Duplexer vs. Receiver combining -
by Fred Daniel

In a perfect world, there would be no need for duplexers. Each repeater would have a separate antenna for the receiver and transmitter, and there would be no interaction between antennas. Since tower space is usually both limited and expensive, this is not practical. From this was born the duplexer. There are two basic types of duplexer. The band-pass and the notch models, and some designs combine both features. Some economy is realized with a duplexer since the repeater only requires one antenna and feedline. This is ideal at a site with only one or two repeaters. 

The least expensive is the notch duplexer, which simply notches the transmit frequency from the receiver line and notches the receive frequency from the transmit line.

Typically, this model is small [see  adjacient  photo], with low insertion loss, and the isolation is typically 70 or 80 dB. In some cases, you may require  additional filtering or a notch duplexer with 100 dB isolation for proper operation. The notch frequencies are very narrow and must be retuned, whenever a different frequency is desired, even 50 KHz away. 

The band-pass duplexer is somewhat more robust in design, It is usually larger in size, has a little more insertion loss as compared to a notch duplexer, and almost always has at least 100 dB of isolation. The operating frequencies are less critical and vibration is less of an issue in regard to tuning. This is the preferred design when expense is not an issue. In addition, it is not unusual to see 2 to 5 adjacent channels on a hybrid transmitter combiner being duplexed within 75 KHz, provided the overall power specification is not exceeded.

The greatest value of a duplexer comes at sites with very few repeaters, so the initial expense and added maintenance of using a receiver combining system, is not normally justified. However, the gold standard of performance is with properly designed and maintained receiver and transmitter combining systems. However, since receiver combining creates a single point of failure, steps must be taken to insure survival, and as we proceed more will be presented on the subject.

Proper receiver combining simply means tailoring and grooming every signal that reaches your receivers, whether you have 5 or 100 in your system. It is common to use a wideband antenna such as the Sinclair 4 dipole as your main receive antenna. These will easily tune 420 to 470 MHz with 9 dB offset gain. 

In your initial design, you must decide what separate bands of frequencies can be combined in each system. For example, it may be simpler to filter 455-459 MHz. and filter 466-470 MHz. seperately, then re-combine to distribute. 

With receiver multi-coupling, the pre-selector is the first filter encountered, by the main antenna and brought to the equipment building by the feedline. The pre-selector is usually found to be one or more "window" filters. In some cases, notch cavities are added to the pre-selector, to reduce the bleed-through of a local transmitter operating near the window band-pass frequency. When building a pre-selector, large high-Q cavities are not required. A series of small low-Q cavities work better to create a low-loss window, with provides very steep attenuation on the sides. If the desired window needs to be wide or operate at VHF frequencies, a series of helical resonators are usually preferred, due to size issues. At UHF frequencies and above, inter-digital filters are often used. The ideal pre-selector loss is zero, but the target is usually less than 2 dB. Keep in mind, any loss from feedline or the pre-selector, before the first amplifier, cannot be made up in gain from the amplifier. That is why on tall towers, and other critical applications, the pre-selector and first amplifier is placed close to the antenna, at the top of the tower. More on this later. A thorough discussion of the various filter designs are beyond the scope of this discussion, but can be found elsewhere on this web site.

The first or only amplifier (amp) in a distribution systems must have low internal noise and a high-intercept rating. The noise figure in dB will add to the feedline and pre-selector losses to reduce the level of weak receive signals. The intercept point is a threshold measurement, of the TOTAL level of all signals entering the amplifier before internal mixing occurs. This mixing should be avoided at all cost, as this will create interference from an effect called intermodulation, or intermod for short. Receiver pre-amps used in most radios, or built for external amateur radio use, typically have a low intercept point. That is why connecting receivers or transceivers directly to an antenna, will cause intermod to occur at locations with as few as 3 other repeaters. This intermod will not only appear in the "barefoot" receiver, but will also cause interference in other receivers at the same site, even receivers served by a quality multi-coupling system. The reason is because the intermod creates signals on the same frequency you desire to receive, and transmits them on the companion receive antenna, so it is not practical to filter. The most preferred amplifiers available are made by www.AngleLinear.com or www.advancedreceiver.com on the web.

Signal splitters are the preferred device to divide the output of the amplifier, for distribution to the various receivers. These are similar to cable TV splitters, but are of a better quality and are 50 ohm, instead of 75 ohm. They are available in splits of 2 and up to 24 and have a typical insertion loss [beyond the power split] of less than 1 dB.  The MiniCircuits.com ZFSC-6-110 shown is an ideal splitter.  
In most cases, the typical amplifier has enough gain to drive up to 36 receivers, with several dB of surplus gain to compensate for cable losses from the distribution panel to the individual receivers. When your design worksheet indicates you will have surplus gain, beyond zero dB, a simple inline attenuator should be inserted on the OUTPUT of the amplifier, to reduce the gain. Any surplus gain beyond several dB will reduce the 3rd-order intercept point of your receiver and bring you closer to the point where that receiver will create internal intermod. However, if you use a calibrated step attenuator such as a Kay model 839, in front of the receiver, you will notice several DB of additional gain does yield a noticeable improvement [using SINAD measurement].. However, you will reach a point where additional gain no longer shows an improvement. Again, I will stress the need for double-shielded silver-plated cable such as RG-400, with silver-plated crimp connectors in your receiver distribution system. Ordinary coax cables such as RG-58/AU appear to work good, but only have a little over 60% shield coverage. This will allow local transmitter signals to enter the distribution system cabling and add to the overall RF energy appearing at each receiver, thereby reducing the point where overload can occur. This warning also applies to the connectors used. Quality silver-plated crimp connectors such as Kings or Amphenol are essential, as tin-plated connectors with poor plating [parts from China] will allow external RF to enter the distribution system. If you have done your installation correctly, you should be able to remove the main receive antenna and terminate the pre-selector input with a 50 ohm load, and not receive a on-channel hand-held in the same room. This is a tough test, but controlling what enters your receiver distribution system is the objective.

In order to make system testing easier, it is suggested that a non-directional 30 dB coupler be inserted between the main receive antenna and the pre-selector. This will allow a technician, without disturbing the main receive antenna for the site, to: 

1) inject a signal for testing the system or, 

2) testing a specific receiver, or 

3) connect a spectrum analyzer to see what is coming down your main receive antenna. 

By using a 40 dB coupler, the calibration on the signal generator is off by a factor of 1,000, thereby representing 1 microvolt as 1 millivolt, and 10 microvolt as 10 mill volts. This reduces the confusion found when using a 20 dB coupler. The use of a non-directional coupler such as the Bird model 4275-020 is desirable so that the coupling of 30 dB remains uniform for both signal injecting and monitoring what is coming down the main receive feedline. If the location of the coupler is some distance from the receivers at the site, you can use a 20 dB coupler and extend the coupled port across the building using RG-214 or Heliax, then inserting an inline attenuator of the appropriate value to compensate for the extension cable losses. This will insure you remain coupled at 30 dB for ease of calibration.

We can now discuss the issue of single-point-of-failure introduced above. At the very least, the person charged with the maintenance of a receiver distribution system should carry on-hand either at the site or in the service vehicle, a complete set of all replacement amplifiers and splitters. This is in case you take a bad lightning strike. Spare filters are rarely required, as all inputs and outputs are simple wire loops to ground. However, spare pass and notch cavities are handy to have at the shop to experiment with, and used to train technicians as to the proper tuning methods. Some commercial tower top pre-selector/amplifier designs, claim increase reliability by using two amplifiers, phased in parallel. This is a good idea, as the output will only drop 3 dB should one amplifier fail. The pair of amplifiers can be monitored by constantly measuring the current that powers the assembly going up the output coax, from the equipment building. If funds are available, a better solution might be a second primary receive antenna, with its own pre-selector, and amplifier. a remotely controlled coax relay could switch the splitter system between the two systems, should a need arise. A few repeaters are built with dual diversity receivers, so the second antenna system could also feed that. I have seen repeaters equipped with a JPS Communications SNV-12 voting system, where both receivers are at the same site, on different antenna systems at least several wavelengths apart. The objective was to eliminate flutter and picket-fence effects on the repeater input. It worked very well for that.

Some ideas for remotely performance testing the receiver distribution system, or any repeater, range from the simple to the sophisticated. The simplest approach is install a beam antenna at your shop, fed with Heliax cable to assure consistent and repeatable test result, over time. Simply use your service monitor signal generator, with one or two amplifiers, to provide just enough signal to perform a 20 dB quieting or SINAD test. By recording the results, even long-term changes can be revealed. If a quality signal generator or service monitor is not available, then a small control station can be used, with a 10 watt 20 dB attenuator, together with a step attenuator like a Kay model 839, will work fine.

Continuous on-site monitoring is easily achieved by selecting a unused frequency, within the band pass of your receiver distribution system pre-selector, and then transmit a very low power signal, just enough the meet the SINAD test, on a companion receiver. It is important the only way the receiver can hear the low-powered transmitter, is through the receiver distribution system. So proper shielding of all components is necessary. For example, a simple 1/2 watt exciter and receiver is available for $16 from www.NiceRF.com as the model SA818-V for VHF or the SA828-U for UHF bands. These can be fully shielded using Bud aluminum boxe. The power and other control leads must be RF bypassed using bypass feed-through caps. You want the radios to be very narrow-band, since the only modulation will be a single CTCSS tone [i.e.: 151.4 Hz at 1 KHz. deviation], so that loss of your test signal can be easily detected. Use attenuators to reduce the exciter output to the minimum required signal level required. The frequency might be the split between regular channels or a guard channel between licensed services. Remember, the output will only be enough to reach you primary receive antenna, and may not even be heard on the ground below the tower. Once placed in service, you can connect the receiver output to your alarm panel, or some other alarm facility you have, to alert service personnel. If space or budget is limited for the transmit antenna, an alternative is to couple the signal to another existing transmit antenna, at the site. The same 30 dB coupler described above will work fine, and will not cause any mixing.