[linux-dvb] Re: VES1820.c question

[Holger Waechtler <holger@convergence.de>]

Hi,

Andreas Oberritter wrote:
> Hi,
> 
> Am Mit, 2002-10-09 um 15.30 schrieb Holger Waechtler:
> 
>>Please don't patch your driver, fix the application to pass the correct 
>>inversion setting to the driver.
> 
> 
> Why not? Documentation says, that this bit is zero when inversion is
> off, one otherwise. As I already mentioned, this works on the dbox2.
> Software auto inversion did not work until yesterday, because
> FE_SET_INVERSION was missing in our driver version. That makes me think
> that the documentation is right.

If a driver implements the inversion parameter correctly heavily depends 
on how the PLL is connected to the demodulator. When I and Q pins are 
reversed you have to exchange the settings for the Inversion bit as well 
even if the spec tells you something different.

We had the same problem for the Stv0299 based Alps/LG tuners. Both 
contain essentially the same hardware and work with the same driver, 
well -- when you swap the meaning of the inversion bit.

I'm not sure if Alps or LG were right in their interpretation of the 
stv0299 spec and who of them layouted the PCB correctly.

Unfortunally we have to 'invent' now some kind of probing to distinguish 
LG from Alps tuners and VES1820's on DVB-C cards from the ones in the dbox.

Maybe we can get out of this when we implement an inversion autoprobing 
in software again.

If the transmission is spectrum inverted or not is a different problem 
-- and this one should definitely be solved in channel lists on the 
application side...

Holger



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