The New Zealand Association of Radio Transmitters Incorporated
Developed: 1998, 1999 and 2000 by Doug Ingham, ZL2TAR
Version 2.5, April 2000
This booklet aims to provide advice on the construction of reliable, high performance, low maintenance, repeaters, and to condense, into a few pages, the RF design experience of several professional communications engineers.
All signal levels in this booklet are expressed as power levels, in dBm, in other words, decibels relative to 1 milliwatt. This is the most common calibration of modern test equipment. Occasionally, where this helps with understanding, the dBm signal levels are also expressed as the equivalent voltages across a specified impedance, usually 50 Ohms.
A modern, well designed, 16 kHz wide FM receiver (16KF3 emission designator) will deliver good but noisy audio, corresponding to 20 dB "SIgnal to Noise And Distortion" (SINAD), at -123 dBm input (0.16 µV in 50 Ohms). At this level of wanted signal it is possible to detect interference from unwanted signals of less than - 136 dBm.
The output of a 25 W transmitter corresponds to +44 dBm. This is 180 dB more than the weakest detectable signal. Put another way, any unwanted repeater transmitter outputs, such as white noise, on the receive frequency, need to be suppressed by 180 dB. Similarly, repeater receiver spurious responses, on the transmit frequency, also need to be suppressed by 180 dB.
Modern mobile and home transceivers, and equipment certified to RFS25, are designed to only receive OR transmit at any one time, so do not need to be, and, in fact, are not designed for the 180 dB dynamic range needed for repeaters.
The requirement for repeater equipment is more severe than RFS25. A repeater has to receive and transmit, on nearby frequencies, simultaneously. Special means, of isolating the transmitter and receiver from each other, have been devised, to ensure a transmit-receive dynamic range of more than 180 dB, while simultaneously receiving AND transmitting. These include special low-noise transmitter and receiver designs, and high-isolation duplexer filters. Cost-effective repeater designs divide the 180 dB
dynamic range requirement into about equal portions: typically about 90 dB dynamic range for the repeater transmitter and receiver, and about 90 dB isolation for the duplexer.
There are special requirements for receivers and transmitters used in the National System.
Transmitters and receivers using synthesized oscillators have much worse, by typically 30 dB, phase noise performance than those using direct crystal oscillators. Repeaters made from synthesized receivers and/or transmitters have inferior performance and/or require more expensive input and output RF filtering to overcome this inherent performance limitation.
Transmitter oscillator phase noise introduces two performance limitations.
First, transmitter noise radiated in adjacent channels "noises-up" nearby receivers tuned to weaker, more distant, adjacent channel transmitters. Second, and more important for repeaters, transmitter phase noise extends to frequencies beyond the associated receiver frequency, "noising-up" the repeater's own receiver. Extra filtering is required, between the transmitter output and the antenna, with a pass response at the transmit frequency and a notch at the receive frequency.
Receiver down-conversion oscillator phase noise also introduces another "noising-up" performance limitation. Strong transmitters on nearby frequencies (including the associated transmitter, in a repeater) mix with the phase noise of the down-conversion oscillator, to produce noise at the receiver's first intermediate frequency (IF), "noising-up" the receiver. Extra filtering is required, between the antenna and the receiver input, with a pass response at the receive frequency and a notch at the transmit frequency.
Transmitter and receiver phase noise (and lack of duplexer isolation) is one of the main causes of "kerchuffing" on weak input signals. The other main cause is de- sensing of the receiver input stages, by excessive amounts of transmitter energy, causes a change in the DC operating conditions, or clipping, in the receiver pre- amplifier or mixer devices. The mechanism is as follows:
Initially the repeater transmitter is off, so there is no noising-up or de-sensing of the received signals. A weak signal appears on the receive frequency. The repeater receiver is operating at full sensitivity. The control system turns on the transmitter; causing noising-up and de-sensing of the receiver, to the extent that the control system thinks the input signal has disappeared. After the "tail" time-out the transmitter is turned off, the receiver resumes full sensitivity, and the cycle repeats indefinitely. The only cure is to fix the transmitter, duplexer or receiver performance deficiencies.
Crystal oscillators are simple and reliable.
Synthesized oscillators are more complex, affecting their reliability, and come in two types, called parallel load and serial load.
The parallel load synthesizer has higher reliability than the serial load type, because of the way in which the frequency data is loaded into the programmable frequency divider IC. In effect, the frequency data is continuously applied, as DC voltages, to the input pins of the divider IC. As a result the divider IC needs to have a large number of input pins, requiring a large package.
The serial load synthesizer has lower reliability than the parallel load type. The frequency data is loaded once, at equipment switch on. Usually, only two pins are required to serial load the frequency data. The IC package is small, which makes it popular for inclusion in compact equipment. Virtually all Amateur transceivers use serial load synthesizers. The power fail detection circuitry MUST be 100% reliable, otherwise a partial re-load, and corrupted frequencies can result, following brief power interruptions.
Serial load synthesizers have been found to be particularly sensitive to Electro Magnetic Pulses (EMP) produced by nearby lightning strikes. The resulting off- frequency operation interrupts service, causes interference to other communications services, and requires a visit to the site to reset. Our hilltop sites can experience many nearby strikes (cloud to cloud or cloud to ground) per year.
Many commercial site owners prohibit the use of serial load synthesizers.
All synthesizers should have lock detectors, but many don't, to mute the transmitter and/or receiver output when the synthesizer is out of lock.
Legend has it that crystal oscillators are difficult to frequency modulate; this is not true. Most transmitter manufacturers haven't bothered to perfect frequency modulated crystal oscillators. Instead, they have attempted to use phase modulators. Unfortunately, phase modulators produce high levels of distortion at normal deviation (5 kHz) and at low audio frequencies (<1 kHz). The performance of phase modulators is unacceptable for FM voice or data use.
Properly designed, crystal oscillators can be designed which produce low distortion and 6 kHz deviation at 6 metres, 16 kHz deviation at 2 metres, and 50 kHz deviation at 70 cm. Voice and data transmissions in all these bands normally use 5 kHz peak deviation, so there is plenty of margin.
A spectrum analyser must always be used.
A spectrum analyser allows inspection and minimisation of unwanted multiples of the crystal oscillator, parasitic oscillations and unwanted carrier harmonics. The wanted output is maximised when the unwanted outputs are minimised.
Unfortunately, SWR bridges, and other broadband power detecting devices, such as the Bird wattmeter, tell a different story. Both devices include voltage sum, or similar, detectors. They falsely indicate higher output power when the transmitter output is impure. Therefore, maximising the apparent output with one of these devices, without using a spectrum analyser, is a sure way to maximise unwanted (spurious or harmonic) out-of-band products.
The centre frequency of the transmitter is easy to measure with a frequency counter. Ensuring that the centre frequency of the receiver is accurate is just as important, to ensure maximum sensitivity and minimum distortion.
A frequency counter can be used to check the receiver down-converter oscillators. The receiver frequency can also be checked, indirectly, by means of a high-accuracy, synthesized, sinewave frequency modulated, signal generator and a SINAD meter. Here is the method.
Set the signal generator to a RF output of -67 dBm, on the nominal receiver input frequency (+/-500 Hz accuracy). Turn on the 1 kHz frequency modulation, and adjust the deviation to 5 kHz peak. Reduce the RF output of the signal generator until the receiver SINAD becomes 12 dB. Alter the centre frequency of the signal generator in 1 kHz steps, until the SINAD degrades by the same amount, either side of the nominal frequency. Note the two frequencies. The receiver centre frequency is approximately the average of the two frequencies.
Distorted audio, and peculiar mute operation, can be produced by miss-aligning the receiver IF, particularly the tuning and matching coils either side of the crystal IF filter. These coils can only be aligned by means of a high-accuracy, synthesized, sinewave frequency modulated, signal generator and a SINAD meter. Here is the method. Do not attempt this if any of the above equipment is not available; otherwise you will worsen the performance.
Set the signal generator to a RF output of -67 dBm, on the nominal intermediate frequency (I.F.). Inject the signal into the I.F. test point often found on the input of the down-conversion mixer. The signal generator should not be connected to the output of the mixer, as this will cause a miss-match at the input of the crystal I.F. filter.
Turn on the 1 kHz frequency modulation, and adjust the deviation to 5 kHz peak. Reduce the RF output of the signal generator until the receiver SINAD becomes 12 dB. Adjust the tuning and matching coils to maximise SINAD. Reduce the signal generator RF output to maintain the SINAD at 12 dB. Continue until there is no further improvement.
Modern "communications" FM receivers feature a 455 kHz final IF IC (MC3357 and similar) including a demodulator consisting of a 4-quadrant multiplier and phase-shift network. The distortion of the recovered audio depends on the linearity of the phase- shift versus frequency characteristic of the phase-shift network. Two-section phase- shift networks have better linearity than one-section.
Phase-lock-loop demodulators can give higher signal-to-noise ratios, and lower distortion, than the phase-shift type, when the loop-filter time constants are correctly chosen. Once again, a high accuracy signal generator and SINAD meter are required to maximise performance.
Cross-band repeaters have worse inherent performance than normal (co-band) repeaters and are wasteful of spectrum. They are a, largely unsuccessful, attempt at avoiding the expense of proper input and output filters. Cross-band repeaters, implemented by internal cross-connection of dual band transceivers, have very poor performance, because of the insertion losses and poor isolation of the simple internal duplexers used, as simple sensitivity and desensing tests show. They also have very poor cross-connection audio quality, and poor stability of mute settings. They also feature serial load synthesizers, previously discussed.
Cross-band repeaters have been responsible for many hours, or days, of lock-up of the associated repeaters, including the National System.
FM demodulators have, what is termed, a triangular noise spectrum. The audio noise density is found to be proportional to the audio frequency. For example, the noise density at 500 Hz is 6 dB less than at 1 kHz, and 6 dB more at 2 kHz.
Therefore, receiver de-emphasis, reducing the high frequency audio response in a closely controlled manner (6 dB per octave) improves the audio signal-to-noise ratio. For flat overall audio response the associated transmitter audio circuits should include corresponding pre-emphasis, where the high frequency audio response is boosted, in a closely controlled manner (6 dB per octave). Transmitter pre-emphasis also improves voice communication efficiency, since the human voice has high energy at low audio frequencies, and low energy at high frequencies.
Care is needed when adjusting transmitter deviation levels of non-voice audio, such as modemed data. Pre- and de-emphasis should never be bypassed when dealing with modemed data, since the signal-to-noise ratio of the audio, and the error rate of the data will suffer, and the transmission/reception will be incompatible with that of every other station.
Commercial repeater transmitters and receivers are made in relatively small quantities, and, for economic reasons, their circuit design is usually based on that of mobile transceivers, from the same manufacturer. This is a pity since, as we have already seen, mobile transceivers are made down to a price.
Professional repeaters are usually designed as two separate units, the receiver and transmitter. Repeater receivers include de-emphasis; repeater transmitters include pre- emphasis. If the repeater is to be connected to a linking system, the de-emphasis and pre-emphasis networks should accurately follow the 6 dB per octave curve from below 300 Hz to above 3 kHz.
Don't forget that the signal generator used to test the receiver de-emphasis, and the deviation meter used to test the transmitter pre-emphasis, have flat frequency response.
Bandwidth restriction (low pass filter) is necessary prior to frequency modulation, and after frequency demodulation, to avoid distortion or noise caused by out-of-band audio. The low pass filter should be designed to have no effect for audio frequencies below 3 kHz.
Receiver de-emphasis is measured with a calibrated FM signal generator, and audio voltmeter.
Connect the (50 Ohm) signal generator to the receiver input, and the audio voltmeter, preferably calibrated in decibels (dB), to the receiver output.
Adjust the signal generator frequency to the centre of the receiver passband; adjust the output level to -67 dBm (100 uV in 50 Ohms); select FM; adjust the modulating frequency to 1 kHz; adjust the deviation to 1 kHz. Note the audio output level in dB. This is the reference level.
Change the modulating frequency to 250 Hz, note the receiver audio output, and add the correction factor from the following table. Change the modulating frequency, in turn, to each of the frequencies listed. If the receiver de-emphasis is correct, the audio output, plus the correction factor, will always be the same number; any change indicates the extent of the error in the de-emphasis network.
The amplitude/frequency response of the receiver IF stages also affects the frequency response of the demodulated audio. "Round shouldered" ceramic IF filters introduce additional de-emphasis at the higher audio frequencies. "Ripples" in the amplitude/frequency response of the receiver IF stages, due to mismatched termination of the crystal filter, also causes distorted audio and unwanted mute operation on weak signals.
Transmitter pre-emphasis is measured with a calibrated audio oscillator, RF signal sampler, dummy load, FM deviation meter and audio voltmeter.
Connect the audio oscillator to the transmitter audio input, and the other equipment to the transmitter output.
Adjust the audio oscillator frequency to 1 kHz and the output level to give 1 kHz deviation. Note the audio output level, from the deviation meter, in dB. This is the reference level.
Change the modulating frequency to 250 Hz, note the audio output, and add the correction factor from the following table. Change the modulating frequency, in turn, to each of the frequencies listed. If the transmitter pre-emphasis is correct, the audio output, plus the correction factor, will always be the same number; any change indicates the extent of the error in the pre-emphasis network.
It is generally accepted that a repeater system should have no more than 3 dB audio frequency response error, relative to the response at 1 kHz. The greatest errors tend to be response reductions at the lowest and highest audio frequencies.
For a stand-alone repeater, this means that the total audio frequency response error, of the receiver plus the transmitter, should be no more than 3 dB. Designers of stand- alone commercial repeaters aim to have no more than 1 dB error in each transmitter and each receiver.
For systems of linked repeaters, such as the National System, the 3 dB specification applies from end to end. The response errors, at the lowest and highest audio frequencies, tend to be in the same direction, and hence are additive.
For example, for a system made up of ten transmitters and ten receivers, the individual audio frequency response errors have to be less than 0.15 dB.
Receiver sensitivity is measured with a calibrated FM signal generator, and audio voltmeter.
Connect the (50 Ohm) signal generator to the receiver input, and the audio voltmeter, preferably calibrated in decibels (dB), to the receiver output.
Adjust the signal generator frequency to the centre of the receiver passband; adjust the output level to -67 dBm (100 uV in 50 Ohms); select FM; adjust the modulating frequency to 1 kHz; adjust the deviation to 5 kHz. Note the audio output level in dB. This is the 100% or 0 dB reference (signal) level. Switch off the modulation and note the audio output (noise) level. The audio (noise) level should have reduced by more than 60 dB, giving a signal to noise ratio of better than 60 dB.
Reduce the signal generator output level in 5 dB steps, noting the (noise) output and calculate the signal to noise ratio. Compare the results with the following table. Signal generator leakage may become significant, and give over-optimistic results, at levels below -120 dBm; this may be partly overcome by locating the signal generator some distance away from the receiver, and by inserting one, or more calibrated attenuators part way along the intervening coaxial cable.
In this particular receiver, "click noise" started at signal levels below -120 dBm. The point at which "click noise" becomes significant is often called the FM "threshold". For signal levels stronger than the FM threshold, the audio signal to noise ratio improves by 1 dB for each 1 dB increase in RF input level; below the threshold the audio signal to noise ratio decreases by about 2 dB for each 1 dB decrease in RF level.
The transmitter keying threshold should be set at the same receiver input level as produces 12 dB SINAD (see earlier). This input level provides reliable service and minimises false trigger. Unfortunately, many repeaters are adjusted to key the
transmitter at a far lower receiver input level (where the audio is virtually unintelligible), just above the noise level, producing intermittent false triggering. Subsequent changes in ambient temperature and DC supply voltage then produce continuous false triggering. This can be very annoying on a stand-alone repeater, but disastrous in a system of linked repeaters, such as the National System.
This technique was developed as a method of selective calling, so that listeners, on busy channels, weren't disturbed by messages not intended for them.
Unfortunately, selective calling has many bad side effects, which reduce the performance of any system in which it is used. The CTCSS tone is transmitted continuously, and is chosen from 38 frequencies between 67.0 Hz and 250.3 Hz.
High-pass filters are in both the transmitter and receiver audio stage. The filter in the transmitter is necessary to prevent low frequency audio contaminating any CTCSS tone. The filter in the receiver is necessary to prevent the users being disturbed by the CTCSS tone. Both filters affect the overall audio frequency response up to, and beyond 1 kHz.
The use of a low frequency tone necessarily increases the time to detect, and the time to release. The tone takes up some of the available deviation and, therefore, the audio deviation has to be correspondingly reduced. Audio non-linearities anywhere in the system, plus multipath between transmit antenna and receive antenna, cause the traffic audio to be modulated by the tone. The tone inserted at each station propagates throughout a repeater system, and the multiple high-pass filters, required to remove them, also degrade the low frequency end of the audio frequency response.
The deviation of the individual tones is additive. So, in a linked system of repeaters, such as the National System, the accumulation of transmitter deviation, caused by CTCSS tones, leaves little deviation for the real traffic!
For example if CTCSS tones are adjusted to 500 Hz deviation, at each transmitter, only half of the available deviation remains available for traffic after five repeaters, or no deviation after ten repeaters.
CTCSS selective calling has been proposed as a magic cure-all, to eliminate unwanted repeater triggering, due to interference and intermodulation. CTCSS selective calling can eliminate some types of unwanted triggering. But, if the intermodulation generation mechanism involves the CTCSS-equipped transmitter, the CTCSS tone shows up as a component of the intermodulation, is accepted by the receiver's CTCSS tone decoder, and the interference persists.
Unfortunately, selective calling only eliminates some of the unwanted triggering, merely delaying the time when the real cause of the interference and intermodulation has to be found, and cured.
Intermodulation is usually created at the repeater site, by a wide range of different mechanisms. Some mechanisms are due to poor equipment design (receivers and/or transmitters), or the use of inappropriate equipment (the wrong type of duplexer); other mechanisms are due to the degradation/ageing of previously properly working equipment, such as antenna systems (antennas and support/tower hardware).
The most frequent cause of intermodulation is overload of the receiver by one, or more, transmitters on the site. As previously mentioned, a high degree of transmitter/receiver isolation is required.
Aluminium based antennas are the second most frequent cause of intermodulation.
The necessary transmitter/receiver isolation can be achieved by two methods: antenna isolation and suitable filters, or single antenna and a suitable filter system called a duplexer. The use of antennas alone, without filters, cannot achieve sufficient isolation. Vertically polarised antennas must be mounted one above the other; they have insufficient isolation when mounted side by side.
There are three basic types of duplexers in common use, having three different levels of performance, and three different prices to match.
The lowest cost duplexers are the so-called pass/notch or mobile duplexers. The type or model number often starts with MD....
As the name implies, they are intended for use where duplex operation is required in a vehicle. Out of band attenuation is non-existent. They are not suitable for use at repeater sites, though many misguided amateurs initially try them. Their poor performance, in repeater use, and their contribution to interference to/from other services, has led many communications site owners to prohibit their use.
At three times the price and giving 30 to 50 dB better performance, are bandpass/bandreject duplexers. These are the most commonly used duplexers (by intelligent Amateurs). They typically feature four 150 mm or 200 mm diameter quarter wavelength-long coaxial resonators, two each in the receiver and transmitter paths.
There is no substitute for using large diameter resonators. Six of the smaller 100 mm diameter resonators are required to achieve the same transmitter-receiver isolation as four of the larger diameter resonators. But their insertion loss is more than double that of a duplexer made of four of the larger resonators.
The bandpass/bandreject duplexer may appear to feature two bandpass filters, but a subtle modification has been made to create additional response notches, giving improved transmit/receive isolation, on the two frequencies. These modifications ruin
the out of band attenuation on one, or both, sides of the pass band. You don't get something for nothing.
The lack of out of band attenuation, on one, or both, sides of the pass band, usually renders this type of duplexer unusable at a multi-operator site. For example, use of this type of duplexer in a 2-metre repeater, may give unsatisfactory performance, if the other services, include VHF-FM broadcasters (90 MHz), or aeronautical radio (120 MHz), or maritime radio (156 MHz), or Es (140 MHz), or E-band (160 MHz) land mobile.
The professional bandpass/bandpass duplexer is three times more expensive and has a corresponding improvement in performance. In this type, true bandpass filters are placed in both the transmit and receive paths of the duplexer.
The correct adjustment of duplexers requires very high quality test equipment, not normally available to repeater trustees. Duplexers must not be adjusted if this equipment is not available.
A repeater receiver, overloaded by the associated transmitter, is an obvious generator of intermodulation interference. Not so obvious is the fact that the transmitter output stage also acts as very efficient intermodulation generator, when external signals are coupled to it, even when protected by a duplexer.
An isolator (3-port circulator with dummy load on one port), placed between the duplexer and the transmitter, attenuates unwanted external signals trying to reach the transmitter, and thereby reduces intermodulation. An isolator is not a cure-all, since it also produces low levels of intermodulation distortion, due to nonlinearity of the ferrite. It has to be placed between the transmitter and the transmit port of the duplexer; the duplexer filtering reduces the interference generated by the ferrite.
Some manufacturers offer UHF duplexers that have co-axial resonators about three times longer than normal. These work on the principle of the 3/4 wavelength resonance. Unfortunately, they still have the normal quarter wavelength resonance at one third of the desired frequency, and the normal, additional, responses at 5/4 and 7/4 wavelength. Given the harmonic and sub-harmonic relationship between many of the VHF and UHF Amateur bands, the use of this type of duplexer is risky.
Antennas, and their mounting hardware and support structures (poles/towers), are subject to corrosion, causing loss of gain, poor SWR and intermodulation generation. Non-corrosive materials, such as stainless steel or hot-dipped galvanised steel, must be used. Aluminium is so reactive, in the salt- and water vapour-laden New Zealand
atmosphere, that antenna systems made of aluminium have a very short life, often measured in months, before bad SWR and/or intermodulation paralyses the repeater.
Some commercial site owners prohibit the use of aluminium antennas.
SWR is the most sensitive test of antenna condition; the absolute value of SWR is not that important, but CHANGES in SWR are. Therefore, the results of the regular SWR tests should be recorded, so that SWR CHANGES can be detected.
Antenna SWR is now measured in decibels and called Return Loss. Return loss is the ratio (in dB) between the forward power and reverse, or return, power, hence Return Loss. Feeder loss affects the measured return loss of an antenna. Consider the following example.
Assume, for example, that the antenna has a return loss of 16 dB and the feeder has a one way loss of 2 dB. The transmit power at the bottom of the feeder is attenuated by 2 dB on the way to the antenna. The return (reflection) of the antenna is 16 dB below this, or 18 dB below the power fed to the bottom of the feeder. The return is attenuated another 2 dB, on the way back to the bottom of the feeder, where the return loss is measured, for a total loss of 20 dB.
Commercial antennas are specified to have a return loss of better than 16 dB.
Therefore, return loss at the bottom of the feeder
= antenna return loss + twice the one-way feeder loss.
For example, if 20 Watts is fed into the bottom of the typical feeder and antenna, assumed above, the return power must be 20 dB (a power ratio of 100), or better, below the forward power, or less than 0.2 Watts. More reflected power than this indicates an antenna or feeder fault. In the absence of a Network Analyser (which is directly calibrated in dB) a professional power meter, such as Bird, should be used to measure the forward and reflected power.
The most common cause of antenna or feeder failure is ingress of water. It doesn't need to be liquid water. Water, in the form of vapour can easily find its way inside the antenna or feeder through the tiniest of holes, such as Possum claw marks in the outer jacket of RG-213/U.
A wide variety of water-resistant (not necessarily waterproof) connectors and adapters are available for most coaxial cable types. However, no waterproof connector exists for separately terminating the coaxial braid and inner conductor of RG-213/U, for use in a J-pole antenna, or for terminating the two inner conductors of a halfwave balun.
One of the best methods of waterproofing RG-213/U feeder or balun ends, to stop water creeping along the braid or along the gaps between the seven strands of the inner conductor, is to pot the assembly in old-fashioned Araldite (tm), or similar slow setting epoxy. Rapid setting epoxy must not be used.
The cable (or balun) end to be potted is placed, bare ends down, in a disposable leak- proof container, which has the shape of the desired end result. The Araldite is mixed in accordance with the instructions and placed in the container such that its level is 10 to 20 mm higher than that of the cable jacket.
The temperature of the Araldite is increased, by means of a fan heater, or hot air gun until it becomes transparent. It becomes very fluid and will leak out of any hole in the container. It also fills the gaps between the strands of the braid and between the strands of the inner conductor. Trapped air bubbles also float to the surface. Top-up the level of the Araldite, if necessary, to compensate for the air bubbles. Allow time for the Araldite to set.
Rotunda is the best brand of PIB self-amalgamating tape. Other brands have inferior amalgamation between adjacent layers of tape and permit the ingress of water vapour.
Apply the tape in accordance with the instructions. Remove the shiny backing tape and maintain a tension in the PIB tape such that its width drops to half. Wrap the PIB tape around the coax and connectors with a half width overlap. Repeat until three complete layers have been built up.
PIB cold flows under tension. Sharp edges of connectors, particularly the back nut of N connectors, can, over time, tear through the PIB tape. Therefore, a cone-shaped transition piece of PIB should be built up against the back nut, before the top layers of PIB tape are applied.
Keep the working surface perfectly clean; a minuscule trace of oil or silicone will prevent amalgamation between adjacent layers of PIB tape and permit entry of water.
Coaxial cable is manufactured by a large number of companies at a wide range of prices. You get what you pay for.
First check that the return loss of the cable is better than 30 dB. That is, if the cable is terminated in a perfect dummy load, the reflected power is 1/1000 of the forward power (20 milliwatts reflected for 20 Watts forward).
The longest lasting flexible cable features a non-contaminating jacket (having 10 to 25 years life or more), at a slight increase in price over contaminating jacket types (having 1 to 2 years life). RG-8/U has a contaminating jacket, whereas RG-213/U has the same inner structure but a non-contaminating jacket.
It is easy to tell the difference between the two jacket types. Sniff a new roll of cable. The cable has a contaminating jacket if it smells of "plastic"; non-contaminating jacket cable has no smell.
It is generally accepted that a repeater system should have no more than 5% audio frequency harmonic distortion at full deviation, and no less than 30 dB signal to noise ratio.
As for the previous discussion on frequency response errors, the contribution of both the transmitter and receiver must be considered. For systems of linked repeaters, such as the National System, the specification applies from end to end of the system.
In the case of harmonic distortion, the addition of errors is more complicated: some components of harmonic distortion are voltage additive, some components are power additive. Commercial operators design systems on the basis of worst case: voltage addition.
For example, for a system made up of ten transmitters and ten receivers, the individual harmonic distortion, at full deviation, needs to be less than 0.25%, if the system is to meet the end-to-end specification of 5%.
In the case of signal to noise ratio, the individual noise components are power additive. So, to meet the 30 dB specification, the transmitter and receiver, if contributing equal amounts of noise, should be better than 33 dB each. However, economic design allows the transmitter noise to be far lower, say 40 dB, so, in practice, the receiver noise becomes the limiting factor, perhaps about 31 dB.
For systems of linked repeaters, such as the National System, the 30 dB specification applies from end to end. For example, for a system made up of ten transmitters and ten receivers, the individual signal to noise ratios need to be 10 dB better: transmitters 50 dB and receivers 41 dB.
The receiver signal to noise ratio is mostly determined by the input signal level. As we have seen above, a good quality receiver requires about -104 dBm (1.4 uV) to achieve a 41 dB signal to noise ratio.
This is the minimum input for every receiver in the system, and takes no account of path fading. Some of the paths are over 200 km in length, which requires more than 40 dB fade margin. These long paths therefore need to be designed for a nominal input signal of -64 dBm (140 uV) to ensure that path, and system, performance objectives are met during a 40 dB fade.
The analogue National System is made up of an alternating succession of stations of two types:
A repeater transmitting near 439 MHz, plus one, or more, UHF Linking Stations (ULS) also transmitting near 439 MHz. The receive-transmit offset is always minus 5 MHz.
A "tail-less" inverted repeater transmitting near 434 MHz. The receive-transmit offset is always plus 5 MHz.
The technical requirements for the inverted repeater are identical to those of ordinary repeaters, previously discussed, except that they do not have a trigger "tail". The technical requirements for the "repeater plus ULS" site are much more onerous, as will now be discussed.
At an ordinary repeater site there is only one transmitter and one receiver, in any one Amateur band. The transmit/receive duplexer filtering requirements have been described above. At a "repeater plus ULS" site there are many transmitters and receivers, in the same Amateur band, required to operate simultaneously without mutual interference.
The equipment consists of a normal repeater, one or more UHF Link Stations (ULS), and a controller. The repeater has a trigger "tail"; the ULS does not have a trigger "tail". The ULS consists of a transmitter and receiver. Each ULS links to a distant, inverted repeater.
The controller manages the operation of the various transmitters and the routing of the audio, from one receiver to all transmitters, in the following two modes:
When the repeater receiver requests control, its audio is routed to all transmitters and all transmitters are turned on.
When a ULS receiver requests control, its audio is routed to all transmitters, and all transmitters, except the one associated with the particular ULS receiver, are turned on.
The equipment behaviour in these two modes can help determine, either on-site or off-site, which receiver has requested control. The signal is arriving via the repeater receiver, if all (repeater and ULS) transmitters are on; the signal is arriving via a ULS receiver, if its own transmitter is off and all the other transmitters are on.
If there is a conflict, with more than one receiver requesting control, a priority encoder decides which receiver is given control. The local repeater receiver is given first priority.
Unfortunately, on-site intermodulation, arising from various causes previously mentioned (corroded antennas or antenna mounts, de-tuned filters, inter-transmitter coupling, receiver overload, water in connectors, and so on), can also generate false signals on the receiver frequencies. These can cause false operation of the station, and interfere with signals from the distant inverted repeaters. In some cases, repeater trustees have falsely accused the distant station of being faulty, when local intermodulation, on the same frequency, is the actual cause.
Similarly, any test equipment being used to identify the sources of intermodulation must not, in itself, generate intermodulation - Amateur transceivers generate considerable intermodulation; handheld transceivers are particularly bad sources of intermodulation; multi-band handheld transceivers, with wide tuning range receivers, are the worst of the lot.
At Belmont there is the 900 repeater, the 8975 ULS, the 9025 ULS and the 9050 ULS. This is four transmitters and four receivers, required to operate simultaneously without mutual interference.
If the operating frequencies were widely spaced, or chosen at random, the duplexer filtering requirements would be horrendous: each transmitter would need four notches, one on each receiver frequency, and each receiver would need four notches, one on each transmitter frequency. That is a total of 32 notch frequencies, some of which would need two filter resonators.
Fortunately, FMTAG has been able to select frequencies that are closely spaced, to considerably reduce the number of notches required. For example, at Belmont, the four frequencies are on adjacent, 25 kHz-spaced, channels.
For example, in the 900 duplexer, the transmit path is designed to pass the 439.000 MHz transmitter frequency, while notching the 434.000 MHz receiver frequency. With Wacom duplexers, the 434.000 MHz receiver notch is found to be quite deep at the other receiver frequencies: 433.975 MHz, 434.025 MHz and 434.050 MHz. Similarly, the receive path is designed to pass the 434.000 MHz receiver frequency, while notching the 439.000 MHz transmitter frequency. With Wacom duplexers, the 439.000 MHz transmitter notch is found to be quite deep at the other transmitter frequencies: 438.975 MHz, 439.025 MHz and 439.050 MHz.
Therefore, provided the transmit and receive frequencies are carefully chosen, only one Wacom duplexer is required for each transmitter/receiver pair, and all the transmitter noise and all the receiver de-sensing is taken care of. However, if the frequencies are not chosen carefully, for example if they are more than, say, 100 kHz apart, considerable performance degradation can occur, and additional filtering will be required to regain full station performance.
However, successful operation on closely spaced frequencies requires engineering discipline. Antenna corrosion is the most common cause of intermodulation problems; antenna return loss should be measured frequently, say once or twice a year. The receiver and transmitter frequencies also need to be checked frequently. The receiver IF passband filters need to be narrow enough for 25 kHz spaced operation. The payback is the ability to obtain full station performance, while using the minimum number of standard Wacom duplexers, at minimum cost.
"Intercept point and undesired responses", pages 121 to 133, February 1983, Vol.VT- 32, No.1, IEEE Transactions on vehicular technology.
"Planning for the National Link", pages 6 to 8, November 1987, Break-In.
"Repeater intermodulation" and "RTA25", pages 4 to 6, July 1991, Break-In.
"Interference between co-sited equipment" and "Duplexers versus separate antennas", page 37, March 1993, Break-In.
"Data repeaters and digipeaters" and "Receiver and transmitter performance", page16, August 1993, Break-In.
"National System - discipline or anarchy?", page 41, April 1998, Break-In.
"All about phase noise in oscillators", QEX, Part 1: December 1993, pages 3 to 6, Part 2: January 1994, pages 9 to 16, Part 3: February 1994, pages 15 to 24.
"Designing low-phase-noise oscillators", QEX, October 1994, pages 3 to 12.
"Programmable frequency dividers", Q-Bit, September 1997, pages 6 to 9.