Where should I begin

Hey all,

Recently at work I've come upon an interesting challenge (luckily it's
not something I need to do FOR work, so there is no hard deadline).
Some of you might be aware of the old "teletext" system that was used
to convey reasonably static "pages" of information to television sets,
during the Vertical Blanking Interval - a period of time in which the
electron gun is moving from the bottom corner of the screen back up to
the top corner to commence drawing the next field of vision - because
there is no need to transmit visual information at this point, other
forms of information (such as lines of teletext pages) may be
transmitted.

Very very briefly, there are approx. 16 video lines per field of
vision available to carry data.  For PAL-SD video, there are 720
pixels (or samples) per line.  The EBU-WST specification defines the
parameters for encoding 45 bytes (360 bits) of data into these 720
samples (actually, a number of samples at the beginning and end of the
line are not utilised).  I'm keen to learn how to generate the 720
samples from the original 45 bytes (and preferably to also work
backwards from a line of 720 samples to 45 bytes of data).

Previously I have seen this conversion done using an FIR filter.  This
has lead me to start reading "Understanding digital signal
processing", as I think this is the right place to start.  Having
started to read this book, I've realized my general knowledge of
mathematics is pretty poor these days.

I guess the questions I have right now are:
- What are the mathematical topics generally required when learning /
working with DSP?
- Can anyone recommend a good book for re-learning these topics?
- Am I on the right track, studying DSP (esp. FIR filters) to
eventually understand how to go from 360 bits to 720 samples?

Cheers,

James

james.fysh <james.fysh@gmail.com> wrote:
 
> Recently at work I've come upon an interesting challenge (luckily it's
> not something I need to do FOR work, so there is no hard deadline).
> Some of you might be aware of the old "teletext" system that was used
> to convey reasonably static "pages" of information to television sets,
> during the Vertical Blanking Interval - a period of time in which the
> electron gun is moving from the bottom corner of the screen back up to
> the top corner to commence drawing the next field of vision - because
> there is no need to transmit visual information at this point, other
> forms of information (such as lines of teletext pages) may be
> transmitted.

Some of that is still there.   One use is for closed captions.

Also, I have a DVD recorder connected to the output of a DTV
converter box, and it knows the name of the programs it records.
That must be somewhere in the analog signal, and VBI seems a
good place.

-- glen

On Mar 7, 11:41=A0am, glen herrmannsfeldt <g...@ugcs.caltech.edu> wrote:
> james.fysh <james.f...@gmail.com> wrote:
> > Recently at work I've come upon an interesting challenge (luckily it's
> > not something I need to do FOR work, so there is no hard deadline).
> > Some of you might be aware of the old "teletext" system that was used
> > to convey reasonably static "pages" of information to television sets,
> > during the Vertical Blanking Interval - a period of time in which the
> > electron gun is moving from the bottom corner of the screen back up to
> > the top corner to commence drawing the next field of vision - because
> > there is no need to transmit visual information at this point, other
> > forms of information (such as lines of teletext pages) may be
> > transmitted.
>
> Some of that is still there. =A0 One use is for closed captions.
Indeed - that's what we're using it for at work.  I think closed
captions are one of the few reasons that VBI has held on for so long.
Recent standards for digital broadcast (DVB) include methods for
carrying the 45-byte VBI data in a digital representation, and
presumably most modern DVB-Set-top boxes have software/hardware in
them for re-creating the 720-luma samples to output as composite/
component video so that the tv set may then capture and decode those
samples back into teletext pages/captions for display!
>
> Also, I have a DVD recorder connected to the output of a DTV
> converter box, and it knows the name of the programs it records.
> That must be somewhere in the analog signal, and VBI seems a
> good place.
Very interesting - so it is connected via composite / component and no
digital (e.g. EIT / EPG) information can possibly be making its way to
the DVD recorder?
>
> -- glen

On Sun, 06 Mar 2011 13:55:26 -0800, james.fysh wrote:

> Hey all,
> 
> Recently at work I've come upon an interesting challenge (luckily it's
> not something I need to do FOR work, so there is no hard deadline). Some
> of you might be aware of the old "teletext" system that was used to
> convey reasonably static "pages" of information to television sets,
> during the Vertical Blanking Interval - a period of time in which the
> electron gun is moving from the bottom corner of the screen back up to
> the top corner to commence drawing the next field of vision - because
> there is no need to transmit visual information at this point, other
> forms of information (such as lines of teletext pages) may be
> transmitted.
> 
> Very very briefly, there are approx. 16 video lines per field of vision
> available to carry data.  For PAL-SD video, there are 720 pixels (or
> samples) per line.  The EBU-WST specification defines the parameters for
> encoding 45 bytes (360 bits) of data into these 720 samples (actually, a
> number of samples at the beginning and end of the line are not
> utilised).  I'm keen to learn how to generate the 720 samples from the
> original 45 bytes (and preferably to also work backwards from a line of
> 720 samples to 45 bytes of data).
> 
> Previously I have seen this conversion done using an FIR filter.  This
> has lead me to start reading "Understanding digital signal processing",
> as I think this is the right place to start.  Having started to read
> this book, I've realized my general knowledge of mathematics is pretty
> poor these days.
> 
> I guess the questions I have right now are: - What are the mathematical
> topics generally required when learning / working with DSP?

Algebra, trigonometry, and calculus.  Knowing differential equations 
doesn't hurt.

> - Can anyone recommend a good book for re-learning these topics? -

That's a Really Good question -- and I have no clue for the answers!  If 
you find any titles with previews, post them here and if I'm not 
inundated with work I'll see if I can help out.

Here's the title you want: "All That Math You Thought You'd Never Need, 
and Now You Do".

> Am I
> on the right track, studying DSP (esp. FIR filters) to eventually
> understand how to go from 360 bits to 720 samples?

Without knowing just how they're massaging the data to get samples -- 
maybe.  It's probably simpler than that.  Certainly if you have the 
grounding to know enough DSP to understand FIR filters, you'll have the 
grounding you need to understand the modulation of digital data.

-- 
http://www.wescottdesign.com

The data is just binary encoded and resampled to just under 2 samples per b=
it (check the standard for the exact number, it won't be the entire 720 sam=
ples).

What you need to do for generating this stuff is just create the binary rep=
resentation of the data and whatever headers, etc. are necessary. Then resa=
mple it to the right rate. That's about it. It's a fractional upsampling ra=
te, so you can either upsample by (lets say it's 1.9 samples/bit) 19 then d=
ownsample by 10 (read up on requirements for filters for resampling) or lea=
rn about polyphase filters if you want to do it on the fly. The standard mi=
ght also call for something like manchester encoding - see wikipedia - but =
I don't believe teletext uses that, I think it's just binary - check the st=
andard to be sure.

Chris

james.fysh <james.fysh@gmail.com> wrote:
(snip, I wrote)

>> Also, I have a DVD recorder connected to the output of a DTV
>> converter box, and it knows the name of the programs it records.
>> That must be somewhere in the analog signal, and VBI seems a
>> good place.

> Very interesting - so it is connected via composite / component and no
> digital (e.g. EIT / EPG) information can possibly be making its way to
> the DVD recorder?

S-video, so the resolution isn't so bad as composite.

It would be nice to know the name for the actual signal, as
the local PBS station doesn't seem to use it.  It isn't the
one used for the DISPLAY feature of HDTVs.

-- glen

Tons of it is still there. The broadcast world is chock full of data packet=
s for all sorts of stuff, AFD, closed captions, timecode, audio, various au=
tomation standards, etc. SMPTE, the standards body for most of this stuff, =
has 75 different packet types registered. Except audio, closed captions, an=
d a couple others, much of it never leaves the broadcast facility, but it's=
 critical in that space. H and V blanking are never going away at least par=
tially for that reason.

In the standard definition world, some of that stuff is encoded as analog w=
aveforms (i.e. closed captions), effectively as PCM, in what is otherwise h=
andled by broadcasters as a digital stream.

Chris

Thanks for all the info - I think I'll just bury my head in books for
a while.  Speaking of that, I'm trying to work through the problems
sections in Understanding Digital Signal Processing and getting
stuck.  There don't seem to be any solutions in the book, or elsewhere
that I can find (googling for "understanding digital signal
processing" just brings up hundreds of sites offering pirated copies
of the book.. sad :().  Is there anywhere in particular I can look for
solutions or hints - otherwise my self-study sessions are going to be
very long and slow!

On Mar 6, 4:55=A0pm, "james.fysh" <james.f...@gmail.com> wrote:
> Hey all,
>
> Recently at work I've come upon an interesting challenge (luckily it's
> not something I need to do FOR work, so there is no hard deadline).
> Some of you might be aware of the old "teletext" system that was used
> to convey reasonably static "pages" of information to television sets,
> during the Vertical Blanking Interval - a period of time in which the
> electron gun is moving from the bottom corner of the screen back up to
> the top corner to commence drawing the next field of vision - because
> there is no need to transmit visual information at this point, other
> forms of information (such as lines of teletext pages) may be
> transmitted.
>
> Very very briefly, there are approx. 16 video lines per field of
> vision available to carry data. =A0For PAL-SD video, there are 720
> pixels (or samples) per line. =A0The EBU-WST specification defines the
> parameters for encoding 45 bytes (360 bits) of data into these 720
> samples (actually, a number of samples at the beginning and end of the
> line are not utilised). =A0I'm keen to learn how to generate the 720
> samples from the original 45 bytes (and preferably to also work
> backwards from a line of 720 samples to 45 bytes of data).
>
> Previously I have seen this conversion done using an FIR filter. =A0This
> has lead me to start reading "Understanding digital signal
> processing", as I think this is the right place to start. =A0Having
> started to read this book, I've realized my general knowledge of
> mathematics is pretty poor these days.
>
> I guess the questions I have right now are:
> - What are the mathematical topics generally required when learning /
> working with DSP?
> - Can anyone recommend a good book for re-learning these topics?
> - Am I on the right track, studying DSP (esp. FIR filters) to
> eventually understand how to go from 360 bits to 720 samples?
>
> Cheers,
>
> James

You can try my website:

www.fourier-series.com

It covers fourier series, complex numbers, DFT, Fourier transform,
convolution, digital convolution and some basic FIR theory.

On Mar 6, 10:55=A0pm, "james.fysh" <james.f...@gmail.com> wrote:

> I guess the questions I have right now are:
> - What are the mathematical topics generally required when learning /
> working with DSP?
> - Can anyone recommend a good book for re-learning these topics?

Any maths book by Gilbert Strang:

http://www.amazon.com/s/ref=3Dnb_sb_noss?url=3Dsearch-alias%3Daps&field-key=
words=3Dgilbert+strang

Rune