More on Video Preemphasis and Deviation Calibration

by

J. R. Mathison, WB9OQM

GO TO Home Page

My following references to European standards is actually to CCIR Recommendation 576-1 and my references to North American standards are to EIA/TIA 250-C. Make special note that the European standards for all measurements on 'video' is not composite as it does NOT include sync. The North American standard for measurement on 'video' is composite. Component values contained in the ATVQ figure on page 17 indicate CCIR Recommendation 405-1 for the preemphasis standard.

In the figure containing the graphs of curves on page 19, the deemphasis curve is indeed the compliment of the preemphasis curve. However, the fact that it had to be shifted upwards by approx 6 db in order to get it to cross the preemphasis curve at 762 Khz should indicate to us that something is wrong! That something is the fact that the crossover point is NOT at 762 Khz. This is a myth so prevalent that it is believed by many experts in FM TV. The crossover is at 400 Khz and actually has no significance for FM TV deviation calibration procedures. The 762 Khz point on the curve is called the 'relative 0 db reference level' which only has deviation calibration significance for European standards. Any other significance for the 762 Khz point is that it is a published point on the preemphasis curve.

A sine wave does not represent the complex video waveform. A much more appropriate waveform is a square wave with rise and fall times of 125 nanoseconds. The 125 ns is the standard for the pedestal - white transitions of both European and North American standards. The square wave should be 100/140 volt or 0.715 volt pk-pk assuming a standard composite video level of 1.0 volt pk-pk. The pulse and bar with sync from a video test signal generator is even better. Be sure to strap the unit for 125 nanoseconds rise and fall times.

A sine wave which has significance for preemphasized video is 1.0 volt pk-pk at 2.333 Mhz. It is no coincidence that this test tone at 1.0 volt pk-pk goes through the preemphasis network at exactly the same level as the 0.715 volt square wave. Yes, the passive preemphasis network has gain!. It is a differentiating network and has gain for the square wave. If you display the resulting waveform on a waveform monitor, set the bottom edge of the square wave portion to 0 IRE, and you will find the top edge of the square wave portion is at approx 23 IRE and the leading edge is a narrow spike standing up to 80 IRE and the trailing edge is a spike dropping down to minus 60 IRE. The preemphasis network has a voltage gain of 140/100 for the 0.715 volt pk-pk square wave making it exactly equal to the 1.0 volt pk-pk at 2.333 Mhz sine wave through the preemphasis network. One must take special note that the addition of sync to the above square waveform will NOT cause the peak to peak amplitude to increase nor the deviation to increase. See my drawing in Fig 1 (A) for a video signal generator waveform prior to preemphasis. Fig 1 (B) is this waveform after passing through the preemphasis network.

It is also no coincidence that the first Bessel null of the 2.333 Mhz test signal represents a deviation calibration of 5.6 Mhz peak which is the 'standard' for composite video under European standards whether preemphasis is used or not. Those of us who work with commercial FM TV have been using that test signal for years to calibrate microwave radio systems which carry FM TV under European standards, which are extensively used even in this country. The deviation for video with sync removed is 100/140 of the 5.6 Mhz peak or 4.0 Mhz. The deviation of the preemphasized signal does not decrease from the 5.6 Mhz peak even if the sync is removed because the presence or not of sync does NOT change the peak to peak amplitude of the preemphasized signal. The problem now is how does one get 4.0 Mhz peak deviation for 'video' when removal of sync changes nothing? One of my expert advisors for a previous article was a high ranking official on the EIA Subcommittee for Television Transmission Systems. This gentleman had been a member during the early 1970's when North American standards adopted CCIR Recommendation 405. He said it was the need to get 4.0 Mhz peak for 'video' out of the above which led to the 'relative 0 db reference point' on the preemphasis curve. He also said it was an attempt at comparison of a sine wave to a square wave. A 762 Khz test signal through the preemphasis network is 100/140 of the square wave or composite video or the 2.333 Mhz test signal. 100/140 of 5.6 Mhz is the necessary 4.0 Mhz peak deviation. Had the European standard for measurements on 'video' included the sync, this devious 'relative 0 db reference level' would have been unnecessary. It is the cause of universal misconception concerning the true nature of the deviation of an FM carrier modulated by 'video'.

ATVQ page 18 describes a well known deviation calibration procedure for FM TV. The first Bessel null of the 0.46 volt pk-pk at 762 Khz test signal through the preemphasis network yields a deviation of 1.833 Mhz peak. Increasing the level to 1.0 volts pk-pk causes a deviation of 4.0 Mhz peak for the 762 Khz test signal. A test signal of 1.0 volts pk-pk at 2.333 Mhz through the preemphasis network will be 140/100 higher or 5.6 Mhz peak deviation. This is exactly the same result as the calibration for European standards which I described above. The results for the square wave or for composite video are also the same. Why not use the first Bessel null of the 2.333 Mhz test signal to begin with? Why are we using European standards at all when North American standards are more conservative of RF spectrum?

Preemphasis for video is fully as important as an RF bandwidth reduction scheme as it is for noise reduction. Video as an asymmetrical waveform has a DC component. The change from a white picture to a black picture changes the DC component by approximately 0.54 volt. The deviation calibration described above results in a deviation of ll.2 Mhz pk-pk per volt. The change in DC component causes a deviation of approximately 6 Mhz pk-pk in addition to that required by a static picture. Use of preemphasis will reduce the level of the change in DC component by a factor of approximately 6.2 and the modulation by the DC component to approximately 960 Khz pk-pk. This is a savings of over 5 Mhz pk-pk deviation and approximately that much in RF bandwidth. This is a very significant RF bandwidth reduction for the dynamic waveforms of video moving pictures.

ATVQ pages 17 and 18 contain a 'formula' for RF bandwidth. This is actually 'J.R. Carson's Rule' which is a 'rule of thumb'. In this case it is presented as a mathematical formula yielding results correct to a one-tenth decimal place. This 'formula' would also lead one to believe that it is the high frequency components of video which cause the widest RF bandwidth. The numbers which we would plug into this 'formula' are the same as above whether preemphasis is used or not. We find that it is actually the reduction in the lowest frequencies (by using preemphasis) which cause the RF bandwidth reduction NOT the highest frequencies as implied by this 'formula'. Its application is more suited to a baseband signal which has more nearly equal amplitudes of sine waves from its lowest to highest frequencies and no changing DC component. Its use may accurately predict RF bandwidth in some cases but be in significant error in others.

Please see my articles Comments on Commercial FM TV Standards in ATVQ-Winter 1994 and FM Video Deviation in ATVQ-Spring 1994 for more on the above.

Please send E-Mail to: mathison (aatt) sdf-eu.org

.

Last revision 2012-12-15