2
\$\begingroup\$

I need to somehow multiply a 15734.26Hz signal by 227.5 times. Those with a keen eye (or a calculator) will notice the result is very close to NTSC subcarrier frequencies. 3.579545MHz or close as damn it.

I realise I could just use a crystal that's that exact specification, and I have been, but the problem is, this crystal is not on the same domain as the rest of the video signal as I'm encoding RGB into composite video.

This is where the dreaded dot crawl comes in. If the subcarrier frequency is derived from the same domain, the dot crawl is gone completely.

I've no way to get on this clock domain BUT I know that the HSYNC x the line clock cycles = the subcarrier frequency. I just don't know how to get there.

Most ICs or circuits don't seem to be able to multiply that much and it's likely for a good reason looking at how they do it.

I assume I will need something that can sample the frequency and generate a clock, based on my criteria, from it.

Assumedly, I'm going to be looking at a microcontroller or FPGA which I'd really rather avoid.

Any ideas?

\$\endgroup\$
12
  • \$\begingroup\$ Multiply by 555 and then divide by two. You may have to multiply by a larger factor if there is a minimum oscillator frequency for you VCO. \$\endgroup\$ Mar 2 at 17:09
  • \$\begingroup\$ @KevinWhite Exactly 455. You could also use a prescaler to divide /2 and then multiply by 455. \$\endgroup\$ Mar 2 at 17:26
  • 1
    \$\begingroup\$ Take a look on PLL IC: MC145151. If the number is rational then you have to divide and then multiply if you want a single IC solution. It's the way the PLL works, it has an input prescaler then phase detector, VCO and then you divide the output frequency and feed back to phase comparator. \$\endgroup\$ Mar 2 at 17:43
  • 1
    \$\begingroup\$ I've used the Sil 532x for generating MHz clocks from similar frequency references. Something in that product line would likely work for you, and their evaluation boards are readily available if you want to test. \$\endgroup\$ Mar 2 at 18:08
  • 1
    \$\begingroup\$ Does this answer your question? How to get a high precision sine wave not available from standard crystal? \$\endgroup\$ Mar 2 at 18:18
5
\$\begingroup\$

There (was) an IC for that: https://www.renesas.com/us/en/general-parts/1523-video-clock-synthesizer-i2c-programmable-delay. Obsolete now... no thanks to Renesas.

Basically, multiply up by 455 to get 2fsc or 910 to get 4fsc (that is 14.31818MHz.) The latter is available as a crystal.

Another option is to 'crash lock' a crystal oscillator to regenerate the subcarrier. That's how old TVs did it. This IC does that: http://synkie.net/wordpress/wp-content/uploads/2013/11/MC44144P.pdf

Another option is forget about making Fsc, and go straight to CCIR656 sampling (13.5MHz) using a chip called a Digital Multistandard Decoder (DMSD). Here's one from TI that I've used in the past: https://www.ti.com/product/TVP5150AM1. You'll need to add a microcontroller to initalize it, but that's not so difficult nowadays.

\$\endgroup\$
2
  • \$\begingroup\$ Thanks for the options. The ICS1523 indeed looks like it would have done the job, though as it's obsolete and board space is minimal, that and a microcontroller would probably not fit too well. \$\endgroup\$ Mar 3 at 10:09
  • \$\begingroup\$ I'm not sure what use the MC44144P would be though, doesn't it only accept a composite video input? Or am I missing an alternate use for it? Not 100% sure what a "crash lock" is. Also no clue what CCIR656 is but will investigate, thanks \$\endgroup\$ Mar 3 at 10:11

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.