Expanding DTV coverage III

The last “Transition to Digital” tutorial covered the reasons for implementing single-frequency networks (SFNs), including some information about the ATSC A/110 SFN standard. This tutorial will cover how two different implementations for achieving SFN work and what equipment is needed.

How an SFN works

The basic requirements for an SFN are synchronized, identical bit streams with adjustable timing coming out of every transmitter. Anywhere in a station’s coverage area where two or more SFN transmitter signals are received, the bit streams must be synchronized and identical.

To synchronize the outputs, the transport stream must be locked to a very stable reference, which it is when it leaves the studio. But when traveling over digital studio transmitter (STL) and other types of links, this precise timing can change. When several links, either in parallel or in a series, are used to feed the separate transmitter sites, the TS will no longer be synchronized. This calls for a local timing reference at each SFN transmitter site that can resynchronize the TS as is arrives.

Next is the creation of exact duplicates of the ATSC data stream coming from all the transmitters on the SFN. The problem here is the trellis coding that takes place within the DTV exciter. Every DTV exciter starts up the trellis coding differently and could create a jamming signal in an SFN, because it is not an exact match to the other transmitter’s TS. Trellis coding begins with an initial 36-bit word that feeds the beginning sequence, and the DTV exciter also makes an arbitrary choice of where to insert frame sync (every 624 MPEG packets). This means there are more than 42 trillion different ways a DTV exciter can transmit the exact same TS. Without a method to synchronize the beginning of the trellis coding and frame sync insertion of multiple ATSC modulators receiving the same TS, they will transmit different bit streams causing them to become a jammer.

Special data must be sent to each DTV exciter to preload the trellis coders with the same initial state and synchronize the insertion of the frame sync, so they will transmit the exact same bit stream.

The last requirement is the ability to control the relative timing of the output bit stream from the DTV exciter. The timing adjustments are used to keep the transmitted bit streams within the multipath/echo range of DTV receivers. This requires field tests where the relative timing of multiple SFN transmitters are monitored and adjusted to be within the DTV receiver’s range.

Another important component to setting up an SFN is the ability to measure the individually transmitted signal’s level and timing. This requires differentiating the multiple RF carriers that occupy the same frequency. To do this, an RF watermark is added to the transmitter’s output about 30dB down. Because all data streams must be exactly the same in an SFN for it to work, no identifying information can be added to the actual data. Development of this part of the system is ongoing.

The main difference in SFN system design is in where the timing adjustment takes place and how the synchronizing and timing data gets to the DTV exciters.

A/110 SFN systems

At least two companies have A/110-compatible SFN systems out now, Axcera and Harris. Even though they implement them in different ways, they are essentially the same, but specifically the Axcera system will be addressed here.

After the TS has been created within a multiplexer, either M/H or legacy, the TS is sent to a distributed transmission adapter (DTxA), where synchronizing and timing data is added. The DTxA adds individually addressable, adjustable timing information to the ASI output, so the timing of all SFN transmitters can be adjusted from one location. The timing and synchronizing information is contained within an operations and maintenance packet (OMP), which has been assigned a PID of 0x1FFA. This is the only OMP PID assigned currently. (See Figure 1.)

At each SFN transmitter location, there is a GPS receiver for frequency and timing reference, as well as a compatible DTV exciter that will accept the trellis coding data so the 36-bits setting the initial trellis state will be the same at every SFN transmitter. Using the GPS input along with the timing information from the DTxA, the relative timing of each SFN transmitter’s bit stream is adjusted relative to the GPS input, and, thus, to the other SFN transmitters, allowing for control of the multipath/echo timing received in the field. (See Figure 2.)

Rohde & Schwarz SFN system

Rohde & Schwarz recently announced its SFN system at the 2009 NAB Show. The company is working on having it adopted as an ATSC companion standard to A/110B. The company also wants to make it an open standard so other companies can contribute to their proposed system. At this point, all the data sent to the SFN DTV exciters is carried in the M/H data and will not work with legacy DTV, but Rohde & Schwarz is working on an SFN system that will work with legacy DTV systems. (See Figure 3.)

Its approach inserts all synchronizing and timing data at the M/H multiplexer using network time protocol (NTP); this data is placed within the M/H data. The modified ASI TS is sent on to all the SFN DTV exciters with no individual timing information, only a overall time reference sent to all SFN sites. All timing (multipath/echo) adjustments are carried out at the individual SFN sites either directly or via remote control.

The Rohde & Schwarz multiplexer also outputs the M/H ASI signal in encapsulated IP form, allowing the use of a high-speed data network to distribute the signal to the SFN sites. The use of NTP at the multiplexer allows for future changes and options that are not possible with GPS alone. Each SFN site also uses a GPS receiver (See Figure 4.)


Representatives from Axcera and Rohde & Schwarz contributed to this tutorial.