Voting Comparators - an overview

Up one level
Back to Home What is a Voter?
An overview of receiver voting systems
By Mike Morris WA6ILQ
(callsign)@repeater-builder.com
Date printed: 21-November-2008

A "normal" repeater system has a receiver and a transmitter, connected by some form of repeater controller, and located at an optimal site for reception and transmission. The system coverage is usually dependent on the geography of the surrounding terrain.

Early public safety dispatch systems had a problem - they had high power transmitters and good receivers and antennas, but the lower-power mobile units (i.e. patrol cars) would be able to hear the dispatch transmitter, but not be heard by the dispatch receiver.

Early attempts at improving the reception included multiple outlying receivers and wire-line connections (i.e. leased dedicated phone lines) back to the dispatcher, who would have to develop selective hearing on multiple speakers.

Enter the "voting system" - a method of automatically selecting the best signal. Early voting systems used multiple audio tones to indicate receiver carrier signal strength, but these were superceded by ones that worked by measuring the signal-to-noise ratio on the fly. Due to the fact that a full quieting signal without speech sounds like a squelched (muted) receiver, an idle channel (i.e. one that was squelched) was indicated by sending a constant precise frequency audio tone down the wire-line - this was, obviously, called the idle tone, or the idle marker. GE used 1950 Hz, Motorola systems used 2175 Hz just to be different. The status tone identified to the voting panel that the wireline between the receive site and the voter input was intact, its presence or absence told the voter that the receiver had squelched or unsquelched, and the level of the tone was used as a convenient reference for setting audio levels.

One "gotcha" was when the voting unit was located at the same site as the dispatch base station - most voters are designed for voting wire-line-connected receivers and the base station receiver wasn't. Some installations wire-lined the base station to the dispatch center and put the voter there, others left the voter at the base station site and used a special interface card to allow a local receiver to talk to the voter channel assigned to it.

Vocabulary:

Receiver voter / receiver voting selector / voting panel / voting site / voter: The equipment, usually rack mounted, that interprets the audio from several outlying receivers (connected by leased phone lines, microvave links, or standard point-to-point FM links) and picks the best quieting signal on the fly and routes it to a single output. Voting is continuous and the switching can be faster than 10 times per second. Control options allow forced selecting a specific receiver, and to manually disable specific receivers (sometimes referred to as "failing a receiver" or "force failing a receiver"). The leased lines aren't cheap - the inband signalling uses upper audio frequencies above 2700 Hz. The lines are equalized to be flat from 300-3400 Hz. and usually the phone company prices the wireline as $X plus $Y per cable mile.
Voting receiver / remote receiver / satellite receiver: an outlying receiver that feeds the voting panel. The term is sometimes used to refer to all of the equipment at the outlying site - the antenna, receiver, controller, link transmitter or microwave interface panel, etc.
Channel: The number of receivers that a single voter unit (sometimes called a voter shelf) can handle is usually enumerated as a channel count. Voter shelves come in 2, 4, 6 or 8 channels (depending on the manufacturer), and some manufacturers have kits that allow multiple untis to be tied together. For example, GE has a kit available that ties two 6-channel shelves together into a single 12-channel voter. And voters can be stacked - the output of one shelf can be used as a channel input to another shelf. It's not unusual in large public safety dispatch systems to have a receiver at each city building (police station, fire station, library, etc) with a wireline back to the voter site. Now duplicate that for each radio channel.
E & M: a generic term used to describe a type of telephone trunk (inter-office) circuit (copper wireline, fiber or microwave)... The letters "E" and "M" are labels for a status signal, or a "busy" lead that was transmitted down the path along with the voice. When the "M" lead (for "Mouth") was made active at one end the "E" lead (for "Ear")on the other end went active. Think of the "M lead" as being the PTT on one end of the circuit and the "E" lead as the COR on the other end. Depending on the type of equipment used, the E&M circuit can be implemented over connections that have DC continuity between the two ends, or connections that require using tones to indicate status (which are filtered out at the receiving end). Some types of E&M circuits (mostly on fiber or microwave systems) allow multiple status signals in each direction (one could be, for example, COR and a second could be PL decode). E&M circuits consist of the balanced audio pair plus the status signals, which frequently are -48vDC and ground referenced. E&M circuits can be one-way, or bi-directional. Overall, E&M technology is much, much more complex than this, but the above is enough for this write-up. Just remember that E&M circuits can "ride" on a wire (i.e. telco), fiber or on microwave links.

Basically from the point of view of the repeater controller, the audio output port of the voting panel chassis is the local repeater receiver audio output. The PTT signal from the voting panel is treated as the COR signal of the repeater controller. Each input channel of the voter is connected, either by wire links, E & M links (wireline, fiber or microwave) or by point-to-point RF links (usually in the amateur radio service on 420-430 MHz, 900 MHz or 1200 MHz bands) to a satellite receiver somewhere inside the repeater transmitter coverage area. When a user keys down one or more of the receivers un-squelch and the voter does an on-the-fly comparison of the signal-to-noise ratio of each of the incoming signals and feeds the best one to the repeater transmitter. It is not unusual in a properly configured voting system to have a multi-syllable word "assembled" from multiple sites.

A good voting system is completely transparent when it's working properly - you can't tell it's there except by the superior coverage. Obviously the voter has to be voting identical simultaneous signals, which is why using VOIP (voice over IP, a.k.a. the Internet) as a link to bring in the audio from one or more outlying receive sites usually does not work - there is too much delay and jitter, causing the voter at one instant to be voting the noise between two words on channel one and voting the middle of the previous word on channel two. Manually adding a fixed delay to the one receiver in an IP network does not work either, as the per-path internet packet switching delay is not consistent, even within one transmission, and definitely not from one transmission to the next. Note that there is a lot of incentive to "fix" this characteristic, and a revision of this paragraph may be needed at some point as the technology advances.

Different voters use one of two designs - either noise voting or a valley voter. Noise voters work by simply rectifying the high frequency noise into a DC level and compare those voltages (i.e. picking the lowest noise). Valley voters work by comparing the instantaneous channel noise at the times that speech energy is not there - i.e. looking for noise inside the speech bandwidth between words, even between syllables. My personal preference is a valley voter since you can't count on the squelch noise above 4khz since it's not carried by some link facilities - microwave or point-to-point RF links. There are some systems that use narrowband RXs and couple the discriminator audio into the modulator of a wideband links transmitter and run audio frequencies up to 15khz or even 20khz on the link, then use a wideband receiver and run that discriminator audio into a noise voter.

The manuals on the Motorola and GE voter systems have a theory section in the front, ahead of the schematics. The manuals on the two vintages of GE voters is available on the LBI master index page at this web site. Look for LBI-30002 for the older, grey-painted unit (all discrete components) and LBI-38676 for the newer black painted unit (with a mix of discretes and ICs). It's worth grabbing either one for the theory part. You will want to grab LBI-31981 for the idle marker encoder (i.e. at the outlying receive site). The GE voting panels operate on 120v/240vAC, or 24/28v DC. If you will be powering it from 12v you need the DC power converter described in LBI-4583. Another voter manual is on this web site that covers the LDG Corp RVS-8 Voter (it's from the April 1996 issue of QST). It's one of the more common noise voters. Read the writeup more for the technical overview than for the sales pitch.

Voting systems:

Positives:
1)Improved coverage and customizable coverage : This is the big selling point in the public safety marketplace. Not only can you can position receivers where the users are, but also you can provide hand-held coverage in places that only mobiles worked before. If you have a "black hole" building where nothing works, put a receiver inside the building and wireline it back to the voter. With a multiple-receiver system you can have a coverage situation where a handheld can provide communications quality that would normally require a 60w or 110w mobile, and if you have police officers or sheriffs deputies that have only handhelds (a popular configuration), perhaps on foot on a beat, on a bicycle or on , horseback that's important.

If you have a great transmit site (like on a TV tower) in a high RF environment that totally numbs your receiver you can put the repeater transmitter there and put the receivers elsewhere. Here in Los Angeles there is a 6 meter repeater system that has the system transmitter on the same 5,500 foot mountaintop as the Channel 2 TV transmitter... none of the 6m receivers are nearer than 20 miles away. The voting panel and the main repeater controller are at ground level in the system owner's garage - and that makes it easy to work on it.

You can add as many main input receivers as you have voting panel ports... and if need be you can cascade voting panels: the Los Angeles County Sheriff's old 39 MHz dispatch system used a 12-channel voter (implemented with the option kit that tied two GE 6-channel shelves together) with each voter channel fed by the output of another 12-channel voter. And this configuration was duplicated for each dispatch talkback radio channel ! I once counted over 60 six-channel voter chassis in one room of open frame 19" equipment racks, and had to stop counting when the tour moved on - there was at least a third more to count, and knowing how many radio channels and receive sites the Sheriffs Radio Center had then, I suspect that there was at least one more room of voting panels...

2)Redundancy: You can lose one receiver site and the system isn't completely dead. Yes, you will have a receive coverage hole (assuming non-overlapping receivers) but the rest still works, and those users that can reach another receive site (i.e. those with real mobile radios as opposed to handhelds) may not even notice the coverage hole... and many public safety systems have more overlap in their receive site coverage that you would mormally expect... Many Fire departments have satelite receivers at every fire station and at every hospital. You can also place standby (backup) transmitters at any of the receiver sites if you want (all it takes is money for the backhaul link, the transmitters and the duplexers...). Simulcast systems - those that use more than one system transmitter sending the same audio signal on the same (repeater output) frequency simultaneously is another can of worms best dealt with in another article.

Negatives:
1)Expense: rent on the outlying receiver sites - both building rent and additional antenna space (many sites include one antenna location with the rack space rental and charge extra on a per-antenna basis). The voting receiver site needs a omni antenna for the system input receiver (with one exception covered later) and a beam for the link transmitter that talks back to the central voter site. Sometimes you can get a feed from someone else's omni receiver antenna. If you plan on ever adding a voter to your system (at any point in the future) you might want to juggle the wording of your site rental agreements as far in advance as you can to allow you to add additional antennas later on without an increase in rent. Or you can do as we did at one site: we mounted a 3-foot (1 meter) piece of thick-wall galvanized steel pipe on the tower where the single antenna would have gone, then mounted a Stationmaster onto the top of the pipe and at the bottom of the pipe we side-mounted an end-mount 4-element 420 MHz beam. Despite the two feedlines this met the site agreement terms for "one antenna".

The exception mentioned above is where the voter receiver site is located near the periphery of the transmitter coverage with a 120 degree (or 180 degree) coverage antenna pointed inwards. The directional coverage can be configured easily by side mounting an omni antenna a carefully selected distance from the tower. See the "Antenna Systems" page at this site for articles on that topic.

2)Expense: Multiple sets of duplicate equipment are generally required. In order for the system to be "audio transparent" every voting receiver site needs to have the exact same audio characteristics - and for two reasons: one, the voter does it's voting totally on audio characteristics and quality - if the sites sound different that will negatively affect the voting process. Second is that it affects clarity and intelligibility - if the voter were to select a different receiver in the middle of a word, or even change receivers multiple times in the middle of a multi-syllable word (which happens much more often that you'd expect) and the audio from the first receiver was normal, the second receiver was bassy and the third receiver was tinny it would be very hard, if not impossible, to understand. The simplest way to guarantee consistent audio quality is to have identical equipment at each site. Having identical equipment also makes for simplified repair - just swap a dead chassis with a good spare. Or swap a complete cabinet. If you plan on adding a voter to your system in the future you may want to start to keep your eyes open for multiple identical sets of 420, 900 or 1200 MHz continuous duty equipment with direct FM exciters (no phase modulators) for the links and a number of identical main input channel receivers. Yes, there is a use for 406-420 MHz or 900 MHz radios! Make friends with the two-way techs in your area, you never know when some water or power utility company will be surplusing a dozen (or more) 900mhz link radios.

3)Expense: Point-to-point links... each satellite receiver needs a audio and COR link back to the voter, and that means multiple sets of link hardware needs to be acquired - be they wire-line, microwave, 420 MHz, 900 MHz, 1200 MHz or whatever. And the linking methods can be mixed as long as the audio characteristics are matched (which can be harder than it sounds). One local amateur system here in Los Angeles goes all three routes on it's voting receivers: two sites talk back to the voter via E&M channels on a private microwave system (with written permission). Several additional sites are brought in via one-way 420 MHz or 900 MHz point-to-point RF links. At the main repeater site there is the main repeater receiver, plus there is another receiver in another building at the same mountaintop site (think "diversity reception") and linked in by 200 feet (about 70 meters) of wire (they found an abandoned antenna and feedline on the other building's tower so they installed a receiver and ran their own DC wire-line between the two buildings) for a total of six. Yes, diversity makes a big difference and is relatively cheap to implement. If you were to listen very carefully you could tell which type of site is the last site voted - you'd hear a) the normal squelch crash then almost immediately a very, very short chirp when one of the microwave links squelch, or b) the normal squelch crash then a little bit of a second squelch tail when the 420 MHz RF link squelches or c) the normal squelch crash then absolutely nothing when the main site - either the in-cabinet receiver or the DC wire-line (diversity) receiver - squelches. And if the user is using a radio with reverse burst, the first squelch crash is completely missing.

Note that the links that carry the received signal back to the voting panel have to be noise free - if RF based they have to be absolutely full quieting ALL THE TIME, even during rainstorms or snowstorms (which can affect a 1200 MHz link pretty bad... and 900 MHz to a lesser extent. The water in the air absorbs RF much more readily than at 420 MHz). Any link noise will confuse the voting process. You could have a perfectly quieting signal at a satellite receiver, but if the link path is noisy the voter will do just what it is supposed to do and vote the best quieting signal it sees at that moment.

The satellite receive site, if using a RF link, can be almost as complex as the main repeater - it is, after all, a complete cross-band repeater, in some cases with both main channel and control receivers. And a constant 20db quieting on the RF link is a good start, usually you end up needing quite a bit better. One corporate microwave system that I am familiar with uses 40db quieting as the minimum accceptable link.

You can cheat a little on the equipment at the main site (the receive end of the links): The link receiver antenna is usually an omni followed by a preamp / multi-coupler feeding multiple receivers. If you have a good antenna with a band-pass "window filter" like those made by DCI and others the link receivers can be mobiles: I have seen stacks of Micors, Mitreks, MaxTracs, Radiuses and even 1970s vintage Motrac / Motran receivers used as 420 and 900 MHz link receivers. One system uses a Down East Microwave block converter that "translates" a set of 1200 MHz link channels down to the 150 MHz range feeding a stack of 1980s vintage high band E.F. Johnson mobiles as link receivers.... Another system uses the receive chassis from an old UHF-to-HF transverter to convert a block of the 438-439 MHz range to a number of (otherwise pretty useless) 36-42MHz band Motrac / Motran receivers (they had salvaged the receiver chassis and the control heads and junked the rest). They made it easy on themselves: to make the conversion just subtract 400 MHz... 439.925=39.925, 439.950=39.950, 439.975=39.975, etc. Equipment for the middle range of low band can usually be had for low prices or sometimes even for the taking: 30-36 MHz gear ends up on 10m, and 42-50 MHz gear ends up on 6m. Absolutely nobody (except volunteer fire departments) wants the 36-42 MHz stuff in between, and the VFDs want more recent (synthesized) stuff.

4)Complexity: The overall system design must preclude some failure modes from taking down the entire system. A failed link receiver (i.e. no audio) appears as a full quieting signal to the voter and can completely lock up the system unless you have an auto-disable timer on each receiver site. This is a built-in feature of the GE voting panel. The more complex the overall system the more you have to think ahead of Murphy's Law. And Murphy was an optimist.
Example 1: One group built their own repeater controller. They installed it on a hill and a year or so later the -12vDC supply died, which via an unexpected design quirk caused the repeater transmitter and all three point-to-point link transmitters to simultaneously key up - continuously. And the failure happened during the worst rainstorm in over 10 years. They had to hike in to the repeater site, over 12 miles on dirt roads (read: mud deep enough to suck your boots off) in sheeting cold rain and snow. By the time the transmitters were shut off it was about 30-36 hours later.
Example 2: A link receiver blew its fuse. The receiver COR was active high, and open collector. The pullup resistor was in the voting panel, which was still powered. The COR went high, the audio was dead quiet which wa interpreted as full quieting. Fortuately, the GE voting panel designers thought of that situation and the panel auto-"failed" the receiver after several minutes of perfectly quiet audio. You can also manually disable a channel (i.e. keep it from being voted) by grounding a pin on the connector.
Moral of the stories: don't forget to consider (and test) all the failure modes, such as a locked on link transmitter (one trick to keep a locked on link from taking down the system is to pass unmuted RX audio (i.e. non squelch gated audio) to the link TX audio input... think about it...
Still, have two ways to shut off the transmitters (both link and repeater).

5)Expense and Complexity:
A voter-based repeater has additional complexity and expense that increments with each satellite site that is added:

every receive site needs a main channel receiver, (and maybe a pass cavity), plus feedline and an antenna (usually omnidirectional)
every receive site needs a link channel transmitter (continuous duty), feedline and an antenna (usually directional)
every receive site needs a controller of some sort

Sometimes the above three can be combined. The RF-linked sites in one of the systems mentioned above have a remote "package" consisting of a system input receiver a link transmitter (on a 420-439 MHz frequency) and a system output receiver. Some folks will tell you to just strap the local input receivers COR to the link PTT and let it go, but there are several good reasons to use a low-end controller (homebrew or commercial) on your cross-band links: timeout, ID, remote control, and telemetry. Each outlying receiver site is by definition a cross-band repeater that has to follow FCC rules as well as good engineering practice guidelines.

The local controller in the above mentioned system is a very simple PIC processor based board that couples the main input receiver to the link transmitter, and has a touchtone decoder that listens to the main output channel receiver. The PIC controller is built on a plug-in card, for easy replacement, and the site number is set by a DIP switch. This allows the control sequences for the remote sites to be set up so the sequences do not overlap - the site number is part of each control sequence, and the site number is included in the link ID (i.e. w6xxx/n where the "n" is the site number). The local controller listens to the system output so that if you can't be heard by the site you are trying to control you can still control it as long as you can get into any receiver of the system. Picture a situation where you need to shut down the north voting site and you can only be heard by the south site.

The controller code has the following commands:

Site master off - this is not just a TX PTT disable, it actually powers down everything but the main output receiver and the controller.

Site normal - master on and select main channel input receiver to 420mhz link transmitter (default mode)

Force site ID and send site number (after bumping the IDer tone out of the audio notch for this transmission only) - basically an "is that site there and working?" test

Lock TX on with main input audio (same as default but locks on the link TX). If there is no carrier on the main input it will blow squelch noise down the link since there is no squelch muting

Lock link transmitter on with no audio (i.e. emulate a dead carrier on the receiver - used for checking path noise). The "no audio" state is accomplshed with the transmitter deviation pot wiper grounded by the normally open contacts of a reed relay

Tone generate mode: there are selectable test tones at several frequencies and each can be sent at different levels... 300Hz, 600Hz, 1khz, 3KHz, 4Khz, and at 1Khz deviation, 3Khz deviation and 4Khz deviation)

Site suicide - This command drops power from everything - just as if a little hand had come out of the cabinet and it unplugged the power plug from the AC wall outlet. As you would expect, it requires a site visit to bring it back to life. This command was added after a parasitic oscillation in a receiver preamp trashed the main input receiver of a commercial system at one of the sites.

Note:
All of the "lock TX" modes and the test tone mode reset to default mode after 20 minutes - just in case.

Each voting receiver assembly was implemented by taking a GE Mastr-II "E" case (the double-high case) 420 MHz radio and replacing the UHF 420 MHz receiver with a main channel input receiver and adding a main channel output receiver into the second tier. The leftover space inside was occupied by the controller hardware. It's a nice package: one radio chassis, one Astron power supply, a short Stationmaster mounted about 8 feet above an end-mounted 6 element 420 MHz beam. Most of the receiver sites involved with this system are at low profile / low RF level residential sites.

6) Complexity:
The central site (where the voting panel is located) becomes more complex. You need a link receiver and the associated equipment for each link channel. Usually an omnidirectional link antenna, a DCI filter, and a multicoupler with a port for each link receiver is sufficient to get you started.

You need additional control facilities. The main system controller (the one at the site where the voting panel is located) needs additional digital outputs, and some inputs that can be sensed. To remotely control a 6-chanel GE voter you need 6 lines that go to ground (each one will be used to fail a receiver), and some kind of interlock logic so that a brain-faded control operator at 3am does not disable ALL of them at the same time (sometime ask me how I remembered to mention that....) locking everyone out completely until someone goes out to the voter site to re-enable them.
You will also need 6 more lines that go to ground, with an interlock to only select one at a time (so you can force select a single receiver), with an interlock so you can't force select a locked out (i.e. dead) receiver. To sense a controller-auto-failed channel you have six inputs that you can read, plus six more that indicate the currently selected channel.
Very few controllers have enough digital outputs to be able to dedicate six (or eight or even twelve) to controlling a voter, and fewer have enough smarts in their "programming language" to prevent all of the disable functions from being turned on at the same time.
Another "it would be nice" feature that few controllers have is enough digital inputs to be able to sense which receiver is selected (again, anywhere from 6 to 12 depending on the manufacturer of the voting panel) plus a way to "totalize" the amount of time the system spends on each receiver. If yours can, then it's worth doing the programming for that as the statistical data that is gathered helps the system tech committee determine which receiver is used the most, and helps spot failure trends... if, for example, all of a sudden the prime receiver is voted 50% less it may be time for a site visit to check the receiver sensitivity.

Do not think that your situation is such that you will not have to control the voter. As one example, I know of one voting system in the Midwest that disables their northeast receiver during summertime band openings as users from the nearby same channel system are heard by the receiver just enough to key it up but to not be understandable. For 9 to 10 months of the year those users are not a concern, just during troposphere ducting periods.

7) Complexity:
Carrier Squelch systems are much simpler than PL systems, and if you can start there, are much easier to implement, and with enough forethought you can implement them so that PL can be added easily in the future. The considerations include the technique employed to switch the entire system from carrier to PL and back, or do you run a single-mode system? (If you have three remote receivers plus the main receiver that's four sites where you may need to change the mode. It all depends on how you design the system. And the last thing you want to do is to cascade PL decoders.

Design questions:
On a PL system there are several ways to set it up - each of which offers some options and deletes others.
Do you put a PL decoder at the voting RX site?
Disadvantages: lots of PL decoders and identical reeds required. If you use PL'd links you have to filter out and re-encode the link PL, which creates a situation called "cascaded PL decoders" - this dramatically raises the system response time which can be a big deal with a mobile or portable swishing in and out of one or more sites, and also effectively locks out users that are answering a question with a quick response i.e. "yes", "no", "roger", etc. - they won't be heard.

Do you put a PL decoder on the output of each link RX ? (it may already be there if you use a receiver that has a factory PL decoder. The receiver may also provide a PL-filtered repeat audio output, which makes the connection to the voting panel simple)

Disadvantages: Again, lots of PL decoders and identical reeds required, plus you have to make sure that the entire path from the antenna jack of the remote receiver through the local controller then through the link transmitter then through the link receiver can pass the user's PL tone itself without any measurable distortion ! Hint: a distorted PL tone results in slow PL decode action (if it decodes at all).
Hint: multiple cascaded PL decoders cause unacceptable pickup delays - figure a worst-case scenario of 1/2 to 2/3 of a second per decoder - do you really want your users to have to squeeze the PTT and wait a full second to a second and a half before they can talk? Unless you are desperate never put more than one PL decoder between the user and the system transmitter.
Once you have an individual decoder behind each link receiver you can use the output line to enable that voting channel. You will also OR all the decoders, and the OR'ed result of all of the link receiver PL decoders becomes the repeater controller PL decode input.
...or do you put one on the output of the voting panel?
Only one PL decoder and reed is required here, but you have to make sure that the entire path is distortion free!... plus you have to avoid three major "gotcha"s: ...
the voting panel has to pass the PL tone without distortion

The voting process in the panel must not affected by the PL tone. Since valley voters need to see a true valley they a high pass audio filter in front of each RX port to remove the PL tone.
The phase relationship of all the receivers's audio, as presented to the CTCSS decoder, must be the same. The users PL encoder is the source and as such each link receiver will deliver it to the voting panel in the same timing/phase, but if there is one extra phase inversion anywhere in the path from remote receiver to the voter then you will find that when the voting panel switches from any of the several in-phase receivers to the out-of-phase receiver the tone decoder interprets the instantaneous phase changes as a reverse burst and will quench the tone decoder... in other words it would look like a reverse-burst and slam the squelch closed. This can be impossible to find if you don't know what to look for. If you put a simple audio transformer in the path between each RX and the voter input you can fix an out-of-phase condition by swapping leads.

If you are constructing PC boards to simplify your multi-link-RX assembly, then a simple layout trick with jumper blocks can be used to fix this problem: two jumpers horizontally select normal phase, two jumpers vertically select reverse phase.
------------)(----------O O---------
audio from  )(          O O
link RX     )(          ! !       Audio to voting panel
------------)(----------!-+
      audio transfomer  !
                        +-----------
No matter what you do, you will find that the level of the PL encoder in the average user's radio fresh out of the box is not consistent from radio to radio, and many of the Alinco/Yaesu/Icom/Kenwood radios have excessive PL deviation right out of the box...
One trick on link PL tones... if you absolutely have to use a PL tone on the link then consider notched audible or "supersonic" PL... the decode time is much much faster on audible tones than sub-audible tones...
Notched audible: The regular Motorola wire-line AC remote control tone panel uses 2175hz as a PTT marker tone (tone on during PTT and notched out of the transmitter audio). You could use that same frequency as a "PL" tone, with a notch filter at the voting panel input - it would have to be at the input because the tone would confuse the voting function. This same idea is used by microwave E & M systems - the M lead switches an audio tone that is decoded at the far end as the E lead, and the tone is notched out of the link receiver by a deep narrow notch and never heard by the end users. That tone notch technique can also used to ID the links in multi-hop linked systems - one common ID frequency is 1064Hz, right in the the gap between low group and high group touchtone... (the IDer tone thus has a minimal effect on any downstream touchtone decoders)

Supersonic PL: In normal usage the term supersonic refers to a sound frequency above the range of human hearing - 30KHz to 35KHz for example. In 2-way usage this term refers to above-the-audible, but in this case that's limited to 4 to 5khz. If you look at the audio performance of the average FM receiver you will see that audio frequencies above a certain cutoff point are treated as squelch noise. You want to pick a frequency just a hair below that point (so there is no attenuation of it) and use it as a PTT indicator, and then notch it out before the voter - you have to do that since the supersonic tone will be right smack in the middle of the squelch noise range that the voter is trying to use to vote with... and you want a narrow enough notch that the voter still gets enough noise to work properly.

8) Complexity: Fine Tuning of the system: The audio levels at the voting panel between receivers are very critical as are the levels between the audio and the idle status tone (if used). Everything needs to be exactly the same for the comparator to properly vote the best reciever signal. Make sure the audio response of all reciever paths are adjusted the same, using all the same remote recievers and all the same link transmitters is the easiest/simplest way. The audio at the voting panel has to be exactly the same.

9) One thing that hasn't been covered yet is the budget: Who is going to fund:

The voting panel
The rent for the additional receiver site(s)
The additional input channel receivers
The link transmitters
The site controllers
The site antennas (user input and the link beam) and feedlines
The fuel for the repeater committe members to drive to the outlying sites

As far as the voiting panel, you can pick up a 6-channel GE voting shelf on eBay today (mid-2005) for about US$100 dollars or less and you can find auctions where the seller will pay the shipping costs. Just make sure that the left most 6 slots have receive cards, the next slot has the common audio card (it has the monitor speaker volume control) and the rightmost slot is the speaker / power supply module. The later panel was painted black. I've seen auctions where the card slot were filled with random cards and some a mix of grey and black. It takes the right manual, a logical brain, some good test equipment, some serious integration and adjustment time (and think carefully - what is your time worth?), and a few interface tricks to get a voter to play well. And the early grey ones are easier to fix for some folks - they used all discrete components where the later black ones used a mix of ICs and discretes. As mentioned above you can download the manuals on either of the GE panels from this web site.

Motorola Spectra TAC voters go for similar prices on eBay, but manuals have to be acquired, and many are no longer available.

Closing Notes:

Remember that acquiring the voting panel is only the start - you need to acquire the multiple sets of RF point-to-point link hardware (and coordinate each RF channel) for each outlying site back to the voting panel site. Taking a single-site system and adding a voting panel and maybe two outlying receivers has just raised the hardware complexity (and maintenance requirements) from the original one repeater to more than three separate repeaters. The odds on being able to access someone elses spare E&M ports on any already-in-place microwave or fiber system is HIGHLY unlikely. And you will find that the linking equipment costs a LOT more time and money than the voting shelf - and then there is the ongoing monthly site rental and increased maintenance costs (and don't minimize that aspect - a system that has a main site and multiple outlying receive sites has a lot more exposure to the elements than a single site system).

And if you are going to deploy N satellite receiver sites you will need N+2 or N+3 sets of identical linking equipment (the +1 is to let the magic smoke out of during R&D, and the +2 and +3 is for hot spares)...

Each outlying receiver site is a repeater (usually cross-band) by itself (unless you luck out and have access to in-place E&M links), and the main site is now an extra-complex multi-port repeater (local main channel receiver plus all the link receivers plus the voting panel). It may take more time than you think (or even more) to rebuild the original repeater to place the voting chassis in line and get the local receiver talking through the panel (i.e. as a single channel voting system), then more time to get the first link receiver working, then sorting out all the adjustments... and simultaneously the other team of club members is getting the first outlying receiver site operational and talking back to the main site. Then you get to do a major chunk of that work over and over again for each new voting receiver site. Can your repeater committee handle that workload? Do you have the technical competence in your group to keep such a system running if the key person suffers repeater project burnout? or has a heart attack? or gets broadsided by a drunk at 3am?

As far as I know the most common method is to run the remote receive sites in carrier squelch and passing 67hz to 4.5khz flat to the link transmitter. Just take discriminator audio and pipe it flat through a limiting audio amp (set to keep the link deviation under 4.5khz) right to the transmitter modulator. No audio muting is needed. At the main site they either put the user PL tone decoder behind the voting panel or they use separate decoders per link receiver.

At the main site you need a stack of link receivers, each of which has a link ID notch filter thats a few cycles wide and deep enough to take out the ID. Be very careful on this filter's design, very narrow filters tend to ring and that does weird harsh things to the users audio.

If you use the system design that runs the RX sites carrier squelch, and passes the user PL tone through the voting panel, with one PL decoder it, you can make use the PL decoder in the link RX (if there is one) in an interesting way... just wire it to the "force select" line on the voter, and diode-OR it to the PL decode input of the controller. Then plug a different PL decode reed into each link RX. This implements one common tone for the entire system (the decoder after the voter) and one unique tone per site (using the link receiver decoder). The users program their radios for the common tone, and the tech crew can do quick go/no go coverage checks and site sensitivity checks by dialing up each site's unique tone.

If you run a carrier squelch system then you can implement the same trick by connecting the link receiver PL decoder output to the "force select" line of the voter channel.

One suggestion: Find somebody who has set up and runs a voting system (public safety, commercial or amateur) on a daily basis, buy him dinner and a few beers and get him talking. You will learn a lot - maybe more than you wanted to... For example, on voting systems that are used on trunked networks the rf links pretty much need to be up all the time as the handshaking between a mobile and the trunking controller is less than 200ms which would exceed the key-up time of an RF link. After all of the technical and financial concerns are discussed you may even decide to not implement a voting system, and to improve the single site system you already have. It might be easier to add a pass cavity and a good preamp ahead of the main receiver, or to convince the users to use 30w to 50w mobiles (maybe even with dual band mobiles that crossband repeat their dual band handhelds) into the improved existing system than to contribute to the implementation of a multi-site voting system so they can use flea-powered handhelds.

If you do go with a ground up new project, you have to toss around the choice of a true signal to noise (valley) voter (GE or Motorola) versus the rectified high frequency noise voters that tend to be more popular in the amateur world. Both work fine for many applications but the true signal to noise type work much, much better, especially over point-to-point radio or microwave links (and the interfacing effort is pretty much the same). Personally, I'd go with a surplus GE, just from the manual availablity aspect, then Motorola.

Another coverage improvement method is to multicast using a hub repeater. Take N number of 2m (or 440MHz) repeaters, each with it's own coverage area and user base, and add a 220MHz, 420Mhz, 900MHz or 1200MHz radio to each one that talks back to a central or hub repeater (i.e. a one-channel remote base on a beam antenna). The only "users" of the hub repeater are the outlying 2m (or 440MHz) repeaters. Each individual 2m (or 440MHz) repeater has three modes in the remote base: local only (i..e link radio off), hub monitor (link radio in receive) and hub active (link radio in transcieve). In local mode the repeater operates normally. In "hub monitor" mode the remote base receiver is active and the local repeater repeats the local traffic and all of the hub repeater traffic. In "hub transcieve" mode the remote base receiver is active as is the remote base transmitter and the local local traffic is repeated as well as all of the hub repeater traffic.
The N3KZ/R system is a good example of this technique - they have over a dozen UHF systems talking back to a hub system.
Hint: do a Google search on N3KZ.

Another method is the one that the Cactus system has used for over thirty years is called "backbone linking". Take a number of local repeaters, each on their own user channel, and add a point to point link to the next site down the road. For example, let's say that you have four sites, called A, B, C and D. Each has a UHF repeater, and there is a link from A to B, another from B to C, and a third from C to D. Each site has a multiport controller: A and D have two-port, and both B and C have a 3 port (assuming that none of the sites have remote bases, or other equipment that needs a port). The controllers are programmed that the link ports are in repeat mode... a carrier coming in to Bs link port from site A causes the link from C to key up and talk to D. Likewise a signal coming in Cs link port from site D causes the link from B to key up and talk to A. Each controller is programmed to treat the entire link system like a single remote base... the local UHF repeater can be disconnected from the collective links, monitor the links or transcieve to the links. So let's say that all are set to transcieve. A carrier on the receiver of the Site A user port the site A repeater to comeup while simultaneously the A-B link keys up, the B-C link keys up, and the C-D link keys up. Each sites local UHF repeater now relays As audio. When A unkeys the other repeaters drop out as well. If a local system wants to drop off link it can by simply disconnecting from it's links, but the link side of the controller continues to relay... i.e. if the local system at site B drops off the links at site B continute to connect A to C. This system model was invented in the 1960s and works very, very well. You can look at the map of the Cactus system here, and remember each dot is a system, and each solid line is a point-to-point RF link, so to determine how many radio ports are on each site controller you need to count the lines entering each dot and mentally add one for the local user repeater, and one more for the local remote base. Yes, the backbone linking method ties up more spectrum, but at that time the thinking was "use it or lose it", and spectrum between 420 MHz and 430 MHz, from 902 MHz to 928 MHz and in the 1296 MHz band wasn't being used by the average ham.
The same model works today and is in use on systems other than Cactus: The Condor Connection is a 220MHz system that goes north from Los Angeles all the way to San Francisco, the CMRA system is a 1200Mhz system that covers a very large portion of California (but the website is stale and has not been updated in a long time), and the WinSystem has over 50 repeaters online fulltime on UHF, 220Mhz and 2m in California plus a fulltime IRLP reflector link (and their web site has a streaming audio function that you can listen in to from anywhere). The WinSystem "insomniac net" is at 11pm Pacific time every night and there are frequent checkins from Australia, Japan and Europe. If you have an IRLP node near you just connect to the reflector at around 10:45pm Pacific time and join in.

With either the hub repeater method or the back bone linking method the users do need to keep track of which local repeaters coverage area they are in, and change channels in their radio to compensate for their travel between local system coverage areas. Some manufacturers, including Tait, make mobile radios designed for this type of operation and when properly configured will vote the best frequency in the multicast group - in other words the user turns the mobile on and it does all the work.

Mike WA6ILQ

This page created 22-Apr-2002

Back to the top of the page
Up one level
Back to Home

This web page, this web site, the information presented in and on its pages and in these modifications and conversions is © Copyrighted 1995 and (date of last update) by Kevin Custer W3KKC and multiple originating authors. All Rights Reserved, including that of paper and web publication elsewhere.