[linux-dvb] [PATCH] Tuner calibration for some Nova-T devices

José Oliver Segura primijos at gmail.com
Mon Dec 18 16:17:20 CET 2006


       I've been able to run the tests and also to test SPURCHECK
definition ir order to see if it improves. As said, seems that the TVE
mux (818000000 Hz) refuses to tune (actually, to "re-tune") even with
the proposed patch and the SPURCHECK enabled.

       As a side effect, I've also found that some of the "reliable"
channels now fail. Femon shows a perfect lock, no UNC/BER, but the
image is almost frozen (very very slowmotion), and unsinchronized with
audio (audio seems to run ok)

        Best,
        Jose

On 12/18/06, José Oliver Segura <primijos at gmail.com> wrote:
>        Hi Treza,
>
>        thanks a lot for the patch. I've been able only to do some
> quick tests (hope to be able to perform some more this afternoon), and
> I've found that this patch seems to give some improvements (great!),
> but some other "strange" effects still appear (thus, I start to think
> that they have no relation with signal strength). I wrote a little
> program that performs a tzap for each of the channels in
> channels.conf, gets output about signal strenght, ber, unc, etc. and
> writes out a summary. Below you could find the compared results (pre
> and post patch):
>
> [root at mediacenter ~]# tail -11 output.prepatch.signal.test
> Summary statistics:
> Frequency       Signal  Ber             Unc
> =========       ======  ========        ========
> 514000000       a151    00000000        00000001
> 570000000       96f9    00000000        00000000
> 794000000       8308    00000000        00000001
> 818000000       7539    00193640        00000fae
> 834000000       79b8    00044993        00000bd6
> 842000000       7b72    00004394        00000009
> 850000000       79ce    00051580        000023f1
> 858000000       7843    000694ee        00001d2b
>
> [root at mediacenter ~]# tail -11 output.afterpatch.signal.test
> Summary statistics:
> Frequency       Signal  Ber             Unc
> =========       ======  ========        ========
> 514000000       9ae6    00000000        00000000
> 570000000       9a0e    00000000        00000001
> 794000000       940e    00000000        00000000
> 818000000       734b    0003d40d        000026e8
> 834000000       8a8e    00000000        00000000
> 842000000       8c8f    00000000        00000000
> 850000000       843d    0000002c        00000000
> 858000000       8154    000000df        00000001
>
>
>        As can be seen, output for post-patch looks better related to
> Ber/Unc, despite the fact that some of the signal strenghts are not
> improved (but others are, that looks good). There are still problems
> with the 818000000 Hz band (note the high Ber/Unc), but I think that
> these problems doesn't have relation with signal strenght, as long as
> with the same signal strenght I can get better or worse results (just
> loop tzapping / ^C some channel in this frequency causes the increase
> of BER/UNC numbers). Could it be related to some management issue in
> the frontend instead of some signal strenght problem?
>
>        I'll try to perform some more tests this afternoon (maybe also
> try the SPURCHECK flag) and post results. If any of you is interested
> in some specific number/information/log, just let me know and I'll do
> my best to get them.
>
>        Best,
>        Jose
>
> On 12/15/06, treza at chez.com <treza at chez.com> wrote:
> > Hi folks
> >
> > Here is a patch for Hauppage Nova-T-Stick and Nova-T-500 users.
> > It sets the MT2060 IF1 frequency according to the
> > calibration values stored in the EEPROM.
> >
> > It is supposed to enhance the signal quality, but, hey,
> > there is no guarantee.
> > Feedbacks would be much appreciated, to know whether it
> > deserves being applied.
> >
> > For those who have a MT2060 based device, very low signal
> > levels and feel _very_ lucky,
> > they can also activate MT2060_SPURCHECK in the mt2060_priv.h
> > file. At worst case it will change nothing.
> >
> > Patrick, the (dib0700_state.mt2060_if1 field in dib0700.h
> > becomes useless with this patch, is it right or shall I move
> > the code to the "frontend_attach" functions ?)
> >
> > Signed-off-by: Olivier DANET <odanet at caramail.com>
>



More information about the linux-dvb mailing list