On Sun, Feb 08, 2004 at 09:43:28PM +1100, Hamish Moffatt wrote:
On both cards I am getting a lot of transport errors reported by Xine
and MythTV on certain TV stations. We have five stations here and one or
two work perfectly but at least two give me regular transport errors.
Looking at the source code for Xine and Myth, they inspect a bit in each
packet of the MPEG-2 transport stream and declare a transport error if
that bit is set.
[..]
Could anybody give me any suggestion about what these transport errors
mean? Are Xine/MythTV inspecting a bit in the MPEG-2 TS header or is
that a header added by the DVB drivers? I don't see why an error bit
would be included in the on-air protocol so I assume it must be updated
by the hardware or drivers.
So I did a bit more digging. It seems that that IS a bit in the MPEG2
transport stream header. The tda1004x is transport stream-aware, so
either it's setting the bit because it detected an uncorrectable error,
or the transmitting station set the bit (which would seem to be a bit
daft).
I would assume they are set by the tda10045. Have you tried the
experiment to ensure a quasi-errorfree stream, either by using a better
antenna or testing on another transponder?So, if the tda1004x is detecting uncorrectable errors and flagging them
in the received packets, why doesn't it return a non-zero count when
asked for its uncorrected error count? (eg when running femon, which
appears to ultimately query the tda1004x over the i2c bus).
I'm having problems with the sp887x card too. I have seen that report
uncorrectable errors in the counters, but I get transport errors from
that which aren't counted, just like the tda1004x.
The error counter values reported of the sp887x always seemed bogus to
me, I'm not sure whether you can count on them.