Spectra TAC Gang Voting Interface Modification

Spectra TAC, Digitac, MICOR, CDM1550, CDM1250, CDM750, and the Motorola Logo are products and trademarks of Motorola.
Used with admiration.

Revisions:
3/22/11 - clarity
3/30/11 - detail for Audio Control Module types and modifications
4/18/11 - optimizing for both 25 kHz and 12.5 kHz bandwidth and matching receiver audio spectral shapes
5/7/11 - added before/after schematics of audio module modifications at article end.

1. When a receiver unsquelches (the combination of carrier squelch and PL decode), there has to be a 15ms period of silence between the end of status tone and gating of receiver audio onto the line. This pause causes the AGC in the comparator to hold at the value last determined from status tone level and enables the activity checker which differentiates silence from speech. Without the activity checker, a fully quieted stuck mic would remain voted and lock a system up.

2. When the receiver resquelches, there has to be a 40ms period of silence to prepare the AGC to resume sampling and to disable the activity checker.

This is true in either Legacy S-Tac comparators, as well as in Digitac comparators. I have both in different systems.

Without these two pauses, the comparator AGC will try to track the received audio as constantly varying idle tone and will level hunt and cause bizzare voting behavior, primarily that a fully quieted receiver may be "voted off" by the comparator in favor of a noisier receiver. This can also be observed as a receiver being voted as long as there is no speech, but voting off when speech is present.

The Audio Control and Encoder modules of a legacy S-Tac Receiver, or current products such as MTR repeaters and MTR voting receivers provide these timings. Most aftermarket "adapters" do not. I had experienced these behaviors when trying to use mobiles with DHE RLC cards as voting receivers. In systems with both S-Tac receivers and receivers with DHE cards, this behavior was occuring. I could not resolve it no matter how carefully I aligned the system.

I came into a set of manuals with extensive notes and bulletins from Motorola field technicians with detailed discussions of these problems when attempting to integrate GE voting receivers into S-Tac systems in the Chicago PD system back in the day. The GE voting receivers did not provide this timing, and the systems behaved badly.

[Revision - audio spectral shape with dissimilar receivers was also a factor. Discussed below]

These system notes revealed why the DHE cards did not work correctly. The DHE cards are designed to work in the GE format, and they would work well in that format; but not in Spectra TAC.

Since I did not want to spend $4K per receiver in a 40 receiver system in the migration to narrowband, this modification uses a Spectra TAC comparator chassis as a card cage for 4 sets of the critical S-Tac receiver modules; and they can be used with current mobile radios as Spectra TAC format voting receivers.

Stock S-Tac receivers use a MICOR receiver deck, and discriminator audio is sent to the audio control module where a bi-level squelch chip provides squelch function, and also to a PL decoder module for the PL decode logic signal. A third module generates and gates 2175 Hz idle tone.

In my case, I'm using CDM series mobiles as receivers. Like all Motorola mobiles with the 16 pin accessory connector, these radios can provide unsquelched flat discriminator audio as well as a logic signal for PL detect. The flat receiver audio correctly drives the Motorola bi-level squelch chip in the S-Tac audio control module so that correct 'AND' squelch behavior can occur. 'AND' squelch is also critical to the S-Tac format, that is how the proper SINAD point is established for receiver muting.

The Spectra TAC receiver modules are positive 12V logic, so a buffer transistor is needed to invert and level shift the TTL level logic signal from the mobiles to the circuitry of the modules. No PL decoder module is required, and we have the added benefit of being able to use either PL or DPL.

The discriminator audio level from the mobiles is hotter than from a MICOR deck, so some simple padding is required to get it down to a level that the squelch chip likes.

[Revision - this pad is not necessary. See below]

Shown are CDM1550XLS mobiles, they will gradually be replaced with CDM750 or CDM1250 mobiles.

Each Chassis contains 4 channels of interface. My system is an 8 channel modulus, so each of my deployable sites gets 2 chassis and 2 receiver panels.

In subsequent re-works, other components have been added to the backplane not shown in the pictures:
** a base resistor for the level shifter transistor.
** a 680 ohm termination resistor for each audio module to terminate the "local audio" winding of the output transformer to eliminate noise pickup.
** a small fan for the chassis power supply. While the added current draw of the receivers is well within the safe operating area of the 13.8V pass transistor, the heat sink is inadequate for the added dissipation. The brushless DC fan also creates electrical noise, which required a noise filter.
** The pesky PC board 2175 Hz level pot on the encoder modules, which tend to get noisy and hard to exactly set, have been replaced with a panel mounted 10-turn trim pot. An extender card is no longer required for basic system alignment.
** All Spectra TAC chassis are designed to mount on deeply recessed rack rails, which is fine if they are being racked with other Motorola equipment, but a pain otherwise. The end plates of the chassis are riveted rather than being spot welded, so it was a simple matter to turn them around. The chassis are now flush with the rack rails.

The comparator power supply is electrically identical to the Spectra TAC receiver power supply. This is handy, because the audio control modules require 9.6VDC for the squelch chip. 9.6V is not used in the normal Spectra TAC chassis, but the 9.6V regulator exists in the comparator power supply.

These receivers are mounted in deployable shelters and rolling racks that take some abuse when they are transported and placed, so a rugged rack panel was essential.

Note the panel mounted 2175 Hz adjustment trimmers at the bottom of the encoder modules.

The CDM receivers now allow me to program all the frequencies for all the events we do arranged in zones for each event and its license. Some of those licenses are still 25 kHz, some have had the 12.5 kHz modifier added, one license was 12.5 kHz to begin with.

8 receivers and 8 interfaces now fit in half the rack space of 8 stock Spectra TAC receivers and weigh half as much.

Audio Control Module
There are four variants of the Audio Control Modules (three newer versions with the three TT jack test points on the panel) and one older Audio Control Module (those with two full-sized lock-nut potentiometers and no test points). Any of them will work well.

The "older of the newer" shiny silver face modules are all non-800 MHz variants (TRN6080B) as far as I have experienced. However, the "newer of the newer" grey hammertone modules can be either TRN6080's or TRN6956's.

In the TRN6956's the squelch IC is tailored with component values for the 800 MHz receiver. Aside from the model number, the visual giveaway for the 6956A is that there are more components in the lower left corner of the board, with the board held so that the panel is to the left looking at the component side. Also, the circuit board material is bright green instead of dark green (usually).

Conversely, the 6080B will have missing components in that area and there will be one capacitor position that is jumpered with either an insulated wire jumper OR with a zero ohm jumper that looks like a resistor with only a single black stripe.

The operational difference is that the squelch circuit is less sensitive in the 6956, and you will not be able to get it to close at signals greater than about 0.3 to 0.35µV.

You want to use "and squelch" and you want the noise squelch to override PL and to close at about .6µV, which is -close- to the 20 dB quieting point.

--This assumes that you have enough receivers to have sufficient diversity, and that you want a receiver that is "trending noisy" to shut down and accelerate the voting process.  If not, the PL will hold the receiver open far after it has become unreadable --.

To modify the 6956 to a 6080, the manual calls out 8 component changes. All you really need to do to a 6956 is change R7 from 33K to 22K, remove C41 and R85, and remove and jumper C40. This increases the sensitivity of the squelch IC so that it will close at higher signal levels, the same as a 6080.

If you want to "do it right" and you have the manual and can identify the components, the component values for the change are in a table on the extreme left end of the schematic.

If you do not have a manual (shame on you) photos below.

Spectra TAC Encoder Module

It is highly recommended that you replace the 2175 Hz level pot which is mounted on the board with a front-panel mounted 10-turn pot. The board mounted pots become noisy with time and they become difficult to set accurately. Accurate 2175 Hz level settings are CRITICAL to correct operation.

Make sure you have the encoder module jumpered for the correct 2175 Hz level range. Either -13 dB system format (JU3 and JU5 out) or 0 dB system format (JU3 and JU5 in); and JU11 on the Audio Control Module out for -13 dB.

Idle (2175 Hz) tone levels are absolute. -13 dB means -13 dB. Not -12.9 or -13.1.

According to the Motorola manuals, these levels are measured against full deviation of a 1000 Hz tone at full quieting. For a 25 kHz receiver with PL this is 5 kHz deviation including PL deviation. (i.e., 4.5 kHz of 1000 Hz plus 500 Hz of PL.)

For a carrier squelch receiver, this is 5 kHz.

In any event, whatever your full deviation value is, that is what the 0 dBm point has to be, with idle tone 13 dB below it.

This must be measured with a proper RMS voltmeter with a calibrated dBm scale, with the receiver audio output terminated with 600 ohms.

Receiver audio spectral shapes and 25 kHz vs. 12.5 kHz Bandwidth

It's critical in voting systems that all the receivers are identical in their audio quality. Ideally, if you have MTR2000 repeaters, you also have MTR2000 voting receivers or Astrotacs that are designed to work with this repeater.

As I began to test this new configuration, I noticed that there was a "difference" in sound quality between the voting receivers and the MTR2000 repeaters that they would work with. I first noticed it when comparing open squelch noise from each. The MTR2000 receiver noise sounded "warmer" than the voting receivers. Being an old concert sound guy this means something to me, but to everybody else, it means that the 300-400 Hz segment of the MTR audio was higher than the voting receivers.

Conversely, the voting receivers sounded "colder" ... they had a more moderate slope on the high frequency part of their response as it rolled off approaching 3 kHz.

I confirmed this with a 1/3 octave audio spectrum analyzer.

Some of this is likely due to the MTR2000 doing internal audio conditioning with Digital Signal Processing internally, and the Spectra TAC modules being analog audio conditioning. Different filter slopes, different characteristics.

The result would be that the voting comparator would evaluate the voting receivers as noisier than the repeaters because of the additional high frequency response. I confirmed this in some experiments.

I was also getting some additional circuit noise from somewhere, and getting bleed-through of a little bit of 2175 Hz idle tone, just enough for the comparator to also evaluate it as noise when the receivers were actually fully quieted. The comparator would consistently favor the MTR receiver over a voting receiver with the 2175 Hz bleed-through when both were fully quieted.

The circuit noise and tone bleed turned out to be from a high impedance buffer part of the audio module circuit that buffers the audio after it passes through the passive boost filter section which is normally used to compensate for high and low frequency loss in phone lines. This section turns out to be VERY sensitive to noise, and is normally not a problem in the stock spectratac receiver chassis.

I use digital microwave or landline T1's to network my system, those circuits are very consistent, and extra line equalization is just not necessary for me.

However, in this modification, the audio module and the tone encoder module are in close proximity to each other and this high impedance section of the audio module picked up tone from the adjacent tone module. It was also prone to pick up of electrical noise from the brushless DC cooling fans, which also do not exist in the stock spectratac receiver chassis.

The fix for all the above was to bypass the equalization section, and add a shunt capacitor to make the high frequency roll-off match the MTR2000 receiver characteristics. This was very simple, involving two capacitors. Drawings are provided below.

The final result is that the voting receivers using the mobile radios very closely match the audio characteristics of the MTR2000 repeater receivers, and each other.

Then I did some testing in 12.5 kHz bandwidth..........

....and had a problem with the squelch circuit. I could not get the squelch to close and cut off the receiver at the desired quieting point. In fact, I could not get it to close at anything above about 0.25µV. it would be way too noisy.

At first I thought it had to do with the noise voltage available from the discriminator in the narrowband mode.

It turned out to be the added resistor to pad the discriminator output of the mobile receiver to match the level to that of the old MICOR deck. It is not necessary. In fact, it pads the discriminator level enough that the squelch chip does not have enough noise to work with in narrowband mode.

I got rid of the pad resistor and the squelch chip started to work correctly. I also removed and bypassed another resistor before the squelch control on the module and the squelch action is now in a useable part of the squelch pot's travel, with the 25 kHz range of adjustment in the lower 1/3 of its travel and the 12.5 kHz range of adjustment in the upper 1/3 of its travel.

I can get the squelch to close at 0.6µV in both bandwidths and I have the latitude to take it up to 0.7µV. This corresponds to a little better than 20 dB of quieting. 

All the tone bleed is gone. The noise pickup is gone. The comparator is able to differentiate between two or more receivers that are very close in quieting to each other.

Reliable and consistent voting behavior takes place at very low SINAD levels with very small SINAD differences between recievers.

--------

Depending on what you use for receivers and repeaters, this will be a variable for you to sort out in your system if you really want voting to work really well for you.