HD encoders...
NAB Review
Hot off the floor
Marvin Born
06/30/1999
Broadcast Engineering
Copyright 1999 by Intertec Publishing Corporation, a PRIMEDIA Company. All
rights reserved.
In the beginning, there was digital (Morse Code is by definition digital; on and
off is digital in its purest form). Then came analog with the introduction of voice
modulation. That was called radio. Later, with the addition of a few more tubes,
more bandwidth and something called a lens, radio with pictures was
discovered. It was called television. Since that early beginning, radio with
pictures has been analog. For more than 50 years it remained that way, until
some sharp engineering type discovered the analog TV signal could be broken
into a number of smaller samples and measured. The sample value could be
used to reconstruct the radio-with-pictures signal and it looked just like it did
before it was sampled. This revolutionary technique was called digital. It could
be considered somewhat different than the first digital signal. Also, it was
transmitted at faster rates.
Now with more than 50 years of television experience behind us, we have HD
radio with pictures. A group of US companies called the Grand Alliance
developed the system that became a standard as defined by the Advanced
Television Systems committee. This system is divided into three parts:
coding/compression, multiplex/transport and transmission. The system
conforms to the MPEG-2 standard of video compression at a data rate of
19.4Mb/s and provides for the transmission of audio via the Dolby AC-3 audio
compression system, which provides six channels of surround sound audio.
The ATSC system provides for HD and SD television. This 19.4Mb/s of HD or
SD data is modulated into a 6MHz TV channel using a modulation system
known as trellis-coded 8-VSB. This entire three-part system is known as
encoding.
Encoding By definition, HD encoders must be ATSC compliant for DTV
broadcasting. There are several manufacturers who build compliant equipment.
One of the earliest was General Instruments . Its current product, DigiCipher
II, houses a number of features. One feature is panel overlap processing. This
allows the passing of motion vectors from one panel to another, eliminating the
artifacts caused by the changing of motion across panel boundaries. Encoders
typically have six panels. Inter-panel multiplexing assigns different bit rates to
different panels, depending on the complexity of the picture information in that
panel. This allows more data to be delivered to the panels with high demand
picture elements. The result is fewer artifacts occurring in those high-demand
panels. The system also has scene change detection and fade detection.
Scene change detection allows the software to detect two adjacent video
frames that are either different video scenes such as a camera switch or that
the adjacent frames are of the same scene but contain high motion (such as a
pan across a football field). At points such as these, the MPEG group of
pictures can be closed and a new one started. This results in better bandwidth
control and higher video quality. The fade detection is similar to scene change
but detects the fade to black that results in a change in bit rate for the black.
Other interesting features of the DigiCipher II include the internal AC-3 stereo
processing, rapid switching from SD to HD service and a DS-3 output port. The
AC-3 process is somewhat self-explanatory. No external audio encoder is
necessary. Up to three stereo pairs can be multiplexed into the transport
stream, such as stereo, single mono left and right or any combination from six
audio inputs. The system will, of course, pass through
pre-compressed/encoded audio such as AC-3/5.1 from an external source.
There is also matching between the time stamps in the audio and the timecode
of the video to ensure correct lip sync.
The rapid switching between standard-definition and high-definition may be a
very useful function in the future. Since there is presently no return on
investment from high definition, one of the possible income streams available
would be to transmit high-definition in the prime time hours and some type of
multichannel SD programming in nonprime time. A quick and smooth
transmission will be necessary to accomplish this change. Broadcasters can
not wait off air while the encoders relock to a new format. Additionally, the DS-3
output has many possibilities, allowing ATSC video to be fed via microwave,
fiber and satellite paths.
Another very popular encoder is the Harris-Lucent FlexiCoder, whose design
(and thus the name) has the ability to house multiple digital format cards. The
FlexiCoder allows six SD channels, or a SMPTE 259 digital SD and a HD data
stream at the same time. The FlexiCoder also houses various decoders to
monitor the transmission. FlexiCoder allows facilities to operate a data service,
PSIP, closed captioning and the SNMP manager. FlexiCoder is the present
workhorse of HD, serving the majority of on-the-air HD stations. Lucent
introduced a second line of single-channel HD encoders, model HDP-100 for its
new VideoStar multislot chassis. The HDP-100 is a smaller, less expensive
version of the FlexiCoder technology that supports 720p and 1080i HD formats
at 60, 59.94, 30 and 29.97 frame rates. Fiber, OC-3 and DS-3 interfaces are
also available. As for audio, the 100 supports one stereo AES/EBU input or one
precompressed IEC-958 audio. It also supports closed-captioning, with an
RS-232CC input. There is also support for four streams of ancillary data.
Worthy of note is the Lucent Terastream that uses Internet protocol streaming
to enable personal computers to receive broadcasts. The Terastream Server will
receive video and data from a VideoStar encoder and stream information into a
standard PC. The new Unity Motion HD satellite feed is based on a Lucent
VideoStar HD encoder. Unity Motion provides three channels of programming,
HD-1, HD-2 and HD-3 to DBS receivers.
Lucent conducted interoperability tests demonstrating a HD video signal
simultaneously with closed-captioning and PSIP (Program and System
Information Protocol). The purpose of this demo was to simulate an actual
at-home environment for the viewer.
Thomson Broadcast was showing its System 5000 Multi-service encoder MSE.
The MSE 5000 supports 13 ATSC input formats from 1080P to 480I. Using
Thomson's own encoding chipset, the encoder uses six panels and can be
used to compress either HD or multiple SD feeds.
NDS introduced a new product, a HD ATSC encoder E5820 that supports
480i@30, 480p@60, 720p@60 and 1080i. It has additional expansion module
space for future data support. The E5820 has a special noise reduction system
that increases picture quality and reduces bit-rate overhead by not digitizing
noise. It supports closed captioning using an RS-232 interface.
COFDM for ENG Although not HD related, but extremely important to
broadcasters, NDS announced its system of COFDM-based digital ENG. With
the loss of spectum space in the 2GHz band coming and the reduction in either
the number of channels or the bandwidth of the individual channels, something
is needed to support the present and future expansion of electronic news
gathering operations. COFDM is a multicarrier system of carrier modulation. It
reduces or eliminates ghosting and multipath problems, as well as being
spectrum efficient. From the field trials it appears to allow much better mobile,
meaning moving live shots with no multipath, as well as reduced mast height.
The MPEG-2 data rate is 5Mb/s for mobile use, while the clear path mode is up
to 28Mb/s. Encoding is 4:2:2 or 4:2:0. Inputs include composite or digital video
plus four channels of audio and an RS-232 or RS-422 data channel.
The COFDM modulation consists of 2000 carriers in 6-, 7-, or 8MHz of
bandwidth, selectable QPSK, 16QAM or 64QAM carrier modulation. The
system provides a 70MHz interface to microwave radio systems. This system
was the subject of much discussion. A remaining test should include multiple
systems operating at the same time in a breaking news environment where the
users are typical ENG operators and direct competitors.
On another front, Scientific-Atlanta showcased its PowerVu Plus MDR (Multiple
Decrypting Receiver), which allows programmers to deliver a digital multiplex
easily down a cable system to any OpenCable set-top by simultaneously
receiving and decrypt up to 16 digital channels.
On the subject of set-top boxes, Philips Semiconductor had their TriMedia
processor on hand. This VLIW chip is ideal for multimedia tasks such as
MPEG-2 video, DVD, AC-3 audio and real-time 3D graphics. It is very likely that
alarge number of DTV set-top boxes will be using powerful ICs such as this
one.
There were quite a few statistical multiplexing and encoding systems including
one from Tadiran Scopus. The system is based on the Codico E1000 series
encoders and RTM-3600 multiplexers. Another was Divicom's Divicast which is
capable of handling up to 35Mb/s of data with as many as 60 MPEG-2 PIDs
(for more information, see Pick Hits, p.74).
Last is a look at Divicom's MV-400 HD encoder. Divicom has a different
approach to HD encoding. Rather than dividing the screen into six tiles and
using six different SD encoders (one for each tile). Divicom divides the screen
in a number of horizontal strips called image slicing. There are 10 slices, each
with its own processing engine and encoder. Some of the highlights of the
MV-400 include: compact size (1RU); support for 1080i, 720p and 480p; use of
motion tracking to reduce artifacts; and integrated Dolby Digital audio support
as well as closed caption encoding. There is also an inverse telecine mode that
eliminates the 3:2 pull-down, reducing bandwidth on film-based program
material.
All manufacturers in the encoding business show a number of different HD
encoders including those for DBS and cable as well as those for broadcast.
The only disturbing display was one manufacturer that showed the same
material in 480, 720 and 1080, so the attendee could see the difference
between the formats. All the monitors were the same, of course, but they were
Gas Plasma units with 700 line maximum resolution. Many people could not
see much difference between 720 and 1080 in that booth. You would think they
would know. |
