Measure DTV Transmissions from Start to Finish

Digital TV measurements cover everything from packet jitter to RF emissions.

Martin Rowe, Senior Technical Editor -- Test & Measurement World, 4/1/2000

Three years ago, I reported on some of the important measurements you need to make when developing and verifying equipment that creates and transports digital video programs.¹ At that time, all digitized TV programs travelled over computer networks—in studios and other production facilities. Today, digital television (DTV) programs travel over the airwaves. More than 120 DTV broadcasting stations are currently on the air. By the end of 2006, all TV broadcasts in the US will be in digital form.

Figure 1. Each step in (a) the digital television transmission path and (b) the reception path requires test equipment and measurements.

After a complex spatial and temporal compression process, encoded video and audio elementary streams (ESs) are organized into packetized elementary streams (PESs). For broadcast, PESs from many programs are packetized into 188-byte packets; that’s the MPEG-2 transport stream. These packets then may be encoded for transport over a fiber-optic delivery system such as an ATM network or encoded by an ATSC modulator before being RF modulated for broadcast over a terrestrial, satellite network, or cable-TV system.

MPEG-2 transport streams contain tables that associate data packets with programs. ATSC transport streams, particularly those on cable and satellite systems, contain programs taken from many MPEG-2 transport streams. To keep track of the programs, ATSC modulators add a table called program and system information protocol (PSIP).

After the data receives PSIP tables and the MPEG-2 transport streams are multiplexed into ATSC transport streams, the modulation process is ready to begin.

The term “modulation” in DTV actually refers to two processes that take place within the ATSC modulator. Although labeled “modulation,” the first process isn’t modulation in the traditional sense. Rather, the ATSC modulator manipulates the data using a scrambler, a Reed-Solomon encoder, and an interleaver.

The second modulation process is typical modulation. Within the ATSC modulator is a true modulator that converts the ATSC transport stream (data) into an 8VSB (8 vestigial sideband) modulated analog signal.

Over the Air
Because terrestrial DTV signals travel over the air, they are subject to impairments from EMI and from multipath reflections. To minimize bit errors, the ATSC modulator uses a Reed-Solomon encoder to add forward error correction to the transport stream, which increases the size of each packet from 188 bytes to 204 bytes.² After Reed-Solomon encoding, the ATSC modulator interleaves the data. That process mixes data from different video frames, which minimizes the visible errors from a burst of interference.

Interleaved data then goes through a trellis encoder, which sets up the data with eight possible symbol outputs. The transport stream, which until this point was a single serial stream of 204-byte packets, becomes a stream encoded in 3-bit segments. A multiplexer adds data-segment sync and frame sync signals to the stream. Finally, the transport stream is ready for RF modulation.

The 8VSB modulator adds a pilot signal, which helps the receiver phase lock to the transmitted signal. This process shifts up the counts of the 3-bit-wide transport stream. A root-raised cosine filter reduces signal power outside the 6-MHz channel bandwidth. A DAC then converts the digital signal to analog I and Q outputs for modulation using the upper sideband only. An upconverter and power amplifier send the RF signal to the transmitter.

On the receive side, a digital TV set or set-top box demodulates the RF signal, retrieving the trellis-encoded bit stream. Then, it reverses the Reed-Solomon encoding, the scrambling, and the interleaving to recreate the ATSC transport stream. After reading the PSIP tables, the receiver can decode the selected MPEG-2 transport stream (in effect, selecting the program channel). Finally, the receiver drives video and audio DACs, producing the video and audio information that viewers see and hear.

What the Test Equipment Does
If you design MPEG-2 encoders and decoders, you can use MPEG-2 analyzers and transport-stream generators to test your products. Analyzers capture packets from an encoder’s output and test them for compliance to the MPEG-2 standard. Analyzers also decode MPEG-2 transport streams into their components—video programs, audio programs, and information that tell decoders how to decode the transport stream and associate the video and audio programs correctly. Analyzers can measure parameters such as packet jitter , and they help you decide how much buffer memory a decoder needs.

MPEG-2 transport-stream generators can record and play back MPEG-2 encoded video. You can use these instruments to test ATSC modulators (Fig. 1a) and MPEG-2 decoders (Fig. 1b). The test streams can come from encoders or multiplexers.

If you’re an engineer who works in manufacturing test, in video program production, or in broadcasting, you can use the new equipment on the scene—MPEG-2 monitors. There’s some overlap between MPEG analyzers and monitors. Both can indicate if a transport stream complies with the MPEG-2 standard and both measure packet jitter. But while MPEG-2 analyzers provide the details of a transport stream, they typically analyze a single stream at a time. Some MPEG analyzers and monitors also test the ATSC transport streams. Test equipment will decode the transport stream and check the validity of the PSIP.

Monitors check for MPEG-2 compliance on multiple transport streams while logging compliance errors. MPEG-2 monitors test the quality of a transport stream after it’s been transmitted through a TV studio’s LAN, WAN, the public network, or a microwave link to a TV transmitter.

MPEG-2 analyzers and monitors test transport streams for errors. Jitter is a significant contributor to a transport stream’s errors. At any point along the transport stream, jitter can add errors just as it can in any datacom transmission. In a transport stream, jitter occurs at the bit level and at the packet level.

At the bit level, engineers look at jitter with an oscilloscope. If the opening of an eye diagram gets too small or vanishes, then a decoder won’t recreate the encoded program.

Get the Clock
MPEG-2 decoders must extract a 27-MHz program clock reference (PCR) from the encoded data so it can phase-lock to the PCR. Encoders must add bits to the transport stream and then scramble the bits so the stream contains sufficient bit transitions for a decoder to extract the PCR. Figure 2 shows a screen from an MPEG-2 monitor with PCR jitter measurements.

Figure 2. MPEG-2 monitors and analyzers decode transport streams into tables (shown as PAT and PES), then calculate PCR jitter. (Courtesy of Xyratex.)

PCR jitter has a direct effect on video and audio quality because the decoder uses the extracted PCR to trigger its video DAC and audio DAC. Typically, clock jitter shouldn’t exceed 0.5 ns.³ Jitter that exceeds 0.5 ns will cause blurred video and distorted audio.

Jitter contributes to a transport stream’s bit errors. You can use MPEG-2 analyzers to measure bit-error-rate (BER) on transport streams. Your goal as a product designer, test engineer, or broadcast engineer is to produce products and programs with BER at levels below that which viewers will notice. Just one bit error per TV frame is too many.

At the packet level, jitter refers to the timing differences between the arrival times of packets. An MPEG-2 decoder contains buffers to hold incoming packets before outputting them at a constant bit rate so a decoder can process them. MPEG-2 analyzers can test the buffers to ensure that the decoder has enough storage space. Decoders must have a minimum of 512 bytes for storing incoming transport streams.

After the ATSC transport stream gets RF modulated, it is effectively an analog signal modulated to represent bits. The modulation and demodulation of a DTV signal add opportunities for more data errors to occur.

Unlike analog TV, where the signal gradually decays, DTV reception is either perfect or nonexistent. The transition from perfect to no reception is called the signal’s cliff. When a signal reaches the cliff, it contains too many bit errors for the receiver’s decoder to recreate the program. That occurs when the received transport stream contains more than 10 bit errors in an MPEG-2 packet.

Figure 3. A modulated DTV signal’s frequency spectrum must fit within an FCC-defined mask. (Courtesy of WHDH. Boston, MA.)

Figure 4. An 8VSB constellation diagram should appear with eight narrow, vertical sets of symbols. (Courtesy of Agilent Technologies.)

The 8VSB modulated signal must fit into a 6-MHz bandwidth, which can carry a transmitted data rate of about 19.39 Mbps. A DTV signal monitor or vector-signal analyzer can measure the signal to make sure it fits into a FCC-mandated mask, shown as red in Figure 3. The pilot signal is clearly visible at 309 kHz above the lower frequency of the channel.

The modulated signal must fit into the mask to ensure that signal power stays within the allotted channel bandwidth. The signal power should not exceed –47 dB (called the “knee”) at 63 MHz from center channel (shown at 0.0 MHz in Fig. 3). Outside the channel, the signal power must be at least 71 dB below the average signal power within the channel.⁴ Otherwise, the signal may interfere with signals on adjacent channels.

When you measure the frequency response for flatness and group delay, you’ll find out how well the raised-cosine filter works. If the spectrum isn’t flat or if the signal has too much group delay, then distortion might cause the spectrum to exceed the limits of the mask.

The demodulated DTV signal needs measuring, too. A useful measurement is a constellation diagram (Fig. 4). A perfect 8VSB signal will look like eight vertical columns of dots in which each dot represents the presence of a symbol. If the lines widen, then the signal contains excessive noise and distortion. Curves in the lines of dots indicates phase noise.

DTV signal monitors can evaluate error vector magnitude (EVM), which reveals the overall quality of the broadcast signal. Think of EVM in terms of a constellation diagram. EVM is the rms difference between the measured symbol locations and the ideal symbol locations expressed in percent.

As the EVM measurement increases, the distance from the transmitter to the cliff decreases and viewers in some fringe areas may lose DTV reception. Typically, broadcast engineers try to maintain an EVM of 7% or less. T&MW

FOOTNOTES
1. Rowe, Martin, “How Do You Test MPEG-2 Transport Streams? ” Test & Measurement World, January 1997. p. 20.

2. Application Note: A Guide to MPEG Fundamentals and Protocol Analysis (Including DVB and ATSC), document number 25W-11848-2, Tektronix, Beaverton, OR, March 1999. p. 34.

3. Application Note: A Guide to Digital Television Systems and Fundamentals, document number 25W-7203-3, Tektronix, Beaverton, OR, September 1997. p. 37.

4. ATSC Document A/53: Transmission and Measurement and Compliance for Digital Television. ATSC, Washington, DC. www.atsc.org/standards/a_53b.doc.

FOR FURTHER READING
Balch, Mark, “High-Definition TV: MPEG-2 Transport and ATSC Data Infrastructure,” Circuit Cellar, December 1999. p.62.

Balch, Mark, “High-Definition TV: DTV System Architecture,” Circuit Cellar, January 2000, p.62.

“A Guide to Maintaining Quality of Service for Digital Television Programs,” Application Note, Document Number FL5582, Tektronix, Beaverton, OR, September 1999.

“A Guide to Picture Quality Measurements for Modern Television Systems,” Application Note, Document Number 25W-11419-1, Tektronix, Beaverton, OR, March 1999.

“DTV Fundamentals and 8VSB Measurements,” Video tape, Document Number 25A-13312-0, Tektronix, Beaverton, OR.

“MPEG-2/DVB Digital Television System Integration,” Wavetek Wandel Goltermann, San Diego, CA. www.wwgsolutions.com/appnotes/mpeg/mpegapp.html.  

Reed-Nickerson, Linc, “Understanding and testing the 8VSB signal,” Broadcast Engineering, November 1997, Intertek Publishing, Overland Park, KS.

You can contact Martin Rowe at m.rowe@tmworld.com.