Just curious... do you lose much signal quality with this approach? DVD
component video output runs something like 4-7MHz bandwidth with something
like 60-90dB SNR (depending on how good the DVD player is), which seem like
'high' enough numbers that you really would have to use high quality
(precision) resistors in your R2R ladder, pay careful attention to
timebases, sample/hold circuits, etc.?
Well, the 'high precision r2r' is made of 10k resistors from the same batch.
It was an experiment, and it worked, the intent was to find out if the input
comparator on a Xilinx Spartan XC2S200 FPGA could be used, and it can.
The SN is bad.
In this real thing I am using Philips TDA8708 AD (it can do 8 bits at
32 MHz).
And it needs no sample hold (like the r2r).
Sampling is then done at 3 x color sub carrier, or 6 (about 26.5 MHz).
The test work perfectly, but to also do the color I will sample Y Cr Y Br
with a HCT4053 CMOS switch, (first 3 low pass Nyquist).
Then Y is sampled effectively with 3 x Fc and the color difference signals
with half of that.
First test had a LM 1881 sync separator, but now automatic sync slicing
and H and V detection, and a 2 Fh PLL are working OK.
In my opinion(!) hehe the signal to noise of 8 bit video is sufficient.
Given the fact that you are probably watching this on a monitor driven
by a RGB video card that has (very likely) 3 x 8 bits DA.
Even if it was 10 bits DA, the software *I* write only does 8 bits RGB
(24 bits video), X11 (Linux Xwindows) does not support more then that,
I dunno of any system / video card that does actually.
(There is a 32 bit X11 mode, but the last byte is always zero).
Given the viewing distance (I am close up) on a 19 inch monitor you
will really see more detail then on a big projection LCD.
Using 720x576, but look at the angle, sit in front and measure the angle.
That, and the price of a new bulb, is why I have not bough a LCD projector
yet. (and bad black level / light distribution).
The real problems are in time base correcting composite VHS PAL,
not only is it noisy (VHS) what causes all sorts of problems in the very fast
FPGA.
You cannot simply shift chroma, as the color sub carrier regenerator
(4.43 MHz) in the receiver will go out of lock with even a few degrees phase
error per line.
So you can jump a whole period (360 degrees), but that is more then 200nS,
and causes visible skew (horizontal).
In the old day studio machines used a variable analog delay line.....
(Amtec Colortec for example Ampex), to do it digitally is a challenge.
Is an FPGA exercise for me
There is some commercial potential for a small (1 FPGA + some stuff)
converter, digitizer, the dream is to also do DivX coding in the chip, and
then Ethernet stream as output at 1500kbps... But I still have do do a lot
and probably need a bigger FPGA do do that.
Cheap mpeg2 coding chips become more available, so perhaps why bother....
In the FPGA the stability of the 2 x horizontal PLL must also be very very
good (no noise), as it seems these monitors have really fast horizontal sync
PLL response, the slightest noise gives noisy edges on the picture.
Imagine even a few nanoseconds jitter on 32 kHz oscillator is visible on
my Samsung syncmaster.
Jan
