In my last column I discussed transmission technology at NAB2003. This month I'll wrap up NAB coverage with a look at DTV reception technology discussed at the convention.
DTV RECEIVER PERFORMANCE
DTV coverage predictions are based on the DTV receiver sensitivity and interference rejection performance defined by the ATSC and outlined in FCC OET Bulletin 69. Ever since the first DTV tuners appeared on the market, broadcasters have complained that performance failed to meet the planning factors that are the basis for the U.S. DTV allocations.
Dr. Charles Einolf, Jr. presented a paper he authored with Oliver Sichelschmkit and C. Thomas Boyer that described tests conducted at the Advanced Television Technology Center (ATTC) in Alexandria, Va., to measure performance limitations in consumer DTV receivers. Tests of six DTV receivers available in the Washington, D.C., metropolitan area in May 2002 (five set-top boxes and one DTV set) at ATTC showed current DTV receivers do not meet the FCC DTV planning factors. Although the sample of DTV receivers may seem low, the same tuners are often used in set-top boxes and receivers sold under different brand names or model numbers.
First, the good news: Under the FCC planning factors, the minimum signal required at the DTV tuner input is -84.0 dBm. Five of the receivers tested had sensitivities close to this level, but one was 9 dB worse.
DTV tuner performance with NTSC interference was not as good. For co-channel NTSC interference tests, three desired signal levels were used: weak (-68 dBm), moderate (-53 dBm) and strong (-28 dBm). The FCC planning factors set the D/U ratio for reception at +1.8 dB. Four of the six tested DTV tuners failed to perform at this ratio under all three signal levels. One performed 11 dB worse, but the at the weak signal level all but two were close to the FCC planning factor level. At moderate and strong signal levels, five of the six receivers worked at a D/U ratio less than 5 dB, with three of them within a dB of the planning factor.
In the presence of upper or lower adjacent channel NTSC interference, all six consumer DTV tuners failed to meet the FCC planning factors at weak and moderate signal levels. Measurement of compliance with the FCC planning factor thresholds of -48 dB D/U for lower adjacent NTSC interference and -49 dB for upper adjacent NTSC interference at the strong signal level (-28 dBm) was not possible in most cases because the undesired power level went above 0 dBm as soon as a D/U ratio of -28 dB (lower adjacent) or -29 dB (upper adjacent) was reached. Two of the receivers worked at lower adjacent D/U ratios around -40 dB at both weak and moderate signal levels, but one performed 18 dB worse than the threshold. Two DTV tuners missed the planning factor upper adjacent D/U ratio of -49 dB by 15 dB.
Equalizer correction capability of the tested receivers for pre-echoes varied from 4 to 10 microseconds. Post echo correction capability varied from 19 to 46 microseconds. There was a wide variation in multipath performance. Some receivers tested poorly, but overall, significant improvements were found in equalizer performance compared with first generation receivers.
During the question and answer period after the paper was presented, it was noted that even the "Blue Rack" receiver used in the original ATSC testing did not meet all the planning factors. It was questioned whether it was even possible for a receiver to meet all the factors simultaneously because a double conversion receiver will have problems meeting the noise figure specification, while a single conversion receiver won't be able to meet the taboo specifications.
What are the implications for broadcasters? Based on the ATTC test results for current consumer receivers, it is best to assume an NTSC adjacent channel D/U ratio of -30 dB or less, not -- 48 or -- 49 dB, when determining real-world DTV coverage. In strong signal environments, a D/U ratio of -- 20 to -- 25 dB may be more appropriate.
The ATTC study concluded there is a need for specific target requirements for DTV receivers. These need to address, at a minimum, RF performance (interference rejection, sensitivity and AGC capability), equalizer performance (pre -- and post -- echo range, static and dynamic multipath, single echoes, echo ensembles) and receiver functionality (PSIP).
DTV RECEIVERIMPROVEMENTS
At last year's NAB, Richard Citta, chief scientist for Linx Electronics presented the company's receiver technology and reported on initial test results at the IEEE Broadcast Symposium in October. Since its introduction, the receiver has been refined and tested in challenging DTV reception locations. At this year's NAB, Citta and Victor Tawil from MSTV described the results of Linx receiver field tests in Washington, Baltimore and Philadelphia.
Here is a brief description of the tests and the results. For the tests, the Linx receiver was compared with two reference receivers-the reference receiver (A) used in the VSB Enhancement Project and an earlier generation reference receiver (B) used by Linx in its Chicago measurements. Measurements were taken concurrently on all three receivers. The sites selected were ones where it was expected DTV reception would be difficult or impossible -- areas surrounded by high -- rises and multistory buildings. The tests were conducted using a dipole antenna tuned to Channel 34 mounted five feet above ground, usually on a sidewalk. The antenna was optimized for best reception on the first channel measured at each site. It was not rotated for measurements on the other channels.
Four in-home locations were tested in Washington; 206 measurements were taken during the tests at 13 outdoor sites in Washington, four in-home sites in Washington, eight outdoor sites in Baltimore and 12 outdoor sites in Philadelphia. In Washington and Philadelphia, six channels were used for the tests. In Baltimore, only four channels were used.
How did the receivers perform in these difficult locations? Reception failure was defined as the number of hits in a three-minute period exceeding 50. Reception was considered successful if the number of hits in three minutes was less than three. Intermittent reception was defined as three to 50 hits in three minutes. Preliminary results from the Washington tests showed a 17.7 percent failure for the Linx prototype compared with 48.1 and 64.7 percent failure rates for the reference receivers A and B, respectively. In Baltimore, the Linx prototype failed in 12.5 percent of the measurements, while receivers A and B failed 62.6 and 63.6 percent of the measurements, respectively. In Philadelphia, the failure percentages were 26.3 for the Linx prototype and 75 percent for both receivers A and B.
Copies of the NAB presentation, including photos of some of the test sites, are available at the Linx Electronics Web site, http://www.linxelectronics.com/technology.htm. See the reports for more details, including station-by-station reception statistics. In most cases, reception performance didn't vary widely from station to station with the Linx prototype, but some stations clearly had a harder time with some of the reference receivers.
Citta said he was looking at the use of diversity antennas selected using smart electronics to improve reception. A dual tuner would not be required, but he said his technique would respond to rapid changes in the multi-path environment much better than smart antennas such as the NxtWave Nxtenna.
ATI'S NEW CHIP
Last year I described how NxtWave's Nxtenna and associated 8-VSB decoder chips with EIA/CEA-909 smart antenna interface worked together to optimize DTV reception. Since then, NxtWave was acquired by ATI Technologies. Fortunately, ATI has continued work on the chip. At NAB, ATI showed it in use with a smart antenna developed by RDI, Inc. RDI's president, Barry Miller, said the DTV-5000 is the first practical smart antenna.
All of ATI's current line of ATSC receiver chips-the NXT2003, NXT2004, NXT2005 and the new THEATER 310 chips-have a EIA/CEA-909 antenna control interface and use Nxtenna technology to characterize the received channel and adjust a smart antenna for best reception. ATI said the newer chips have increased equalizer span, new semi-blind and blind equalization algorithms, faster equalizer update algorithms and improved synchronization algorithms. The THEATER 310 chip combines VSB, QAM, NTSC and BTSC stereo decoding in one device and eliminates the need for separate analog IF strip.
An ATI press release prior to NAB said the THEATER 310 chip "is able to deal with strong co-channel NTSC interference, burst noise, adjacent channel interference and single frequency interference in addition to its multipath canceling capability." Samples of the chip should be available to manufacturers in the third quarter of this year.
DTV PROPAGATION
Most engineers know software analysis of DTV propagation often fails to match real-world conditions. In his paper titled Longley-Rice 101 at NAB2003, William Meintel described the limitations and underlying assumptions that are the basis of the popular Longley-Rice propagation model.
He explained that the Longley-Rice model calculated field strength using a theoretical propagation model that was tweaked using test data from the field. It is not suitable for determining signal levels in urban areas or in dense forests. In addition, the test data used in the model is an average of measurements taken over a year and does not capture short-term variations that can have a significant effect on propagation.
Longley -- Rice considers three different propagation modes -- line of sight, diffraction and scatter. The model allows selection of different propagation environments-maritime and continental temperate are the two most used in the U.S. However, Meintel said there is little difference between the two for paths less than 100 km.
Longley-Rice also requires setting the desired statistics for percent of time, location and confidence. For broadcast use, 50 percent confidence is always used. The statistical adjustments are based on empirical data.
When looking at Longley-Rice studies, it is important to remember that the coverage predictions are statistical and based on medians! All models, Longley-Rice included, are a compromise and there is always room for improvement.
Peter Ludè of iBlast described his company's experiences with real-world DTV reception. The iBlast data receivers have the ability to send reception data to iBlast, where it can be analyzed, making it easy to see how propagation models perform. During his NAB presentation, Ludè said he preferred the Anderson 2 propagation model for evaluating DTV coverage. He did not go into details about the model, but I was able to find information about it on the EDX Web site: http://www.edx.com/resources/pub.html.
The model was developed by Harry Anderson at EDX and uses deterministic ray-tracing to predict the time delay and fading characteristics for the channel in a hypothetical urban area. In Anderson's paper in the IEEE Transactions on Broadcasting for September 1993, he suggested polarization diversity "may be a useful technique to mitigate some of the channel impairments predicted by the propagation model." The paper is available at the link above. I'll be taking a closer look at it in a future column.
As always, your comments and suggestions are welcome. Drop me a note at dlung@transmitter.com
