[linux-dvb] Re: New UK DVB-T channels testing

[Klaus Schmidinger <Klaus.Schmidinger@cadsoft.de>]

Holger Waechtler wrote:
> 
> Hi,
> 
> Klaus Schmidinger wrote:
> > Dave Chapman wrote:
> >
> >>On Tuesday 08 October 2002 22:19, Klaus Schmidinger wrote:
> >>
> >>>Are these frequency offsets always +/-167kHz? Will that be the same for
> >>>other countries, too?
> >>
> >>In the UK the offsets (if present) are always +/- 167KHz.  I haven't heard
> >>about offsets being used in other countries.
> 
> Only +/-166667 offsets are specified in the DVB standard.

Ah, that would make offset handling possible with just O0, O1 and O2.

> >>>I'm asking because I would like to make the frequency for DVB-T channels
> >>>in VDR's channels.conf also be in MHz, and if that offset thing is in any
> >>>way standardized we could just add another DVB-T specific parameter to
> >>>cover the offset (like O0 = 0 kHz, O1 = +167 kHz, O2 = -167kHz).
> >>>
> >>>I wonder, though, why that offset is necessary, and whether the driver
> >>>shouldn't be able to tune to a channel at, say, 490 MHz - 167 kHz
> >>>automatically if it is given an inital frequency of 490 MHz. I mean
> >>>167 kHz isn't _that_ far off for a transponder with several MHz bandwith.
> >>>Or am I missing something important here?
> >>
> >>No, trying to tune to a frequency of 488.833MHz by approximating to 489MHz
> >>doesn't work - I've just tried it.
> >
> >
> > That's odd... IIRC with DVB-S you can be off by even 1 or 2 MHz and the driver will
> > tune to the transponder correctly, so I really wonder why this isn't possible with
> > DVB-T. I mean it's just a mere 166 kHz! Is there no AFC in the DVB-T part of
> > the driver?
> >
> > Can one of the driver developers comment on this?
> 
> That was probably the zigzag scan. It's range is 1000 times wider for
> DVB-S since frequencies are there measured in kHz, not Hz to prevent an
> overflow.
> 
> For DVB-C/T we need a finer stepsize since the frequencies there are
> much smaller. If you want to achieve fastest possible tuning you should
> pass the exact frequency.
> 
> All demodulators have built-in AFC's too, but their range is sometimes
> pretty limited - and even there you can speed up tuning when you pass
> the exact frequency.
> 
> >>I think the "O" option would just make the channels.conf more confusing - it
> >>is much simpler just to keep the frequency the way it is, in KHz.
> >
> >
> > The problem is that I would like to be able to use the transponder frequency
> > as a parameter in a "unique channel ID" (I know, there are people who would
> > like that to be the NID, TID whatsoever - there will be a switch to configure that!).
> > Anyway, since the "official" frequencies are apparently given as integer MHz values
> > (for instance 23 * 8 + 306) and the "offset" is mentioned as an additional parameter,
> > see http://www.itc.org.uk/uk_television_sector/reception_advice/digital_trans_guide/show_transmitter.asp?siteID=81.
> > So I would prefer _all_ frequencies in channels.conf to be in MHz, so that the numbers
> > can be expressed in 16 bit values. And regarding channels.conf being "confusing": well,
> > that's mainly because DVB-T needs so many parameters! I don't know why they have to
> > have _that_ many of them, but that's just the way it is. And apparently the "offset"
> > is just another of these parameters.
> 
> Why don't you want to use 32-bit frequencies?

Because I would like to use the "natural" values, and those are typically values
like 12345 MHz for DVB-S, and 567 MHz for DVB-C and DVB-T. The "center" frequencies
are apparently given as integer MHz values (even though the DVB standard allows
32 bit values). See Lauri Pesonens most recent posting:

  The NIT reports frequencies like 283000000Hz when I'm using 282750000Hz
  with VDR. So The NIT reports the 'official' freqs without the offset.
  Offsets are -250kHz for QAM_128 freqs and -125kHz for QAM_64 freqs.

Therefore it should be sufficient to store the "center frequency" and allow for
an offset handling, which in case of DVB-T could be done by an 'O' parameter
(note that it's the character between 'N' and 'P', not the number 'zero'), and 
in case of DVB-C a table of offsets for the various QAM modes should do the trick.

> > Can somebody with DVB-T please run a program that parses the NITs and reports
> > the transponder frequencies given there? I'd say if these are given as the "real"
> > values (like 488.833MHz), then VDR should store them as such in channels.conf.
> > But if they are given as integer MHz values (like 489MHz) then there's no point in
> > having them in kHz in channels.conf.
> 
> The NIT's report frequencies in 10Hz steps for DVB-T, in 100Hz steps for
> DVB-C and in 10kHz steps for DVB-S.

That is apparently true, but I have yet to see an example where it really uses
that high resolution. From the postings I've seen so far the values are only
given in MHz resolution, not finer. Sorry I can't contribute myself to this,
but I don't have DVB-C or DVB-T equipment.

Klaus
-- 
_______________________________________________________________

Klaus Schmidinger                       Phone: +49-8635-6989-10
CadSoft Computer GmbH                   Fax:   +49-8635-6989-40
Hofmark 2                               Email:   kls@cadsoft.de
D-84568 Pleiskirchen, Germany           URL:     www.cadsoft.de
_______________________________________________________________


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