Digital wireline STLs

March 1, 2009

The link from the studio to the transmitter is a criticial, but often overlooked part of the transmission system. There are generally two options: wired or wireless. There are some common reasons behind building a wireline STL system as opposed to using a radio system:


  • A move to a new studio location where there is no line-of-sight to the transmitter.
  • A move where there aren't any channels in the 950MHz band that can be coordinated.
  • The radio station already has a radio link established, but wants a wireline STL as an alternate or backup.
  • The station has data or program audio (such as a satellite receiver) that needs to be backhauled from the transmitter site.


    There could easily be additional reasons, but these offer a good foundation. I'll look at what is available in the equipment marketplace for wireline STLs in the categories of what we used to call audio loops; equipment that makes use of T1; and finally, I'll expand the capability greatly and see how to use T3.

    Audio lines

    Ordering 15kHz loops from the telephone company used to be standard operating procedure for STL purposes (and remote broadcasts of course). The results were hard to predict ahead of time: it depended very much upon the quality of the telco techs you just happened to get to align the system from end to end. If you ordered a stereo pair, you were really willing to test the limits of your own patience.

    Pulsecom PCAU

    Pulsecom PCAU

    Here in New York City, our local phone company (Verizon) still offers 15kHz loops but fortunately for us, the modern version. These circuits are built around the Pulsecom PCAU. The PCAU card looks very much like the old Tellabs 4008 cards, and it accepts analog in, and puts analog out on the far end. That's pretty much where the resemblance ends though. The reality is that the PCAU is an A/D converter and communicates with the far end via a digital path through the phone company. By making use of Apt-x coding, the bandwidth requirement is lowered (making telco happy). According to Pulsecom, the units automatically align themselves with one another, for flat frequency response and zero loss on the far end. Two units can be made into a stereo pair by means of a short interconnect cable on both the near and far ends.

    Pulsecom also makes the HD PCAU, which is suitable for HD Radio purposes. This card has 20kHz of audio bandwidth (once again relying upon Apt-x coding); accepts analog or AES; has provision to accept the sample rate reference clock; and finally, it has built in provisioning to transport PAD and SIS data to the far end as well.

    Moving on to T1

    The explosion in data requirements for both the cellular telephone system and other types of wireless data have caused local phone companies to greatly expand their infrastructure into remote mountain tops and other tower sites in order to accommodate these customers. Fortunately broadcasters have been able to take advantage of this. Several manufacturers offer equipment designed to use the now-ubiquitous T1 for transport.

    There are several compelling reasons to use T1 for transport of an STL system:


  • Buying in quantity reduces the unit price. Obviously I don't know every tariff in every state, but my experience in California, Washington and New York is that the cost of a stereo pair is usually at least as much if not more than an entire T1.
  • Not only do you buy the A to Z direction with a T1, but you get the Z to A direction as well. This makes it easy to configure a TSL system should you need it.
  • The TDM nature of T1 makes it easy to combine multiple types of service in to one link: audio, telephone, serial data, and ethernet can be combined into one system.



    Let's take a look at some of the equipment out there.


    Moseley Starlink SL9003T1

    Moseley Starlink SL9003T1


    Moseley offers the Starlink SL9003T1. The heart of this system is the 3RU intelligent multiplexer into which daughter cards of various functions are installed. (An entire system is made up of two of these frames, of course, with sets of cards.) The cards for audio transport will accept analog input (+18dBu limit of headroom) or AES (or S/PDIF) via 110-ohm balanced XLR; sample rate 32, 44.1 or 48kHz with built-in SRC; and an auxiliary port for RS-232 that will run up to 9600 baud. If you want to accommodate HD Radio with multicast (or any of the other reasons to have a LAN extension at the transmitter) you would add the card that functions as an 802.3 Ethernet bridge; and you could build an off-premise telephone extension by adding the voice module data cards. The multiplexer frame can accommodate up to two T1 interfaces (one for redundancy), each of which has a built-in CSU. Management of the system is by a windows-based GUI that goes on a client computer; remote management is done via a built-in communications channel that operates over the link.


    Harris/Intraplex STL Plus

    Harris/Intraplex STL Plus

    Probably the most well-known manufacturer of T-1 based equipment is Harris/Intraplex. The STL Plus (previous page) is a system made up of two frames into which daughter cards are installed. For an HD Radio application, this would consist of a PT353 (encoder card) and PR353 (decoder card) and a pair of DS64NC cards (making up the LAN bridge). The audio cards accept analog audio, or an AES data stream (selectable sample rate, with built-in SRC). Because the audio cards have both AES and analog inputs and outputs, the end-user can select the type of interface card that plugs in to the rear apron of the frame; analog only, digital only, or analog plus digital (XLR connectors all the way) are available as options. The plug-in interface card (known by Harris as module adaptors) for the DS64NC pair has an RJ-45 connector. Harris makes other card sets for OPX, and for data-reduced audio paths as well. Management of the system is via a serial connection, and remote management can be over the link using one of the ds0 timeslots.


    Musicam Team

    Musicam Team

    Musicam USA makes a frame-based system as well, known as Team. The frame is 4RU and can accommodate up to 14 encoder or decoder modules in one frame. (Up to eight frames can be integrated to make up one system.) The frame can accept up to four T1s since each T1 interface card will connect up to two separate T1s. The audio modules accommodate analog and AES (via XLR connectors on the standard modules, or via D-connectors on the slim modules). MPEG layers 2 and 3, as well as Apt-X and Enhanced Apt-X are the options available for audio transport.


    The local or remote units can be managed via Ethernet or RS-232, and have the ability to dynamically change the number of timeslots allocated for network communications on its LAN bridge, as well as the configuration of the audio cards (whether analog or AES is picked, and the audio coder in use).


    Worldnet Oslo

    Worldnet Oslo

    One of APT's products that uses T1 for transport is the Worldnet Oslo. Like all the others discussed so far, the Oslo is a system made up of a mainframe, with plug-in modules that accomplish various functions. The frame can accommodate up to 12 audio channels in either direction: Individual audio modules come as two-channel duplex analog cards, four-channel analog simplex cards (one input and one complementary output card per system of course) and two-channel duplex cards with AES and analog inputs/outputs. Audio can be sent as linear PCM, or via compressed data codecs MPEG layer 2 or via Apt-X or Enhanced Apt-X. System management is by an Ethernet connection and a GUI called Worldnet NMS that lives on a client computer. I should note also that the mainframe will hold a redundant power supply.



    So to review quickly: Each of these systems is mainframe-based, modular and configurable. Each system has 24 timeslots (24 ds0s) or a grand total of 1536kb/s of payload capability.

    Beyond T1


    What do you do if a single T-1 isn't enough for you?


    I wrote earlier that if you buy in quantity, you get a better deal on a per-unit basis. Turns out here in NY that the cost of five T1s is the same as an entire T3, which is 28 T1s. Because we have five stations in New York we went with T3 all the way.


    Let's say though for the sake of argument that you don't need 28 T1s, although you do need more than one. What could you do in that case? Let's make up an example and then look at one potential solution. Say you have two stations on a mountain top, and through budget analysis, you've determined you can afford four T1s through your local telco. You want to use linear PCM for your audio on both stations, and you also want the most bandwidth for the transmitter site LAN as you can get a hold of. Of course you don't want the failure of any particular T1 to take either station off the air; in fact, it would be really sweet if you could stay on the air even if two of the T1s were to fail — no matter which two.


    For station A you purchase a frame-based T1 system. In that frame, you configure two audio cards to use 18 timeslots. That becomes your primary STL for station A. In the same frame, you configure a data-reduced audio path that uses four timeslots. That becomes the backup path for station B. For station B, you do the same thing after buying a second frame-based system; 18 timeslots as the main station B STL, and a data-reduced path of four timeslots that is a backup for station A. Station A uses T1-1, and station B uses T1-2.


    Adtran 4305

    Adtran 4305

    For T1-3 and T1-4 you may need some help from your IT people (unless that of course is you). For this example, you could spec a set of routers that have multiple T1 interfaces. One example of that is the Adtran 4305. With a router such as this, you can literally bundle multiple T1s in to one network link. The cool thing also is that the link will continue to operate even if one of the T1s goes down. Obviously the data throughput will be reduced.


    The two bundled T1s make up your LAN extension now, giving you a LAN bandwidth of 3072kb/s.


    Taking this one step further, you could develop an audio stream for stations A and B (though the data rate can't be that high unless you want the streams to hog your network) that go through the routers to reach the far-end site.


    A system such as this can easily continue to operate (albeit at reduced audio quality) through the failures of two of the four normally connected T1s.


    I wrote earlier about using a T3 data circuit for data transport between the studio and the transmitter site (assuming that makes financial sense for your station). Again, let me point out that Adtran makes equipment for just such purposes. The MX2800 (as the name implies) takes 28 T1s and aggregates (or muxes) them together to build up a T3 data stream, and provides the network clock reference. The interface for both send and receive directions (since this is also a full-duplex link) are in the form of 75-ohm BNCs. You'll hand off your T3 to your local telco via these cables.


    The great thing about T3 is the scalability; just because you have 28 T1s to start doesn't mean you have to turn all of them on right way. You can use what you need at first, and turn more on as needed. Remember, the cost is the same whether you use 5, 10 or 25 of the T1s. Using a router such as the one I mentioned earlier allows you to bundle multiple T1s together to form a high-bandwidth link.


    If you were to build a high-bandwidth LAN extension to the transmitter site, another set of possibilities comes in to being for your station's STL (and TSL of course) systems.


    For example, the APT Oslo, mentioned earlier, can also be built for IP transport by the addition of the dual IP MUX card (the dual part referring to its separate Ethernet ports, which can live on separate networks).


    Harris/Intraplex offers the Net Express, which is very similar to the system I wrote of earlier, but T1 transport is replaced by IP transport.


    Audio TX STL-IP

    Audio TX STL-IP

    Another possibility for the transmission of audio over a network such as this is the STL-IP from Audio TX. This is a 1RU box with analog or AES inputs and outputs, wordclock in, and of course the network interface (RJ-45 100baseT). Sample rate up to 96kHz and word length up to 24 bits; linear PCM, AAC, layer 2, layer 3 and G-722 are the most well-known of its available coding schemes. Management is done via a Web browser or a telnet session using command line.


    Likewise, many of the codecs commonly used for remotes, such as Tieline, are capable of communicating via an IP link and can be used to interface via a network.


    And finally, you can take the approach that your transmitter site isn't really far away after all, and that it's just another node on your network. This would be your approach should you chose to go with Axia and one of its AES nodes. This device has eight AES inputs, and eight AES outs, and makes use of an Ethernet network for transport of audio from point A to point Z. (An entire full-bandwidth 100baseT connection is required between the two nodes.) If you have that much bandwidth available to your transmitter site, this is an approach you may want to consider. Alternatively, if the site is in the general proximity of the studio, you may want to consider a fiber optic run between the two points. You could extend your network by adding two Ethernet switches (one at point A and the other at point Z) that are trunked via a fiber interface.


    Wireline STLs have come a long way in the last 10 years or so. As the proliferation of networking has changed the way radio stations are put together, so has it expanded the possibilities for STL systems. I hope I've given you some ideas about how powerful and useful the current technologies are, along with some ideas to consider in the future.


    Irwin is transmission systems supervisor for Clear Channel NYC and chief engineer of WKTU, New York. Contact him at

    T1 Basics


    When studying the specs of T1 equipment you need to keep a couple of things in mind:


    The various services will use up some number of timeslots in the T1, each of which is 64kb/s. At minimum, any one service will use one 64kb/s timeslot (or ‘ds0') and at maximum, 24 timeslots.


    The grand total of all timeslots used cannot exceed 24 since that is the most data that can be carried by a T1.


    For example, if you wanted to use 18 timeslots for your audio path, and six for the Ethernet bridge, there would be nothing left over for any other services. Six timeslots would give your LAN extension a maximum bandwidth of 256kb/s, which is rather slow considering all that goes on at transmitter sites nowadays. Read on in the main article to see how to get around that problem.