[mythtv] mythtv-setup - DVB channel scanning

David Matthews dm at prolingua.co.uk
Fri Mar 3 14:28:02 UTC 2017

On 02/03/2017 11:31, roger wrote:
> On 01/03/17 18:19, John Pilkington wrote:
>> On 01/03/17 15:30, David Matthews wrote:
>>> On 27/02/2017 12:41, roger wrote:
>>>> On 27/02/17 10:22, Stuart Auchterlonie wrote:
>>>>> On 23/02/17 14:34, roger wrote:
>>>>>> I have been delving into DVB-T channel scanning in mythtv.
>>>>>> When using "full scan (tuned)" and the transport port scans with
>>>>>> "search
>>>>>> new transports" with a DVB_T tuner we often see the scanner trying to
>>>>>> tune transports that are not physically receivable by the hardware. I
>>>>>> looked into this and found that the code was not honouring the
>>>>>> other_frequency_flag in the TerrestrialDeliverySystemDescriptor
>>>>>> contained in the per transport stream loop in the Network Information
>>>>>> Table (NIT). This flag means that this particular transport stream is
>>>>>> also available on frequencies other than the one specified in the
>>>>>> TerrestrialDeliverySystemDescriptor itself. Information on these
>>>>>> other
>>>>>> frequencies information should be found a FrequencyListDescriptor for
>>>>>> this transport. The ETSI EBU standards mandate that this flag must be
>>>>>> set if a ChannelListDescriptor is present but not that a
>>>>>> ChannelListDescriptor must be present if this flag is set, which
>>>>>> seems a
>>>>>> little weird.
>>>>> Not information but some theories. Since the flag is meant to indicate
>>>>> that the mux is available on other frequencies, but it doesn't include
>>>>> the frequencies, then it's probably trying to signal that this mux
>>>>> could
>>>>> be seen on multiple frequencies when doing a full scan.
>>>>> I'm guessing it is trying to give the receiver a heads up that it
>>>>> needs
>>>>> to do some differentiation and decide on the "best" mux. Although
>>>>> why it
>>>>> would need a flag to do that i dunno....
>>> My feeling is that a tuned scan is right for DVB-S but not for DVB-T in
>>> the UK and that the only way to pick up all the multiplexes is a full
>>> scan over all the frequencies.  In particular the definition of the T2
>>> delivery system descriptor in the NIT allows the frequency to be
>>> optional and last time I looked it wasn't being included.  That means
>>> that a receiver has to do a scan in order to find T2 multiplexes and
>>> can't find them from the NIT.
>>> See
>>> https://www.dvb.org/resources/public/standards/a38_dvb-si_specification.pdf
>>> table 129.  The bandwidth, guard_interval and frequency information are
>>> only included if descriptor_length > 4.
>>> David
>> A comment as a user:
>> My pctv-290e autoswitching tuner, given only basic parameters for the
>> main DVB-T2 mux from Waltham (East Midlands), automatically tunes all
>> the DVB-T muxes but not the second DVB-T2 mux used for BBC FOUR HD.
>> That needs to be specified separately. Now I see why.
>> But sometimes I relocate and use the Sandy Heath transmitter. The
>> signal I get there isn't usually as good and that procedure has
>> occasionally yielded 0-byte recordings.  A full scan followed by
>> manual mux selection seems to give more reliable results there.
>> Mostly I record SD channels.
>> John P
> Hi John,
> Looking through some old docs. I found that the frequency list
> descriptor had actually been removed from UK radiated DVB-T signal in
> the 2007 switchover. At a guess that was because it was not really
> proving useful in the UK DVB set up, and they wanted STB and DTV
> manufacturers to switch to full frequency scans by default.
> The old version 7 D book, which you can find online, also has useful
> information in it.
> A idea for myth might be to add a geographical scan option and use the
> ofcom dtv transmitter data table
> https://www.ofcom.org.uk/spectrum/information/transmitter-frequency.
> <https://www.ofcom.org.uk/spectrum/information/transmitter-frequency>
> to scan for frequencies in  a particular radius.
> This should easy to implement. (But not by me!)
> Roger

I have wondered in the past about improving the support for DVB-T scans. 
  At present if there are duplicate entries the code seems to just 
select the first it finds.  It would be better if it could select the 
entry with the stronger signal based on S/N ratio or signal strength.

I've had problems with scanning myself in the past.  I can pick up most 
multiplexes from both Craigkelly and Black Hill but the current aerial 
gets a better signal from Craigkelly.  It's further complicated because 
Craigkelly appears to just relay everything from Black Hill and, from 
what I recall, that includes the NIT.  So the multiplex frequencies in 
the NIT from Craigkelly are actually those for Black Hill.  Perhaps this 
is what John was finding with Sandy Heath.

All this suggests that for DVB-T in the UK the only way to do a scan is 
by a full scan and the NIT should be completely ignored.  I'm not 
convinced that geographical information would be very helpful because 
there can be low-power relays that cover very localised areas and in 
hilly areas such as Scotland the choice of transmitter to use can depend 
on which side of the hill one is located.


More information about the mythtv-dev mailing list