Can we discuss DC restoration circuits?

[ChrisW6ATV]

Originally Posted by old_tv_nut

No, it's stated backwards, as I explained.

Oops, you are right, good catch. Somehow I missed the description of black level going UP in higher-contrast scenes, which is something I have not ever seen.

Miniman82-

Yes, you would need to build a circuit that would be driven by the LM1881.

Thanks for the posts, Jerome.

[miniman82]

I figured.

That being the case, the problem now becomes where to get the correct voltage to clamp the signal to? I read somewhere that to do it accurately/reliably enough, you would need a DAC. Idea is that you use a microcontroller to set the DAC to an accurate black level, then use the LM1881 to clamp the signal to the DAC generated voltage. The McGyver in me says there's another way, though.

The article I referenced in my first post mentions a comparator in the feedback loop of the video amp: why not use the LM1881 to trigger the comparator, thereby stabilizing the black level before the beginning of each horizontally scanned line?

[cbenham]

Originally Posted by miniman82

why not use the LM1881 to trigger the comparator, thereby stabilizing the black level before the beginning of each horizontally scanned line?

This may work, however you might need to experiment with the value of R set on the LM 1881.

Its nominal value is 680K for NTSC signals, but I have had problems with the resulting pulse being slightly too wide to properly drive a clamp. I've reduced R set to 470K or 560K with better results.

The pulse must start and end during the back porch time and must not extend past the end of the H back porch.

If it does, the clamp will be clamping to a voltage other than ground, the usual voltage level of the back porch portion of the waveform.

Hope you are successful.

Cliff

[miniman82]

Originally Posted by cbenham

This may work, however you might need to experiment with the value of R set on the LM 1881. Its nominal value is 680K for NTSC signals, but I have had problems with the resulting pulse being slightly too wide to properly drive a clamp. I've reduced R set to 470K or 560K with better results.

Cool, thanks for the heads up! Can you tell me what set you are working with, and what circuit you drove with the LM1881?

The pulse must start and end during the back porch time and must not extend past the end of the H back porch.

The LM1881 datasheet lists a 4µs gate time, which is shorter than the 4.7µs alotted for the entire pulse. I assume this is so what you are describing does not happen, but naturally there will be some setup with an o-scope involved to get things perfect.

If it does, the clamp will be clamping to a voltage other than ground, the usual voltage level of the back porch portion of the waveform.

See, this is where I get confused. If clamping to ground were sufficient, everyone would be doing it. However, my understanding is that the true black level is not at ground potential, but rather slightly above it.

http://en.wikipedia.org/wiki/File:Composite_Video.svg

In the above illustration, you can see that there is a blank level at .285v (the so-called 'blacker than black' level), and then slightly above that you have the 'true' black level at .339v (7.5 IRE). This is where studio monitors are calibrated to:
http://www.outside-hollywood.com/200...tudio-monitor/
So is it also the voltage you would ideally clamp to, or am I wrong?

[old_tv_nut]

Originally Posted by miniman82

Cool,
See, this is where I get confused. If clamping to ground were sufficient, everyone would be doing it. However, my understanding is that the true black level is not at ground potential, but rather slightly above it.

http://en.wikipedia.org/wiki/File:Composite_Video.svg

In the above illustration, you can see that there is a blank level at .285v (the so-called 'blacker than black' level), and then slightly above that you have the 'true' black level at .339v (7.5 IRE). This is where studio monitors are calibrated to:
http://www.outside-hollywood.com/200...tudio-monitor/
So is it also the voltage you would ideally clamp to, or am I wrong?

I'm sure Cliff meant "blanking (back porch)" rather than ground. Fact is, there is no place in the signal where there is an NTSC black to clamp to. PAL solved this by making black equal to the back porch level. In an NTSC receiver, you must also clamp to back porch, and then adjust the black level down slightly to be correct. I repeat: it is not totally terrible to clamp to the sync tips, it may just require occasional adjustment when changing channels/sources. The back porch clamp is necessary in pro gear, since there will not be someone constantly adjusting for changes in incoming sync amplitude.

Regarding the clamping pulse width: the value in the IC is likely a compromise between burst gate and clamping pulse width. Ideally, the burst gate should be wide enough to include the whole burst, while the clamp pulse should be narrow enough to never include sync or active video. So, changing the value as Cliff suggests is a good idea.