ANTENTOP- 01- 2004, # 005 A Simple SSB Transceiver

Credit Line:
by Ashhar Farhan
http://farhan.net.co.nr/ computercorp@hotmail.com

A dual-band transceiver with a crisp receiver and a used in cable TV equipment and has a good HF
clean SSB signal is described. It started its life as an performance: both as a low noise small signal transistor
investigation of the excellent S7C receiver described as well as driver up to 1 watt level. 2N3866 is expensive
in EMRFD. This transceiver was specifically (about Rs.20 each, but well worth the expense). It is
designed to use components that are easily used in a number of critical places.
available in TV and Radio spares shops. The
receiver sports an above average dynamic range, ► Measure what you have built. We used a 12 volt 1.5A
very clean signal and noiseless performance. power supply, a frequency counter, a test oscillator (to
Although the components are easily available, and measure the crystals and coils) and a high impedance
every detail about making it is covered here, this is voltmeter with an RF probe to test and measure the
not a weekend project. The design is elaborate and design. All these test equipment were homemade. The
invites improvisation. transmitter design did require a PC-based oscilloscope.
It helped us identify the spurs and harmonics using the
We decided to pursue the following rules in in-built FFT functionality. But now that the design is
designing this transceiver: complete, just an RF probe and a 14MHz receiver are
enough to align the rig.
► Use what is easily available. Very often, we find
designs that look good but they use exotic parts like ► Quality over quantity. A better signal is preferred to a
TUF-1 mixers that are simply impossible to get hold bigger signal. This is a 6 watt design that will work off a
of in India and other countries. Instead, we have simple 12V, 1.5A supply (using a single 7812).
tried using those spares that are universally
available. The ladder crystal filter

► Keep impedances and gain low: Often, we try A good filter is central to the crispness of a receiver and
coaxing maximum gain out of a stage making it the quality of the transmitter. There are two types of
difficult to duplicate and stabilize. We chose to take crystal filters possible, the lattice filter and the ladder
only modest gain out of each stage, using extensive filter. The lattice filter requires ordering crystals with 1.5
feedback to make the circuit stable. Most of the KHz frequency difference between them. This was ruled
interconnections between modules are for 50 ohms out, also procuring readymade filters from BEL India and
termination. In fact, the rig was a number of discrete other sources was ruled out as it is too expensive to do
board connected using RCA audio cables and that. Instead, a ladder filter was chosen. The ladder filter
sockets before we hooked it all up together to work. offers results as good if not better than a lattice filter.
However, the design is crucially dependent upon internal
► No PCB. We directly solder the components over parameters of the crystals used. It is not possible to
a plain copper clad board (un-etched PCB). It is an suggest any generic values for the capacitors to be used
excellent way to experiment, physically robust and in the ladder filter. Rather, a method to measure each of
has a quick and dirty appeal. You can usually solder the crystals and calculate the capacitor values has been
up a whole circuit as you think it out in a few worked out. We present this here. This design procedure
minutes. See the pictures. will work only for 10 MHz crystals. 10Mhz is the chosen
IF of our filter as the crystals are easily available and it
► Broadband. We wanted to be able to use sits comfortably between 7 and 14 MHz amateur bands.
broadband design where applicable. We have found We have followed the Butterworth design methodology
that the television balun cores are an excellent and given in the new ARRL book ‘Experimental Methods in
very cheap (about Rs. 2 per balun, that is 5 cents) RF Design’.
way of making broadband transformers.
The circuit centers around a four crystal ladder filter.
► Modest cost. While we didn’t want to use very Each lot of crystals from each manufacturer differs from
expensive components. We didn’t want to the others. We will describe a way to experimentally
compromise the performance either. You will see calculate the values of the capacitors for the filter. You
that we have used 2N3866 exclusively. This was should probably buy 10 crystals and select 5 of them.
because we found that the BF195/BF194/2N2222
series transistors available in the market were
consistently inferior in the HF range and performed
below their stated specs. The 2N3866 is commonly

http://www.antentop.bel.ru/ mirror: www.antentop.boom.ru Page 64

ANTENTOP- 01- 2004, # 005 A Simple SSB Transceiver

Circuit of the transceiver

http://www.antentop.bel.ru/ mirror: www.antentop.boom.ru Page 65

ANTENTOP- 01- 2004, # 005 A Simple SSB Transceiver

VFO of the transceiver

For this purpose, construct the test circuit of figure 2. C1 = 21 * F, C2 = 40 * F. Choose the nearest
1. This is a simple Hartley-style crystal oscillator. available fixed capacitor. If you can’t find a fixed
You will require access to either a frequency counter capacitor within 10% of this value, then parallel two
or a general coverage receiver (ask a neighborhood capacitors to achieve the capacitance.
ham to allow you to bring over your crystals to his
shack and test them for few minutes). Mark each For instance, in the case of the first prototype, we
crystal with a number and solder it into the circuit measure an average of 5KHz of shift. Thus, the
(don’t use a crystal socket). Connect the 9 volt capacitors calculated were 107pf and 200pf. We used
battery and measure the frequency. If you are using 100pf and two parallel 100pfs as a substitute for 200pf
a receiver, find out the frequency on which the capacitors. These calculations are for 200 Ohms
crystal is absolutely zero. Note the frequencies with termination. For a complete discussion of this design
the 33 pf capacitor in series and shorted. You will method, you are referred to the excellent paper by
have a pair of frequencies for each crystal. Select Craver in the Communications Quarterly of 1993, Winter.
four crystals with pairs of frequencies that match
within 50-40 Hz of each other. A fifth (for the carrier) Broad-band design without Toroids
oscillator crystal should be within 100 hz of the other
four selected. It was decided to use broad-band techniques where
suitable and keep the circuit free of too many critically
tuned circuits. We decided to investigate the TV baluns
as cores for broadband transformers. The TV baluns as
small ferrites as shown in the picture.

Almost all the broadband transformers are bifilar. Two

(the modulator and the transmit mixer cum product
detector) are trifilarly wound. They are simple to produce.
Making a bifilar transformer:

► Take two lengths of 36 swg copper enameled wire.

►Hold them together. Tie one end to a nail.

► Twist the wires together so that they cleanly have

about 8-10 turns per inch.
Calculate the value of the capacitors of Fig.2 like
► Check that the wires are evenly twisted (although
this:
there will be more twists towards the ends).
1. Calculate the average frequency shift of the four
► If the balun core is mounted on a PCB, cut it out with
chosen crystals as F (in KHz).
a cutter and remove all the original windings.

http://www.antentop.bel.ru/ mirror: www.antentop.boom.ru Page 66

ANTENTOP- 01- 2004, # 005 A Simple SSB Transceiver

► Pass the twisted pair through one hole to the the same value as the terminating filter impedance
other side, bend the wire back and pass it back across the input and output ends of the filter. This is
through the other hole (like a U turn). This is one incorrect. This looks like a resistor that is paralleled with
turn, like this, make similar 10 turns. a reactive impedance of the rest of the circuitry attached
to the filter. When the crystal filter is not properly
► Cut out the remaining ends of the windings terminated and sees reactive termination, ripple and
leaving about half an inch of the twisted pair on each ringing are introduced. This will spoil the crispness of the
end. receiver and spoil your on-the-air quality.

► Scrap the enamel off to about quarter inch, and The crystal filter is terminated on both sides by ‘strong’
tin the leads. RC coupled amplifiers based on 2N3866. This is slightly
unusual. The 2N3866 is used mostly as a VHF power
► Using a VOM at low ohms setting, identify the amplifier. It has excellent low-noise characteristics, good
two separate windings of the twisted pair. If we call gain and using it as a small signal device is now an
the two wires X and Y, each will have two ends A established practice. The 2N3866 is an expensive
and B. This you will have four ends AX, BX, AY and transistor. It costs about Rs.20 in the open market. We
BY. Short AX and BY together and use this as the think it is a good investment.
center point of the transformer in the circuits. Use
AY and BX as the two opposite ends of the Using RC coupled broadband amplifiers makes the IF
transformers. system a ‘no-tune’ affair. The output of the post-filter
amplifier is coupled to a two diode mixer. The two-diode
Making a trifilar transformer is similar, except that mixer uses a broadband bifilar wound transformer. It is
you have to use three wires twisted together. next to impossible to get toroids in India. We have
Separate out the three wires as before, use the first evaluated using TV baluns as substitutes for toroids.
two as described above, and the third winding as the These baluns are available at most TV spare shops.
secondary.
Most designs we have studied couple the RF input to the
IF sub-system diode detector through the transformer and inject the
BFO at the center of the transformer. This is a wrong
The crystal filter and its associated IF circuitry is practice. The diode mixer requires a minimum of 5mW of
shared between the receiver and transmitter. energy from the transformer input to operate properly.
Although the crystals are inexpensive enough to be There should be enough energy to switch on both the
able to afford separate filters for the transmitter and diodes. This means about 1.2 v peak voltage. The
the receiver, we noted that each filter would have a received signals are rarely this level. As a result, the
different center frequency. This would make zero- product detector operates like a regular envelope
tuning difficult for SSB operation. Therefore, it was detector and the diodes act as distortion devices to mix
decided to share the same crystal filter, carrier the BFO with the signal. The correct configuration is to
oscillator and the VFO between transmit and receive inject the BFO across the transformer
functions.
An unusual approach is taken here. The IF amplification
The crystal filter requires 200 ohms impedance gain is just enough to maintain good noise figure and
matching at both ends to provide the correct recover the losses in the ladder filter. We measured
bandwidth and low ripple. A regular practice among almost 10 dB loss in the filter.
hams is to strap a resistor of approximately

http://www.antentop.bel.ru/ mirror: www.antentop.boom.ru Page 67

ANTENTOP- 01- 2004, # 005 A Simple SSB Transceiver

The Receiver (requiring about 0.6 across each diode, that is, 1.2 volts
across the winding). This roughly translates to about 5
The receiver is minimal. By keeping the number of mW power. The diodes switch the low level signal
active devices low (3 devices between the antenna coupled at the center-tap of the coil to the detector
and the audio amplifier), very good fidelity is output. Therefore, in our design we have applied the
achieved. The circuit is kept at a low impedance and local oscillator through the primary of the transformer
broadband everywhere except the front-end. This and the incoming signal from the IF stage to the center
helps in stability. tap.

The front-end uses a low-noise FET. We have used There is a 100 ohms preset used to null the local
a BFW11 (because the local component shop ran oscillator from appearing at the output. This is of
out of BFW10). They have slightly different importance during transmit where the balanced detector
characteristics. Almost any FET can be used if it is also doubles up as the transmitting mixer.
biased properly. The FET should be biased for An audio pre-amplifier follows the detector. The capacitor
exactly half the pinch-off voltage. Wes Hayward of 220 pf between the base and the collector ensures
(W7ZOI) has described the proper way to bias a that the ‘hiss’ is kept down. The audio amplifier used is
FET Mixer for proper operation: an LM380. Almost any audio amplifier can be used. We
have tried everything from the PC’s ampli-speakers to a
► Short gate and source and measure the current Sony amplifier to a TBA810 amplifier. We would
that flows through a 560 ohms resister connected to recommend using a high fidelity, low cost amplifier like
+12V through the drain. This gives the exact Idss. the TBA810 if you plan using a speaker. If most of your
work is with headphones (to save your companion from
► Place a 10K resistor between the source and the the late night QRM), we recommend the LM386.
ground. Keeping the gate grounded and the drain
still connected through the 560 ohms resister, The Transmitter
measure the voltage between the source and the
ground. This gives you the pinch-off voltage. The transmitter starts with the modulator using a 741.
There is a three resistor network that biases the electret
► The FET has to be biased such that the voltage microphones. We use a Phillips ‘walkman’ style
on the source is exactly half the pinch-off voltage headphone with built-in microphone for our work. The
and there is half of Idss current flowing through the electret microphone requires a bias that provides 5V as
FET. Such a scheme assures you that the FET is given by the circuit.
driven between pinch-off and maximum drain by the
VFO injected at the source. This gives the proper The balanced modulator also had two 22pf trimming
switching action for the mixer to operate as well as capacitors for nulling the carrier. They were later found
maximum gain. We measured the pinch-off voltage unnecessary (as long as both the diodes are purchased
for BFW11s as 2.1 volts and Idss as 5mA. A from the same roll) and removed. If you do find balance
standard 1K resistor at the source gives the proper a bother, feel free to add a 22 pf trimmer to one side and
bias. a 10 pf fixed to the other side as indicated in the
schematic.
It should be noted here that we first tried a double
balanced diode ring mixer at the front-end. It has a The output of the balanced modulator is routed to the
number of spurious responses that literally made it common IF amplifier through a buffer amplifier using a
impossible to use the receiver. We tried to properly BF195. This serves to keep the carrier leak from the
terminate the diode ring mixer by inserting modulator out of the IF string during the reception mode.
attenuators between the mixer and the Ladder filter’s
pre-amp. It didn’t cure the problem. When we The balanced detector of the receiver also doubles up as
changed to the FET mixer, the noise figure a mixer during transmit. It is important to balance out the
improved, the receiver’s dynamic range, while VFO energy at the output by setting the 100 ohms
unmeasured, was never found lacking in the last one trimmer properly. We noticed a 50mW residual out-of-
month of extensive usage at VU2PEP. band output from the transmitter when the VFO is
unbalanced. The power chain is an interesting broad-
The output of the IF amplifier is detected in a band amplifier. You can use this in virtually any
balanced detector using just two diodes. Here gain, transmitter of up to 7 watts (and higher with more than
we break a common myth. You will see most of the 12 volts supply to the final stage). Three stages of
HF receivers employing a two diode balanced broadband amplifiers feed an IRF510 PA. It is an
detector with the BFO fed to the center tap and the interesting twist that the driver 2N3866 transistors cost
incoming signal applied through the primary winding more than the IRF510! The IRF510 should be biased for
of the detector transformer. This is wrong. The signal 80mA of standing current during transmit with the
applied through the primary winding should strong microphone disconnected (no modulation) and carrier
enough to switch the diodes on and off nulled by the trimpot of the balanced modulator.

http://www.antentop.bel.ru/ mirror: www.antentop.boom.ru Page 68

ANTENTOP- 01- 2004, # 005 A Simple SSB Transceiver

Construction ► Keep the bias trimmer totally down towards zero.

Attach VOM in series with point X in the power amplifier.
We heavily recommend constructing over pieces of Apply power in transmit mode and slow increase the bias
un-etched PCBs. They are cheaply available until you have 80mA flowing through the IRF510.
everywhere. See the pictures as a guide to ► Connect the RF probe across the dummy load.
component layout. We recommend the following ► As you whistle, You should get about 20-24volts of
rules: peak RF on the probe. When you pull out the
microphone from the jack, the RF output should drop to
► Keep your leads short. Short connections are complete zero. What if your transmitter is unstable?
more important than components that are at right
angles to each other. What might look neat to you • Don’t curse your fate. All transmitters start out as
might look unstable to the RF design. unstable beasts. Relax.
► Keep the outputs and inputs isolated from each • Start disconnecting power from the stages
other. We have taken care to keep the high starting from final IRF510 and working
impedance points down to a minimum. But still, backwards. When you have located the unstable
maintain design hygiene. stage, there are a number of things you can do
► Make one module at a time, test it completely, to fix it.
then move to the next one. Construct the transceiver • Try increasing the value of the 10 ohms resistor
in the following steps: used in the emitter degeneration OR
► Make the VFO. Check the RF output using an RF • Strap a resistor of about 1K across the output
probe. Check the stability on a regular receiver or a transformer of the unstable stage to ‘load’ it.
frequency counter. With the tuning capacitor fully • Move the linear amplifier away from the rest of
closed (the plates inside each other), set the trimmer the circuitry.
so that the VF0 frequency is exactly 3.9995 MHz • Redo the board. This time spread the stages out.
(keep 5 KHz margin at the band end) We guess that the linear chain should occupy
► Make the BFO. Check the output on the RF about 6 inches of space, all laid out in one line.
probe.
► Calculate the ladder filter values and make the IF Substitutions
strip along with the audio preamplifier.
► Connect the BFO, VFO, IF strip and an external The BF195 transistors can be substituted with any other
audio amplifier together. When you power on and HF transistor like 2N2222 etc. The 2N3866s are best not
attach a piece of 2-3 meter long wire to the input of substituted. The circuit works with slight increase in the
the IF amplifier you should be able to hear the noise figure if BF195 or equivalents are used in place of
atmospheric noise. Tune the BFO coil by fully 2N3866s in the IF stages. The output power on the
screwing the slug in and then slowly tuning it out transmitter absolutely needs the 2N3866s. Substituting
until the IF noise sounds right (not too shrill and not them with other switching transistors didn’t give good
too muffled). performance.
► Wire up the receiver mixer, connect the VFO.
Peak the mixer output and the RF input coils for The IRF510 should not be substituted with any other
maximum output. Then tune to a weak signal on the transistor. The other IRFs, though rated higher, have
band and tune for the best signal. Be careful to tune higher input capacitance which makes them a bad
for best quality of signal and not for maximum choice for 14MHz operation.
loudness. Take a break, spend a day or two listening
to the band with your receiver. Nothing is more The LM380/LM386 can be substituted with almost any
enjoyable than using a crisp receiver that you have other audio amplifier. Our first amplifier was Cambridge
homebrewed. SoundWorks Sound System. If you turn down the bass,
► Wire up the modulator. If you have an they are an excellent system for the shack. We have
oscilloscope, you can check the modulation. The tried a TBA180, an LM386, an LM380 and even a glow-
modulated output will be too low for you to be able to bug guitar amp. Feel free to experiment.
measure on the RF probe.
► Wire up the linear chain. DON’T solder the Final Notes
IRF510 yet.
► Put the transceiver in transmit mode. Whistle into The first contact we made using this rig was DF6PW. He
the microphone and peak the transmit mixer output reported us 57. Within the first evening we had worked
coils for about 6 volts peak RF voltage on the probe four continents. The rig is regularly used at VU2PEP.
at the 56 ohms resistor where the gate of the People are often surprised at how the transmitter quality
IRF510 would be. is ‘just like a commercial rig’. Many refused to believe
► Solder in the IRF510. ATTACH A DUMMY that it is a seven watt rig.
LOAD. We used four 220 ohms two watts resistors
paralleled together.
72/73!

http://www.antentop.bel.ru/ mirror: www.antentop.boom.ru Page 69