[linux-dvb] DVB-C QAM128 errors and signal strength

[Jarkko Lavinen <jlavi@iki.fi>]

Hello DVB folk,

I bought 2 Siemens DVB-C cards with analog module from Dvbshop for 260
euros a pair. I am now running one card on an old K6 machine with
2.4.21 kernel, dvb-1.0.0-pre3 driver, vdr-1.2.1 and dvbtune from CVS
head.  Dmesg says my card being rev1.5, ves1820 frontend, spXXXX
tuner.

I am in HTV cable network in Helsinki.  I am able to tune and receive
both QAM64 and QAM128 modulated frequencies.  The difference between
these two is that QAM64 has little bit errors and no block errors
whereas QAM128 has lots of bit errors and too many block errors to be
usable.

Here is an example from QAM64 channel "Nelonen" @ 162MHz. I am using
channels.txt from Jaakko Hyvätti's web page.  I use modified Dvbtune
to give more statistics, and I ran Dvbtune -m for 30 seconds. The
averages are rounded upwards because there is so much error in the
measurement.

  +--------+--------+----------+----------+--------+
  |        | Streng | Bit errs |      SNR | Uncorr |
  +--------+--------+----------+----------+--------+
  | avg:   |  65535 |     2800 |    60200 |      0 |
  | sdev:  |    0.0 |    231.1 |    571.4 |    0.0 |
  | min:   |  65535 |     2430 |    58339 |      0 |
  | max:   |  65535 |     3150 |    60909 |      0 |
  +--------+--------+----------+----------+--------+

If I try the same with QAM128 channel "TV1" @ 283MHz, results are
awful:

  ...
  |     30 |  65535 |   579560 |    56797 |     48 | (S|L|C|V|SY|)
  ^C
  +--------+--------+----------+----------+--------+
  |        | Streng | Bit errs |      SNR | Uncorr |
  +--------+--------+----------+----------+--------+
  | avg:   |  65535 |   530000 |    57600 |     30 |
  | sdev:  |    0.0 |  26444.9 |    805.3 |   12.8 |
  | min:   |  65535 |   486940 |    55255 |      7 |
  | max:   |  65535 |   579560 |    58339 |     54 |
  +--------+--------+----------+----------+--------+

I don't understand why the signal strength is stuck at maximum. It
shows different values when there is no signal and cable is removed
from the card.

To me this looks the signal is saturated and perhaps too strong, and 
overdrives the tuner and the frontend.  Could the errors be caused
more by distortion than by noise?

Has anybody had similar problem and tried signal attenuator?

I could try to build a resistor based signal attenuator like in this
link: http://www.epanorama.net/documents/wiring/cable_impedance.html,
"Impedance matching network using resistors" at the end of document.

I am using passive signal splitter to share the antenna signal to
other devices (cable modem, TV, VCR). Without the splitter the signal
improves a bit for QAM128, but the difference in bit error is smaller
than the measurement error.

Changing the PC position few meters away from potential noise sources
helps a little.  This would suggest better shielding might help.
Currently I have to run the PC case open because the DVB card is quite
warm.

Jarkko Lavinen


--
Info:
To unsubscribe send a mail to ecartis@linuxtv.org with "unsubscribe linux-dvb" as subject.