Randy Hoffner /
01.09.2008 12:00 AM
Will the End of NTSC Be the End of 59.94?
Television engineers have been living with the noninteger frame and field rate of color NTSC since 1954. When NTSC does finally go away, the broadcast necessity for 59.94 Hz will also go away, but will this result in its end? Where did this strange parameter come from, anyway, and why do we need it?
In order to minimize interference from the electrical power system, the original black-and-white NTSC had a field rate of exactly 60 fields per second and a frame rate of exactly 30 fps. Each 525-line frame was made up of two fields of 262.5 lines each. Multiplying 60 fields times 262.5 lines produced a horizontal scanning frequency of exactly 15,750 Hz. When NTSC was modified to accommodate a color subcarrier, the total number of lines remained at 525, 262.5 lines per field, but the frame rate changed.
TROUBLE AT FIRST
According to a knowledgeable RCA engineer who was involved, the original choice for the color subcarrier frequency was (455 ÷ 2) x 15,750 = 3.583125 MHz exactly. This subcarrier frequency caused a problem; in that it was spectrally located 0.921375 MHz (roughly 921 kHz) below the sound carrier of exactly 4.5 MHz, and this spectral spacing caused a visible beat at that frequency, the difference between the sound carrier frequency and the color subcarrier frequency.
The RCA engineers decided that the visibility of this beat would be minimized if it, like the color subcarrier itself, were an odd multiple of one-half the horizontal line rate, which would cause it to interlace line to line and frame to frame. In order to achieve this relationship, the sound carrier frequency either had to be moved up by 0.1 percent, or the color subcarrier frequency had to be lowered by 0.1 percent.
The RCA engineers understood that the FCC would not agree to change the sound carrier frequency, so the color subcarrier frequency was lowered to 3.5795454... MHz. In order to maintain the 455:2 relationship between the horizontal line frequency and the color subcarrier frequency, and the 525:2 relationship between the horizontal and vertical scan frequencies (ƒH = 525:2 x ƒV, or 15,750 = (525 ÷ 2) x 60 in monochrome NTSC), these related parameters had to be changed: the line rate frequency became 15,734… Hz, the frame rate became 29.97… Hz, and the field frequency became 59.94… Hz. The compatible reception of monochrome NTSC (now long gone and irrelevant) and color NTSC (used even to broadcast monochrome material today) was not compromised by the slight one-tenth of 1 percent difference in their frequencies, but the 920 kHz beat problem was solved.
The noninteger frame rate of NTSC has required us to use dropframe timecode. Because there is not an integer number of frames per second in NTSC, in order to get the timecode frame count to agree with real time, some timecode frame numbers must be periodically dropped.
Although the mechanics of dealing with this are well understood, many people in the video and postproduction industries would be happy to see dropframe timecode disappear with NTSC broadcasting. NTSC 59.94 Hz also required a clever mathematical approach in order to distribute digital audio samples over video frames.
This consideration, in fact, made a substantial contribution to the ultimate standardization of the 48 kHz professional digital audio sample rate. The two earliest digital audio recorders both used a 50 kHz sample rate, but 48 kHz became the standard in large part because it bears a better mathematical relationship to NTSC frequencies.
A significant advantage of the 59.94 Hz rate is that it permitted the worldwide standardization of the 13.5 MHz digital video sample rate in ITU-R Rec. 601, for both 59.94 Hz systems and 50 Hz systems. 13.5 MHz is exactly 6 x 2.25 MHz. The NTSC horizontal line rate is exactly 2.25/143 MHz (15,734… Hz); the PAL horizontal line rate is exactly 2.25/144 MHz (15,625 Hz).
So by using a sample rate that is an exact integer multiple of 2.25 MHz, we can achieve an identical number of samples per line in both NTSC and PAL. This facilitates easier conversion between 525/59.94 and 625/50 images, and makes it much easier to build 601 machines that may be switched between the two video systems. This would not work using 60 Hz NTSC parameters.
The FCC broadcast channel allocation tables are based on the 29.97 Hz frame rate, so we cannot simply do away with noninteger frame rates in NTSC broadcast. Noninteger frame rates are not a problem or a requirement in DTV and HDTV broadcasting. However, broadcasters are transmitting DTV with 59.94 and 29.97 Hz frame rates. Why? Because broadcasters and networks need these noninteger rates for NTSC, and it would be next to impossible to operate a large television plant in which two different time bases are employed. Therefore, both DTV and NTSC are operated with noninteger frame rates.
WISH YOU WERE GONE
When NTSC goes away, the broadcast requirement for the noninteger frame rate goes away, too. Well, almost. We know that there will be NTSC on cable for some time after NTSC broadcast disappears. There is, additionally, over 50 years worth of 59.94 Hz video material in the tape vaults of the United States, much of it in daily use. Besides the massive inventory of composite NTSC material, both analog and digital, there is also a very large inventory of material recorded in the 59.94 Hz ITU-R 601 component format, both analog and digital.
While anyone who works in the postproduction industry would not hesitate to cheer the passing of dropframe timecode, we are not likely to see its disappearance in the foreseeable future. There are three principal reasons for this: (1) over 50 years worth of legacy material; (2) the well-entrenched ITU-R Rec. 601 component video standard; and (3) beyond doing away with dropframe timecode, there is little compelling reason to move to the integer frame rate.