[sdiy] Video Synth...

[Magnus Danielson]

From: np <np at planetarc.de>
Subject: [sdiy] Video Synth...
Date: Fri, 21 Dec 2007 22:21:41 +0100
Message-ID: <476C2E65.1060002 at planetarc.de>

Hej Nils!

> I'm playing with the idea to build a little video synth related circuits over the next days and I have a couple of questions. Hope you fine guys can help me even if it's a bit off-topic:
> 
> First problem: Signal generation for a PAL BAS signal.
> 
> For my video signal generator I need two sync-signals at relative stable frequencies. VSync is at 50Hz (well - that's easy). HSync at 15625 Hz.
> 
> I did some math and found out that I can get both frequencies by dividing down a 27Mhz crystal oscillator. That would be a cheap way to do this, but I haven't ever done anything that runs at these high frequencies. Will prototype board (the one with the dots, not the strips) work at that high frequencies or am I asking for trouble? For the frequency division job I plan to use the 74HCT4059.

Be sure to have the standard 100 nF ceramic caps decoupling the power of the
chips and you will be fine. Most importantly you want to ensure that the caps
sits very near the chip and connects with smallest possible traces over to the
power pins. The key is to minimize loop area between the chip and the
capacitor. For 27 MHz you will be fine if you just put it on the short-end of
the chip and make traces to go directly to the power pins.

As for the frequency itself, 27 MHz is a standard video frequency, so you can
get the classic CCIR 601 (now ITU-R BT.601) sample rates, which matches both
PAL and NTSC thanks to math magic. For the frequencies you required, you can
do with a significantly lesser frequency if you only require PAL compatibility.

> Is there a simpler way to get my frequencies? Any idea that gets me out of the need to do HF stuff is welcome. Maybe I've overlooked another crystal that runs at a lower frequency, or I could do things with PLLs, or there is a even simpler solution. I don't want to use a microcontroller btw.

Your frequencies are

f      = 25 = 5^2
 frame

f      = 50 = 2*5^2
 field

f      = 15625 = 5^6
 line

Common term is 5^2 so you end up having the multiples of 1, 2 and 5^4 and a
common source for these is 2*5^4 thus, the lowest common frequency becomes

f    = 2*5^4*5^2=2*5^6 = 31250 Hz
 com

f      = f    / n      => n      = f    / f      = 2*5^6/5^2 = 2*5^4 = 1250
 frame    com    frame     frame    com    frame

f      = f    / n      => n      = f    / f      = 2*5^6/(2*5^2) = 5^4 = 625
 field    com    field     field    com    field

f      = f    / n      => n      = f    / f      = 2*5^6/5^6 = 2
 line     com    line      line     com    line

Recall that both PAL and NTSC is interlaced, so the two fields only hold half
of the lines in each frame.

> Next problem: for the image synthesis I need a bunch of two quadrant multipliers. I've looked up the datasheet of the SSM2024 as I have lots of them lying around here but I found nothing about their slew rate in the spec. _That_ makes me wonder how they perform at higher frequencies at all.

It only tells you that it was not a prioritized area since their main target
was the audio products. You should be fine.

> The worst thing they will do frequency wise is a VCA job for a 15khz saw-wave. I do not want much ringing and smearing at the sawtooth reset though.

If you use a high frequency like 27 MHz, your sawtooth can be the counter
output through a DAC. 13,5 MHz is a particularly nice frequency to use as a
step in that equals the length on the screen matching the heigth of a line.
You might want that. That divisor would be at 864 or 1728 if you go from
27 MHz.

>  From experience, can you tell me how high the SSM2024 can practically go?

Don't have a SSM2024 to play with. I'd expect several MHz of BW, but then I
would expect that to change with control current. I also expect it to depend
on the hookup and especially the capacitive output loading.

In general, I recommend you to learn the more HF stuff as they will become a
limiting aspect for you.

Cheers,
Magnus