bieling@terra.es wrote:
Michael J. Mahon wrote:Jorge ChB wrote:Michael J. Mahon <mjmahon@aol.com> wrote:I'd suspect that the colour burst being absent in the middle of a frame would produce pretty bizzarre results.Like for example ? I ran the program above and nothing "bizarre" came on the screen.I didn't say something bizarre *would* happen, only that it *could* happen, because it is completely outside the NTSC specification.Yep, the most "bizarre" thing that could happen is a switch from color to monochrome, or viceversa...No--actually, loss of color sync could result in some "rainbow" displays for part of a frame as the monitor color reference oscillator tries to get back into phase lock. The "trajectory" of phase lock during reacquisition can be chaotic, so, with the "right" timing, wierdness can happen. Some monitors may incorporate circuitry to detect solid phase lock of the reference and disable chroma processing until it is achieved, but I'd be surprised to find this in non-professional monitors.So you think that 8 cycles is not enough to achieve phase lock ? Then how does it phase lock if there are *just* 8 cycles in the color burst ? You think that it takes several lines/bursts to achieve phase lock ? Because if not, it really doesn't matter how much "chaotic" the phaselock is, as long as it has completed by the end of the color burst...
Yes, it takes several lines to achieve stable phase lock. The uncertainty principle requires that a 13-cycle burst only provides enough information to sync within 1/13th of the burst frequency, so it takes correlation over a longer interval to achieve good lock. The color reference oscillator in an analog monitor is a servo controlled by a gated reference signal. It is only by making multiple corrections to the phase that the oscillator is brought into stable lock. This could range from a few lines to many lines, depending on the initial state and the parameters of the servo loop. The bandwidth of the phase locked loop control signal is relatively long, averaging over several lines. There is a relationship between the time constant of this loop and the acquisition time and phase stability of the resulting lock. If the loop time constant is short, lock can be achieved more quickly, but the phase of the oscillator will be more perturbed by noise, resulting in more phase noise. If the loop time constant is long, acquisition time is longer, but the phase noise of the resulting lock is lower. Sometimes PLLs provide variable time constants, so that the loop can adapt quickly to a change in input, but then settle with low noise on a stable signal. -michael Parallel computing for 8-bit Apple II's! Home page: http://members.aol.com/MJMahon/ "The wastebasket is our most important design tool--and it is seriously underused."