View Single Post
  #12  
Old 01-12-2020, 11:44 PM
old_tv_nut's Avatar
old_tv_nut old_tv_nut is offline
See yourself on Color TV!
 
Join Date: Jul 2004
Location: Rancho Sahuarita
Posts: 7,815
The September 1970 issue of IEEE Spectrum has a cover and 12-page article that explains a lot. Unfortunately, it is covered by IEEE copyright, so I cannot share a copy.

https://ieeexplore.ieee.org/document/5213552

I'll try to answer briefly:

1) The film runs at 60 frames per second, locked on average to the AC line frequency, but the scan is normal 60 field per second interlace (actual, "random interlace" - the vertical and horizontal are not locked, the same as in a simple closed-circuit TV camera of that era).

2)In order to not require the CRT scan lines to fall exactly on scan lines on the film, each film frame is a blend of two successive NTSC fields. Furthermore, the scan used in recording the film master is wobbled slightly up and down at four times the NTSC subcarrier rate to fill in the blank spaces between the lines. Therefore, ther is no chance of moire' patterns due to the flying spot raster not matching the film raster. The resolution of the film is actually so great that the 14.31818 MHz pattern can be resolved under a microscope. The flying spot diameter is too large to resolve that, but when the electrical and optical focus are perfect, video response goes to 4 MHz, the same as broadcast TV.

3) The blending of fields required a video field memory for the recording process, which was a very expensive thing in the late 60s. I believe CBS used mercury acoustic delays at one time, and eventually went to magnetic discs.

4) When the film is standing still, the raster is 3 units vertical height to 4 units width, the same as NTSC. During motion play, the film moves one film frame height (3 units) in 1/60 of a second, so the raster height is increase to 6 units to scan 3 units of picture frame while simultaneously chasing the film position by 3 units in a 60th of a second.

5) The chroma frame contains two stripe patterns added together. One pattern is a constant pilot sine wave for frequency and phase reference, at 1/4 NTSC chroma frequency, or about 900 kHz. The other stripe pattern is the chroma subcarrier signal translated from normal 3.58 MHz to half that, about 1.8 MHz. The chroma sidebands run from about 1.2 to 2.4 MHz. In the model you probably have, these frequencies are separated by using band reject filters. In a redesign I did at Motorola, the frequencies are separated by linear-phase band-pass filters. These are much less sensitive to hue changes due to horizontal scan width and non-linearity. I'm not sure if any of the modified players got into the market. So far, every one I've seen has a schematic packed with it that indicates the older arrangement.

6) A little bit more about the chroma frame: The pilot signal is in phase with the NTSC
"I" signal, not color burst. This is so that intermodulation between the two sine waves (due to tolerances in film contrast) will not cause skin hue to deviate. It also means that the peak values of the combined waveforms are plus during skin tone and negative during cyan. Some baseband I signal is subtracted from the two sine waves (stripes) to compensate these extreme peak values so the film can handle them better.

7) The pilot waveform is doubled in frequency to be a frequency and phase reference to recover the chroma. They both vary together due to horizontal scan rate (width or linearity variation). By a sequence of frequency mixing steps, one of which involves a local 3.58 Mhz crystal oscilator, the variation in the chroma signal phase and frequency variation is cancelled out and is restored to a standard stable NTSC chroma signal. My later redesign used a first demodulation step using the doubled pilot, followed by remodulation on the crystal frequency. Either way works, but their are details about sensitivity to multiplier chip imbalances and so on.

8) You will see a "gamma" adjustment in each video channel. This is not the usual gamma correction in a video camera, but correction for deliberate stretching of the highlights in the recording. The idea is that compressing those highlights after scanning reduces the amplitude of CRT phosphor grain noise. Caution: it also affects the pilot/chroma intermodulation in a chroma frame, changing the hue of green and magenta colors.
__________________
www.bretl.com
Old TV literature, New York World's Fair, and other miscellany

Last edited by old_tv_nut; 01-12-2020 at 11:51 PM.
Reply With Quote