Index of /hauppauge

Contents
========

Hauppauge diffs for v4l-dvb hg repository with kernel 2.6.24

These diffs affect the following devices:

HVR4000, HVR3000, HVR1300, NOVAS+, PVR350.

News 18 August 2008
===================

With updates supplied by CARLO I have created an updated
patchset for kernel 2.6.26. This is called scratchpad.
This is a test version of the patches and incorporates
the elements found in the experimental directory which
also means that it supports the mythtv patch found there.

News 23 feb 2008
================

Added an experimental section for DVB-S2 on the hvr4000
in the `experimental' sub directory.

News 8 feb 2008
===============

Some mfe bugs are being looked at now, meanwhile, James
Brown reminds us that you cannot use both DVB frontends
on the hvr3000/4000 cards as only one is supported at
any one time even when both are available. If you use
one the other will stop functioning. This is a hardware
limitation not a linux limitation and works that way
in windows too.

After reading the ubuntu forum I have noticed some
misunderstanding about the hvr4000 driver. The driver
has been tested and built on AMD64 and was previously
working on i386. It will work for both 32 and 64 bit
linux. The firmware file loads into the device and
has no relation to the host system CPU and is host
platform independant. There is no such thing as
a 32 or 64 bit linux firmware. The firmware is for an
unknown processor on the card, probably an ARM.

News 7 feb 2008
===============

Fixed patches to work in 2.6.24 kernel on the hvr4000.
(bug and kernel changes).

These patches are named as per today and yesterday's
are now removed.

News 6 feb 2008
===============

There are now two patches:

v4l-dvb-hg-2008-02-06-sfe.diff

This is as previous patch and allows a single frontend
(hvr3000/4000) to be selected on module load.

v4l-dvb-hg-2008-02-06-mfe.diff

This patch includes work by Fabio M. Di Nitto which
merges hvr3000 multi-frontend support into the above.
This gives both frontends (hvr3000/4000) on module load.

This is work in progress and under test.

News 9 dec 2007
===============

Fixed and added comments for future reference.
Removed ISL6421 hack and instead moved configuration
bit to device attachment.

This WILL be the last diff for some time as I will now go
back to other things :-)

News 8 dec 2007
===============

This diff adds the frontend selection module param
for HVR3000 devices. If you have such a device and
wish to use DVB-S in v4l-dvb hg then test this
and report success or failure. Thanks.

Update:

This ought to fix analog demod detection and diseqc
issues with 4x1 switches on hvr3000 cards. It will
also add remote control filtering for hauppauge
only devices and allow use on the hvr3000 in kernel
2.6.23. You will not have both DVB devices enabled
simultaneously, yet may select which you require
at module load time.

Update 2:

Remote control filtering and analog demod detection
also works for the HVR1300.

Update 3:

Remote control filtering works on the PVR-350 and
probably others in that range.

Update 4:

Remote control filtering works on the NOVAS+ and
will fix 4x1 diseqc switches on that device also.

News 7 dec 2007 (older news relevant to hvr4000 only)
=====================================================

Tidied up code a bit and fixed a bug introduced yesterday
retuning DVB-S.

Will commence work testing DVB-T since installing a feed
yesterday for this very purpose. All London muxes available
for test.

Comment: With DVB-S2 testing, you don't need to initially
get the pilot bit correct as the driver will retune with
it's opposite value should the lock fail when it's wrong.

Update:

DVB-T is now working. There can only be one frontend enabled
on the device (at this time at least) and a hack has been
introduced as a module parameter for cx88-dvb to select
the frontend you require.

Eg.

This option will load the DVB-S/S2 frontend and is
the default:

	options cx88-dvb frontend=0

This option will load the DVB-T frontend:

	options cx88-dvb frontend=1

Update 2:

A small fix was introduced for switching back between DVB-T
and DVB-S/S2 when unloading then reloading modules.

Update 3:

Module paramter hack has been renamed to frontend after
success with a number of users loading DVB-T frontend.

Update 4:

To aide with confusion, it must be made clear that when
testing DVB-S2 with the v4l-dvb patch, just the regular
szap is required for tuning and not szap2 after the
necessary sysctl vars are entered.

News 6 dec 2007
===============

Some SYSCTL vars have been added for those wanting to
test DVB-S2 yet do not want to or cannot use the abandoned
hvr4000 repository. To use this feature you will need to
setup the modfec/rolloff/pilot as described in the source
file cx24116.c using SYSCTL(8) prior to running szap.

Eg. for Pro7 HD on Astra 19.2E:

# 8psk 2/3
sysctl dev.cx24116.modfec=0x0d
sysctl dev.cx24116.pilot=1
sysctl dev.cx24116.rolloff=2

# channels.conf
Pro 7 HD:12722:h:1:22000:255:259:10200

To reset the driver to DVB-S use:

sysctl dev.cx24116.modfec=0xfe30

Update:

sysctls moved from kernel. to dev. for 2.6.24-RC (?)

option added to ISL6421 driver to turn off dynamic
current protection. this fixes issues with loaded
switches.

News 5 dec 2007
===============

This is the final diseqc configuration for this driver.
See the img/ folder for screenshots of operation.

A kernel module parameter now exists named "toneburst". Previous
parameter named "desc" has been removed. One new option which
completely disables toneburst has been added.

The "toneburst" module parameter can have 1 of 3 values:

0: NEVER send tone burst
1: Cache tone burst or derive tone burst from command (default).
   In some cases a diseqc message may need transmitting twice,
   yet this is the preferred method.
2: Cache message and send with known toneburst.
   This method sounds fine except when some applications omit
   sending the burst when is now needed.

The above parameter is also exposed as a sysctl:
 dev.cx24116.toneburst

This can be set whilst the frontend is offline without reloading
the driver to determine what works best for you, eg.

sysctl dev.cx24116.toneburst=0 
sysctl dev.cx24116.toneburst=1
sysctl dev.cx24116.toneburst=2

It should be understood that the cx24116 firmware can never
send a toneburst on it's own, it always has to be sent with
a DiSEqC message, else never at all (until proven otherwise).

Very old hvr3000+4000 patch has been removed, as has pilot.diff

News 21 nov 2007
================

Imported signal readings from reelbox code. Re-calibrated
against a FORTEC receiver (originally dvb2000). This
include BER, UCB, strength and quality. Previous strength
reading is now quality as confirmed in reelbox code and
on FORTEC.

News 19 nov 2007
================

Diseqc command strategy has been reverted to an older method
which caches message prior to burst IOCTL. This is now the
default and this ought to fix some issues with szap found
recently with some switches. The updated derived burst
hack strategy still exists but now must be enabled with a module
parameter called 'dsec'. See modinfo cx24116

FYI. You may need to enable the hack method for mythtv. Either
method ought to work for kaffeine and szap. The status of
VDR is unknown, try either.

News 4 nov 2007
===============

An updated patch for v4l-dvb has been created adding a missing 
definition for the tuner module which is now necessary for proper
attachment.

Slightly older and unreported news is that the stoth/hvr4000 repo
was removed in October from v4l-dvb and a clone of that can be found
locally in the hg directory.

It has been brought to the attention of the v4l-dvb mailing list
that the open source REELBOX has support for the DVB-S2 demodulator
used in the hvr4000 (cx24116). The source for that is located
in the reelbox directory. It is not known if this code contains
reverse engineered operational knowledge or is based on factual
device data from an NDA with conexant.

News 2 sep 2007
===============

A solution to DVB-S2 tuning problems can be now be found in pilot.diff
This will probably be merged soon against the latest hvr4000 repository.

The solution to the rarely reported zero signal strength reading
may be resolved by inspecting register 0xa3. This register possibly
contains an overflow bit which indicates 100% signal. Testing is needed.

News 1 sep 2007
===============

The hvr4000 stoff-15.diff has been merged with the hvr4000 repository
and is no longer needed.

Due to the above, the hvr4000+3000 diff is now out of date and
in need of maintenance. This will be done some time later.

News 31 aug 2007
================

The patch for v4l-dvb has been updated to cope with a change in cx88 core
structure.

DVB-T has yet to be tested due to lack of time.

Diffs for the HVR-4000
======================

v4l-dvb-hg
----------

This diff is against the V4L-DVB main repository and adds the HVR-4000
minus DVB-S2 support to that.

The diff supports diseqc, signal strength, analogue TV, analogue inputs and
remote IR support on the HVR-4000 as well as DVB-S.

It is possible to watch satellite TV and analogue TV simultaneously if
you so wish.

This diff won't have development on the DVB-T part any time soon.

hvr4000+3000
------------

NOTICE: This diff is now out of date and needs maintenance.

This diff is against the HVR4000 repository and is a merge of the stoth
hvr4000 and hvr3000 repositories. The aim of this  diff is to add DVB-T
support to the HVR4000 which will come later if a solution is found.

It is work in progress but has been tested on a hauppauge NOVA-S-PLUS
which continues to work and will be tested against the HVR-4000 at a
later time.

Building for v4l-dvb
====================

hg clone http://linuxtv.org/hg/v4l-dvb
cd v4l-dvb
patch -p1 < v4l-dvb-XX-XX-2007.diff
make menuconfig
make
make install

Known problems
==============

1. IR sometimes doesn't start from a cold boot. Reloading modules fixes this.

2. Some DVB-S2 transponders fail to tune. This has been investigated and
   the reason discovered. pilot.diff resolves this problem when applied
   to the latest hvr4000 repository.

3. The hardware signal strength register sometimes reports zero. Register
   0xa3 may actually contain an overflow bit which indicate 100% signal. This
   needs to be inspected.

--

 // /
{:)==={ Darron Broad <darron AT kewl DOT org>
 \\ \