Thursday, September 13, 2012

Two repeaters, One frequency (Part 3)

For a follow-on article in this series, see Part Four (link)  for a discussion of how the voting receiver system works.

In parts One and Two the general overview of a "synchronous" (or "simulcasting") and voting repeater system was discussed.  In a nutshell:
  • Both repeaters operate on the same frequency saving spectrum and simplifying the system's use since the user doesn't have to remember which particular frequency of a "normal" linked system covers a certain area best.
  • The coverage of the two repeaters overlaps to a degree.
  • Because of precise frequency control, the two transmitters don't really clobber each other in overlap areas, particularly in a moving vehicle.
  • Because of voting receivers and multiple transmitters, the users can seamlessly move between coverage areas with no intervention on their part.
  • The total coverage is greater than the sum of the parts owing to the increased likelihood of one or another site hearing the user and/or being heard - particularly if in an area where coverage is spotty to one or both sites.
 Originally (back in the late 90's) the idea was to frequency-convert the received signals from the 2-meter frequency to a subcarrier-baseband and send them to the main site where they could be voted upon and then a master modulator would then ship back (via a microwave link) a subcarrier which was then up-converted to the transmitter frequency.

The details were worked out and some of the equipment was actually built and tested - and it worked!  However, the magnitude of the task bogged things down and one thing led to another and the project languished - until 2009.

By then I'd already put together 2 voting systems and one multi-transmitter synchronous system (using GPS frequency references) and had other ideas on how to do things a bit more simply which translated to "being more likely to get completed!" The project got underway in earnest in mid-July of 2009 where the plans were re-draw and tasks divided as appropriate.

Instead of building the transmit and receive gear from the ground up it was, instead, decided to modify off-the-shelf GE MastrII radio gear to fit the bill.  This equipment is readily available on the surplus market and the individual pieces could be used with little or no modification - which meant that spares of those same pieces (receiver, transmitter, power amplifier, etc.) could be kept on hand as spares!  What's more, for the most part these units used common, off-the-shelf parts (resistors, capacitors, transistors) and were thus field-repairable now and for the foreseeable future.  Finally, a lot of information is available on these radios on the web so if, in the future some trouble shooting is required, there's plenty of advice to be had online.

What modifications were required to the radios were fairly simple:
  • Instead of a standard crystal module (called an "ICOM" by GE) a simple, plug-in module (using a "gutted" ICOM) was plugged into the exciter instead.  This was connected via coax to an external module that provided the low frequency (at 1/12th of the transmit frequency) at the precise frequency.
  • Transmit audio was fed into the subaudible tone input port.  This was done because it did not have the highpass and lowpass filters that the normal microphone inputs had:  We would do the high/low pass filtering externally!
  • The receiver modification (for the Scott's Hill site) simply involved obtaining discriminator audio.
There were some additional modifications done to provide interfacing to the rest of the system - namely an outboard de-emphasis, a low-pass filter and a switchable notch filter (for a "quirk" we later discovered) but these were mounted on the backplane - a more-or-less passive board that would likely never require replacement!  Pretty much everything else was "stock" and could be tuned up and adjusted according to the original manuals!

Transmit frequency control:

The most important aspect of a multi-transmitter (simulcasting) repeater system is that the transmitters be where they are supposed to be, frequency-wise!  While there are several ways of doing this, we took a somewhat unique approach.

A standard transmit crystal (at 1/12th of the VHF transmit frequency) was ordered and placed into an "EC" type ICOM.  This is, in effect, a self-contained oscillator module that has provisions to be frequency-controlled with an external voltage.  This module is completely standard and off-the-shelf and it could be plugged into any GE MastrII VHF transmitter and work normally.

In our case, however, this "EC" module is plugged an external module - called a "Disciplined Oscillator" - that takes the crystal frequency (which is 12.2183333 MHz for a 146.620 MHz transmit frequency) and locks it to a reference based on a 10.0 MHz oven-controlled crystal oscillator.  This is done by synthesizing an audio frequency, using a PIC microcontroller clocked to the 10 MHz oscillator, that has a resolution of a few parts per billion and with a bit of dividing, mixing and comparison has the result of locking the 12.2183333 MHz oscillator to the 10 MHz reference to within a tiny fraction of a Hz.  Essentially, the frequency accuracy is that of the 10 MHz oscillator!

The 10 MHz oscillator is an oven-controlled crystal oscillator (OCXO) pulled from scrapped satellite gear and is well-aged (made in about 1990) and has a stability of about 10E-8 - within 1 Hz or so at the 2 meter transmit frequency.  This OCXO also has an external voltage control tuning line that is under control of the PIC microcontroller and with it, the 10 MHz frequency (and thus the transmit frequency) can be tweaked to set each transmitter on the desired frequency - which also means that the frequency difference between the two transmitters may be precisely controlled.  In the nearly 3 years since the system was made operational we've observed that the transmitters have stayed within about 1 Hz of their intended frequencies relative to each other over the course of the temperature excursions during the year!

This "Disciplined Oscillator" module also has another function that, since it was computer-based, was easy to implement, and that's as a simple dual cross-band repeater.  On Scott's (the remote site) it simply cross-bands the 2 meter receiver to the 70cm link transmitter - taking care of thinks like proper IDing, timeout timers, etc. and it also takes the 70cm link receiver and controls the 2 meter transmitter coming back the other direction:  Both operate independently of each other...

Squelch control and voting:

It also does one more thing:  COS (Carrier Operated Squelch) signalling.  The 146.620 repeater is one of the few repeaters in the area that does not have a subaudible tone requirement, this being because it's an "open" repeater and that extreme care is taken at all receiver sites to keep the receive frequency as clean as possible - a task that is arguably easier since the demise of analog television in the U.S.!

Since the Scott's Hill transmissions are relay to/from the master site via a UHF link there would be an extra squelch tail (the "ker" in "ker-chunk") if the loss of a signal at Scott's were signalled simply by its UHF transmitter being keyed/unkeyed.  Instead, the loss of squelch is signalled by the appearance of a strong, 3.2 kHz tone sent over the link which performs two functions:
  • It signals to a decoder at the master site that the squelch as closed at the other end.
  • It signals to the voting controller at the master site that the signal being received is "bad" and should NOT be used.
(This 3.2 kHz tone is "notched" out and its brief appearances in the system audio are not heard by the users.)

So, what happens if a user's signal into Scott's is dropping rapidly in and out?  As the squelch opens and closes, the tone is turned off and on (tone on = squelch closed/signal dropout.)  When the tone is turned on the voter disqualifies this tone, but if that same user is getting into the other receiver (at the master site) then this tone will guarantee that the signal from that receiver will be used, instead.

There is a short "hang time" on the link transmitter which means that when an input signal disappears from the 2 meter receiver at Scott's, the tone will turn on instantly, making the master site ignore the input, and then the UHF link transmitter signal will drop and in this way, the extra squelch tail from the UHF link transmitter dropping is never heard by the user.

As it happens, the signals coming the other way (from the master site to Scott's over the UHF link to be retransmitted on 2 meters) also use this 3.2 kHz tone - this time, to control the Scott's VHF transmitter.  In this case, however, the activation of the tone starts the "Unkey" sequence at Scott's allowing time for the disciplined oscillator to put an extra "beep" on the transmitter (so that users know which transmitter they are hearing!) and then unkey the VHF transmitter.

Since the 3.2 kHz tone being sent to Scott's occurs just before the master site's 2 meter transmitter unkeys, it's possible to set the timing so that both sites unkey at precisely the same time:  If both site's didn't unkey simultaneously, many users would be annoyed by the presence of an extra squelch tail if they could, in fact, hear the "other" transmitter hanging in there for a short time!

At the master site:

As it turns out, the master site's interfacing a was bit easier... sort of...  This repeater's master site is actually split, with the receiver and antenna being several hundred feet away from the transmitter, this being done to put it farther away from the megawatt of RF being emitted from all of the TV and Radio transmitters on the main site!  It is connected via transformer-coupled cables (for lightning protection) and has operated with minimum maintenance since the early 1980's.

Since we already had on-hand local COS (squelch) and audio from the receiver, there was no need for the tone signalling schemes of the remote site but, instead, the audio and COS lines could be input to the voter.  There was a problem:  The "local" receive audio was "too" good!

The way the voter works is that it analyzes mostly the audio above 2.5 kHz and of the receivers being compared, it is the receiver with the MOST audio above 2.5 kHz that is considered as being the one with the worst signal.  The reason for this is pretty simple:  As an FM signal gets weaker, it gets noisier, so it stands to reason that given two otherwise identical signals, the one that is also noisier will have a total signal level that is higher - particularly at higher audio frequencies.

The problem was that the audio from Scott's had already passed through a radio link which tended to scrape off the audio above about 3.5 kHz or so while the "local" audio, being coupled via wire, had no such low-pass filtering, so we had to add some.  What was happening is that the "local" audio - with its additional "highs" (as compared to the audio from Scott's) was being considered to be "bad".  By removing those extra high-frequency components and making the two audio signals pretty much equal we were able to make the more-or-less directly comparable.

Next time:  A bit more about the voting controller and some of the remote control/monitoring capabilities.


This page stolen from

No comments:

Post a Comment


While I DO appreciate comments, those comments that are just vehicles to other web sites without substantial content in their own right WILL NOT be posted!

If you include a link in your comment that simply points to advertisements or a commercial web page, it WILL be rejected as SPAM!