[linux-dvb] Re: Twinhan DST / TS204 and signal quality questions

[Werner Johansson <wj@yodel.net>]

Johannes Stezenbach wrote:

> Well, I was exaggerating a tiny bit, but you get the idea. You cannot
> trust "signal quality" readings from consumer equipment.
Got it. The only interesting piece of information that my Thomson STB 
supplies is the BER value (in exponent number form, not some cheesy bar 
graph), with gives some idea of how the signal looks to the frontend. 
But it would be even more interesting to have the BER calculated both 
before and after the FEC takes place, in that case it would have been 
easy to output some "real" quality data, but as the packets before FEC 
would be hard to come by if not designing new frontend chips, I don't 
think we'll see anything like that soon..

This is the hard part about the digital domain, just think Compact Discs 
which play and play with (virtually) no degradation when getting 
scratched, up until it's unplayable as the FEC cannot cope with it 
anymore. No heads-up warning before that, it just dies with 
uncorrectable errors. So in some cases it would be interesting to get 
the BER of the non-corrected data, for quality measurments.

> (OTOH our frontend drivers could still be crap compared to some STB,
> regardless what the "signal quality" readings say.)
That's a completely different thing, it all comes down to how to get 
access to the frontend data - and it seems that some frontends just 
doesn't want to supply as much measurements as others, then no amount of 
driver hacking can get it back (unfortunately)..

> It's not the fault of the frontend, which signals data/fec output
> on a pin. But I guess Twinhan saved to decode this pin to make the
> hardware cheaper. Which IMHO isn't a real problem.
Right, my mistake. I see the idea, it just never occured to me why it 
would ever be interesting to look at the (already used) fec data, except 
for frontend chip debugging. No real problem, I was just curious.

/Werner