Simplification!

If you have been following this project, you will know that the original plan was to take the received signal, transport it to Farnsworth Peak as a converted IF for processing, and then transport an IF back to Scotts, modulated with the voted receive audio and upconvert that to the transmit frequency.  For information on this original scheme, visit the this page.

As it turns out, several of the modules to do this function were constructed and tested - but then, for the designer(s) and builders, "life" intervened and stalled the project for several years.

A similar project - and its results:

In early 2005, an unrelated project was undertaken to assemble a UHF synchronous/voting system.  This system used a different approach for some of the key components:

• Twisted pair/leased 4-wire lines were used to provide audio linking.
• A surplus commercial GE receiver voting shelf used.
• Unmodified GE Mastr II receivers were used at each of the 4 receive sites.
• Minimally-modified GE Mastr II transmitters were used at each of the 3 transmit sites.
• GPS frequency references were used to provide transmitter frequency control.

The use of GPS frequency control:

There are several reasons why a GE Mastr II transmitter was used over a more modern transmitter:

• They are readily available and inexpensive.  Although these transmitters date from the late 70's and early 80's, they were implemented in large numbers.
• They have well-known reliability characteristics.  These transmitters, having been around for such a long time, have well-known track records.  Knowing how well these units hold up - and knowing what sorts of problems that they are likely to have - allows them to be refurbished and modified to maximize reliability.
• They use Phase Modulation as opposed to having a frequency-modulated oscillator.

This last point is the most important when it comes to having a synchronous transmitter network.  All one needs to do is to lock each unmodulatedcrystal oscillator to a common reference and all of your transmitters will be on-frequency.  While it is possible to lock a frequency-modulated oscillator to a given reference, this causes several problems that complicate the synchronous system and can potentially worsen performance in overlap areas.

When you use a frequency-locked oscillator and provide "true FM" (that is, 6 dB/octave pre-emphasis needs to be applied to the audio feeding it) you must design your PLL such that the "knee" of its loop's time constant is below that of the lowest modulated frequency.  Typically, a slight compromise is made in that the loop time constants are somewhat "fudged" at low audio frequencies (around 300 Hz) with the valid assumption that a slight departure from the pre-emphasis curve at low audio frequencies won't make any noticeable difference in perceived audio quality and intelligibility.  As it turns out, this means that subaudible tones generally fall below (or in the middle of) this "fudged" area.

Having a relatively "slow" PLL tracking response means that the carrier frequency actually "wobbles" slightly with modulation.  This effect can be heard using a multimode radio and tuning in an FM signal using an SSB receiver:  Some transmitters can be heard to have their carrier frequency wander about in response to voice modulation.  Some transmitters do this very noticeably, while this effect isn't obvious at all on others.  Exactly how much this occurs depends largely on the precise time constants in the PLL circuit and how they respond to the spectral content of the modulating audio.

A "straight PM" (or phase-modulated) radio - like the GE Mastr II - does not modulate its crystal frequency at all.  All modulation is imposed, in later stages, on a fixed, unchanging (except for thermal drift, of course) carrier frequency.

It is, of course, much easier to lock an unmodulated carrier to a common reference than a modulated one!  Now, it is certainly possible to satisfactorily use a "true FM" modulated oscillator for a successful synchronized transmitter system, but in the interest of simplicity, we simply chose a phase-modulated transmitter.

Locking to GPS:

As it turns out, the "fundamental" frequency of the GE transmitter is around 12 MHz (being 1/12th of the VHF transmit frequency, or 1/36th of the UHF transmit frequency.)  It is relatively easy to lock this frequency to a common reference.  In the case of our system, we obtained some surplus GPS receivers that provided a disciplined 10 MHz output and locked our 12 MHz crystal frequency to that 10 MHz reference.

Using GPS frequency references, it was observed that the phase of the UHF transmitters would, over the course of an hour, move about 45 degrees with respect to each other under average conditions.  This frequency stability is far greater than was required as being within 1-2 Hz would have been adequate.

The voting receiver system:

A standard GE voting receiver shelf was obtained surplus, along with a "pilot tone board" for each of the four receivers.  In the receiver, the pilot tone is activated when the radio's squelch is closed to indicate that this particular receiver is inactive.

The GE voter shelf, if it detects a tone, ignores that receiver.  When the tone is gone, however, if will look at each of the currently active receivers to determine which one has the "lowest" audio level.  The theory is that a signal that is noisy will have not only the audio of the original audio, but it will, overall, be somewhat "louder" owing to the added "hiss" of the weak signal.  This scheme works quite well, provided that the system has been carefully adjusted so that each receiver provides the same amount of audio for a given amount of modulation.

The failure of one of the receivers or its connecting audio line needs to be taken into account as a line/receiver failure would cause the tone to disappear and cause very quiet audio.  The voting controller will, if this condition persists for more than 30 seconds, declare that this receiver has failed and remove it from the polling.

Observed performance of the radio system:

As previously mentioned, these three transmitters have very significant overlap in parts of their coverage area.  With proper care and attention to detail, the effects of this overlap are similar to those of slight multipath.  With each transmitter being precisely on-frequency, the "capture effect" of FM detection can be used to maximum advantage, with the two signals having to be within 2 dB or so of each other before degradation can occur - and that's only under worst out-of-phase conditions.  In the real world, it is very difficult to hold a handie-talkie in a location to cause such precise match of signals - and the deleterious effects of phase cancellation of the RF carrier are relatively unlikely to occur anyway.

Audio phase matching and delay:

The audio channel for two of the three transmitters were fed via a leased 4-wire line.  This means two things:

• Additional delay caused by the 4-wire circuit
• The inability to transmit a subaudible tone over the 4-wire circuit.

The transmission of the subaudible tone was handled by amplitude-modulation of the tone on a 2.5 KHz pilot carrier on the transmitted audio link.  At the transmitter site, this pilot carrier was removed in the transmitted audio and the subaudible tone modulated upon it was synchronously demodulated and used to produce the subaudible tone to be transmitted.  The subaudible tone phasing, on the other hand, was very critical:  Designed into the system was a means to adjust the phase of this tone and it was determined that keeping the phase of the subaudible tone of each transmitter within 15 degrees of each other (and the amount of deviation within 10%) was sufficient to eliminate a destructive "buzz" that would otherwise occur.

Application of this to the '62 synchronous system:

Having had practical experience with the above project has allowed us to confidently rescale the '62 synchronous system:

• The Scotts Hill site will use a GE Mastr II transmitter.  Only minimal modification is required to convert it to "synchronous" operation at a later time.
• The Farnsworth Peak exciter will be replaced with a GE Mastr II exciter:  The existing VoCom power amplifier will be kept.
• The 33 cm microwave link between the two sites will be used.
• The existing 146.02 receiver on Farnsworth (located at a remote site, connected by wire) will be retained.  This receiver already has a PWM-encoded "squelch-noise" signal encoded with it and this signal will be used to regenerate the "squelch noise" signal and can be used to provide direct indication of signal quality.
• A GE Mastr II receiver will send wideband audio (which will include ultrasonic "squelch noise") over the 33cm microwave link from Scotts to Farnsworth.  At Farnsworth, a "squelch-noise" signal will be derived from the Scotts Hill receiver's audio and the comparison of this signal with the "local" Farnsworth squelch information will allow voting.  This means, of course, that a homebrew "Squelch/Voting" comparator system will have to be constructed.
• A 100 KHz pilot carrier will be sent via the 33cm microwave link.  This will be used in lieu of a GPS frequency reference:  While the accuracy of the transmitted frequency won't be as accurate as that of a GPS reference, they will track each other precisely!
• Because the '62 repeater does not transmit a subaudible tone, we won't have to worry about it!