Xref: utzoo sci.electronics:7314 rec.video:7444 Path: utzoo!utgpu!watmath!att!pacbell!ames!amdcad!rpw3 From: rpw3@amdcad.AMD.COM (Rob Warnock) Newsgroups: sci.electronics,rec.video Subject: Re: sync signal generator Message-ID: <26659@amdcad.AMD.COM> Date: 8 Aug 89 07:07:15 GMT References: <89216.120136BHB3@PSUVM> <21056@cup.portal.com> Reply-To: rpw3@amdcad.UUCP (Rob Warnock) Organization: [Consultant] San Mateo, CA Lines: 77 In article <21056@cup.portal.com> mmm@cup.portal.com writes: +--------------- | I'm not really sure I understand your question, but if you want a source | of a composite sync signal, you can use the National Semiconductor MM5320 TV | camera sync generator chip (if they still make it). It outputs a | composite sync signal. It can be sync'ed up to an external signal. +--------------- Ah, yezzz... used it once in a project I was doing... Jameco (a local/mail_order distributor) still carries it; cost is about $12 each. But it's listed as a "closeout item, will not be reordered". (But it's been listed that way for over a year now...) +--------------- | ... The National chip is much cheaper, | unless you count the cost of a -12V power supply. | ...some tips: | 1) The outputs are TTL compatible, but they're not TTL. They pretty much | swing between near +5 and near -12. This is important to know if you are | designing a resistor summer to mix the sync with the video. +--------------- Or if you want to feed something other than TTL, like CMOS. The current that's sourced from the -12 supply can blow CMOS inputs (or even blow the whole chip by causing the CMOS to go into "latch-up"), so use a "real" TTL or LS-TTL chip as a buffer first. The National chip's outputs are actually *designed* to use the input undershoot diodes of the following TTL input as a clamp on the output voltage. (Ugh!) +--------------- | 2) This chip really surges on the power consumption at certain points in | its cycle. You need a good-size decoupling capacitor on the -12 supply | right near the chip. I used 22 uF. | 3) This chip runs hot. Take appropriate precautions. +--------------- Ottherwise, it works o.k., I guess... But for just those reasons, plus the uncertainty about future availability, I gave up using it and built a PROM/counter-based circuit instead. A naive version would use 455 x 525 locations (clocked at Color-Burst x 2), but you can do what the National chip does and clock at (CB * 4) / 7 = 2.045454286 Mhz which gives 130 clocks per H line, or 130 x 525 = 68250 locations in the PROM. [Note that "CB*4" is supplied on the bus of most IBM PC clones.] Since most EPROMs are 8 bits wide, you can use some of the extra bits that you wouldn't need for outputs to do some run-length compression, which lets you pack the pattern into a 4Kx8 EPROM (2732 or equiv). [A slightly simpler counter circuit uses more ROM but still gets it into a single 2764 EPROM, at about the same price. You also get an extra usable output, which I use for a "test pattern".] Using a 29C64 EPROM and 74HCTxxx counters, the power drain is pretty low. The total parts cost is *less* than the National chip, though the board area is greater. Besides, you get to completely choose the outputs. The outputs I needed were NTSC composite sync, NTSC composite blanking, start_of_line (similar to H_sync, but keyed at the edge of blanking), start_of_field (similar to V_sync, but keys on at line 22), and "test_pattern". (The other three bits go back to the first counter section and control how many clocks to repeat each output pattern.) If you feed the composite sync, blanking, and test pattern outputs to a simple resistive adder, you get an analog output monochrome composite-video test pattern. Using a bigger PROM, you could use an address bit to select between several patterns stored in the ROM, for example, between NTSC and PAL. More details if anybody really needs them... Rob Warnock Systems Architecture Consultant UUCP: {amdcad,fortune,sun}!redwood!rpw3 DDD: (415)572-2607 USPS: 627 26th Ave, San Mateo, CA 94403