[linux-dvb] Re: diseqc switch recommendation

[Josef Wolf <jw@raven.inka.de>]

On Mon, Dec 27, 2004 at 06:22:16PM -0700, Vladimir Shved wrote:

> European diseqc multi-switch is my term, that I came up with, I could not
> come up with anything else for something like the 9x8 diseqc switch. I
> misspelled European in my first message, sorry about that. I don't know
> anyone in North America making those kinds of switches, so those are
> European only switches. I'm not familiar with European multi-switches, so
> wondering if there was a diseqc multi-switch which could drive 4
> satellites,(4 dual LNBs) with at least four receivers/DVB-S cards, etc.

I'm still not sure I understand the difference.  For four LNBs (in germany)
one would usually install a 17x4 or 17x8 switch (four inputs per LNB) and
have quattro-LNBs (do not confuse with quad-LNB) attached to it.  I have
such a setup and it works great.

> Thanks for the hint but I did try that option in 2.4 and 2.6 kernels, I've
> got different issue than yours. The hw_sections value only seems to solve
> reliability problems when transferring the stream to applications and does
> not seem to help with sending diseqc commands.

In theory, you're right.  But the article I mentioned says that hw_sections
seems to have impact on sending diseqc commands.

> Disabling hw_sections sets
> driver to handle certain task in software rather than in hardware/firmware
> mode. Although, I'm not expert, and cant comment much about this.

One of the answers to the mentioned article brought up the speculation
that the hw_sections setting could have some timing impacts to the
firmware which in turn could make diseqc generation fail.  Diseqc
generation seems to be implemented in firmware, not in hardware.

> I think your problem mostly lies not in driver but in software or hardware
> setup.

Definitely not. It can't be in software or hardware because I can make it
fail just by booting 2.4 kernel and make it work by booting 2.6 kernel
without making any modification to software/hardware/cabling.

> Drivers seem to handle diseqc commands very well from the start, I
> tried many different drivers(old and new) and all of them were able to
> handle diseqc switching without problems.

You have FF cards, right?  As I wrote in my previous message, the problem
that diseqc don't work at all with 2.4 kernel was in conjunction with
_budget_ cards.  Please re-read my previous message carefully.  I was
talking about _two_ problems:

1. Budget cards don't send diseqc at all with 2.4 kernel.

2. FF card sends commands _unreliably_.  This problem is _not_ dependant
   on kernel version.  This problem seems to be similar to your problem.

> Maybe you have bad switch, or signal does not reach the switch,

Definitely not.  For two reasons:
1. I can make it work/fail without changing the switch.
2. My diseqc-tester shows failure, too.

> or software sends wrong diseqc commands,

Oh, I tried with:

- VDR
- szap
- dvbstream
- vls
- vlc
- I even wrote my own little test application and posted the diseqc-snippet
  into this list.

All of them show the same symptoms.  Therefore I don't think it is a
problem in the application code.

> or there is lots of noise in cabling which causing to corrupt diseqc
> commands.

But how comes that cabling noise is dependant on the kernel I run?

-- 
Please visit and sign and http://www.ffii.org
-- Josef Wolf -- jw@raven.inka.de --