The TRA967 is a small military FM transceiver, made by RACAL Tacticom in 1981. It covers the VHF frequency range from 36.000 to 75.975 Mhz. There are 4 decade switches used to set the frequency, which is uses a 25 Khz channel spacing. There is a 1 watt version (TRA.967/1) and a 3 Watt version (TRA.967/3). The transmit deviation is 5 Khz. The radio uses a 150 hz pilot tone on transmit and receive. It operates from a 12 volt DC battery, either a MA968A NiCd rechargeable battery, or a MA968B dry cell battery (which uses 10 cells type C). The receiver power consumption is 120 mA and the transmit power consumption is 0.6 A (1.4 A for the 3 watt version). The radio is small, measuring 285 mm tall, 166 mm wide 76 mm high and weighing 3.2 kilos. It is normally carried in a canvas haversack, worn on the hip, and uses a whip and a handset. It can be used in a vehicle, as there is a vehicle mounting frame. There is an external speaker and power supply Type MA988. There is also a 25 watt power amplifier Type TA970 and an Antenna matching Unit type MA872. It can also use a headset. Two TRA967 can be connected (using a MA4009) to provide a rebroadcast or relay setup.

Picture 1: TRA967

The controls are very simple.
The antenna connector is at the top left.
Then there are 4 knobs to set the frequency, from 36.000 to 75.975 Mhz, in 25 Khz steps.
The left most knob has 5 positions to select the 10s of Mhz (3 - 7).
The next knob has 10 positions to select the 1s of Mhz (0-9)
The next knob has 10 positions to select the 100s of Khz (0-9)
The right most knob has 4 positions to select the 25 Khz channel spacing (0, 25, 50, 75).
In the bottom row, at the left hand side is a BNC connector for RF out (for the power amplifier, if used), labeled 50 ohms. Near this is an earth terminal.
Next is the main FUNCTION switch, which selects OFF, ON, WHISPER, REBRO (rebroadcast) and NOISE (squelch off).
Next is the VOLUME control knob.
At the extreme right are two AUDIO connectors (6 pin and 7 pin).

Picture 2: Controls

Picture 3: Nameplate

There are several test points throughout the radio, referred to in the manual, which can be used to examine the internal signals. The MA991 RF transceiver board has connections that start with A, and the component designation starts with 1, (for example, point A15 on component 1TR16). The MA966 synthesiser board has connections that start with B, and the component designation starts with 2, (for example, point B41 on component 2ML4). The AMU antenna board has connections that start with C, and the component designation starts with 3, (for example, point C1 on component 3L9). The chassis has connections that start with D, and the component designation starts with 4, (for example component 4D1). The schematics use the older gate symbols. The NAND and NOR gate symbols use the NAND symbol with either a “&” or “1” inside it. There are some errors in manual. Part 1, Chapter7, Fault Finding and Performance Checks, Section 7.12, Peak Deviation Check (9), should be “not less than 3.5Khz” (the manual says 3.5hz).

The radio is in a sealed case. The manual has a section (Part 1 Section 7.5) on replacing the desiccator, which involves presurising the case, and then immersing the radio in water, to check for bubbles! For repair, the radio can be released from the case, by undoing two screws on the bottom. The case then slides off, revealing the insides. This consists of a frame with two printed circuit boards attached. The copper layer is on the outside, and the components are protected on the inside. Several screws can be removed, and each board will hinge out (with the wiring still connected) so that the components can be serviced. This is a nice mechanical design feature which aids serviceability. There is a wiring harness on the front panel, which goes to all the controls. One board has the RF components, for both receiver and transmitter. The other board has the synthesiser, which produces two RF signals, one for the receiver and one for the transmitter. Near the front of the radio is a small board containing capacitors and inductors. This is the internal AMU (Antenna Matching Unit).

The radio uses a 12 volt DC power supply. This is applied to a switching regulator IC (uA723) to produce several regulated voltages. These voltages are independent of the battery voltage, and as the batteries run down, the voltages do not change. The radio uses several different types of IC (Integrated Circuit), which require several different voltages, so the regulator provides them all. The regulator produces +5 volts, +9 volts, +12 volts, +27 volts and -7 volts DC. The power supply is located on the synthesiser board.

The RF board contains a conventional single conversion superheterodyne FM receiver, with a 10.7 Mhz IF frequency.

Figure 1: RF Board Block Diagram

The aerial input has diode protection and a 30 Mhz high pass filter. There is a two transistor cascade RF amplifier (two BF197), followed by an RF transformer, tuned by two varactor diodes, controlled from the synthesiser board. There is a two transistor converter (two BF197), using the synthesiser as the local oscillator. There is a 10.7 Mhz crystal filter to provide the selectivity (7.5 Khz wide). The IF amplifier IC (CA3089) also provides a limiter and FM detector.

There is a single transistor (ZTX237) audio pre-amplifier, that also acts as a mute during transmit or an “out of lock” fault. Then there is a high pass filter to remove the received “pilot tone”. At this point the “side tone” is injected during transmit. Also here, is a gating transistor (ZTX237) for the squelch. Following this, is an audio amplifier IC (741), driving a two transistor (ZTX212 and ZTX237) amplifier for high level audio output. There is also another audio amplifier IC (741) using the front panel volume control to provide earphone (handset) audio. The audio also goes to another IC (741), which has a 150hz notch filter, and the output is rectified, to operate the squelch circuit. An additional DC input (proportional to carrier level) comes from the IF amplifier and is combined with this. So both carrier level and pilot tone can operate the squelch circuit.

The microphone input goes through a two transistor amplifier (two ZTX237) to an IC (LM703). There is a 20dB increase in gain, when switched to WHISPER. The output is then sent to the synthesiser board to produce FM modulation. The pilot tone filter IC is switched during transmit, and becomes an oscillator, to provide the 150hz transmit pilot tone, which is injected here. The RF transmit output from the synthesiser is applied to three pairs of transistors (2N3639, BLY33, BLY85) which form the power amplifier. There is an ALC circuit to maintain a constant power output level. The PA output goes through a 76 Mhz LPF to the transmit/receive relay. It is then connected to the AMU, located between the RF board antenna connection and the whip antenna socket. The capacitors and inductors are switched into circuit by the two Mhz front panel switches, to match the whip to the transmitter output.

There is also provision for an external AMU type MA972, which can be used in a fixed station arrangement. A voltage is provided on the 50 ohm output BNC connector, depending on the frequency in use. The voltage is 0 volts for 36.000 - 43.975 Mhz, 6 volts for 44.000 - 59.975 Mhz, and 12 volts for 60 - 75.975 Mhz.

Picture 4: RF Board (inside with AMU visible)

Picture 5: RF Board (outside with PA transistors visible)

ELECTRICAL DESIGN (MA966 synthesiser board)
The synthesiser is on another PCB. The design is a PLL type (Phase Locked Loop). It does not use a microprocessor, everything is done with individual logic ICs and discrete transistors. It is basically a VCO (Voltage Controlled Oscillator) which is compared to a reference oscillator, to achieve stability. The RF output on transmit, goes to the RF power amplifier. The RF output on receive (10.7 Mhz higher than the transmit frequency) goes to the receiver mixer.

Figure 2: Synthesiser Board Block Diagram

The frequency standard is a 3.2 Mhz crystal controlled oscillator, using a CMOS IC (type 4001) (Complementary Metal Oxide Semiconductor). This then goes to two binary counter ICs (4020 and 4024) which divide the frequency down, to produce the signals for other parts of the synthesiser. A frequency of 100 Khz is used in the power supply. A frequency of 781 hz is used in the Phase Comparator. The frequencies of 390 hz and 12 hz are used in the lock detector. The frequencies 24 hz, 12 hz and 6 hz are used to generate the synthesiser clock. The VCO covers the frequency range 36.000 – 86.675 Mhz, and is set to the appropriate frequency, by applying a voltage to it, in the range 2 - 20 volts DC. It is a FET transistor TR23 (J300) set up as a Hartley oscillator. The buffered output goes through a coax cable to the receiver mixer. There is second buffer that sends a higher level output to the transmitter via a different coax cable. This circuit requires 9 volts DC.

The VCO output also goes to a “prescaler” that divides (reduces) the frequency by 32. This uses ML1 and ML3 which are divide by 4 counter ICs (8601). These are special ICs consisting of an ECL type (Emitter Coupled Logic). ECL can operate at this high frequency, but are very power hungry. They require 5 volts DC. The ECL devices are turned off (when not required) to save power. The ECL divided frequency is now suitable for the CMOS logic. The remainder of the synthesiser consists of this logic family. CMOS logic is low power and requires 12 volts DC. There is ML4 which is a divide by 2 counter (4013) and this completes the prescaler. The prescaler output frequency is in the range 1.125 – 2.708 Mhz (the VCO frequency divided by 32). The combination of transistors, ECL logic and CMOS logic illustrates the early nature of this synthesiser design. Careful construction, level shifting, and power control, show the design route taken. Later transceivers could achieve the same result with far fewer devices, and with a lower power consumption.

The frequency from the prescaler, is divided again, but this time, by a complicated divider, that can be changed (programmable). It is often called a “divide by N” counter, (N being a number which is the divide ratio). The divider number is controlled by the four frequency switches on the front panel. Each change in the divider number represents 25 Khz (in the case of this radio). The output code from each switch, is loaded into the divider counter, as a start point (to begin counting from). The manual uses the older term “jammed” with a “strobe” pulse. Nowadays we would say “latched” with a “load” pulse. The counter then increases by one count (increment), each time there is a pulse from the prescaler. When the counter reaches the required final count (when it matures), there is a pulse output.

The design of this counter is fascinating. One design method would load the start count, and then count up to 99999 to stop. An alternative, would be to load a start count, and then count down to 00000 to stop. Each approach would work. The RACAL designers have chosen a more elegant and simpler design, but more complicated to understand. The numbers 99999 or 00000 were not important, all they wanted, was the total count (between the start and the stop), with the minimum number of ICs used. So they chose a stop count of 29925 and a second stop count of 40625. The first stop count is required to generate the transmit frequency, and the second stop count is required the generate the receive frequency (10.7 Mhz higher). Why use such strange numbers? To detect the stop count, you have to use several logic gates, to sample the counter output bits and determine when the counter matures. They have chosen 2 stop numbers, which use the minimum number of gates, and the stop numbers can be easily changed by one control line, for transmit or receive. They chose to count up to these numbers, as “up counters” were easier than “down counters” at the time.

They chose different counter ICs. A common counter is a binary counter. As it is clocked (incremented), each binary count bit influences the next count bit, so the count propagates through the counter, sometimes like a “ripple” effect. The output of these bits, go to the gates so that the stop count can be detected. If a decade count is being used then momentary errors may occur. For example, when the count changes from decimal 3 to 4 (binary 1100 to 0010) three bits change. Sometimes, some bits may change before the others. This can be ambiguous, as momentarily, the decimal count may be 0, 1, 2, 3, 4, 5, 6, or 7, before it settles down. It depends on the logic design, and perhaps even the internal design of the counter IC, and maybe even the batch of ICs manufactured on that day. The RACAL designers avoided any possible count errors by choosing a special counter IC, called a JOHNSON counter (or RING counter). It is a binary counter, but the output bits are not 4 bit BCB (Binary Coded Decimal), but 5 bit JOHNSON code. For each count (increment), only 1 bit changes at a time. So following the previous example, when the count changes from decimal 3 to 4 (JOHNSON 11100 to 11110) only one bit changes, so the decimal count can only be 3 or 4, which is precise. This also means that only 2 bits from each decade counter need to be monitored to determine when the count matures. This requires 2 bits multiplied by 3 decade counters plus the 2 bit divide by four counter. This means that only 8 bits need to be decoded (3 x 2 + 2 = 8). If the divider used BCD counters (4 bits each) in which all bits needed to be monitored, this would be a total of 14 bits (3 x 4 + 2 = 14). The gating using JOHNSON code is much simpler.

They also have to load the counters with a start count. One possibility is that the frequency switches (10 position) might be setup to give 10 output lines (decade), or perhaps have them encoded as BCD with 4 output lines. However, these RACAL switches produce a descending JOHNSON code, which is used to load into the counters. This is a very clever counter design by RACAL engineers. The output from the programmable counter goes to the Phase Comparator.

The Phase Comparator, has two input signals. One signal is from the Frequency Standard and the other is from the VCO. The 3.2 Mhz Frequency Standard is changed by the reference divider to 781 hz. The VCO frequency is changed by the prescaler and the programmable divider. The comparator uses two flip flops ML23 (4013) each clocked by the two different signals. This produces two output pulses, depending on the difference in frequency. If the VCO is too low in frequency, the resulting pulse from one flip flop will turn on TR13 which will charge capacitor C76. If the VCO is too high in frequency, the resulting pulse from the other flip flop will turn on TR16 which will discharge capacitor C76. When the input frequencies are the same, the capacitor voltage will be stable.

The capacitor voltage is buffered, and then applied to the VCO. This will make the VCO change to a frequency, which when divided by the prescaler and the programmable divider, produces 781 hz, the same as the divided frequency standard. Thus by changing the programmable divider with the 4 front panel frequency switches, the VCO can be set to the required transmit and receive frequencies. The capacitor voltage to control the VCO frequency, is modulated by the transmit audio from the microphone, so this produces the FM transmission signal.

The radio is designed to operate from batteries. To prolong battery life, a current saving circuit is used. This circuit operates only in receive, and only when switched to WHISPER, REBRO, or ON (when below the squelch cut off level). The circuit generates an asymmetrical square wave, which has the radio on for 20 mS, and at reduced power for 140 mS. The circuit uses 2 flip flops and a few gates. It turns off the prescaler, so the ECL counters do not consume the high current. It also turns off the programmable counter. It stops the Phase Comparator so the VCO control capacitor does not change, and the oscillator drift is minimal. On the RF board, it turns off the supply to the RF amplifier, mixer, and IF amplifier. If a signal is heard, it breaks squelch, and the receiver and synthesiser turn fully on. When the current save circuit is operating, a faint clicking can be heard in the earpiece.

Picture 6: Synthesiser Board (inside view)

Picture 7: Synthesiser Board (outside view)

The radio was in a good clean condition, and the outside case and front panel only had a few scratches. The accessories were difficult to find (handset, canvas, whip, gooseneck, and battery) so that the radio sat on the shelf for a few years, until the accessories were acquired, one by one.

Picture 8: Accessories

The radio was clean inside, but some wires were disconnected. Obviously, there has been a fault, and someone had attempted to repair it. The wires were identified, and checked with a multi meter. There appeared to be low resistance to chassis on the main power supply line. A current limited bench supply was connected, and it was slowly wound up towards 12 volts, but at 2 volts, it drew lots of current, which indicated a hard short circuit somewhere. The main supply was disconnected to the RF board and the synthesiser board, which showed that the fault was on the RF board. A current probe was used to find where the main current was going. A capacitor (TAG tantalum) was replaced, and the short was gone.

The radio was tested, but the frequency switches were stiff to rotate. Upon close inspection, there was a dent in the front panel edge, near the switches, indicating a heavy blow. The knobs were removed, and the switches loosened and repositioned, until they moved more freely. The receiver had a good signal to noise ratio of better than 1 microvolt, for a 10dB signal to noise. Several frequencies were tested, using all the switches. The transmitter gave out 3 watts at the same frequencies. The frequency was slightly low, so the synthesiser 3.2 Mhz reference standard was adjusted with the trimmer capacitor.

The next time the radio was used, it had a beeping in the earpiece, indicating a fault. When checked, the synthesiser was not producing the receive or transmit signals for the RF board. These should have been on B29, B30 (receiver local oscillator) and B30, B31 (transmitter exciter). More checking revealed that the 3.2 Mhz crystal oscillator was not working. The oscillator output is available on test point B60, for viewing on an oscilloscope. The oscillator IC (ML20 which is a 4001) was replaced. The oscillator was now working, but there was still a fault. The Reference Divider was checked (ML22 and ML21), and this was behaving properly, generating the required control signals (100Khz, 781Hz, 390Hz, 24Hz, 12Hz, and 6 Hz).

In the manual, there is a good circuit description and partial circuits, to aid in understanding the synthesiser function. There is also several fault diagnosis flowcharts to help repair. In addition, there are several links on the PCB, which can be used for circuit isolation, to help diagnosis. For example, the “LINK” (B69 to B70) can be cut, which breaks the PLL feedback loop. The PLL now has no DC control. This point can be connected to a bench power supply (through a 10K resistor), and a frequency counter and oscilloscope connected to B74. The voltage was varied between 2 and 20 volts, and the voltage controlled VCO should have varied over its full range. It varied from 39.635 to 86.016 Mhz, so the VCO was working correctly. The ECL prescaler (divide by 32) was checked and found to be working, and the frequency from this was 1.238 to 2.688 Mhz.

The programmable divider was checked, but it appeared to be not working. This is a very complicated counter to fault find in. The counters appeared to be stuck (not counting). The strobe line (load) was stuck high. The strobe flip flop was being clocked, but not resetting. It should do so at the end of each count. Was it the counters not working, or was it the gates not resetting it? (the chicken and the egg situation). The current save circuit was checked to make sure that it was not stopping the counter. It appeared to be working properly.

Because the strobe was stuck, the frequency switches were thus being continually loaded into the 3 decade counters. The individual bit outputs of the counters were checked with an oscilloscope, whilst rotating each frequency select switch. The same was done for the 4 bit counter. They all appeared to be loading and changing satisfactorily. The decade switches were then set to 74.425 Mhz, which loaded all ones into the counters. (refer to the tables in the manual). The inverted outputs were all zeros and monitored with an oscilloscope. Then the 25 Khz switch was turned to each of its 4 positions, and one count pulse was seen on each bit of the decade counters. This meant the counters were probably working. The gates were checked, and gate ML5 (type 4023) appeared to not giving any output, being low under all conditions. The IC was replaced, and the counter was now working. But the fault beeping tone was still present in the earpiece.

Picture 9: Removing an IC

The Phase Comparator was checked and each flip flop had a signal input. The flip flop outputs were checked, and the charging circuit was seen to be pulsing. The two signals were measured with a frequency counter. The divided standard frequency was correct (781 hz). The divided VCO signal was low (480 hz). So the charging pulses were correct. The capacitor should charge, and thus increase the VCO frequency. The transistor TR14 was removed and checked, but it tested good. The transistor TR13 was removed but due to the absence of a layout, the wrong transistor was taken out! It was put back in. When the real TR13 was removed and tested, it was indeed faulty. A new one was fitted. The radio burst into life.

All frequencies were tested on receive, and all worked. There seemed to be approximately a 10 Khz bandwidth at 2 microvolts input. All frequencies were tested on transmit, and all worked into a dummy load, producing more than 3 watts. An aerial was connected, and the local repeaters could be tripped, and the recovered audio seemed very good. It was along and detailed repair session over many days. It was unusual that the radio had so many faults. Two ICs, one transistor, and one tantalum. However, there are many tantalum capacitors still in there, so maybe there will be many more repair sessions in the future!

The radio works well and is quite small and light. It is easy to use, as it has minimal controls. The 3 watts output and pilot tone are sufficient to work the local 6 meter repeaters. The signal to noise in the receiver is better than 1 microvolt over its entire frequency range.

The manual is very detailed, as the radio has several unusual features. The inclusion of Pilot tone operation, current saving, external AMU and PA control, are all useful features. It is an early solid state design, as is evidenced by the unusual mixture of technologies (transistors, FETs, CMOS, and ECL). This unfortunately requires several different voltages, 27, 12, 9, 5 and -7 volts. The synthesiser is a clever design. The folding out PCBs are very helpful for servicing. The test points and links, are also helpful to deduce a possible fault.

The PCBs have a seeming random layout, with ICs oriented up and down and sideways. The PCB silk screened legend is obscured by many components, so identifying the required component takes time. The scanned manual on the internet has no layout diagrams, so fault finding is a challenge. The ability to select illegal frequencies is an interesting but annoying problem. The frequencies below 36.000 Mhz and above 76.000 Mhz can be set with the switches, but the radio will not function, it will give the “out of lock” beeping tone. Is this in the original specification or is the radio incapable of working at these frequencies?

Figure 3: RF circuit

Figure 4: Synthesiser circuit

Technical Manual TRA967 VHF Transmitter Receiver, RACAL TACTICOM VMARS Archive web site http://www.moffatig.com/emers/racal_document_index.html

Technical Manual TRA967 VHF Transmitter Receiver, RACAL TACTICOM DAVE MCKAY G1JWG website http://www.pmrconversion.info/manuals/racal/TRA967.htm

Ray Robinson