[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: IIgs RGB to VGA continued

jsnospam@cimmeri.com wrote:

The basic setup is:
AL875 ADC
AL250 scan doubler
ICS1523 clock generator in a PLL

We tried changing the dot clock sampling rate.   We noted many
interesting effects, but no overall image improvement.

Also considered phasing: it's done by the ICS clock generator chip.
The phasing is adjustable and we tried changing it throughout its range
and it made no difference.   Not during dot clock adjustment.

Now look at:
http://members.aol.com/mdnttrain/vga6.jpg

In this latest photo, a first order low pass filter was added on each
of the 3 color inputs.  This reproduces the lower bandwidth of the
older monitors and makes the modulation bars disappear.   We are
beginning to believe that the modulation bars are really coming out of
the machine like that and not a defect of the converter itself.    A
supporting observation is that the beat bars are quite "solid" looking
(real beat bars we've seen are soft).

Another is that they don't show up on the black and white portions of
the image as Vince noted.   That means that if the bars ARE present,
they are always white bars on white backround, and black bars on black
backround .. which seems inconsistent with the other bars we're seeing.

Do you all agree with our theory, or think something is still wrong in
the converter's phasing circuit / dot clock rate?


Your results prompt a little application of sampling theory.  ;-)

If non-synchronous sampling is to work, the bandwidth of the incoming
signal(s) must not exceed the Nyquist frequency, or half the sampling
frequency.

Since the RGB signals coming from the IIgs are not band-limited, I
would expect some pretty sharp edges, and therefore frequencies well
over 14MHz.  The patterns you see are the result of aliasing of those
frequencies to lower frequencies (the difference between the dot clock
frequency of the IIgs and the sampling frequency).

If you're sampling around 15MHz, then the bandwidth of the RGB signals
should not exceed 7.5MHz, or, for practical filters, 6.5MHz.

The bad news is that using a filter also kills the bandwidth required
for sharp 80-column text.  ;-(

That's why I still recommend *synchronous* sampling.

-michael

New, faster SUDOKU v2.0 solver for Apple II's!
Home page:  http://members.aol.com/MJMahon/

"The wastebasket is our most important design
tool--and it's seriously underused."