On Thu, 2009-07-23 at 17:19 -0400, Tom Ryan wrote:
> Well, it worked good for a day and then started sync issues again. We
> have narrowed the issue down to the particular device connected to the
> analog composite input (a sony SSC-DC374 camera). If we connect a vcr
> or a sony handycam to the cable, all is perfect. If we connect a tv
> (instead of the card) to the cable, the output from the SSC is
> perfect.
> 
> The camera has an AGC (automatic gain correction) as well as a level
> adjustment. We have tried adjusting them to various levels and
> settings to no avail. (We have not worked the full gamut though :)
> 
> Any ideas, an option to the module load, etc that we're missing???

There are no module options to set for this, nor user controls.

The horizontal sync problem that I am aware of is rare but it does
happen.  Normally the first few lines of a frame, espcially when going
from a dark image to a light image, will not sync, but most lines of the
image will be OK.  It arsies due to an unfortunate combination of
circumstances.  It usually exhibits a few "false" syncs, not usually
complete loss of hsync.  It can be fixed with adaptive reconfiguration
of the AGC and +12 dB boost in the CX23418 A/V core.

Complete loss of horizontal sync is something that I would not expect,
unless you have a really weak signal or one with a large DC offset.

If you look in cx18-av-core.c:cx18_av_initialize() there is a comment
that states the droop compenstation, sync tip clamping, and extra +12 dB
is disabled for all three analog input channels.  Turning one or more of
those on may fix your problem.

The analog front end is reconfigured on every input switch near the end
of cx18-av-core.c:set_input(), BTW.

So:

1. What card are you using?
2. What driver version are you using?
3. How did you "up the gain on the cards" before?
4. What is the output of v4l2-ctl -d /dev/videoN --log-status when you
don't have horizontal lock and when you do?
5. Do you have an oscilliscope to look at the video signal coming out
from the SSC-DC374 to compare it with a known working video signal?
(It'll take the guessing out of troubleshooting)

You may also want to try the quick hack of a patch below.  It will make
tuning operations and input switches take ~1/3 of a second though.


Regards,
Andy

diff -r 892610bec939 linux/drivers/media/video/cx18/cx18-av-core.c
--- a/linux/drivers/media/video/cx18/cx18-av-core.c     Wed Jul 22 20:24:53 
2009 -0400
+++ b/linux/drivers/media/video/cx18/cx18-av-core.c     Thu Jul 23 20:25:20 
2009 -0400
@@ -560,6 +560,16 @@
                v |= 0x10;
                cx18_av_write_expect(cx, 0x803, v, v, 0x1f);
        }
+
+       cx18_msleep_timeout(300, 0);
+       v = cx18_av_read(cx, CXADEC_DFE_CTRL1);
+       if (v > 0x35) {
+               /*
+                * Variable Gain Amplifier adjustment settled at > +8.38 dB, so
+                * enable +12 dB for channel 0: CVBS or Y input
+                */
+               cx18_av_and_or4(cx, CXADEC_AFE_CTRL, 0xFFFFFFFE, 0x00000001);
+       }
 }
 
 static int cx18_av_s_frequency(struct v4l2_subdev *sd,





> Thanks!
> 
> Tom
> 
> 
> -----Original Message-----
> From: Tom Ryan
> Sent: Wed 7/22/2009 1:52 PM
> To: [email protected]
> Subject: Re: cx18: horizontal sync issue with analog composite 1 input
> 
> Sorry to post so soon, I fixed it by upping the gain on the cards.
> 
> Thanks!
> 
> 
> On 7/22/09 1:08 PM, "Tom Ryan" <[email protected]> wrote:
> 
> > We have 3 cx18 cards installed in a rhel5 box with the latest
> drivers
> > installed.
> >
> > Audio recording is ok, but the video appears to have no horizontal
> sync.
> >
> > I have searched the archives as well as the ³normal² places and have
> came up
> > short.
> >
> > Any suggestions?
> >
> > Thanks.
> >
> > Tom
> 
> 
> 
> 
> 
> _______________________________________________
> ivtv-users mailing list
> [email protected]
> http://ivtvdriver.org/mailman/listinfo/ivtv-users


_______________________________________________
ivtv-users mailing list
[email protected]
http://ivtvdriver.org/mailman/listinfo/ivtv-users

Reply via email to