Denys Roussel, F6IWF

An Ultra Low-Cost HF SSB/CW Transceiver with 20W Output, an AGC Meter, S-Meter and Audio Filters
Part-1, The design

VHF Communications 2/1996


Please note that none of the circuit or block diagrams, photographs, etc., are included here

Note: Although VHF Communications Magazine, by its very name, normally specialises in VHF and up, this project was found to be of such quality that we decided to include it. We hope that you find it as interesting as we have. The Editor

1. INTRODUCTION

The idea to build an inexpensive transceiver came from a joke on the air with a friend of mine, Francis "F6AWN", about the high prices of modern radio equipment made in Japan. As I do not have any recent commercial equipment, I was not very well informed about cost and I was somewhat surprised by prices. Some of these Japanese rigs cost more than £2000. Consequently, it is sometimes difficult for youngsters, beginners or radio-clubs to invest in the very latest equipment.

To help bring amateur radio activities to every body, it should be interesting to make a transceiver for the value of a banknote. I preferred to build a complete SSB/CW unit with medium power output and satisfactory performance, rather than a QRP station 1 watt crystal-controlled unit. The challenge was set at building such a unit for less than DM 100, or approx. £45.

2. SPECIFICATIONS

2.1 Why SSB ?

A lot of QRP transceivers at low cost have been described in the amateur radio literature, few for phone operation. It is obvious that this mode is more attractive than a CW contact (of course for me!).

Nowadays, it is almost impossible to use a Double Side Band receiver. HF bands are very busy, and the other side band carries signals received simultaneously on a DSB receiver. In DSB transmit mode, the interference often spoils communications on adjacent frequencies.

2.2 On Which Band ?

During the period of sunspot activity, propagation conditions vary, and for this reason the lower frequencies 3.5 and 7 MHz are more reliable than the higher 10 to 28. While the 80 metre band performs well at night, but is not ordinarily useful in the day time, at the same time the 7 MHz band is the only band useful for short or long distance work with a good antenna. In addition, 40 metres is the better test for a receiver. The transceiver should operate on 40 metres, and if possible on 80 metres if it is feasible with the price target.

2.3 The Power Level

The greater part of QRP transceivers are running between 1.5 and 8 Watts. This power level is too low to have easy contacts, a power of about 20 W will give best performance. It is 4 times lower than normal radio sets, i.e. 6 dB down or 1 S point. The power supply for 20 W output is not so big and is still compatible with a battery operation.

2.4 Other points

The minimum requirements of a transceiver for comfortable operation, a transceiver are that it should have a loud speaker, an AGC stage, an S-meter and filters.

The Automatic Gain Control function is subject for discussion, some operators do not like using AGC, because it is impossible to correctly hear weak signals when stronger are present. This is quite true, but it is very difficult to scan the entire 7 MHz band entirely without AGC, the first broadcast station makes the operator jump and wake up neighbours! I have acquiesced to the wishes of both camps by fitting an AGC ON/OFF switch.

2.5 Complete Specifications (objective):

The purpose is not to have the best HF transceiver in the world, but to offer minimal characteristics to use it in actual conditions. The performance of the receiver should also be similarly constructed.

Some operators will consider that the basic specifications as are being considered here by no means indicate a state-of-the-art transceiver, but like all such projects, it is necessary to relate it to the price you want to pay.

General:

Operating range: 7 to 7.1 MHz
Modes: LSB and CW
Stability: better than 500 Hz/hour after 10 minutes warm-up
Supply: 13.8 V DC nominal
Consumption: <4A

Receiver:

Input intercept point: 15 to 20dBm
Unwanted side band rejection: better than 40dB
Noise figure: better than 20dB
IF bandwidth "wide SSB": 300 - 300 Hz
IF bandwidth "narrow SSB": 300 - 1800 Hz
IF bandwidth "CW": 300 Hz centred on 650 Hz tone
AGC Range: from S9 to S9+40dB for 6dB change
Maximum AF power output: approx. 1W

Transmitter:

Unwanted side band rejection: 35dB typical
Carrier rejection:: better than 35dB
Power output: approx. 20W
IM3: better than -30dB at full power output
Harmonics level: better than -40dB at full power output

3. DISCUSSION OF THE CONCEPT

There are two basic systems for generating SSB signals:
a) The Filter method
b) The Phasing method

a. The use of a band-pass filter is the conventional method. A good side band rejection can be obtained with filter using four or more crystals. Probably it is the only 'real' solution to building a modern SSB transceiver today. In our case, the problem remains its price. A non-expensive crystal filter with crystals supplied for LSB and USB costs of the order of £25 at least, over 50 % of the target price of our unit!

After several tests on scale filters using cheap crystals, I was disappointed with the results (too much disparity between elements). At this time, I read some articles by Mr.Oppelt (VHF Communications) and Mr R.CAMPBELL (QST) which persuaded me to find another solution.

b. The Phasing method was very popular in the sixties, especially for transmission. This solution is a good one on paper, but in practice the very accurate phase shifting required and the high precision components associated almost completely eliminated this process.

Today, a 1 % metal resistor is inexpensive, it is possible to find 5% stable capacitors very easily, the development of logic circuits permits clocks to be built generating in perfect quadrature. This allows us to realise a different concept of an SSB transceiver working with phasing technology in direct conversion.

Direct conversion is the most simple (and the cheapest) process to employ to listen to the HF bands. Mr.Campbell and some English amateurs have shown that it is possible to design such a Direct Conversion Receiver which eliminates the opposite side band with the phasing method. In a receiver, all the functions are available to realise a transmitter, it is simply a matter of input/output connections and levels.

From this, I concluded that the design of a phasing transceiver remained theatrically possible. It is a solution to attempt the DM 100 target price and this unusual technology is for me a very attractive project.

4. PROBLEMS WITH DIRECT CONVERSION RECEIVERS

Direct Conversion receivers are the most simple, but are not without problems.

There are five major problems with these designs which are:
- The RF hum
- The AM detection
- The AF hum
- The Microphone
- The AF loop oscillations

All these problems where explained very well in an article by Nic Hamilton (G4TXG) in Radio Communication , April 1991.

4.1 The RF hum

The RF hum comes from 50 Hz leaks which modulate the Local Oscillator, providing 50 Hz side bands. These side bands are detected by the input of the receiver causing a 50 Hz buzz in the loud speaker.

With an external mains power supply, the leakage of the LO is phase modulated in the rectifying bridge of the supply by the 50 Hz, re-radiated by external electric lines and received by the antenna! This phenomena is explained in the ARRL Handbook 93

4.2 AM broadband detection (also called in Europe the "Radio Moscow effect")

When very strong AM signals are presented to the mixer, it rectifies (demodulates) the Amplitude Modulation. This signal is always present in output of mixers but rejected by the IF filter.

In a Direct Conversion receiver, it is the AF signal which is retained and amplified, so it is important to filter only the wanted bandwidth, especially on 7 MHz. The strongest station in western Europe is Radio Moscow on 7155 kHz, which is very difficult to eliminate.

4.3 The AF hum

The AF hum comes from 50 Hz leaks directly into the AF stages. In a Direct Conversion receiver, the gain distribution is completely different from that in superhet one. The AF gain is more than 100dB and the AF circuits are very sensitive to the 50 Hz leakage.

4.4 The Microphony

There are two types of microphony, AF and RF. The AF microphony can be easily cancelled by employing good mechanical construction. The RF microphony effect results from the LO leakage at the receiving mixer input, which is reflected by the input filter. Mechanical actions on the input circuits changes the phase of the reflected LO and gives AF sound in the mixer output. Good capacitors and sealed coils are required to reduce this effect.

4.5 The AF loop oscillations

These oscillations appears when there is a power AF amplifier driving a loud speaker. High currents in the supply wires give voltage differences which are detected by low power stages. Special PCB designs are necessary to eliminate these self -oscillations. This problem is the most difficult to resolve when designing a Direct Conversion receiver with a high level AF output stage. I had to redesign the AF section printed circuit board twice because of this problem.

5. TRANSCEIVER BLOCK DIAGRAMS

5.1 A few words about the Phasing Method

This process was explained very well by Mr.Oppelt in VHF Communications 2/87. In short, two oscillators are generated in quadrature, giving 0° and 90° phase difference to the converted signals after mixing. After this, two AF phase shift networks (+45° and -45°) achieve the phase difference of 0° and 180° to remove the unwanted side band and keep the other.

If the phase difference is maintained to within 1° difference and levels to within 0.1% of magnitude, 40dB of side band rejection is possible theoretically. Table 1 gives the rejection levels for various phase accuracies.

The block diagram in Fig.1 shows the AF Phase shift method for receive and the block diagram in Fig.2 the same method used for transmit. Both diagrams are very similar and it is a simple matter to combine them to achieve receive/transmit operation. The block diagram in Fig.3 shows this method employed in a transceiver.

5.2.1 Presentation

To achieve the target of DM 100 construction price, the maximum number of circuits should work in receive and transmit.

The solution researched during the phase development allows the use of most of the stages in both modes, excepting the transmit power amplifier and two operational amplifiers. Even the AF amplifier is used in the transmit mode for CW sidetone operation. A block diagram of the complete transceiver is shown in Fig.4.

5.2.2 Receive mode operation

The received signal from the antenna is fed by the transmit/receive antenna relay to the transmit low pass filter. Putting the LPF at this point provides a cleaner signal for the pre-mixer bandpass filter for only 0.5dB additional losses.

After a 10dB switchable attenuator (mechanical switch), the received signal passes through a diode switch and through the bandpass filter. More diodes are used to switch on or off the RF Pre-amplifier. This 10 dB amplifier is not absolutely necessary, but it enables the transceiver to be operated using lower gain antennas (e.g. portable or mobile operation). To protect the pre-amp from out of band strong signals, I preferred to place it after the bandpass filter. This solution is not the best for noise figure, but it is for intermodulation products. Because of the large losses produced in mixers, the global noise performance changes by only 1 dB, the preamp may be placed before or after the LPF.

The noise figure characteristic is not the most important consideration on the low HF bands, so it is preferable to place the preamp after the BPF. In normal mode the preamp is by-passed using a diode.

The signal is then divided into two equal components by a splitter. Both mixers receive a local oscillation from the RF processor (in quadrature for phasing operation), and the split RF signal. The mixer outputs are amplified by two receive preamps before the +45° and -45° phase shifting via an AF switch. The Rx preamps are audio ones because with the direct conversion process, the AF signal is already present in the mixer output. A perfect amplitude and phase symmetry is required along this line.

After combining, there is only one side band. A first stage amplifies and coarsely filters the audio signal. A true HPF eliminates the low AF frequencies which are not necessary for voice recognition.

The next stage is an AF low pass filter. To reduce the number of DIL packages (cost), the same filter serves for SSB "WIDE" and "NARROW" and helps the CW filter in narrow mode.

The AF is now filtered and must be amplified. An AGC function is required for the comfort factor discussed earlier. An AF electronic gain control is used to develop the AGC. Before the AGC detection, It is necessary to amplify and filter the audio again. Please note that the positioning of the AGC detection is after the complete filtering, because, so as not to cause desensitisation of the receiver by out-of-band signals. The filter function is split into two parts for better working in transmit mode.

The AGC can be switched "ON" or "OFF" for certain QRM, QRN operations. The S-meter is connected directly to the AGC line.

The AF signal is amplified to loudspeaker level using a standard IC, an AF switch is necessary to cut off the AF amplification during transmit.

5.2.3 SSB Transmit operation

In transmit mode, the signal passes through these stages in reverse. The microphone voltage is amplified by a switched audio amp activated only in Tx. This AF signal is amplified and filtered by the second line of AF filters (the same ones used in receive) and the AGC is now used for the Automatic Level Control (ALC).

The AF switch cuts off the AF to the loudspeaker amplifier, and the AF is switched to the to +45° and -45° phase shifters via the logic switches. The two receive AF amplifiers are changed in impedance and the low impedance signals modulate the mixers. Only one side band RF signal remains after combining.

The receive preamplifier is used as a post mixer amplifier with diode switching. After filtering in the band pass filter, the RF signal is sent to the PA via another diode switch, amplified up to a level of 20W, switched again via the antenna relay and filtered by the output low pass filter. The ALC detection is fed to the electronic gain control and gives 4 to 5dB of compression.

5.2.4 CW Transmit mode operation:

A Tx/Rx circuit detects the keydown, switches on a squarewave sidetone oscillator and switches the transceiver into transmit with an adjustable delay. This squarewave is sine converted by the AF receive CW filter and fed to phase shift networks.

In transmit, there is only one carrier which is the converted AF note in single side band. The ALC function is deactivated during CW Tx operation and the microphone amplifier is switched OFF. The AF tone is also sent to the loud speaker amplifier for sidetone generation.

The advantage is that, because the AF passes through the receive CW filter, the transmit frequency is the same as the receive one, without oscillator shift and "spot" button.

6. CIRCUIT DESCRIPTION

6.1 The Mixer (Fig.5)

The mixer is one of the most important part of a receiver. In a Direct Conversion receiver this stage should be able to offer resistance against intermodulation and also to the AM broadband detector effect. After some tests, I first eliminated active mixers (MosFET's and special IC circuits) which required sharp IF filter in the input circuitry.

Schottky diode ring mixers like SBL 1 or so gave the best results. The problem was that AM stations are on the air very often, even during the day. The low threshold of Schottky is probably not best suited to this problem. (And not for price - remember the DM 100 !) To increase the threshold of the diodes, one possibility is to use silicon diodes and another is to put diodes in series. The performance against AM Broadband detection is then better and the intermodulation too.

The input signal is split in TR2. The impedance here is 25 Ohms (2 x 50 Ohms in parallel), and TR1 is necessary to match to the 50 Ohm RF band pass filter. Each transformer is wound on one ferrite bead and the diodes are 1N4148.

The mixer output is terminated by a 51 Ohm resistor for the RF (R00l and C00l). The filter is constructed with 3 coils (3 ferrite beads) for cost reasons. Also, the core effect of a bead is important for reducing the possibility of 50 Hz leaks. This filter is very simple (too simple) and must be changed for a more sophisticated one in the near future.

Performance of the mixer

This mixer was tested for losses and intermodulation (with an L0 power of 16dBm):

1) LO: 10 MHz; Input: 0.1 to 10 MHz; Loss: -5dB
2) LO: 100 MHz; Input: 0.1 to 100 MHz; Loss: -7dB
3) LO: 10 MHz; Input: 2 carriers near 5 MHz at -8dBm; 3rd Intercept point: +22dBm (in)

6.2 The HF Preamplifier

Requirements:

We need to amplify the signals in receive and also in transmit. In receive we must be able to switch the preamplifier on or off . In transmit a post mixer amplifier is used to increase the S/N ratio before the band pass filter and to present the required level to the Power Amplifier. A solution is to use two separate amplifiers with a diode switch, an other is to use the same amplifier with a bidirectional switch (Fig.6).

A Norton stage is used in the amplifier. This kind of amplifier is very resistant against intermodulation and has very low noise characteristics (ca. 1dB). To save components, biasing of D401 and D402 is provided by the transistor current consumption (15mA).

During transmit +Rx is OFF, +Tx at 12 V, D402 and D403 conduct and T401 is supplied with DC. The mixed signal is output on the emitter of the BFR91 and amplified by about 10dB. D405 is biased in reverse by R401 and R402 to have low capacity and cancel the loop effect.

In receive +Tx is OFF, +Rx at 12 V. If Rx AMP is OFF, D405 conducts, and the other diodes are blocked. The Rx signal is feed to the mixer without amplification.

If RX AMP is ON, D404 and D401 conduct and T401 is supplied. The input signal is sent to the mixer with 10dB of gain.

All chokes are wound on a ferrite bead. TR4O1 is wound on two ferrite beads. Only one additional diode is necessary to change this stage to a dual transistor system, and the price of a second amplifier is saved.

6.3 The AF Amplifier and Mixer

After some tests on various circuits (low noise operational amplifiers, voice recorder preamplifiers, etc.) I reverted to the diagram of Mr.Lewallen and Mr.Hayward which was the most linear under strong signal conditions. This design was modified to ensure an Rx/Tx operation (see Fig.7):

a. Receive mode : +Tx OFF, +Rx = 12 V:

T103 conducts, the base of T101 is AC grounded by Cl0l. D10l is blocked. This ground base amplifier is biased with 500mA to provide a 50 Ohm termination to the mixer

Z = 26/IE (IE in mA)

A low noise transistor (BC549C) is used to ensure low noise performance. The emitter follower T102 provides a low impedance output. The Gain is fixed by R102 at about 40 dB. T104 is an active supply decoupler.

b. Transmit mode: +Rx OFF, +Tx = 12 V:

T103 is blocked, D10l is conducts. The collector of T10l is AC grounded by Cl08. The transistor T10l becomes a common collector with input on the base through C103 and P101. Because the emitter is already connected to the mixer port, there is no need for any other switching. The amplitude TX balance is set by P101.

6.4 The VFO and Phase Shifter (Fig.8)

a. VFO The VFO is a classic Colpits with a J31 0 and a separator stage T802 (BC238B). This diagram benefits from the noise improvements described by Mr.J.Jirmann , DB1NV in VHF Communications 3/93. The working frequency of the oscillator is 28 MHz to give the 7 MHz band after division by 4 in the phase shifter.

b) Phase Shifter:

The RF from the VFO is amplified by 7400 gates and drives a high speed 74F74 dual D flip-flop. The two sections of the 7474 are connected as a digital phase shifter with two outputs 0° and 90°. Because of propagation delay differences in the circuits, outputs are not absolutely in quadrature. Also, in phasing receivers described in the past, the RF Phase Shift Network was realised with passive components and tuneable. I believe that this tuning capability is important to compensate the existing phase differences in and after the mixer and to achieve good unwanted sideband rejection. C902 modifies the propagation delay and adjust the phase difference. Its value is determined experimentally during set up. R904 and R905 provide good termination to the mixer

6.5 AF Phase Shift Network (Fig 9)

The two + and -45° shift networks are made from Hilbert transformers (or "all pass filters") according to the work of Mr.R.Oppelt (VHF Communications, 2/87).

This process works better than polyphase or simple shifters with R and C which are sufficient to transmit but not for receive.

R201 to R219 should be 1% metallic film resistors. C203 to C209 are 5% MKH capacitors, 1 or 2 % would be better but are not easily available. The phase difference is theatrically better than ± 1°. IC's are low noise (and low cost) type TL074. Two CMOS 4053 are used for AF switching. After combination in P201, the AF receive signal is amplified and filtered by two opamps. C210 and C211 form a coarse band pass filter and the last stage is a conventional high pass filter.

6.6 The AF Stages (Fig.10)

6.6.1 SSB I CW filter

The signal from the high pass filter is fed to a low pass filter (IC3O1A). This filter has switchable bandwidth. Most of the time, separate filters are used and switched to provide several bandwidths. In this design, resistors are switched on the same filter by a CMOS 4066 to give wide and narrow positions. This allows saving of an additional quad OP amp package. In CW mode, IC3O1A is always in narrow position and assists the CW band pass filter. All CMOS switches are controlled by the +Rx. During Tx operation IC 301A is on "narrow" and IC302A and IC302B are open, providing isolation from the AF microphone signal.

This stage also serves to sine convert the square CW wave before modulation (CW Tx mode).

6.6.2 AGC/ALC/ AF Transmit filter:

RX operation:

In an SSB transceiver, the ALC detection can carried out on the audio. In a Direct Conversion receiver, the AGC action has to be performed on the AF because there is no IF.

After several tests on specialised IC's, FET's, etc., I found that the most efficient circuit was a standard transistor (T301)mounted in a variable resistor circuit.

After amplifying and filtering (1C301C-1C301D), the AF signal is sent simultaneously to the loud speaker amplifier and the ALC amplifier (T302). The cut-off frequency of both filters is 3 kHz. This function also takes place in Tx mode.

D301 rectifies the AF and C311 produces the ALC delay, the AGC can be set on or off by means of T303. AGC detection is after the complete filtering to prevent desensitisation by unwanted signals.

The S-meter is connected directly on the AGC line. A 100mA model is required and there is no need for "zero" adjustment.

TX operation:

The microphone signal is amplified by T308, which is supplied only in Tx (T306 switch). The signal is amplified and filtered by IC3O1C and D (as with the Rx signals). The AF filtering is important for phasing transmitters because there is no crystal filter to limit the HF spectrum.

The AGC line is grounded in Tx, and T301 is now controlled by the ALC line through D302. With only a few components, this function provides an appreciable comfort during speaking as well as 4 to 5dB compression. The ALC voltage is also used to display a power indication on the S-meter.

In CW, T308 is not supplied and the CW signal is sent directly to the AF phase shifters (CW Tx TONE). There is no AGC action but the power level is still displayed by the S-meter.

6.6.3 Loudspeaker Amplifier

A TBA82OM was chosen for is low price and good performance. The most difficult task is to block AF to the loudspeaker during voice Tx operation, T304 and T305 ensure this mute function.

For CW side tone operation, the CW tone is injected in the chip via the gain setting pin-2. It is an unconventional method but s the audio reproduction is satisfactory.

To be continued.


HOME PAGE