By Date: <-- -->
By Thread: <-- -->

bttv 2.6.16: wrong VBI_OFFSET?

On Tue, 2 May 2006, Daniel [iso-8859-1] Glöckner wrote:
>
> HDELAY is specified as being relative to the falling edge of HRESET. And as
> we are told to set HDELAY to 10.5µs for PAL and 9.4µs for NTSC, HRESET
> seems to coincide with 0H.

In the functional desription of cropping, it says "For video decoding, HDELAY
is programmed with the number of pixels between horizontal sync and the first
pixel of each line to be displayed or captured." In the register description,
the wording is changed to "pixels between the falling edge of HRESET and the
start of active video."

> > An undocumented 7th bit that is set to 1, that does make some sense.
>
> Has someone contact to Conexant? I only found a literature request eMail
> address.

I wonder if Conexant even knows?  The description of VBI capture for the 23880
is just a copy and paste of the original Brooktree docs, even though the
register names, locations, and widths have changed.  The "regions of the NTSC
video frame" figure still is has some of the lines labled wrong.

> > Do you have updated information with a 2-bit
> > HSFMT that allows for an /HRESET width of 16 pixels?
>
> See
> http://www.conexant.com/servlets/DownloadServlet/100600B.pdf?FileId=443
> The 878A appears to have 2 bits.

That's updated from the Brooktree and Rockwell datasheets I was looking at.
My Brooktree sheet says HSFMT is 1 bit, 0 = 64 pixels, 1 = one pixel, and then
has a correction that 1 = 32 pixels.  The Rockwell sheet just has the 32/64.
That conextant sheet has a two bit (depending on which column you look at)
HSFMT.

> Btw. did someone check the behaviour of Bt878(no A), Bt848 and CX2388x?

I've checked my CX23883 card and it's not the same.  That datasheet only
describes a 1 bit HSFMT.  Changing HSFMT appears to have no effect at all.  Do
you know if some kind of chip reset is necessary to make a change to HSFMT
take effect?  Changing VBI_V_DEL has the expected effect.

I don't have a signal generator, but I tried capturing line 21 on a number of
different cable channels.  It looks like the offset isn't 256 pixels, but more
like 202 pixels.  See captures:  http://www.speakeasy.org/~xyzzy/vbi

The first bit after the run-in should start at 27.452 us, or 768 pixels at
Clk*2 from 0H.  It appears to start about 566 pixels in.

--
video4linux-list mailing list
Unsubscribe mailto:video4linux-list-request (at) redhat.com?subject=unsubscribe
https://www.redhat.com/mailman/listinfo/video4linux-list