[PATCH] TDA9887 PAL-Nc fix

[PATCH] TDA9887 PAL-Nc fix

[Gonzalo de la Vega]

The tunner IF for PAL-Nc norm, which AFAIK is used only in Argentina, was being defined as equal to PAL-M but it is not. It actually uses the same video IF as PAL-BG (and unlike PAL-M) but the audio is at 4.5MHz (same as PAL-M). A separate structure member was added for PAL-Nc.

Signed-off-by: Gonzalo A. de la Vega <gadelavega@gmail.com>


diff --git a/drivers/media/common/tuners/tda9887.c b/drivers/media/common/tuners/tda9887.c
index cdb645d..b560b5d 100644
--- a/drivers/media/common/tuners/tda9887.c
+++ b/drivers/media/common/tuners/tda9887.c
@@ -168,8 +168,8 @@ static struct tvnorm tvnorms[] = {
 			   cAudioIF_6_5   |
 			   cVideoIF_38_90 ),
 	},{
-		.std   = V4L2_STD_PAL_M | V4L2_STD_PAL_Nc,
-		.name  = "PAL-M/Nc",
+		.std   = V4L2_STD_PAL_M,
+		.name  = "PAL-M",
 		.b     = ( cNegativeFmTV  |
 			   cQSS           ),
 		.c     = ( cDeemphasisON  |
@@ -179,6 +179,17 @@ static struct tvnorm tvnorms[] = {
 			   cAudioIF_4_5   |
 			   cVideoIF_45_75 ),
 	},{
+		.std   = V4L2_STD_PAL_Nc,
+		.name  = "PAL-Nc",
+		.b     = ( cNegativeFmTV  |
+			   cQSS           ),
+		.c     = ( cDeemphasisON  |
+			   cDeemphasis75  |
+			   cTopDefault),
+		.e     = ( cGating_36     |
+			   cAudioIF_4_5   |
+			   cVideoIF_38_90 ),
+	},{
 		.std   = V4L2_STD_SECAM_B | V4L2_STD_SECAM_G | V4L2_STD_SECAM_H,
 		.name  = "SECAM-BGH",
 		.b     = ( cNegativeFmTV  |

Re: [PATCH] TDA9887 PAL-Nc fix

[Andy Walls]

On Wed, 2012-04-18 at 09:22 -0300, Gonzalo de la Vega wrote:
> The tunner IF for PAL-Nc norm, which AFAIK is used only in Argentina, was being defined as equal to PAL-M but it is not. It actually uses the same video IF as PAL-BG (and unlike PAL-M) but the audio is at 4.5MHz (same as PAL-M). A separate structure member was added for PAL-Nc.
> 
> Signed-off-by: Gonzalo A. de la Vega <gadelavega@gmail.com>

Hmmm.

The Video IF for N systems is 45.75 MHz according to this popular book
(see page 29 of the PDF):
http://www.deetc.isel.ipl.pt/Analisedesinai/sm/downloads/doc/ch08.pdf

The Video IF is really determined by the IF SAW filter used in your
tuner assembly, and how the tuner data sheet says to program the
mixer/oscillator chip to mix down from RF to IF.

What model analog tuner assembly are you using?  It could be that the
linux tuner-simple module is setting up the mixer/oscillator chip wrong.

Regards,
Andy

> 
> diff --git a/drivers/media/common/tuners/tda9887.c b/drivers/media/common/tuners/tda9887.c
> index cdb645d..b560b5d 100644
> --- a/drivers/media/common/tuners/tda9887.c
> +++ b/drivers/media/common/tuners/tda9887.c
> @@ -168,8 +168,8 @@ static struct tvnorm tvnorms[] = {
>  			   cAudioIF_6_5   |
>  			   cVideoIF_38_90 ),
>  	},{
> -		.std   = V4L2_STD_PAL_M | V4L2_STD_PAL_Nc,
> -		.name  = "PAL-M/Nc",
> +		.std   = V4L2_STD_PAL_M,
> +		.name  = "PAL-M",
>  		.b     = ( cNegativeFmTV  |
>  			   cQSS           ),
>  		.c     = ( cDeemphasisON  |
> @@ -179,6 +179,17 @@ static struct tvnorm tvnorms[] = {
>  			   cAudioIF_4_5   |
>  			   cVideoIF_45_75 ),
>  	},{
> +		.std   = V4L2_STD_PAL_Nc,
> +		.name  = "PAL-Nc",
> +		.b     = ( cNegativeFmTV  |
> +			   cQSS           ),
> +		.c     = ( cDeemphasisON  |
> +			   cDeemphasis75  |
> +			   cTopDefault),
> +		.e     = ( cGating_36     |
> +			   cAudioIF_4_5   |
> +			   cVideoIF_38_90 ),
> +	},{
>  		.std   = V4L2_STD_SECAM_B | V4L2_STD_SECAM_G | V4L2_STD_SECAM_H,
>  		.name  = "SECAM-BGH",
>  		.b     = ( cNegativeFmTV  |
> --
> To unsubscribe from this list: send the line "unsubscribe linux-media" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH] TDA9887 PAL-Nc fix

[Gonzalo A. de la Vega]

On Thu, Apr 19, 2012 at 8:50 PM, Andy Walls <awalls@md.metrocast.net> wrote:
> On Wed, 2012-04-18 at 09:22 -0300, Gonzalo de la Vega wrote:
>> The tunner IF for PAL-Nc norm, which AFAIK is used only in Argentina, was being defined as equal to PAL-M but it is not. It actually uses the same video IF as PAL-BG (and unlike PAL-M) but the audio is at 4.5MHz (same as PAL-M). A separate structure member was added for PAL-Nc.
>>
>> Signed-off-by: Gonzalo A. de la Vega <gadelavega@gmail.com>
>
> Hmmm.
>
> The Video IF for N systems is 45.75 MHz according to this popular book
> (see page 29 of the PDF):
> http://www.deetc.isel.ipl.pt/Analisedesinai/sm/downloads/doc/ch08.pdf
>
> The Video IF is really determined by the IF SAW filter used in your
> tuner assembly, and how the tuner data sheet says to program the
> mixer/oscillator chip to mix down from RF to IF.
>
> What model analog tuner assembly are you using?  It could be that the
> linux tuner-simple module is setting up the mixer/oscillator chip wrong.
>
> Regards,
> Andy

Hi Andy,
first of all and to clarify things: I could not tune analog TV without
this patch, or I could barely see a BW image. With the patch applied,
I can see image in full color and with good sound. So it works with
the patch, it does not work without it.

Now, I'm not an expert on TV (I am an electronics engineer thou) so I
am having some trouble trying to put together what I read in the
TDA9887 datasheet and the reference you sent. The thing with PAL-Nc is
that it has a video bandwidth of 4.2MHz not 5.0MHz (page 51) and the
attenuation of color difference signals for >20dB is at 3.6MHz instead
of 4MHz (page 54). You can just search for "Argentina" inside the
document.

So, this works... but now I'm not sure why. I guess cVideoIF_38_90 is
compensating for the bandwidth difference. I need to study this.

Gonzalo

>
>>
>> diff --git a/drivers/media/common/tuners/tda9887.c b/drivers/media/common/tuners/tda9887.c
>> index cdb645d..b560b5d 100644
>> --- a/drivers/media/common/tuners/tda9887.c
>> +++ b/drivers/media/common/tuners/tda9887.c
>> @@ -168,8 +168,8 @@ static struct tvnorm tvnorms[] = {
>>                          cAudioIF_6_5   |
>>                          cVideoIF_38_90 ),
>>       },{
>> -             .std   = V4L2_STD_PAL_M | V4L2_STD_PAL_Nc,
>> -             .name  = "PAL-M/Nc",
>> +             .std   = V4L2_STD_PAL_M,
>> +             .name  = "PAL-M",
>>               .b     = ( cNegativeFmTV  |
>>                          cQSS           ),
>>               .c     = ( cDeemphasisON  |
>> @@ -179,6 +179,17 @@ static struct tvnorm tvnorms[] = {
>>                          cAudioIF_4_5   |
>>                          cVideoIF_45_75 ),
>>       },{
>> +             .std   = V4L2_STD_PAL_Nc,
>> +             .name  = "PAL-Nc",
>> +             .b     = ( cNegativeFmTV  |
>> +                        cQSS           ),
>> +             .c     = ( cDeemphasisON  |
>> +                        cDeemphasis75  |
>> +                        cTopDefault),
>> +             .e     = ( cGating_36     |
>> +                        cAudioIF_4_5   |
>> +                        cVideoIF_38_90 ),
>> +     },{
>>               .std   = V4L2_STD_SECAM_B | V4L2_STD_SECAM_G | V4L2_STD_SECAM_H,
>>               .name  = "SECAM-BGH",
>>               .b     = ( cNegativeFmTV  |
>> --
>> To unsubscribe from this list: send the line "unsubscribe linux-media" in
>> the body of a message to majordomo@vger.kernel.org
>> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>
>

Re: [PATCH] TDA9887 PAL-Nc fix

[Mauro Carvalho Chehab]

Buenos dias Gonzalo,

Em 20-04-2012 11:01, Gonzalo A. de la Vega escreveu:
> On Thu, Apr 19, 2012 at 8:50 PM, Andy Walls <awalls@md.metrocast.net> wrote:
>> On Wed, 2012-04-18 at 09:22 -0300, Gonzalo de la Vega wrote:
>>> The tunner IF for PAL-Nc norm, which AFAIK is used only in Argentina, was being defined as equal to PAL-M but it is not. It actually uses the same video IF as PAL-BG (and unlike PAL-M) but the audio is at 4.5MHz (same as PAL-M). A separate structure member was added for PAL-Nc.
>>>
>>> Signed-off-by: Gonzalo A. de la Vega <gadelavega@gmail.com>
>>
>> Hmmm.
>>
>> The Video IF for N systems is 45.75 MHz according to this popular book
>> (see page 29 of the PDF):
>> http://www.deetc.isel.ipl.pt/Analisedesinai/sm/downloads/doc/ch08.pdf
>>
>> The Video IF is really determined by the IF SAW filter used in your
>> tuner assembly, and how the tuner data sheet says to program the
>> mixer/oscillator chip to mix down from RF to IF.
>>
>> What model analog tuner assembly are you using?  It could be that the
>> linux tuner-simple module is setting up the mixer/oscillator chip wrong.
>>
>> Regards,
>> Andy
> 
> Hi Andy,
> first of all and to clarify things: I could not tune analog TV without
> this patch, or I could barely see a BW image. With the patch applied,
> I can see image in full color and with good sound. So it works with
> the patch, it does not work without it.
> 
> Now, I'm not an expert on TV (I am an electronics engineer thou) so I
> am having some trouble trying to put together what I read in the
> TDA9887 datasheet and the reference you sent. The thing with PAL-Nc is
> that it has a video bandwidth of 4.2MHz not 5.0MHz (page 51) and the
> attenuation of color difference signals for >20dB is at 3.6MHz instead
> of 4MHz (page 54). You can just search for "Argentina" inside the
> document.
> 
> So, this works... but now I'm not sure why. I guess cVideoIF_38_90 is
> compensating for the bandwidth difference. I need to study this.

>From other discussions we've had at the ML, it seems that devices sold in
Argentina with analog tuners sometimes come with a NTSC tuner, and sometimes 
come with an European PAL tuner. They solve the frequency shifts that
happen there via some tda9887 (and/or tuner-simple) adjustments.

It seems that the setup, when using one type, is different than the other.

That's why we need to know exactly what it is the tuner that your device
has.

So, from time to time, we receive patches from someone in Argentina fixing
support for one type, but breaking support for the other type. 

What we need is that someone with technical expertise and with the two types
of devices, with access to real PAL-Nc signals, to work on a solution that
would set it accordingly, depending on the actual tuner used on it.

Regards,
Mauro

Re: [PATCH] TDA9887 PAL-Nc fix

[Ezequiel García]

Hola ;)

>
> From other discussions we've had at the ML, it seems that devices sold in
> Argentina with analog tuners sometimes come with a NTSC tuner, and sometimes
> come with an European PAL tuner. They solve the frequency shifts that
> happen there via some tda9887 (and/or tuner-simple) adjustments.
>
> It seems that the setup, when using one type, is different than the other.
>
> That's why we need to know exactly what it is the tuner that your device
> has.
>
> So, from time to time, we receive patches from someone in Argentina fixing
> support for one type, but breaking support for the other type.
>
> What we need is that someone with technical expertise and with the two types
> of devices, with access to real PAL-Nc signals, to work on a solution that
> would set it accordingly, depending on the actual tuner used on it.
>

I live in Argentina, and have access to digital signal and some
tuners. Perhaps I could
help, provided someone cares to send me the relevant devices?

Gonzalo: What device is that?

Re: [PATCH] TDA9887 PAL-Nc fix

[Andy Walls]

On Fri, 2012-04-20 at 11:01 -0300, Gonzalo A. de la Vega wrote:
> On Thu, Apr 19, 2012 at 8:50 PM, Andy Walls <awalls@md.metrocast.net> wrote:
> > On Wed, 2012-04-18 at 09:22 -0300, Gonzalo de la Vega wrote:
> >> The tunner IF for PAL-Nc norm, which AFAIK is used only in Argentina, was being defined as equal to PAL-M but it is not. It actually uses the same video IF as PAL-BG (and unlike PAL-M) but the audio is at 4.5MHz (same as PAL-M). A separate structure member was added for PAL-Nc.
> >>
> >> Signed-off-by: Gonzalo A. de la Vega <gadelavega@gmail.com>
> >
> > Hmmm.
> >
> > The Video IF for N systems is 45.75 MHz according to this popular book
> > (see page 29 of the PDF):
> > http://www.deetc.isel.ipl.pt/Analisedesinai/sm/downloads/doc/ch08.pdf
> >
> > The Video IF is really determined by the IF SAW filter used in your
> > tuner assembly, and how the tuner data sheet says to program the
> > mixer/oscillator chip to mix down from RF to IF.
> >
> > What model analog tuner assembly are you using?  It could be that the
> > linux tuner-simple module is setting up the mixer/oscillator chip wrong.
> >
> > Regards,
> > Andy
> 
> Hi Andy,
> first of all and to clarify things: I could not tune analog TV without
> this patch, or I could barely see a BW image. With the patch applied,
> I can see image in full color and with good sound. So it works with
> the patch, it does not work without it.

I believe you.  However, I beleive your fix is in the wrong place.
Every M/N tuner datahseet that I have seen, specifies a Video IF of
45.75 MHz.

> Now, I'm not an expert on TV (I am an electronics engineer thou) so I
> am having some trouble trying to put together what I read in the
> TDA9887 datasheet and the reference you sent.

Here's a datasheet for the LG TAPE-H091F tuner (M/N using a TDA9887):
http://dl.ivtvdriver.org/datasheets/tuners/TAPE-H091F_MK3.pdf
Look at the block diagram on page 26.  This design is very typical of
analog tuner assemblies.

The IF SAW filter is fixed.  I could program the mixer/oscialltor chip
and the TDA9887 IF decoder chip so the Video IF in use was 38.90 MHz,
and probably get a viewable TV picture.  That doesn't mean using 38.90
MHz is right, if the IF SAW filter is fixed and centered at 45.75 MHz.


>  The thing with PAL-Nc is
> that it has a video bandwidth of 4.2MHz not 5.0MHz (page 51) and the
> attenuation of color difference signals for >20dB is at 3.6MHz instead
> of 4MHz (page 54). You can just search for "Argentina" inside the
> document.
> 

Yes, that's fine.

The Video IF is what I am concerned about.  Your change could break
other tuners and/or the linux tuner-simple.ko module is programming
some(?) tuners wrong for PAL-N and PAL-Nc

By the way, this document describes PAL Combination N (-Nc) as a
difference from PAL-N:
ITU-R BT.470-6
http://www.itu.int/dms_pubrec/itu-r/rec/bt/R-REC-BT.470-6-199811-S!!PDF-E.pdf


Argentine Law 21.895, October 1978, wasn't much help in describing the
technical aspects of PAL-Nc:
http://www.infojus.gov.ar/index.php?kk_seccion=documento&registro=LEYNAC&docid=LEY%2520C%2520021895%25201978%252010%252030


> So, this works... but now I'm not sure why.

Likely the tuner-simple.ko module is programming the mixer oscillator
component for the 38.90 MHz IF.  That may be right or wrong, depending
on you exact tuner model.

Again, what exact tuner assembly are you using?

>  I guess cVideoIF_38_90 is
> compensating for the bandwidth difference. I need to study this.

Nope.  It is just matching how the tuner-simple.ko module programmed the
mixer/oscillator chip.

Regards,
Andy

> Gonzalo
> 
> >
> >>
> >> diff --git a/drivers/media/common/tuners/tda9887.c b/drivers/media/common/tuners/tda9887.c
> >> index cdb645d..b560b5d 100644
> >> --- a/drivers/media/common/tuners/tda9887.c
> >> +++ b/drivers/media/common/tuners/tda9887.c
> >> @@ -168,8 +168,8 @@ static struct tvnorm tvnorms[] = {
> >>                          cAudioIF_6_5   |
> >>                          cVideoIF_38_90 ),
> >>       },{
> >> -             .std   = V4L2_STD_PAL_M | V4L2_STD_PAL_Nc,
> >> -             .name  = "PAL-M/Nc",
> >> +             .std   = V4L2_STD_PAL_M,
> >> +             .name  = "PAL-M",
> >>               .b     = ( cNegativeFmTV  |
> >>                          cQSS           ),
> >>               .c     = ( cDeemphasisON  |
> >> @@ -179,6 +179,17 @@ static struct tvnorm tvnorms[] = {
> >>                          cAudioIF_4_5   |
> >>                          cVideoIF_45_75 ),
> >>       },{
> >> +             .std   = V4L2_STD_PAL_Nc,
> >> +             .name  = "PAL-Nc",
> >> +             .b     = ( cNegativeFmTV  |
> >> +                        cQSS           ),
> >> +             .c     = ( cDeemphasisON  |
> >> +                        cDeemphasis75  |
> >> +                        cTopDefault),
> >> +             .e     = ( cGating_36     |
> >> +                        cAudioIF_4_5   |
> >> +                        cVideoIF_38_90 ),
> >> +     },{
> >>               .std   = V4L2_STD_SECAM_B | V4L2_STD_SECAM_G | V4L2_STD_SECAM_H,
> >>               .name  = "SECAM-BGH",
> >>               .b     = ( cNegativeFmTV  |
> >> --
> >> To unsubscribe from this list: send the line "unsubscribe linux-media" in
> >> the body of a message to majordomo@vger.kernel.org
> >> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> >
> >
> --
> To unsubscribe from this list: send the line "unsubscribe linux-media" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html