[linux-dvb] A/V sync issues with DVB-S

[linux-dvb] A/V sync issues with DVB-S

[Thomas Börkel]

HI!

Is it normal, that every now and then (1-2 of 10 recordings), I get one
serious (suddenly 30 frames off) A/V sync problem with DVB-S?

I am seeing this with TT 1500 (budget_ci) and Nova-S Plus (cx88). I
tested kernel 2.6.23 and 2.6.25.

OK, when playing this recording with mplayer, I see no glitch and no
warnings.

But when I play it with MythTV, it stutters for a few seconds when it
tries to softly match the sync.

I haven't seen this with DVB-C, so I'd like to know if something is
wrong with my SAT system.

Any help would be greatly appreciated.

Thanks!

Thomas

_______________________________________________
linux-dvb mailing list
linux-dvb@linuxtv.org
http://www.linuxtv.org/cgi-bin/mailman/listinfo/linux-dvb

Re: [linux-dvb] A/V sync issues with DVB-S

[hermann pitton]

Hi,

Am Montag, den 18.08.2008, 07:13 +0200 schrieb Thomas Börkel: 
> HI!
> 
> Is it normal, that every now and then (1-2 of 10 recordings), I get one
> serious (suddenly 30 frames off) A/V sync problem with DVB-S?
> 
> I am seeing this with TT 1500 (budget_ci) and Nova-S Plus (cx88). I
> tested kernel 2.6.23 and 2.6.25.
> 
> OK, when playing this recording with mplayer, I see no glitch and no
> warnings.
> 
> But when I play it with MythTV, it stutters for a few seconds when it
> tries to softly match the sync.
> 
> I haven't seen this with DVB-C, so I'd like to know if something is
> wrong with my SAT system.
> 
> Any help would be greatly appreciated.
> 
> Thanks!
> 
> Thomas
> 

DVB-C is in general much better protected against troubles generated by
changing environmental conditions, but in case of weather conditions
which suddenly create ice on any surface, it is better to know the one
in charge even there.

Means, watch your dish. Maybe birds or a cat like to stay there
sometimes or wind might move branches of a tree in the near of it.

Next, for my still very limited DVB-S experiences, get an idea about
which transponders are critical for you. Kaffeine for example annotates
the SNR to them during a scan and you always can check if you have some
needed minimum when you start to record.

This is card/frontend specific, but for the very few I have in critical
reception conditions, SNR below 50% is a clear indicator that I won't
have to blame any apps later. A thunderstorm comes up.

After that, on normal mpeg2 DVB-S all applications should still do
fairly well, but if you go out for x264 HDTV, it is another and
completely different additional story, especially for what I see on
x86-64 quad CPUs.

Mostly OT, but since I did spent time on it recently, maybe have it too.

Compiling with CPU optimizations can make it even worse and concerning
multithreading the default one thread some have is for sure not enough,
but in best case eight at once might come through, with 32 all is in
vain and so on. 

However, kaffeine for recording and xine for playback from that file do
work. Every 15 to 20 minutes I have to hit break/play once to restore
suddenly lost A/V sync, when it becomes too much.

But it looks even worse on files recorded with full GPU hardware
acceleration on vista trying to play them at GNU/Linux. OK, both play
without visible issues on vista.

With that I can live already very comfortable now with 1080i non S2
HDTV ;) and don't have to reboot to the other stuff anymore.

Cheers,
Hermann
















_______________________________________________
linux-dvb mailing list
linux-dvb@linuxtv.org
http://www.linuxtv.org/cgi-bin/mailman/listinfo/linux-dvb

Re: [linux-dvb] A/V sync issues with DVB-S

[Thomas Börkel]

HI!

hermann pitton wrote:

Thanks for the informations.

>> Is it normal, that every now and then (1-2 of 10 recordings), I get one
>> serious (suddenly 30 frames off) A/V sync problem with DVB-S?

> DVB-C is in general much better protected against troubles generated by
> changing environmental conditions, but in case of weather conditions
> which suddenly create ice on any surface, it is better to know the one
> in charge even there.

OK.

> Means, watch your dish. Maybe birds or a cat like to stay there
> sometimes or wind might move branches of a tree in the near of it.

There is no tree, but could be birds.

> This is card/frontend specific, but for the very few I have in critical
> reception conditions, SNR below 50% is a clear indicator that I won't
> have to blame any apps later. A thunderstorm comes up.

Measuring the strength and SNR is not so easy now. My SAT receiver
always says strength 70% and SNR 80%. My new Nova S Plus cards always
report strength 97% and SNR 99%. Only my TT 1500s reported reasonable
numbers like strength 63-80% and SNR 75-90%.

In the end, I think overall strength and SNR is not my problem. I always
get a lock. It's the glitches that happen every now and then.

> After that, on normal mpeg2 DVB-S all applications should still do
> fairly well, but if you go out for x264 HDTV, it is another and
> completely different additional story, especially for what I see on
> x86-64 quad CPUs.

I am not recording x264 at the moment, meaning I have no DVB-S2 card.

Thanks for the informations.

Thomas


_______________________________________________
linux-dvb mailing list
linux-dvb@linuxtv.org
http://www.linuxtv.org/cgi-bin/mailman/listinfo/linux-dvb

Re: [linux-dvb] A/V sync issues with DVB-S

[Antti Palosaari]

Thomas Börkel wrote:
> hermann pitton wrote:
> Measuring the strength and SNR is not so easy now. My SAT receiver
> always says strength 70% and SNR 80%. My new Nova S Plus cards always
> report strength 97% and SNR 99%. Only my TT 1500s reported reasonable
> numbers like strength 63-80% and SNR 75-90%.

What is the recommended way to report signal strength or SNR by the 
driver? Both of those are 16bit unsigned value. Is it better to scale 
0-0xffff or return as decibels? If SNR is reported as decibels it will 
look like very small value...

Antti
-- 
http://palosaari.fi/

_______________________________________________
linux-dvb mailing list
linux-dvb@linuxtv.org
http://www.linuxtv.org/cgi-bin/mailman/listinfo/linux-dvb

Re: [linux-dvb] A/V sync issues with DVB-S

[Steven Toth]

Antti Palosaari wrote:
> Thomas Börkel wrote:
>> hermann pitton wrote:
>> Measuring the strength and SNR is not so easy now. My SAT receiver
>> always says strength 70% and SNR 80%. My new Nova S Plus cards always
>> report strength 97% and SNR 99%. Only my TT 1500s reported reasonable
>> numbers like strength 63-80% and SNR 75-90%.
> 
> What is the recommended way to report signal strength or SNR by the 
> driver? Both of those are 16bit unsigned value. Is it better to scale 
> 0-0xffff or return as decibels? If SNR is reported as decibels it will 
> look like very small value...
> 
> Antti

May of the drivers I've added recently report their values in db, but I 
typically also have one precision point. So I report 20.4 as 204 decimal.

- Steve


_______________________________________________
linux-dvb mailing list
linux-dvb@linuxtv.org
http://www.linuxtv.org/cgi-bin/mailman/listinfo/linux-dvb

Re: [linux-dvb] A/V sync issues with DVB-S

[hermann pitton]

Hi,

Am Mittwoch, den 20.08.2008, 10:07 -0400 schrieb Steven Toth:
> Antti Palosaari wrote:
> > Thomas Börkel wrote:
> >> hermann pitton wrote:
> >> Measuring the strength and SNR is not so easy now. My SAT receiver
> >> always says strength 70% and SNR 80%. My new Nova S Plus cards always
> >> report strength 97% and SNR 99%. Only my TT 1500s reported reasonable
> >> numbers like strength 63-80% and SNR 75-90%.
> > 
> > What is the recommended way to report signal strength or SNR by the 
> > driver? Both of those are 16bit unsigned value. Is it better to scale 
> > 0-0xffff or return as decibels? If SNR is reported as decibels it will 
> > look like very small value...
> > 
> > Antti
> 
> May of the drivers I've added recently report their values in db, but I 
> typically also have one precision point. So I report 20.4 as 204 decimal.
> 
> - Steve
> 

please allow me to abuse the thread a little for some status I seem to
have on saa7134 and DVB-S so far, not going into x264 issues.

Can only report for the saa7134 driver and DVB-S and what we have for
the dual triple Medion Quad Creatix/944 and the new single triple
Creatix/CTX948 with tda10086/tda8283 and isl6405 dual.

If SNR goes down 50% caused by external influences on some few
transponders (1 or 2), artifacts will become visible from time to time
in the preview. This has effects on the apps trying to keep A/V sync on
the recorded immediately too and can be seen with verbosity enabled for
them in more details.

The apps do differ a lot, how well they deal with the mess, but in
general without such mess for mpeg2 they have no problems.

My external DVB-S receiver is at the very same point with loopthrough to
second DVB-S connector and to the external receiver of the Medion Quad
from its first, which really can actively control voltage and tone.

It was exactly the other way round, before we had support for the
isl6405. Worked fine too. (A new PC had the mobo then upside down on the
left, helped to become aware of the loopthrough ;)

Since the external box runs almost without much buffering, one can see
major glitches there at first and they will happen in the buffers too a
little later on our drivers. This is fully in sync, just that delay.

This works very reliable here.

However, signal strength and SNR reported differ for the external box
too, vista is even worse on BDA drivers and always reports 100% for both
until you are really off and then seems to display some fake.

Also the late 2006 driver seems to behave exactly like what we have now
on this dual triple Medion Quad. Others did say it is all fine
previously.

On the second DVB-S device without i2c-access to the isl6405, powering
it up from the first to 18 Volts and on close disable it works fine.
But the attempted voltage/tone switching in between does nothing on that
one.

On the late 2007 driver they seem to have given up on the second DVB-S
and nothing at all happens any more. That recent .inf is also only
listing the first 16be:0007 as DVB-S capable anymore.

Since we still have loopthrough by default enabled on the tda8263, it is
still usable so far.

For the record, the internal DVB-S loopthrough has some loss for what I
can see, but this does not count for what goes through to an external
receiver. It was the other way round previously the same.

Still no high symbol rates here for testing, but the Oliver/Manu/Hartmut
patch for those is confirmed by others to be well done.

Also for the record, it might take years before anyone comes back to
it :), the manual DVB-T/TV analog switch distributes the RF signal
evenly to both tuners, means works much better than a passive splitter,
don't know why. No external LNA found. Likely the assumed slightly
better splitters are still crap.

All in all, much fun on the Quads now.

Thanks to everybody involved, especially to Hartmut and of course Andrew
previously and to all others who paved this road.

Cheers,
Hermann











_______________________________________________
linux-dvb mailing list
linux-dvb@linuxtv.org
http://www.linuxtv.org/cgi-bin/mailman/listinfo/linux-dvb

Re: [linux-dvb] A/V sync issues with DVB-S

[Simon Kenyon]

Steven Toth wrote:
> Antti Palosaari wrote:
>   
>> Thomas Börkel wrote:
>>     
>>> hermann pitton wrote:
>>> Measuring the strength and SNR is not so easy now. My SAT receiver
>>> always says strength 70% and SNR 80%. My new Nova S Plus cards always
>>> report strength 97% and SNR 99%. Only my TT 1500s reported reasonable
>>> numbers like strength 63-80% and SNR 75-90%.
>>>       
>> What is the recommended way to report signal strength or SNR by the 
>> driver? Both of those are 16bit unsigned value. Is it better to scale 
>> 0-0xffff or return as decibels? If SNR is reported as decibels it will 
>> look like very small value...
>>
>> Antti
>>     
>
> May of the drivers I've added recently report their values in db, but I 
> typically also have one precision point. So I report 20.4 as 204 decimal.
>
> - Steve
>
>
> _______________________________________________
> linux-dvb mailing list
> linux-dvb@linuxtv.org
> http://www.linuxtv.org/cgi-bin/mailman/listinfo/linux-dvb
would it be possible to define a standard and then normalise the results 
to that range.
i realise that it might take a year or more before all drivers adopted 
this "standard".
but it would be worth it to at least declare one.

i see this sort of discussion on this list at least once a fortnight.
i also want to build a signal strength meter for a dish installation 
tool i'm trying to write
--
simon

_______________________________________________
linux-dvb mailing list
linux-dvb@linuxtv.org
http://www.linuxtv.org/cgi-bin/mailman/listinfo/linux-dvb