On Monday 16 of June 2008, manu wrote: > Le 15.06.2008 13:20:30, Ales Jurik a écrit : > > On Sunday 15 of June 2008, Dominik Kuhlen wrote: > > > Hi, > > > > > > On Thursday 22 May 2008, Ales Jurik wrote: > > > > Hi, > > > > > > > > my friend told me that he is sometimes able to get lock by > > > > decreasing > > > > > > (not increasing) the frequency. Yesterday I've tested it and it > > > > seems to > > > > > > me he was right. So I'm able to get (not very stable, for few > > > > minutes) > > > > > > lock as well as by increasing and by decreasing frequency of the > > > > same > > > > > > channel (EurosportHD). > > > > > > The derotator code in the stb0899 seems to be setting the initial > > > > frequency > > > > > to the lowest search range and never changes this (as it should > > > > increase > > > > > steadily to find a lock) > > > > > > > It was also detected (not by me, I don't have riser card) that > > > > when the > > > > > > card is connected not directly into PCI slot but with some riser > > > > card, > > > > > > the needed difference for getting lock is higher (up to 10MHz). > > > > So > > also > > > > > > some noise from PC is going into calculations. > > > > > > > > I don't think the problem is in computation of frequency but in > > > > for > > > > > > example not stable signal amplitude at input of demodulator or in > > > > not > > > > > > fluently changing the gain and bandwith of tuner within the band. > > > > As I > > > > > > see in the code some parameters are changing in steps and maybe 3 > > > > steps > > > > > > for whole band is not enough? Especially in real conditions (not > > > > in lab)? > > > > > Could you please try the attached patch if this fixes the problem > > > > with > > > > > nominal frequency/symbolrate settings? > > > > Many thanks, but I don't see any improvement. Minimum time after I've > > lock is > > 150s. > > > > The only improvement was done 2weeks ago when Manu Abraham fixed some > > headers > > with register addresses. From that time it is possible to get lock > > also at > > problematic channels but very unstable. > > > > But I've found that all channels with lock problems are DVB-S2, 8PSK, > > FEC 3/4. > > So all channels broadcasted in QPSK is possible to tune without > > problems and > > also channels broadcasted in DVB-S2, 8PSK, FEC 2/3 are possible to > > tune to > > without problems. Channels in DVB-S2, 8PSK with different FEC (to 2/3 > > and > > 3/4) I was not able to test. > > For me adding 4MHz solved the lock problems for DVB-S QPSK FEC 3/4, but > did not solve my problems for a DVB-S QPSK FEC 5/6. > I will try Dominik's patch to see if it improves things. > Stay tuned (oh well) > Bye > Manu > So it seems to me that there are more problem in driver that (maybe) could be solved separately: 1. Some channels/transponders could be tuned by increasing/decreasing freq. by 4-10MHz. This solution is not for me usable. I didn't find any transponder on such solution is working. But for my friend this is working on transponders which I'm able to tune to directly. Maybe some problem in tolerance of TT S2-3200 components or in PC case (grounding or so). 2. Some channels/transponders is not possible to tune if this transponders is choosing as first after changing satellite (by diseqc switch). This is generally solved by users by tuning to another transponder at problematic satellite (for me for example this two transponders use the same modulation) - it seems not to be modulation problem (timing problem?). 3. At some transponders above 2 point doesn't help to get lock. For me these transponders are all DVB-S2, 8PSK and FEC3/4. Maybe I'm not right but hope another small step forward. BR, Ales _______________________________________________ linux-dvb mailing list linux-dvb@xxxxxxxxxxx http://www.linuxtv.org/cgi-bin/mailman/listinfo/linux-dvb
