1 Bit digitization of a NTSC Line

[LaserSteve]

Please suggest a comparator suitable for turning an analog video line into an on/off stream. Goal is to digitize one line of a field for a crude autofocus system by optimizing the number of high frequency pixel transitions per line. Ideally the output is TTL.   The subject is a patch of filter material under highly controlled illumination, imaged with a microscope objective.  The filter material consists of a non-woven  random web of  300 nm to 1.5 um Electrospun nanofibers, which are bright against a dark background.  My subject matter results in pixels that are pretty much, on or off over a limited grey scale to begin with.


 I already have the line selection and sync electronics done,  I just want a suggestion of a good comparator for the job in a easy to hand solder package (SOIC?) 

I have bought very few fast comparators in my career, hence the request.

Thanks, Steve

[T3sl4co1l]

So, less than microseconds will do?  NTSC isn't very much bandwidth!

Your classic LM393, 311, 319 will just do, but I'd suggest something MCP6561 or better.  I forget offhand what some common few-ns comparators are, but there's a big step from jellybean ~100s ns response, to "fast" <10ns response.  And you aren't paying much extra for it, no need to be shy.

Expect to need low voltage supplies, 5V or even less.  This should be perfectly fine with respect to composite video, and you can always divide down the signal level if need be.

Mind that you'll need B&W video (you're already well aware, I'm sure).  If color is present, it should probably be filtered out, or sent through a separator or decoder to get pure luminance.

Or using a digital camera, I'd bet you could manage something with shift registers and an Arduino.  Or a faster MCU with DMA acquisition.  Or SPI straight up sampling the raw data (in no particular order?) might even work.  Oh hey, nice thing about a parallel digital interface, you can just use the HSB directly...

...But I digress.

Tim

[LaserSteve]

Thanks,

      I have lots of  the versatile black and white Pulnix Tm-6 and  Tm-7  in the Cn and Ex variants, Thanks to Electronic Goldmine blowing them out for ~10$ a piece years ago.  I'm told they came from indexing a robotic  hard disk storage unit, evidently some company or agency had huge clusters of disk indexer robots.

      While outdated, they are my "go-to" cameras for proof of principle at home and the  University.  Having used them for things like field flipping stereo vision, spectrometer readouts, robotics etc.  They work well , sync perfectly, with  decent dynamic range. They  have all sorts of inputs and outputs brought out, have a internal daughterboard that is hackable for even more control, and just refuse to die.  When you can fit the programing on a thumbwheel switch and a few toggles, they are student friendly as well.  I think we cranked out nine doctoral theses  using them in one materials lab alone.

   This is a proof of principle, use a digital  scope or gated frequency counter on the output, and then ask for funding from the big company who wants the sensor  developed.  It already does a few useful  things other then just imaging the filter material.   Then I can branch into digital video as suggested.

The Microchip part looks like a nice jellybean to have laying around.

Thanks Again,

Steve

[T3sl4co1l]

Hah, cool!

Tim