PCTV 292e support

PCTV 292e support

[Russel Winder]

[-- Attachment #1: Type: text/plain, Size: 2243 bytes --]

From the material on the LinuxTV webpages, there is support for PCTV
292e, and in emails it appears others are using this device. I find
that on Debian Sid and Fedora Rawhide using the distributed kernels and
libdvbv5, dvbv5-scan fails to get activity from the device. PCTV 282e
works fine.

Fedora Rawhide has kernels 4.4 and 4.5, Debian Sid 4.3.

Debian Sid distributes dvb-tools as well as libdvbv5 (1.8.0). Fedora
Rawhide only distributed libdvbv5 as far as I can tell, but I compiled
dvbv5-scao from source from a git repository clone, but it behaves the
same.

On Debian Sid with PCTV292e:

|> dvbv5-scan /usr/share/dvb/dvb-t/uk-CrystalPalace 
Cannot calc frequency shift. Either bandwidth/symbol-rate is
unavailable (yet).
Scanning frequency #1 490000000
       (0x00) Signal= 0.00dBm
Scanning frequency #2 514000000
       (0x00) Signal= 0.00dBm
Scanning frequency #3 570000000
       (0x00) Signal= 0.00dBm
Scanning frequency #4 506000000
       (0x00) Signal= 0.00dBm
Scanning frequency #5 482000000
       (0x00) Signal= 0.00dBm
Scanning frequency #6 529833000
       (0x00) Signal= 0.00dBm
Scanning frequency #7 545833000
       (0x00) Signal= 0.00dBm

Swapping to the PCTV 282e, scanning works entirely as expected: if
there is a lock it happens immediately and the channels are shown after
a while; if there is no immediate lock there is a spinning of the
signal strength value – definitely not just a solid zero.

I note that PCTV 282e has a green light on permanently and works fine.
PCTV 292e lights up blue when plugged in and then the light goes out.
On Linux it never comes on again. Trying the device on Windows with the
distributed software the blue light comes on every so often as activity
happens. The hypothesis is that the device is not being driven
properly. And yet people appear to be using the device?

-- 
Russel.
=============================================================================
Dr Russel Winder      t: +44 20 7585 2200   voip: sip:russel.winder@ekiga.net
41 Buckmaster Road    m: +44 7770 465 077   xmpp: russel@winder.org.uk
London SW11 1EN, UK   w: www.russel.org.uk  skype: russel_winder


[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 181 bytes --]

Re: SV: PCTV 292e support

[Russel Winder]

[-- Attachment #1: Type: text/plain, Size: 988 bytes --]

On Sun, 2016-01-24 at 05:35 +0000, Peter Fässberg wrote:
> You have probably not installed the firmware.
> 
> Check the kernel log (dmesg) and you will find an error and the name
> of the missing file.

Hummm… Spot on. Now I just feel stupid :-)

[62987.648972] si2168 2-0064: found a 'Silicon Labs Si2168-B40'
[62987.649462] si2168 2-0064: firmware: failed to load dvb-demod-si2168-b40-01.fw (-2)
[62987.649467] si2168 2-0064: Direct firmware load for dvb-demod-si2168-b40-01.fw failed with error -2

From what I can tell neither Debian nor Fedora have package with this
firmware. Unless I am just missing it. If this is the case should they
be packaged?

-- 
Russel.
=============================================================================
Dr Russel Winder      t: +44 20 7585 2200   voip: sip:russel.winder@ekiga.net
41 Buckmaster Road    m: +44 7770 465 077   xmpp: russel@winder.org.uk
London SW11 1EN, UK   w: www.russel.org.uk  skype: russel_winder


[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 181 bytes --]

Re: SV: PCTV 292e support

[Russel Winder]

[-- Attachment #1: Type: text/plain, Size: 1409 bytes --]

Hi,

I followed the instructions at http://blog.palosaari.fi/2014/04/naked-h
ardware-15-pctv-triplestick-292e.html

|> md5sum dvb-demod-si2168-01.fw 
87c317e0b75ad49c2f2cbf35572a8093  dvb-demod-si2168-01.fw

which is as expected, and put a symbolic link to this file in
/lib/firmware. Sadly this is not the file the driver is expecting:

[68031.899042] si2168 2-0064: found a 'Silicon Labs Si2168-B40'
[68031.899069] si2168 2-0064: firmware: failed to load dvb-demod-si2168-b40-01.fw (-2)
[68031.899073] si2168 2-0064: Direct firmware load for dvb-demod-si2168-b40-01.fw failed with error -2
[68031.899084] si2168 2-0064: firmware: failed to load dvb-demod-si2168-02.fw (-2)
[68031.899086] si2168 2-0064: Direct firmware load for dvb-demod-si2168-02.fw failed with error -2
[68031.899089] si2168 2-0064: firmware file 'dvb-demod-si2168-02.fw' not found

I am assuming that it is not a question of renaming the file (*), but
that some other firmware needs to be extracted from the discs.

(*) I did try that and it attempted to load the firmware but got an
error -121.

-- 
Russel.
=============================================================================
Dr Russel Winder      t: +44 20 7585 2200   voip: sip:russel.winder@ekiga.net
41 Buckmaster Road    m: +44 7770 465 077   xmpp: russel@winder.org.uk
London SW11 1EN, UK   w: www.russel.org.uk  skype: russel_winder


[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 181 bytes --]

Re: SV: PCTV 292e support

[Russel Winder]

[-- Attachment #1: Type: text/plain, Size: 740 bytes --]

On Sun, 2016-01-24 at 06:56 +0000, Russel Winder wrote:
> […]
> that some other firmware needs to be extracted from the discs.

If in doubt check out http://palosaari.fi/linux/v4l-dvb/firmware/

Firmware loaded and running. Device tuning and working.

Except that the T2 channels cannot be found properly. I guess this is a
"solved issue" as well, so back to the research.

Apologies for the earlier noise.

> 
-- 
Russel.
=============================================================================
Dr Russel Winder      t: +44 20 7585 2200   voip: sip:russel.winder@ekiga.net
41 Buckmaster Road    m: +44 7770 465 077   xmpp: russel@winder.org.uk
London SW11 1EN, UK   w: www.russel.org.uk  skype: russel_winder


[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 181 bytes --]

Re: SV: PCTV 292e support

[Andy Furniss]

Russel Winder wrote:
> On Sun, 2016-01-24 at 06:56 +0000, Russel Winder wrote:
>> […] that some other firmware needs to be extracted from the discs.
>
> If in doubt check out http://palosaari.fi/linux/v4l-dvb/firmware/
>
> Firmware loaded and running. Device tuning and working.
>
> Except that the T2 channels cannot be found properly. I guess this is
> a "solved issue" as well, so back to the research.
>
> Apologies for the earlier noise.

Maybe your uk-CrystalPalace is old.

There's something called w_scan which really does scan.

Try distro install or searching for it.

I don't use and haven't updated the dvb5 tools for ages. Looking back at
experiments it seems I have T rather than T2 in my tuning files. Maybe
that's something to keep in mind, also IIRC for T2 channels you do need
each frequency (unlike the Ts whose details are found by tuning to any
channel).

I don't know your setup or requirements, but using players to
tune/record will avoid the "-p" issue. I run headless and use tvheadend.
If I had an HTPC setup I would probably use kodi (xbmc).

Re: SV: PCTV 292e support

[Russel Winder]

[-- Attachment #1: Type: text/plain, Size: 2467 bytes --]

On Sun, 2016-01-24 at 10:07 +0000, Andy Furniss wrote:

> 
> Maybe your uk-CrystalPalace is old.

It's the one distributed with Debian Sid, I would hope it is up to
date. We haven't had a physical channel revamp here in quite a while.
From what I can tell it correctly describes what should be there.

> There's something called w_scan which really does scan.

It finds all the physical channels, quite happily describes all the
virtual channels in the T1 channels, fails to find anything in one of
the T2 channels and finds unnamed channels in the other T2 channel. The
device itself is fine, as it gets all T1 and T2 channels on Windows.
This implies something awry with it in a Linux context.

> Try distro install or searching for it.

Debian Sid packages it, but I have some source code to compile as well.

> I don't use and haven't updated the dvb5 tools for ages. Looking back
> at
> experiments it seems I have T rather than T2 in my tuning files.
> Maybe
> that's something to keep in mind, also IIRC for T2 channels you do
> need
> each frequency (unlike the Ts whose details are found by tuning to
> any
> channel).
> 
> I don't know your setup or requirements, but using players to
> tune/record will avoid the "-p" issue. I run headless and use
> tvheadend.
> If I had an HTPC setup I would probably use kodi (xbmc).
> 

The whole point of my activity is to rewrite Me TV. This is intended as
a very lightweight DVB player. The idea is not to have MythTV, Kodi,
etc. which are intended to be media centres. I just want a television
window with EPG. Original Me TV was GTK+2, Xine, DVBv3 with direct
access to the kernel API. I am rewriting for libdvbv5, GStreamer,
GTK+3. 

I am starting with scan and tune codes so as to set up dvr0 as the
input source for the rendering. dvbv5-zap -p is an experimental tool to
plug into a gst-launcher-1.0 script just to trial things. My code has
the same problems dvbv5-zap has, describing my problem in terms of
dvbv5-zap behaviour just means it isn't my code that is wrong, there is
an issue somewhere in the libdvbv5 code or the device driver.
  
-- 
Russel.
=============================================================================
Dr Russel Winder      t: +44 20 7585 2200   voip: sip:russel.winder@ekiga.net
41 Buckmaster Road    m: +44 7770 465 077   xmpp: russel@winder.org.uk
London SW11 1EN, UK   w: www.russel.org.uk  skype: russel_winder


[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 181 bytes --]

Re: SV: PCTV 292e support

[Andy Furniss]

Russel Winder wrote:
> On Sun, 2016-01-24 at 10:07 +0000, Andy Furniss wrote:

> It finds all the physical channels, quite happily describes all the
> virtual channels in the T1 channels, fails to find anything in one
> of the T2 channels and finds unnamed channels in the other T2
> channel. The device itself is fine, as it gets all T1 and T2 channels
> on Windows. This implies something awry with it in a Linux context.

OK, I can't reproduce this on Tacoleneston which has three T2 muxes.

I am using some old git version (Jun 10), I'll try current as time allows.

One of the T2s only has 2 channels anyway and as is normal AFAIK
channels that aren't running at time of scan don't get audio/video pids
listed, but the rest of the info is there.

There is a timeout option eg -T 3 trebles the timeouts - maybe try that.


> The whole point of my activity is to rewrite Me TV. This is intended
> as a very lightweight DVB player. The idea is not to have MythTV,
> Kodi, etc. which are intended to be media centres. I just want a
> television window with EPG. Original Me TV was GTK+2, Xine, DVBv3
> with direct access to the kernel API. I am rewriting for libdvbv5,
> GStreamer, GTK+3.
>
> I am starting with scan and tune codes so as to set up dvr0 as the
> input source for the rendering. dvbv5-zap -p is an experimental tool
> to plug into a gst-launcher-1.0 script just to trial things. My code
> has the same problems dvbv5-zap has, describing my problem in terms
> of dvbv5-zap behaviour just means it isn't my code that is wrong,
> there is an issue somewhere in the libdvbv5 code or the device
> driver.

Interesting, so you know a lot more than me about this stuff :-)

Experience as a user of 292/290s - they do need some time/grace to tune
in/stop spewing "junk". IIRC I added 5sec somewhere in TVH in addition
to whatever it already uses.

I guess going from T to T2 is worse - and then factor in that some T2s
are much lower power than others.

It did seem in the thread that I started where EAGAIN worked around,
that the code was giving no grace at all and expecting to be able to
parse stream content straight away. I may mis-remember though!

Re: SV: PCTV 292e support

[Russel Winder]

[-- Attachment #1: Type: text/plain, Size: 3793 bytes --]

On Mon, 2016-01-25 at 00:48 +0000, Andy Furniss wrote:
> Russel Winder wrote:
> > On Sun, 2016-01-24 at 10:07 +0000, Andy Furniss wrote:
> 
> > It finds all the physical channels, quite happily describes all the
> > virtual channels in the T1 channels, fails to find anything in one
> > of the T2 channels and finds unnamed channels in the other T2
> > channel. The device itself is fine, as it gets all T1 and T2
> > channels
> > on Windows. This implies something awry with it in a Linux context.
> 
> OK, I can't reproduce this on Tacoleneston which has three T2 muxes.

I have managed to get some proper T2 tuning. :-)

Someone emailed me privately to tell me about: 

https://github.com/OpenELEC/dvb-firmware

The 292e demod firmware in their is 8 bug fix release further on that
the one I had. It looks like there was a crucial bug fix in there.

> I am using some old git version (Jun 10), I'll try current as time
> allows.
> 
> One of the T2s only has 2 channels anyway and as is normal AFAIK
> channels that aren't running at time of scan don't get audio/video
> pids
> listed, but the rest of the info is there.
> 
> There is a timeout option eg -T 3 trebles the timeouts - maybe try
> that.

I had tried -T 2 but that didn't do any good. I suspect the firmware I
had just wasn't up to it, but the new one was.

> 
> > The whole point of my activity is to rewrite Me TV. This is
> > intended
> > as a very lightweight DVB player. The idea is not to have MythTV,
> > Kodi, etc. which are intended to be media centres. I just want a
> > television window with EPG. Original Me TV was GTK+2, Xine, DVBv3
> > with direct access to the kernel API. I am rewriting for libdvbv5,
> > GStreamer, GTK+3.
> > 
> > I am starting with scan and tune codes so as to set up dvr0 as the
> > input source for the rendering. dvbv5-zap -p is an experimental
> > tool
> > to plug into a gst-launcher-1.0 script just to trial things. My
> > code
> > has the same problems dvbv5-zap has, describing my problem in terms
> > of dvbv5-zap behaviour just means it isn't my code that is wrong,
> > there is an issue somewhere in the libdvbv5 code or the device
> > driver.
> 
> Interesting, so you know a lot more than me about this stuff :-)

I doubt that, but I am learning a lot very quickly!

I think having D, Rust, Python, and C++ bindings to libdvbv5 would help
tremendously getting it traction. Writing stuff in C with all the 1970s
programming paradigms is a real turn off in 2010s.

> Experience as a user of 292/290s - they do need some time/grace to
> tune
> in/stop spewing "junk". IIRC I added 5sec somewhere in TVH in
> addition
> to whatever it already uses.
> 
> I guess going from T to T2 is worse - and then factor in that some
> T2s
> are much lower power than others.

I am finding locking behaviour to be very strange. Sometimes I get an
immediate lock on a -50.0 signal, sometimes -35.0 signal fails to lock.
I am not sure if this is just a timing/sampling thing or whether there
is a quality of signal thing I am missing.

As I am focused on lightweight, I am working with non-fixed aerials so
low signal strengths. Though for testing I have an aerial with powered
high gain.

> It did seem in the thread that I started where EAGAIN worked around,
> that the code was giving no grace at all and expecting to be able to
> parse stream content straight away. I may mis-remember though!


-- 
Russel.
=============================================================================
Dr Russel Winder      t: +44 20 7585 2200   voip: sip:russel.winder@ekiga.net
41 Buckmaster Road    m: +44 7770 465 077   xmpp: russel@winder.org.uk
London SW11 1EN, UK   w: www.russel.org.uk  skype: russel_winder


[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 181 bytes --]

Re: SV: PCTV 292e support

[Andy Furniss]

Russel Winder wrote:
> On Mon, 2016-01-25 at 00:48 +0000, Andy Furniss wrote:
>> Russel Winder wrote:
>>> On Sun, 2016-01-24 at 10:07 +0000, Andy Furniss wrote:
>>
>>> It finds all the physical channels, quite happily describes all
>>> the virtual channels in the T1 channels, fails to find anything
>>> in one of the T2 channels and finds unnamed channels in the
>>> other T2 channel. The device itself is fine, as it gets all T1
>>> and T2 channels on Windows. This implies something awry with it
>>> in a Linux context.
>>
>> OK, I can't reproduce this on Tacoleneston which has three T2
>> muxes.
>
> I have managed to get some proper T2 tuning. :-)
>
> Someone emailed me privately to tell me about:
>
> https://github.com/OpenELEC/dvb-firmware
>
> The 292e demod firmware in their is 8 bug fix release further on that
> the one I had. It looks like there was a crucial bug fix in there.

Ahh, that's good. I ought to get that myself, I am still using 4.0.4!

>> I am using some old git version (Jun 10), I'll try current as time
>>  allows.

FWIW I did build current git and it does work.

> I am finding locking behaviour to be very strange. Sometimes I get an
> immediate lock on a -50.0 signal, sometimes -35.0 signal fails to
> lock. I am not sure if this is just a timing/sampling thing or
> whether there is a quality of signal thing I am missing.

I've only really tested with a "real" aerial, but it did seem like
sometimes it took longer than others to lock.

> As I am focused on lightweight, I am working with non-fixed aerials
> so low signal strengths. Though for testing I have an aerial with
> powered high gain.

Random thoughts on player -

You'll need to handle AAC in LATM that switches between 2 and 5
channels. VLC falls down in this respect. Avoid FAAD as the release at
least had a bug that will cause volume issues with some content if it
has dynamic range control metadata. ffmpeg aacdec handles channel
switching and ignores DRC meta - so you get full range.

On V4l

As I leave mine plugged in a headless box I don't hit this, but I think
there's a chance that if you repeatedly plug and have fragmented memory
that it will eventually fail to allocate some buffers.