Welcome to the PA3CNO website
A 80m QRP CW transceiver
- Created: Tuesday, 15 February 2011 13:01
- Last Updated: Wednesday, 30 October 2013 11:53
- Written by Administrator
- Hits: 13573
The M.A.S., (Minimal Art Session) is a contest which was founded by Dr. Harmut "Hal" Weber, DJ7ST (SK). The idea is to encourage Hams to build and operate a rig consisting of a minimum of components.
The idea behind this contest is to encourage Hams to build and operate a rig using a minimal number of parts. To be eligible, a transmitter must use no more than 50 parts (pretty easy to do) and a transceiver must use no more than 100 parts (a bit harder to do). To encourage the use of a very small number of parts, bonus points are awarded by the percentage of the actual number of parts used less than the maximum allowed number. Thankfully, things like hardware, knobs, connectors, headphones, key and the like are not counted as parts. Also, the transmitter low pass filter is considered to use 3 parts, even if in actual fact it uses more parts to ensure spectral purity of the transmitted signal.
However, if you use any IC's, the number of parts integrated into the IC must be counted. It is impossible to determine how many resistors, transistors and diodes (capacitors are usually not integrated onto a chip, but sometimes there can be) are inside a chip since the manufacture usually does not bother to tell you and if there is a simplified diagram of the insides, it does not show all the parts. And even if this number can be determined, even a simple IC can have dozens of parts integrated. So, this effectively eliminates the use of any modern IC's in a rig designed for the MAS contest. Not being able to design with op-amps, audio amps like the LM386, mixers like the SA612 or CMOS logic gates, is a definite handicap to designing a simple, but effective rig which is actually capable of making contacts.
Despite the limitations of not using IC's as noted above, KD1JV decided to give it a try. Some very simple and very low parts count rigs have been devised over the years, one of the most well known is the "Pixie". The problem with these overly simplistic designs is they simply do not work very well. The chances of actually making a contact with one of these rigs is slim to none. They are a waste of time and of natural resources. What good is a extreamly low parts count rig if you can't make contacts with it? If you're to spend time and effort to design and build a rig, you want something which at least has a good chance of actually talking to someone! If your gonna work a contest, you got to be able to hear people coming back to you!
The rig KD1JV came up with uses 51 parts for a complete transceiver, giving a 50% bonus. The transmitter puts out about 2 watts, the only supr is -50dBc at the second harmonic, and uses just 16 parts. The receiver is a regenerative type detector, with a RF pre-amp and two stage, high gain audio amp, for a total of 35 parts in its minimual configuration. MDS is about 0.5 uV if you have good hearing. Since the MAS contest is an 80 meter event, this rig is designed for 80.
Optional Parts: Parts marked with a "*" on the schematic below are optional parts. These are D3, a reverse polarity protection diode and a fine tuning control. Since the rig does not produce a side tone on its own, an optional side tone generator is also described. These parts are included in the printed circuit board layout to make the rig more useable in general use.
Click HERE to see a full blown version
The transmitter is a simple crystal oscillator using a 3.579 MHz color burst crystal. A 2N7000 MOSFET is directly coupled to the output of the oscillator for the PA. Q3 is used to key the oscillator and PA on and off. Q4 is used as an inverter so that normal, active low keying can be used. Rise and fall time wave shaping is not included to reduce parts count, so this circuit will produce key clicks. C7 provides feedback so the circuit will oscillate. Normally, a second capacitor would be used from the emitter to ground, but the 2N7000 has enough gate capacitance to eliminate the need for that additional cap. Instead of the normal sine waves one would expect from a crystal oscillator, this oscillator was made to produce fairly narrow pulses. This improves the efficiency of the PA so that even though the 2N7000 is in a plastic TO-92 package, it does does not get alarmingly hot producing 2 watts of output. It is advisable the antenna load be preset to a low SWR before transmitting, as the 2N7000 has a 60V break down voltage and a high SWR can easily exceed this, causing the part to fail. The output low pass filter provides some impedance between the output of the PA and the antenna load. C1 in combination with L2 forms a trap at the second harmonic, other wise an additional filter stage would be required to meet FCC spectral purity regulations. Instead of buying a single 1500 pfd cap for C5 in the LPF, two 680 pfd caps could be used instead.
T1 is a bifilar wound transformer, which means two wires are wound around the core at the same time. (5 turns). Use an ohm meter to determine the ends of the wires A-B and C-D, then connect the ends B and D together to form the center tap, as shown in the diagram in the schematic.
The receiver is a regenerative type and is a slightly modified version of the QRPKITS "Scout" regen designed by Charles Kitchen. See http://www.qrpkits.com
Q5 is the QSK switching transistor. During transmit, this transistor is turned off to isolate the receiver input from the low pass filter. Q8 is a common base RF pre-amp to keep the oscillations from the regenerative detector from being transmitted and reduce pulling effects from the antenna. The resonant circuit made up of the secondary of T2 and C16 determines the operating frequency of the receiver. Ideally, the tuning cap C17 should be an air variable with vernier drive. If you don't mind adding a few additional parts, a pot tuned varactor diode can be added for fine tuning and a small value trimmer cap (C27) used to help set the tuning range. Making C16 150 pfd and using a 50 pfd tuning cap (jumper out C28) allows for pretty much full coverage of the 80 meter band, so a vernier dial is needed or the tuning is very touchy. The schematic is drawn showing the use of a polyvariable capacitor with the varactor fine tuning. Polyvariable caps are also available from qrpkits.com.
Stability of the receiver is directly related to the stability of the input tuned circuit. NPO or C0G type caps should be used and an air variable for tuning. Using a powdered iron core for the inductor is a liability, but an air core coil would be much larger and more difficult to manage physically.
In order to receive CW or SSB signals, the regenerative detector must oscillate. A feedback winding on T2 turns the circuit into an oscillator. V1 in combination with C20 is used to control the amount of feedback. Polarity of the feedback winding is important. If you can not get the detector to oscillate, reverse the feedback winding connections. When winding T2, wind the 43 turn primary first and leave as much of a gap as possible between the start and finish of the winding to have a place for the two 6 turn winding to fit onto. Then wind the two 6 turn windings next, continuing in the same direction as the primary turns. Now, pick on end of the 43 turn primary winding as the "hot end" connected to the tuning caps. The start of the 6 turn winding next to the end of the 43 turn winding should go to the j-fet and the other end to the 5.1V supply. The polarity of the winding going to the RF pre-amp does not matter, so you can pick either end for those connections.
Ideally, the regen control is set so that the detector just starts to break into oscillation. This gives the best selectivity and sensitivity. However, this point will change when returning the frequency for receiving, so in practice, set the control so oscillation is sustained over the tuning range.
R9 and C15 form a low pass filter to eliminate high frequency audio and any RF which is present on R10. Note that the drain and source terminals of a J-FET are symmetrical, so they can be interchanged. That is why the schematic looks different from the way it might normally be drawn. Q7 and Q9 form a high gain darlington amplifier. Q6 further amplifies the audio and has the headphones connected in series with the collector, so it is acting as a Class A amp. Doing it this way eliminates the need to make an amplifier which can drive a low impedance load and saves a significant number of parts. NOTE: The mounting sleeve of the headphone jack is connected to the power supply, so must be insulated from a metal front panel!
Keying the transmitter without any kind of audio muting circuit resulted in very loud clicks in the headphones. This was clearly not acceptable, so a mute circuit had to be devised. This resulted in adding R11, C19, C14, Q11 and D2. When the transmitter is keyed, Q4 is turned off allowing R3 to pull the gate of Q3 high, enabling the transmitter. D2 allows the gate of Q11 to also be pulled high, turning Q11 on and connecting C14 to ground, which by-passes the base of Q6 to ground. When the transmitter is un-keyed, the RC time constant of R11 and C19 delays the turn off time of Q11 to allow any voltage transits to dampen out and eliminates serious clicks from being heard. Some minor clicking is still audible, but it is of reasonable level and not at all annoying.
A 5.1V zener, D1 stabilizes the voltage to the RF pre-amp and regen detector. If no reverse polarity diode is used to save a part, one must be careful to observe correct polarity when connecting up power. Powering the rig with a regulated 13.8V supply is recommended, although a 12V gell-cell can also be used, though that might result in chirp, as the supply to the oscillator is not regulated. Minimum operating voltage is about 10 volts, with the power output dropping off to about 500 mw.
Note that a regenerative receiver is effectively a direct conversion receiver, so signals on both sidebands will be detected.
Since one will normally be using the receiver for CW or SSB it will be in the oscillating detector mode. There is enough RF signal present on R10 to add a sensitive frequency counter for a digital readout. This fact could also be handy in initially getting the receiver to tune in the desired frequency range. A frequency counter or a general coverage receiver can be used to help set the tuning range.
I built the RX first, in Manhattan style:
The first thing I noticed is that the receiver is very wide. That is not comfortable in a crowded band, and eligable for improvement. I had a very nice variable capacitor; "nice" referring to size and capacitance... There was a mechanical problem with the smallest section, resulting in jumping off frequency when tapping on it. That's why I temporarily used another variable capacitor.
The transmitter consists of only 2 transistors, so that took considerably less time than building the receiver. I applied a small heatsink to the 2N7000 just to be sure. I skipped the 2N7000 that will key the transmitter: instead I connected a straight key. I soldered a FT243 socket to the base of the transistor and inserted my 3575 crystal. I listened to the transceiver on my FT857 and I heared the tone, but dod not see any output. Oops... Antenna switch in the wrong position. After correcting that, I still had no output. Oops again... Used the cable between BNC adapter and antenna switch for something else. The good news was that the 2N7000 is really not that susceptible to SWR mismatch. Nothing worse than forgetting to connect the antenna, and it's still alive. After correcting all my mistakes, the output was just about 1,5W.
I checked the current consumption and that was just over 300mA. On a 13,8V supply that makes 4W input. That is quite a bad efficiency. So I started experimenting with the transformer connected to the drain of the FET. I started with 8 bifilar turns in stead of 5 to increase the overall inductance. Then I reduced the number if secondary turns. That reduced the current while the output remained the same, thus improving efficiency. The other way to improve efficiency is increase power output while the current remains the same. But that's not what the FET likes...
In the middle of my experiments I heared a very weak signal on my FT857 (the antenna switch connected the antenna to the QRP rig and not to the FT857) calling CQ on 3575. That was Hielke PA3EQX from Alphen aan den Rijn (about 15 miles away). I quickly soldered the original transformer in place and answered the call. And he returned to me! With a 539 report (time to move to Alphen aan den Rijn... At my QTH the QRM is rarely below S7 on 80m) he could copy me FB. But after a while I could smell the PA transistor - it turns VERY hot. Improving efficiency is a must, apparantly.
The variable capacitor I picked from the junkbox was 100pF. To cover the CW part of the 80 band (3500-3600), a deviation of about 18pF is enough. So I put 22pF in series, but that had an annoying side effect: the first 90o of the tuning range covers 3500-3520, and the second 90o covers the rest... That is easy to explain: if the variable capacitor reaches 22pF, the total capacitance is divided by 2, but the variable capacitor is already at 80% of its tuning range! So I checked the second section of the original variable capacitor and that proved mechanically OK. And now I only need 47pF to cover 3500-3600. Another advantage is the 1:3 vernier drive of that capacitor, omitting the need for additional finetuning.
Time to improve the audio section. Since I'm not going to use this rig for a MAS contest, I'm not limited by the number of components... I used an old design for a 75Hz CW filter, recalculated it for a fc of 700Hz and a Q of 4. I replaced the headphone by a 100 Ohm resistor and connected the filter to that resistor. See the schematic:
Click HERE for a full blown version
Transistor Q1 is part of the original design. The 220k LIN variable resistor allows for smooth panning between original and filtered signal. Also, a Volume control is added that lacks in the original design. The LM386-1 is able to deliver 325mW; if you use the 386-4 version, you'll get 1W. But 325mW is more than enough to drive a set of headphones. An additional advantage is that the ground connection of the headphones is now at 0V level and not at power supply level, as in the original design.
Some pictures of the complete transceiver:
I've made some really nice QSOs with this baby. Receiver sensitivity is around 3uV which is OK for 80m if your antenna is not that bad. The audio filter improves the selectivity a bit, but of course as in any DC design, you will hear adjacent stations. As you can see at the picture of the front, I also added a HF attenuator: a 1k variable resistor in series with the RX antenna switch (the 2N7000). That is really necessary to operate the transceiver when very strong signals are present. Otherwise the oscillating preamp will "lock" onto the incoming signal, resulting in DC output hence no sound will be heared....