[linux-dvb] Re: DVB Newbie

[Chris Pitchford <nova-t@intrepid.cx>]

I hope you dont mind me reposting to the list, Paul..

I actually don't have a monitor on the that has the dvb card in it.. all
the TV is played back across the lan. My laptop is my prefered playback
computer, and I've not (though I REALLY should) touched mplayer or most of
the other software on the box for a good while (6 months or so)

Its not just mplayer thats causing the problem..

I'm using the Crystal Palace transmitter in the UK, and I've actually just
moved house about 1.5miles nearer, its about 3 miles from my house and I
have a straight road to the transmitter, perfect for line of sight! :)

More investigation has shown two things:

Firstly, using :

tzap -r CHANNEL4 &
cat /dev/dvb/adapter0/dvr0 > c4.ts
dvb-mplex -t MPEG2 -p 2828 -q 2827 -o c4.mpg -i TS_STREAM c4.ts

will result in a good MPG file I can play with xine or mplayer.. BUT there
are errors after a few minutes.. infact the longer I leave the "cat"
command running, the more likely dvb-mplex will fail. Using ts2ps to play
the ts stream into mplayer again works for a few minutes then errors
appear and get progressively worse until mplayer dies.

tzap -r "BBC ONE" &
cat /dev/dvb/adapter0/dvr0 > bbc1.ts
dvb-mplex -t MPEG2 -p 601 -q 600 -o bbc1.mpg -i TS_STREAM bbc1.ts

will result in gibberish, infact dvb-mplex occasionally drops to the
prompt with or sometimes without errors:
A 10 second grab of video just now resulted in

  Scanning audio stream for access units information
Error in MPEG stream at offset (bits) 2232653: supposed marker bit not found.

Another test just now resulted in:

  Multiplexing file test.file.mpg
    video PTS (3497.00ms) underflow at pack 1051 by 25.94ms
    video PTS (3577.00ms) underflow at pack 1078 by 28.50ms
    video PTS (3617.00ms) underflow at pack 1085 by 9.91ms
    video PTS (3697.00ms) underflow at pack 1121 by 40.00ms
    video PTS (3737.00ms) underflow at pack 1128 by 21.41ms
    video PTS (6017.00ms) underflow at pack 1881 by 44.16ms
    video PTS (6057.00ms) underflow at pack 1895 by 46.98ms
    video PTS (7057.00ms) underflow at pack 2237 by 105.08ms
    video PTS (7097.00ms) underflow at pack 2244 by 86.49ms
    video PTS (7057.00ms) underflow at pack 2263 by 184.59ms
    video PTS (7097.00ms) underflow at pack 2271 by 169.06ms
    video PTS (7977.00ms) underflow at pack 2505 by 4.65ms
    video PTS (8097.00ms) underflow at pack 2546 by 10.03ms
    video PTS (8137.00ms) underflow at pack 2558 by 6.73ms
    video PTS (8097.00ms) underflow at pack 2578 by 107.89ms
    video PTS (8137.00ms) underflow at pack 2587 by 95.42ms
    video PTS (9897.00ms) underflow at pack 3135 by 11.26ms
    video PTS (9977.00ms) underflow at pack 3172 by 44.41ms
    video PTS (10017.00ms) underflow at pack 3185 by 44.16ms
    video PTS (9977.00ms) underflow at pack 3207 by 151.44ms
    video PTS (10017.00ms) underflow at pack 3213 by 129.79ms
    video PTS (13137.00ms) underflow at pack 4186 by 12.85ms
    video PTS (13177.00ms) underflow at pack 4197 by 6.49ms
    video PTS (13137.00ms) underflow at pack 4216 by 104.59ms
    video PTS (13177.00ms) underflow at pack 4223 by 86.00ms
    25 video underflows (SCR >= PTS or DTS)

ts2ps | mplayer produces 100% bad result..

I know BBC1 is a QAM16 channel.. whereas Chan4 isn't.. Ive tried dvbtune's
scan_uk-t.sh script and it fails to see anything on the qam16
multiplexes..

dvbtune has not yet reported block errors, I've even reduced the delay
between samples to see if I can spot an error, but no..

Also dvbstream really does not seem to work, even on channel 4. It
produces equally bad streams for any of the channels.. The problem is, I
have a simple set of shell scripts I have used since about october last
year and they have not changed.. they did work.. but now produce these
corrupted results.. I previously did nothing more than call dvbstream -o
-f<freq> <pid> <pid> with the -qam 16 -cr 3_4 for the qam16 mplexes..

This should be the most basic way to view the video..

Has work stopped on dvbstream and modifications to the driver now made it
incompatible?

Also, should I try running 2.5? On the ftp site it mentions that the
saa7146 code is being rewritten for 2.5 then being back ported to 2.4 when
proven stable. Is this now old news and the new re-written version been
merged in?

Does anyone know of any other things that might need adjusting in a system
to get the card to work? such as IO-Apic which is active on this machine
(its a single cpu p4)? Ive tried moving the dvb-t card between slots
(there are only three in the system, the graphics card is on board (and
not used)

Ive also tried piping the data across the network in a tcp stream incase
the disk io was causing the problem but the results were the same.

Does anyone know of any reason why the QAM16 channels would be so bad in
comparison to the regular multiplex which gets progressively worse the
longer the capture/playback?

Its all a bit confusing, especially since it worked almost perfectly at my
last address (but admittedly in a different computer, with an ancient OS
install of slackware 8.0)

Cheers

cp

On 19 Jun 2003, Paul Andreassen wrote:

> Hi, I'm a newbie too, so don't complain if I'm completely of track.
>
> >>From the dvdtune and tzap output it looks like your card is working
> fine.  My guess is mplayer is broken.  Try xine (or sinek) instead.  And
> also try saving a stream to your hard drive, stopping the DVB drivers
> and playing it back.  This should tell you if its the drivers or not.
>
> Paul
>
> On Thu, 2003-06-19 at 11:39, Chris Pitchford wrote:
> > Hi all,
> >
> > I was just wondering how things were progressing on this problem..
> >
> > Ive just reinstalled my machine after moving house and with the latest cvs
> > snapshots of the linux-tv driver, dvbtune, dvbstream and libdvb I cannot
> > get dvbstream to work using my old script. From the first frame rendered
> > by mplayer (which has not been upgraded, it runs on a laptop) the image is
> > corrupt with squeeking audio. Very different to my last experience at my
> > old address about 3 months ago.
> >
> > Anyway, I have tried using tzap to tune and cat dvr0 to get the ts stream
> > out and it does indeed work much better.. however, and this has been the
> > problem more or less from day one when I get my Nova-t last year, the
> > longer I leave the card running the greater the number of errors that will
> > occur until (after about 5 minutes) mplayer will die
> >
> > Running dvbtune -m occasionally shows blockerrors of +1 for the first line
> > when starting, but other than that its constant at zero..:
> >
> > Signal=11051, Verror=491, SNR=53199dB, BlockErrors=0, (S|L|C|V|SY|)
> > Signal=11051, Verror=494, SNR=53713dB, BlockErrors=0, (S|L|C|V|SY|)
> >
> > As I copied those lines, mplayer was flickering and "squeeking" quite
> > impressively! Ive seen errors caused in the stream when BlockErrors is not
> > zero, but now its constantly zero yet these errors become more and more
> > frequant as the minutes pass..
> >
> > I seem to be in the same boat and was wondering.. ah mplayer just died..
> >
> > @@@ libmpeg2 returned from sig11... (bad file?) @@@
> > A:18448.2 V:18448.2 A-V:  0.051 ct: -0.023  4127/4127  13% 15%  6.4% 0 0 0%
> > Broken frame at 0x3CCF30
> > A:18448.5 V:18449.1 A-V: -0.596 ct: -0.052  4151/4151  13% 15%  6.4% 0 0 0%
> > Successfully enabled DPMS
> >
> > .. was wondering if anyone had any idea why the card seems to gain
> > errors.. If it was a signal strenth problem surely it would be from the
> > moment mplayer started.. not good for 3 or 4 minutes, bad for another 2
> > then fatal 2 or 3 after that?  Is this something to do with the tuner?
> > possibly wondering off "lock"?
> >
> > Here's the first few lines from when tzap is started
> > # ./tzap -r CHANNEL4
> > using '/dev/dvb/adapter0/frontend0' and '/dev/dvb/adapter0/demux0'
> > tuning to 481833333 Hz
> > status 0d | signal 2b2b | snr d2d2 | ber 0000ffff | unc 0000000b |
> > status 1f | signal 2b2b | snr d0d0 | ber 000001f7 | unc 00000000 | FE_HAS_LOCK
> > status 1f | signal 2b2b | snr cece | ber 000001f4 | unc 00000000 | FE_HAS_LOCK
> > status 1f | signal 2b2b | snr d1d1 | ber 000001f1 | unc 00000000 | FE_HAS_LOCK
> >
> >
> > I was wondering if there was something I could tweak or look at. Could it
> > be faulty hardware? i'm using a standard Nova-t and really haven't had
> > sustaineed luck with it. It works for periods then these errors creep in..
> >
> > Any help would be greatly appreciated
> >
> > Cheers
> >
> > cp
> >
> > On 8 Jun 2003, --[ UxBoD ]-- wrote:
> >
> > > > Okay getting a little better now - but I have noticed that when mplayer bails on me
> > > > this is what is showing :
> > > >
> > > > @@@ libmpeg2 returned from sig11... (bad file?) @@@40  10%  5%  0.8% 0 0 9%%
> > > >
> > > > Any idea what causes this? would a dedicated MPEG card help this?
> > >
> > > > --
> > > >
> > > > // --[ UxBoD ]-- // 2.4.20-gentoo-r5 // Pentium(R) 4 CPU 2.53GHz //
> > > > // gpg --keyserver wwwkeys.pgp.net --recv-keys 56ED1CB5 //
> > > >
> > >
> >
> --
>
>



-- 
Info:
To unsubscribe send a mail to ecartis@linuxtv.org with "unsubscribe linux-dvb" as subject.