, the transistor count is in the millions. And these radios do work better than the old 6 transistor AM radios of the 60's.
The all discrete component transceiver described below is a throw back to yesterday, when we didn't have fancy IC's to work with. In the QRP sprit of "doing more with less" and the Minimum Art Session (MAS) DL contest of designing a functional rig with 100 parts or less, I set out to design a no IC, VFO tuned transceiver with super-het receiver with reasonably good performance. It quickly become apparent that a super-het receiver with good performance and a very low parts count were not compatible goals and the parts count quickly began to grow. Then once on air listening tests were started, using a good antenna, even more parts had to be added to keep strong SWBC stations from playing havic.
After all the solder smoke cleared, I found I had used over 80 componets in the receiver and some 40 more in the transmitter, for a total of 120+ componets. But it is a full feature receiver with RIT, audio mute, QSK switching on the input, .5 uV MDS (it takes a 1 uV signal to get Q5 copy) and good selectivity. The audio is very low noise and is almost silent without an antenna connected. The transmitter puts out a respectable 3 to 4 watts, depending on supply voltage. And it all fits on a 4" x 2.75" circuit board.
Receiver section schematic:
(diagrams can be printed by first saving to your PC by right clicking on the picture, then use a graphics viewing program to print the picture using scale to page option. If you try to print from the browser, only the part which shows up on the screen will print)
Value changes 6/25/09
The Receiver section:
The signal from the antenna first passes through the QSK switch, C22 and Q6. Q6 is normally biased "ON" via resistor R9 and Q17 is "off". During transmit the gate of Q6 is pulled low via Q10 and Q6 is turned off, while Q9 is turned on, isolating the input to the receiver from the antenna. At the same time, the gate of Q26 is pulled low, muting the audio. R7 allows some audio to leak past Q26 for side tone. R3 and C10 delays the turn on time for Q23 to prevent audio thumps as the transmitter ramps down. Q17 had to be added to provide additonal isolation of the receiver input, as using only the series QSK switch Q6 allowed too much singal into the receiver and was causing transmitter instablility. This was likely due to the transmitter signal finding its way back into VFO via the 1st mixer. Q10 had to be added to provide isolation bewteen the receivers audio mute line and the transmitter keying transistor, as the AGC works with the audio mute circuit. A direct connection between the audio mute line and the transmitter key input caused the rig to "self key" when AGC action started. The gates of Q10, Q11 and Q9 are connected back to the transmitter keying transistor via diode D7, which casues these to turn on as soon as the transmitter is keyed. An R/C delay on the gates of these transistors keep them turned off for a short while after key up, allowing time for the transmitter signal to decay before turning RIT, audio and the antenna input connection to the receiver back on.
The receiver input is double tuned, using a link coupled toroid inductor and a RFC choke for the second inductor. The second tuned stage was found to be needed to reduce SWBC interference.
The first mixer is comprised of two J-FETs connected in series. These emulate a dual gate MOSFET in operation, with the input signal applied to the gate of the bottom FET and the VFO to the top FET. This is a slightly simplified version of the mixer as shown in "Experimental Methods in RF Design".
The mixer output then going through a 3 crystal filter for reasonable selectivity and opposite side band rejection. L1 and C8 form a trap at 7 MHz to keep strong SWBC stations from passing through the mixer (which has little isolation bewteen the input and output) from getting into the crystal filter and bleeding through.
The two transistor IF amp circuit was found in "Solid State Design for the Radio Amateur" and provides plenty of gain. Initially, a single transistor IF amp was tried, but overall receiver sensitivity was poor, despite also having a RF pre-amp and more audio gain.
The product detector uses the same dual J-FET mixer design as the 1st mixer, with C1 by-passing any RF present on the output of the mixer. The BFO signal if provided by Q28, which initally was a simpler oscillator. Unfortunately, the crystal frequency could not be pulled well enough to get it to provide a 600 Hz beat note, so a few additional parts had to be added.
The audio from the product detector passes through the audio mute switch, Q23, then into a high gain audio amp consisting of the darlington pair, Q2 and Q3. Q1 buffers the output of the audio amp so it can drive a fairly low impedance headphones. R21 reduces the amount of current flowing into the headphones and keeps Q1 from getting hot.
AGC: Earlier prototypes of this rig used a simple audio AGC signal to limit the amplitude of strong signals. I decided to remove the AGC system as it really wasn't all that effective. Instead, it was replaced by a simple volume control. The circuit board is laid out so either a volume control or a fixed resistor can be used. In the latter case, a headphone with in-line volume control would be used.
The VFO is a Colpits configuration with large value C feedback caps. C31, C41 and C38 should be polystyrine types for best results, though C0G caps could be used instead. This VFO is much more stable than the original VFO, which was a Hartley configuration using a J-FET. There was just enough drift in the original VFO circuit to be annoying.
Q7 amplifies and buffers the VFO signal before going to the first mixer. This stage was found to be required as tuning the input tank circuit to the mixer would pull the VFO frequency and strong SWBC stations could cause the Digial Dial connected to actually jump around.
The VFO, BFO and AGC control pot voltages are regulated with a 6.8V zener diode, D3.
The schematic shows a single 50 pfd tuning cap, which would be ideally an air variable with reduction drive. The pcb board uses a poly-variable for board mounting and an additional cap, C30 to limit the tuning range. A value of 22 to 47 pfd would be used for C30 depending on how large a tuning range you'd like. I found when using the poly-variable for tuning, the tuning is non-linear. The first 50% of rotation changes the frequency by 15 kHz, but the second 50% changes it by 50 kHz, making the tuning a bit touchy at the higher end. I found a solution to this was to put a fixed value cap across the tuning cap. 22 pfd worked with the poly-variable, but of course, this reduces the tuning range.
Q11 is used to disable the RIT control by shorting it out. This causes R29 and R29 to bias the RIT tuning diode at the center voltage of the RIT control. The gate voltage for turning Q11 comes from the transmitter keying voltage.
Transmitter circuit discription:
The transmitter mixer gets its VFO signal from the Q7 VFO buffer to provent pulling of the VFO when the mixer is keyed.
The BFO signal is provided by the cyrstal oscillator Q19. (NOTE: C57 and C60 should be made 47 pfd to better keep the transmit frequency the same as the receiver frequency) The mixer is a discrete implementation of the old CA3028 transistor array and provides a high level output. The mixer output is filtered by two 10.7 MHz IF cans, retuned for 40 M by the addition of 47p caps, C55 and C56. Mouser part number 42IF123-RC or 42IF122-RC can be used.
Ideally, the gates of the BS170 PA MOSFETs, Q14 and Q16 want to see a narrow signal with fast rise and fall times to work most effectively. Not having any logic gates to "square up" the signal driving the gates, we can only approximate this type of drive signal. Q15 is a non-linear amp being driven by the fairly large output signal from the transmitter band pass filter output. The output of Q15 is 6 V peak pulses at 7 MHz, which is then buffered by the emitter follower Q13. The buffer is required to keep the gate capacitance of thePA FETs from loading down the output of Q13 and drastically reducing the output.
The PA puts out about 4 watts with a 13.8 V supply. Because the drive is not ideal, these do get fairly warm. A small heatsink clipped across the PA transistors will keep the power out from droping as the PA gets hot. The drain break down voltage is also fairly low, 60 volts, so they can not tolerate high SWR conitions which can casuse the drain voltage to increase above 60 volts and to cause them to draw excessive current, resulting in thier distruction. Therefore, one needs to insure the load has a low SWR before transmitting! D9, not shown on the schematic but is on the circuit board, can be added to help protect the PA from high SWR voltages. (1N4753B, 47V 1W Zener)
The mixer and driver stages are keyed on by a PNP transistor, Q12. R32 and C45 provide an R/C time constant to ramp the keying voltage up and down to provent key clicks. The non-delayed keying voltage is taken from the collector of Q12 and sent to the receiver muting and QSK circuits.
Flood fill ground plane is used on the board layout to maximize the ground area. Care must be taken when etching the board to make sure the copper around the tracks and pads has been fully etched away. Also, do to the close proximity of the ground plane to tracks and pads, extra care must be taken when soldering in parts as to not make shorts to the ground. Using 0.020" solder instead of 0.032" solder is a big help in preventing solder shorts and using excessive solder.
Note the 6 point to point jumers indicated by the black lines. In addition, there is a long jumper across the board required to connect the pads labled "A" (the gate of Q11 to the C39/R23/D7 junction) Also a short jumper is needed to connect the gates of Q16 and Q14 together. Put this on the bottom of the board. This jumper was required as to not isolate the source connection of Q16 to the ground plane.
A pdf file with the board layout which will print to actual size can be down loaded by clicking HERE. This is a through board view and can be printed directly onto toner transfer film. The image gets reversed when you iron the pattern onto the board.
If you wish to add the Digital Dial, which also makes getting the VFO adjusted to the right frequency easier, they can be obtained from Hendricks Kits, http://www.qrpkits.com
R46 and C61 are not shown on the schematic and are used to filter the supply voltage to the digial dial. R46 is 100 ohms and C61 is 100 ufd. Without these parts, mulitplexing noise from the display gets into the receiver audio. Also, C63, a 10 pfd cap, is used to couple the VFO signal to the Digital dial counter input.