H.F. Repeater Project:
M.A.Pinfold ZL1BTB
This project has come about to improve mobile to mobile communication. Signal strength and readability between dipole based systems in daylight hours running 70-80 W P.E.P and about 100 Kms apart is all of the time 5+20 over 9. The copy between the above base station and a similar powered mobile with a 2 meter length helical is usually strength3- 5 though varying but is almost all of the time readability 5. However, the problem arises when two mobiles want to communicate over the similar rf path. Signals are much weaker the QSB is much pronounced , the effect of local and vehicular generated noise does much to reduce overall readability thus making it hard work to copy most of the time.
The only way around this is to improve the received signal strength at the mobile. This can only be done by increasing the erp of the distant mobile station and since we cant easily increase the mobile antenna size or increase the RF output of the mobile transmitter enough to make a sizable difference then we have to work on the signal between mobiles .This is where the repeater concept emerges . we pick up the mobile signal on a full size dipole, amplify and frequency shift it and retransmit it on another full size dipole . Thus we have both mobiles working into a set up that provides very useful signal to noise both ways.
In order to do this successfully, there are some factors that must be taken into account.
The input and output frequencies used must be relatively close together eg at upper 80 meters no more than 0.5 MHz this rule of thumb is because the helically wound mobile antennas fall off in performance radically when they get a few 100,s of KHz off resonance. Go too far away from resonance and the system becomes quite deaf, and you defeat the purpose of the system. The closer you can get the transmit and receive frequencies then the better the mobile antennas will perform, However in general, the quality of the received copy is a function of the received signal to noise and not just signal level. The other important factor that you must consider is the physical separation of the base transmitter and receiver dipoles .In my setup at home I have two dipoles, One in the front of the house and the other way down the back of the section are end to end in configuration to minimise coupling, using a HP spectrum analyser and Wavetek 3001 signal generator ,I have measured the isolation between the two dipoles at around 43 dB
This is no where near enough isolation to run 70W PEP, 0.2 Mhz separation and have a receiver as sensitive and vulnerable to strong signal overload, as is the ubiquitous 7727 series transceiver.
You need good low loss filters to remove unwanted signals in both the transmitter and receiver.
THE ABOVE FREQUENCIES ARE NOT THE ACTUAL ONES USED .
Base Station Description:
One Codan is used for the transmitter and one for the receiver. You can couple them together in two ways.
I favor the 1.6 MHz if coupling as there is no demodulating or remodulating of the , therefore the potential for introduced distortion should be less ,hopefully enabling a few more useable dBs signal to noise at the receiver when the signal gets weak ??
The transmit unit needs to be turned to enabled by either hard wiring the ptt line or via the use of the most excellent squelch units the Codan receiver possesses . the squelch can drive a "ptt relay" via suitable simple transistor /op amp interface unit. The squelch unit operation is to be desired as it operates very well indeed. It is a real audio squelch, the kind that all ham radio transceivers should have installed. I have seen static crashes at s9+ and the Codan squelch stays closed, only to open if someone spoke!!!. Also this means all the transmit circuitry is quiescent and not consuming power and not transmitting broad band noise at repeater output frequency.
Filters:
Effective isolation between a good transmitter and receiver is the key to effective repeater performance
The Codan has a broadband linear power amplifier stage. The output spectrum is passed through bandpass filters to reduce higher order harmonics only. If you are transmitting on 3.60 Mhz (heaven forbid) and the receive frequency is 3.8 MHz then the transmitter will produce tens milliwatts of broadband noise on frequencies half a meg either side of the carrier frequency.
This noise will be modulated by the carrier signal at 3.60 MHz, remember the receiver at 3.8 MHz will hear this modulated noise and attempt to demodulate it, you will get feedback between receiver and transmitter, this shows as a swirling twisting sound to the noise on your received signal. Too much feedback and the signals will be almost unintelligible.
We have to notch out the noise coming out of the transmitter on the receive frequency (3.8 MHz) and to pass the wanted frequency.
Notched broadband noise out of TX unit with no 1.6 MHz if applied , MILLIWATTS of it !!
To stop the transmitter overloading the front end of the receive Codan, we have to notch out the strong 3.60 MHz transmitted signal. I made up some simple high Q tuned circuits in the form of a single series resonant tuned circuit at the unwanted frequency of interest, 3.60 and 3.8 MHz . these were placed effectively across the respective receive and transmit coax to "short circuit" unwanted signals to ground.
I used a brand new 4 litre paint tin as my container and placed a coil of 20 turns of 6mm enameled copper wire, 3 inches in diameter inside the tin one end soldered to the bottom of the paint tin. The top end of the coil had a high quality 230 pf mica capacitor combination soldered to it., the other end to the center pin of a flange mount BNC socket. Don’t use a capacitor that is too small or you notch depth will be too shallow, or a capacitance that is too big , your notch will be deep but it will also be wide and begin to absorb some of your wanted signals. I had to squeeze and expand the coils to get to the tuned circuit frequency I wanted. It is a lot of jiggling to get a good rejection notch depth and still have it narrow enough such that it does not absorb too much of the wanted signal .The best compromise I could get was 17 dB!! And not loose too much wanted signal, but wait there is a better, easier more reproducible way!
Aim for a total figure of at least 40dB between the receive filter and the transmit filter . If you are making a straight suck-out notch filter to put across the coax, You don’t have to use high voltage caps. The transmit filter will only be carrying a few tens of milliwatts of unwanted signal. But do go for good quality components to achieve as high a Q as possible.
After I had experimented enough with the paint tin filter and had proved its worth. it was time to spend some money. . I had a local engineering aluminium fabricator make up some "paint tins "out of 1.5 mm Aluminium sheet. They were about 200mm long and 150 mm in diameter with a 180mm square sheet welded to the bottom leaving them open on top . Four of these were made ,two for transmit and two for receive. However on hindsight, I should have had them made square ie in the form of two 150mm square boxes 150mm and 200 high so I could have sat two tuned circuits together with a coupling aperture between ,thus I could have experimented with aperture and capacitive top coupling as well link coupling.
What I found was that if you are using resonant tuned circuits with anything more than 10w of RF applied to them at resonance, is that you must use wide spaced (more than 2mm spacing between plates) air dielectric tuning capacitors
Those small 100pf E.F Johnson air spaced receive capacitors will arc over with the trusty Codan power transmitting through them. I mounted the tuning capacitor at the base of the container The Hot end of the tuned circuit is down inside the container and shielded from stray pick up or radiation of signal , this enables easy access to the coils cold earthy end to try different coupling loop configurations.
I used very thick 7mm enameled copper
wire to try and achieve a high "Q". the coil of 29 turns 3" in diameter
close wound was soldered to the top of the tuning capacitor and the ground end of the coil
flattened out with a hammer to improve contact at the high rf current earth end f the
coil, and bolted tightly
to the wall of the aluminium container. SO239 connectors were attached to the side of the enclosure about the level of the end of the tuned circuit and about 3 inches apart, close enough to enable capacitors or inductors to be placed between the SO239 center connectors. I made two (slightly larger than the tuned circuit diameter),1 turn coupling loops out of 6mm enameled cu wire and solder them to the center and earth lug connections of the SO239’s, and placed them both over the cold end of the coil to achieve good tight but adjustable coupling.
I set up the sweep generator and the trusty Spectrum analyser (as I do not have a matching tracking generator ) using the sweeper at a high rate 100 Hz or so and the spec an at 100 kHz per division at a much slower sweep rate enables one to see a usable graphical representation of the tuned circuit response. I used a digital camera fujifilm MX500to generate the pictures, covered over the flash aperture and light sensor and took a photo of the screen in dim light, hoping the shutter will stay open long enough to do the job and I have some sort of acceptable result as you can see .
The sweep above shows the rejection notch below the band-pass about 40 dB deep with an insertion loss of under 1dB. This is fine for the wanted signal above the unwanted signal .
I found that the coupling loop position with respect to the tuned circuit controlled the distance of the notch from the peak and you could move the notch towards and away from the pass-band peak .only one pickup loop was more responsible for a majority of the response. also there was a minimum distance ,pass to notch that could be achieved by this was about 150 KHz . moving the loop away from the coil pushed the antiresonant notch away from the pass-band peak.
However there arises the problem of how to place the antiresonant point above the pass-band peak as would be required for the other frequency filter. I worked on this for many nights trying the standard technique as applied to full length coaxial cavity filters , the phasing inductance/capacitance across the input and output of the filter ( see coaxial cavity filters article ) no real joy either. However at the final 5 minutes of frantic "dickering in the shed" after being called inside by "her indoors" I struck the jackpot .
The antiresonant notch can be placed on the opposite side by using a high value series mica compression trimmer of value 50-800 pf in series with the 1-2 turn coupling loop. Tune the shape of the pass-band response with the mica compression trimmer.! This technique enables you to cancel out inductive loop reactance and substitute capacitive reactance ,and place the notch on the high side. It seems only effective in the single loop Tee connected tuned circuit.
100 Khz per Div Tee connected filter (one port only)
ISOLATION:
You require at least 80 dB of notch depth at the receive frequency , in the transmit passband filter . This is to remove from the transmitter noise on the receive frequency. Remember the Codan 7727 is a very sensitive receiver and can easily hear 0.2 uV , so you want as little interfering signal/modulated noise from the transmitter as possible. I used two filters in series to achieve this with a combined loss of about 1 dB. In the receive coax, you need a receive passband filter with a notch to remove as much of the transmitter signal. This is to stop the transmitter from overloading the receiver and causing receiver intermod and desensitising. I used one receive passband filter with a notch depth of about 35 dB. Perhaps two in series would be better?
Another source of unwanted coupling between transmitter and receiver is the via the two antenna feed lines Do use a balun at the antenna of both transmit dipole and receive dipole. A balun will help stop rf coupling to the outside of the feedline and getting back into the set and causing potential isolation problems.
The other source of potential coupling is via the coax that carries the signal from the transmitter to the receiver, use some large toroids, one at each end, with 5 or 6 turns of the coax through them to choke any unwanted induced rf coming along the outside of the interconnecting 1.6 MHz coax.
Modifications to the transmitter and the receiver.
You really need circuits of the Codan 7727 series to do these modifications and please be aware that there are subtle variations in the actual circuitry and component values between the various versions although the major circuit layout is similar.
The 1.6 Mhz I.F needs to be tapped off and terminated in a BNC at the rear of the
transceiver . This is done via a simple high impedance unity gain j-Fet amplifier with a
Bipolar low impedance output stage suitable to drive a 100 metres of 50 ohm coaxial cable.
The Codan has a convenient test point TP5 that brings the 1.6 MHz I.F for measurement. I
coupled signal out of this point with a 10 pf capacitor and into the high impedance unity
gain buffer stage. The output of the amp was fed by good quality coax to the BNC socket on
the rear of the set, the coax being connected by a .01 capacitor to the center of the
connector.
You need the coax D.C isolated from the I.F as I use the coax for DC signaling the distant Codan PTT line to enable transmit. I do not use the Codan squelch to control the transmitter although it is very good, in fact excellent, it does still misfire now and again. Too unreliable on weak signals. You may like to work on this concept, let me know how you go.
Transmitter modifications.
At the transmit Codan, the 1.6 MHz signal needs to be fed into the up mixer just before the crystal filter. This is such that the relatively broadband signal from the I.F TP5 , is filtered down to a bandwidth of 2.7 KHz, before it passes into the upmixer and is converted to the final output frequency before amplification. There is no advantage having the upmixed I.F, fed to the power amplifier stage, wider than the 2.7 KHz signal you are receiving down the coax from the receiver. This also helps to minimise the amount of broadband noise transmitted either side of the upmixed carrier signal. It can help to improve frequency isolation between transmitter and receiver. In theory you should really have some form of impedance transformation from the low impedance of the coax/level resistor combination of 50-100 ohms, connecting to the crystal filter with an impedance of 1.8-2.0 K ohms. The purists could use a step up stage in the form of a link coupling into a 1.6 MHz tuned circuit across the input to the crystal filter to attempt correct impedance matching and thus maintain a rectangular shape factor and minimise passband ripple.
I used a small 2 pole relay to switch the ptt line for transmit and at the same time to short out the zener diode that stabilises power to the 1.650 KHz carrier oscillator . This enables the Transmit Codan to "hear what the Receive Codan hears when it is in receive mode, However when the transmit relay pulls in and shorts the ptt line I also short out the zener diode of the oscillator , it stops oscillating as it is not required in so there can be DSB signal out of the Double balanced modulator ,or residual carrier leakage at the if frequency, into the upmixer. mode (unless you are going to connect the two Codans by audio link from TP8 to TP8 then you will require the transmit carrier oscillator to function in transmit mode.). I turn my repeater on via a DTMF code ( Yes the Codans with the specified high stability crystals are stable enough to use DTMF) At the transmit Codan, 12 volts is fed through the PTT relay and into the centre of the interconnecting IF coax, when a valid "on" tone sequence is received , a transistor is switched on in the receiver and grounds the center of the RF decoupled 1.6 MHz interconnecting coax. This effectively switches the PTT relay on and the repeater function is enabled . When we have finished using the repeater, the simple "off "DTMF sequence is sent and releases the PTT relay
.
RECEIVE CIRCUIT MEASUREMENTS
I measured the 1.6 IF output level with a high impedance oscilloscope probe on TP5 whilst injecting the correct HF signal into the antenna connector
Input level Test Point 5 Level
uV dBm 1.6 MHz IF
0.1 uV -130 0.05V
0.3 uV -120 0.1V
1.0 uV -110 0.3V
3.0 uV -100 1.0 V
10 uV -90 1.5 V
30 uV -80 1.5 V
There is a good level of IF signal available. Unfortunately when it is all tuned up there is only 30 dB of signal headroom due to the effects of the AGC and the amount of IF gain. If you de-tune the I.F. such that the gain goes down by 10 dB, you can get 40 dB of potential signal headroom. This is to allow a greater variation in rf output power when transmitting SSB and it doesn’t sound so "full" and compressed.
The Codan receiver needs a simple modification to the AGC stage to improve the intelligability of the retransmitted signal . It took a lot of tracking down to fix this problem of poor signal to noise on the received signal as it always sounded scruffy and noisy ,despite excellent received signal strength. I found that it was due to the very fast attack and release of the AGC system. The receiver gain was rising back up to full between voice syllables, and noising up the retransmitted received signal , The Codan has a dual time constant AGC which is great for mobile use but turned out hopeless for the repeater function. The fast time constant is provided by a 1 uF tantalum found on the outside edge of the First IF transformer. Change it to about 20-40 uF tantalum , this gives a much longer time constant and hang on the AGC rail , it works a treat and made a huge improvement to intelligibility.
TRANSMIT CIRCUIT MEASUREMENTS
I then looked at the 1.6 MHz TX SSB IF level just after the crystal filter at test point TP5 while in transmit mode . A whistle into the microphone produces 0.3 V of DSB at 1.6 MHz measured with a high impedance probe
This shows that the receive output IF level is more than adequate to drive the TX up mixer with minimal coupling circuitry. On the transmit Codan, I injected the 1.6 MHz I.F signal into the post SSB crystal filter amplifier ,capacitively (.01uf ) into the emitter of tr16 , technically I should really feed the High level 1.6 MHz IF signal from the receiver through the transmitter SSB filter to clean up any broad band noise contributed by the receiver IF , but it was not tried, I may try this next.
TRANSMITTER MODS
The 1.6 MHz carrier injection oscillator must be disabled ,this can be done several ways ,I short circuited the oscillator bias zener diode ,or lift one of the resistors that power the oscillator .
You can also lift the appropriate component so as to stop any RF noise coming through the crystal filter from the DSB generator . The rear of the transmit Codan has a BNC Flange socket fitted on the rear. The centre of this is capacitively coupled (.01uf) to a 100 ohm variable pot such that there is 100 ohms always across the connecting coax. The centre wiper connects to the centre of a short piece of coax that feeds the emitter of TR16 via another .01uf cap.. The pot enables variable rf power output .
I have a small transistor (BC548) driven, low current relay that connects the PTT line when energised. the D.C signaling for this is via the long 1.6 MHz IF coax via good high impedance RF chokes (I wound my own on small surplus ferrite toroids) connected to the centre of the BNC at the transmitter and receiver.
PERFORMANCE
I injected 1.6 Mhz from the signal generator (HP8640) into a point R113 (820 ohms)just before the input to the crystal filter. via a DC blocking capacitor and then measured to RF output power into a 50 ohm dummy load using a Bird Thruline power meter..
Input Level to upmixer RF Power Out in Watts
-50 dBm (1 mV) 1
-40 (3mv) 8
-39 10
-38 14
-37 18
-36 24
-35 30
-34 37
-33 45
-30 (10mV) 50
There is ample signal level from the receive IF amplifier to drive the up mixer, so you can have a really long run of coax between sets and afford to loose quite a bit of signal .
My two dipoles configured end to end are only 200 feet apart and have a measured isolation of 43 dB this is why you must have extra filtering to try to achieve as much isolation as possible . If you don’t, feedback between the transmitter becomes apparent in the form of swirling ,phased distortions on signals coming through the system .My test transceiver base station 150 kms away used to reckon I was transmitting from deep space or somewhere behind Mars . You will also notice a pronounced increase in "modulated phasing sounding noise on the channel and not a random broad band typical HF sound.
There are a number of ways to use this system . In the experimental stage, I had the ptt line on the Transmit unit permanently hard wired, and used to leave the receiver running all the time. I would manually switch on the Tx unit when I wanted to use the system before going outback or into town, this has the slight disadvantage of being a little wasteful on rf spectrum as the machine is going all the time relaying all static crashes pops whistles and signals out on another frequency. You can hardly hear its RF band noise over 20 Kms away mobile, beyond that nothing at all until someone puts a signal into it.
The latest version uses DTMF to enable the system, The Codans are more than stable enough on SSB to use DTMF reliably! Remember how I use the DC down the coax to switch the PTT relay in the transmitter . When the DTMF chipset receives the correct tone codes, it turns the PTT line on via the IF coax connection and the system is enabled. I use another code to turn it back into standby mode. I have resisted using the squelch to trigger the transmitter as it can be somewhat unreliable and mute out on a readable but weak signals, also it reserves the use of the system to those who know the access codes. I or other users turn the system on, use it, then switch it off.
The other idea is to use a 1750 tone burst to enable the system and a different tone tail at the end of each transmission to disable the system, or perhaps a 1750 Hz tone burst to enable the system and reset a five minute timer which is reset every time some one talks, but turns itself off after a preset period if no-one re-triggers it. You can’t use CTTSS as it will not go through the receive SSB filter (they roll off rapidly below 300 Hz! It just wont get through at a level to be useful.)
How does the system go ? very well, I’m pleased to say. I have tried it over a range of 200Kms away from the base and conversed with a friend who was aeronautical HF mobile, loud and clear. There is of course some fading and distortions as is apparent on any HF SSB channel but it has been quite a useful device to make that usually scruffy hard copy mobile to mobile signal a pleasure to copy.