Simple crystal (4.43MHz - PAL) oscillator + PLL multiplier circuit wanted.

[Lionel]

Per the subject line. I want to experiment with direct PAL video
generation from a microcontroller, with the minimal number of parts.
I'm thinking that a good starting place might be to use a CPU clock
that's based on the PAL subcarrier frequency (4.43MHz) multiplied with
a PLL+VCO+Divider to give me phase reference points, & directly
generate the composite colour signal via an R2R ladder.

So, some questions:

1) Is this a reasonable concept? (I'm comfortable with per-cycle level
timing & coding)

2) Any tips/suggestions/thoughts as to a better approach?

3) What's a good (simple, cheap, stable) circuit to oscillate a
standard 4.43MHz PAL sub-carrier xtal? (Ideally just an inverter or
two)

4) Suggestions for a cheap + simple PLL frequency multiplier (x8? x4?)
for the above to give me multiple cycles per PAL subcarrier cycle?

5) How many analog samples do you actually /need/ to generate per
subcarrier cycle to get good quality PAL colour?

6) Any recommended online resources/references that cover direct
digital synthesis of medium to high quality PAL colour video?

Thanks in advance for any help or advice.

 

[martin griffith]

Per the subject line. I want to experiment with direct PAL video
generation from a microcontroller, with the minimal number of parts.
I'm thinking that a good starting place might be to use a CPU clock
that's based on the PAL subcarrier frequency (4.43MHz) multiplied with
a PLL+VCO+Divider to give me phase reference points, & directly
generate the composite colour signal via an R2R ladder.

So, some questions:

1) Is this a reasonable concept? (I'm comfortable with per-cycle level
timing & coding)

2) Any tips/suggestions/thoughts as to a better approach?

3) What's a good (simple, cheap, stable) circuit to oscillate a
standard 4.43MHz PAL sub-carrier xtal? (Ideally just an inverter or
two)

4) Suggestions for a cheap + simple PLL frequency multiplier (x8? x4?)
for the above to give me multiple cycles per PAL subcarrier cycle?

5) How many analog samples do you actually /need/ to generate per
subcarrier cycle to get good quality PAL colour?

6) Any recommended online resources/references that cover direct
digital synthesis of medium to high quality PAL colour video?

Thanks in advance for any help or advice.

Normally PAL/NTSC is samples at 4 times subcarrier. There are some
genlock ics around see here:
http://www.intersil.com/products/pt/parametric_table_10395.asp
You may have to find out all the joys of the "eight field sequence",
fortunately, I have forgotten most of it, it's messy.

Maybe you should think about working in YUV/component/RGB, which has
no subcarrier, but is sampled at 27M/sec, and is easier than PAL

List of chips: http://www.epanorama.net/links/videochips.html


martin

 

[David L. Jones]

Per the subject line. I want to experiment with direct PAL video
generation from a microcontroller, with the minimal number of parts.
I'm thinking that a good starting place might be to use a CPU clock
that's based on the PAL subcarrier frequency (4.43MHz) multiplied with
a PLL+VCO+Divider to give me phase reference points, & directly
generate the composite colour signal via an R2R ladder.

So, some questions:

1) Is this a reasonable concept? (I'm comfortable with per-cycle level
timing & coding)

A few people have had (limited) success with direct composite PAL
video generation directly from a micro, like this one:
http://www.rickard.gunee.com/projects/video/pic/gamesys.php
Ordinary quality, but shows that the minimum parts count is a PIC and
two resistors!
Much faster micros are available would obviously do a bit better than
the PIC.

Dave

 

[[email protected]]

A few people have had (limited) success with direct composite PAL
video generation directly from a micro, like this one:http://www.rickard.gunee.com/projects/video/pic/gamesys.php
Ordinary quality, but shows that the minimum parts count is a PIC and
two resistors!
Much faster micros are available would obviously do a bit better than
the PIC.

Note that his _PIC_ solution is BW only, he does color on a SX micro:
with an about 53MHz clock(12x Fc )
http://www.rickard.gunee.com/projects/video/sx/gamesys.php

There are several cheap PAL modulator chips, so a PIC and
for example a MC1377 modulator chip would be 'low chip component
count'.
MC1377 is old, and uses a 4.43MHz xtal, quality is good enough for
games.

The alternative is a PIC and a sparate 8.86 MHz (2xfc) oscillator,
some 4053 switches, some 4013 FF, some inverters and xor gates to make
a PAL modulator.
I'd use the MC1377
Or you need a very fast PIC !

 

[René]

er the subject line. I want to experiment with direct PAL video
generation from a microcontroller

In the dark recesses of my past, I remember owning a ZX80....

http://en.wikipedia.org/wiki/Sinclair_ZX80

 

[Tam/WB2TT]

......................................that a good starting place might be to
use a CPU clock

that's based on the PAL subcarrier frequency (4.43MHz) multiplied with
a PLL+VCO+Divider to give me phase reference points, & directly
generate the composite colour signal via an R2R ladder.

You might want to consider the converese of that. 20+ years ago it was
common in cheap computers here to start with an oscillator at 14.318 MHz,
which was then counted down to the NTSC color subcarrier frequency of 3.579
MHz. 14.318 is also an integral multiple of the horizontal and vertical sync
frequencies. I see that 17.734475 is a standard crystal frequency. Is that
4x the actual PAL frequency?

Tam

 

[Jan Panteltje]

In the dark recesses of my past, I remember owning a ZX80....

http://en.wikipedia.org/wiki/Sinclair_ZX80

I had one too, and it was BW.
It had a Z80, at 3,58 Mhz or so, and a custom ship with the shift register in it.
One of the best BASICS ever, the Z80 refresh register was also used for the
display.
It was followed by the ZX81 (Timex1000 in the US), and I still have the book
Timex1000 / ZX81 ROM dissasembly by Dr Ian Logan and Dr Ohara.
Even controlled an audio switcher with it....
ZX81 was BW too.
Later came the Spectrum, never had one, but _that_ one was color I think.

 

[Lionel]

.....................................that a good starting place might be to
use a CPU clock

You might want to consider the converese of that. 20+ years ago it was
common in cheap computers here to start with an oscillator at 14.318 MHz,
which was then counted down to the NTSC color subcarrier frequency of 3.579
MHz. 14.318 is also an integral multiple of the horizontal and vertical sync
frequencies. I see that 17.734475 is a standard crystal frequency. Is that
4x the actual PAL frequency?

Yes, it is. However, I don't know if it's high enough to give me
enough samples to create a clean Chroma, etc. I'm pretty sure that
I'll need to sample at a minimum of 8x (35.468950MHz) the base colour
subcarrier frequency. I /hope/ I won't have to go as high as 16x
(70.937900MHz).
That said, using 17.734475MHz means I'd only have to multipy it by 2
or 4 to get my higher multiples, which may make life easier.

 

[martin griffith]

Yes, it is. However, I don't know if it's high enough to give me
enough samples to create a clean Chroma, etc. I'm pretty sure that
I'll need to sample at a minimum of 8x (35.468950MHz) the base colour
subcarrier frequency. I /hope/ I won't have to go as high as 16x
(70.937900MHz).
That said, using 17.734475MHz means I'd only have to multipy it by 2
or 4 to get my higher multiples, which may make life easier.

http://www.nxp.com/pip/SAA7121H_V2.html
may be of interest


martin

 

[Lionel]

On Tue, 06 Mar 2007 05:37:18 +1100, in sci.electronics.design Lionel


http://www.nxp.com/pip/SAA7121H_V2.html
may be of interest

Unfortunately, it's not accessible:
------------
Forbidden
You don't have permission to access /pip/SAA7121H_V2.html on this
server.
Additionally, a 403 Forbidden error was encountered while trying to
use an ErrorDocument to handle the request.
------------

Fortunately, Philips have it on their shiny new marketeer-flavoured
website:
<http://www.nxp.com/pip/SAA7121H_V2.html>
---------
General description
The SAA7120H; SAA7121H encodes digital YUV video data to an NTSC or
PAL CVBS or S-video signal. The circuit accepts CCIR compatible YUV
data with 720 active pixels per line in 4 : 2 : 2 multiplexed formats,
for example MPEG decoded data. It includes a sync/clock generator and
on-chip DACs.

 

[Frank Raffaeli]

Per the subject line. I want to experiment with direct PAL video
generation from a microcontroller, with the minimal number of parts.
I'm thinking that a good starting place might be to use a CPU clock
that's based on the PAL subcarrier frequency (4.43MHz) multiplied with
a PLL+VCO+Divider to give me phase reference points, & directly
generate the composite colour signal via an R2R ladder.

So, some questions:

1) Is this a reasonable concept? (I'm comfortable with per-cycle level
timing & coding)

2) Any tips/suggestions/thoughts as to a better approach?

I've generated PAL and NTSC CVBS in a Cyclone II (5K) FPGA, $12.90 in
singles. I load the FPGA with a 0.68 micro. Total parts cost: ~ $16.00
in singles w/ 8-bit DAC, ~ $17.00 for 10-bit DAC. Parts total in the
1000+ would be about $8.00. USB is about 1.50 extra if you need
computer control.

3) What's a good (simple, cheap, stable) circuit to oscillate a
standard 4.43MHz PAL sub-carrier xtal? (Ideally just an inverter or
two)

Do you need genlock? If not, a simple xtal osc at 4x or 8x with
inverter will work fine.
Are you familiar with the PAL color subcarrier phase shift wrt the
field sequence?

Frank Raffaeli
http://www.aomwireless.com/

 

[Tam/WB2TT]

Yes, it is. However, I don't know if it's high enough to give me
enough samples to create a clean Chroma, etc. I'm pretty sure that
I'll need to sample at a minimum of 8x (35.468950MHz) the base colour
subcarrier frequency. I /hope/ I won't have to go as high as 16x
(70.937900MHz).
That said, using 17.734475MHz means I'd only have to multipy it by 2
or 4 to get my higher multiples, which may make life easier.

*****************
That would still simplify things because you could use a frequency doubler,
or two in series, and not have to worry about phase noise and such of a PLL.
A 2N2369A, or the like, biased class C will work nice as a doubler at these
frequencies. You would need a limiter on the output.

Tam
*************