V.27ter information needed and what are the main phases for decoding v.27 faxes

where can I find more information about V.27ter other than ITU
standard.

And what are the main phases for decoding V.27ter faxes???(i.e:
demodulation descrambling....?) and what else.

amoor_netx@yahoo.com (amoorS) wrote in message news:<8903a9c7.0311110547.2a22b032@posting.google.com>...
> where can I find more information about V.27ter other than ITU
> standard.

Modem Books, articles, papers, TI Application notes. And not necessarily on V.27.
 
> And what are the main phases for decoding V.27ter faxes???(i.e:
> demodulation descrambling....?) and what else.

- Demodulation
- Equalizer !
- Timing and Carrier Recovery
- Demapping/Symbol Decoding
- Descrambling

Thats basically it. Hope this answers your other question as well. 

Sachin

scngupta@yahoo.com (Sachin Gupta) wrote in message news:<d2308724.0311122127.7e8a44a7@posting.google.com>...
> amoor_netx@yahoo.com (amoorS) wrote in message news:<8903a9c7.0311110547.2a22b032@posting.google.com>...
> > where can I find more information about V.27ter other than ITU
> > standard.
> 
> Modem Books, articles, papers, TI Application notes. And not necessarily on V.27.
>  
> > And what are the main phases for decoding V.27ter faxes???(i.e:
> > demodulation descrambling....?) and what else.
> 
> - Demodulation
> - Equalizer !
> - Timing and Carrier Recovery
> - Demapping/Symbol Decoding
> - Descrambling
> 
> Thats basically it. Hope this answers your other question as well. 
> 
> Sachin






Thank You Sachin for your help.
And I have  questions about your answer:

1- Is it ordered correctly or not (i.e. Demodulation then Equalizer
then Timing and Carrier Recovery then Demapping/Symbol Decoding then
Descrambling) .

2- what is Timing and Carrier Recovery (If you have web site to my
answer please write it)

3-does all these phases can be implemented by software?
4- IS AGC(autmatic Gain Control) necessary or not?
5- I have Descrambler and demodulator for V.27ter, Can You trace it to
see it is correct or not because I have some Problems in fax decoding.

Thank You Sachin Gupta

> > > And what are the main phases for decoding V.27ter faxes???(i.e:
> > > demodulation descrambling....?) and what else.
> > 
> > - Demodulation
> > - Equalizer !
> > - Timing and Carrier Recovery
> > - Demapping/Symbol Decoding
> > - Descrambling
> > 
> > Thats basically it. Hope this answers your other question as well. 
> > 
> > Sachin
>
> Thank You Sachin for your help.
> And I have  questions about your answer:
> 
> 1- Is it ordered correctly or not (i.e. Demodulation then Equalizer
> then Timing and Carrier Recovery then Demapping/Symbol Decoding then
> Descrambling) .

Depends on your design. Carrier Recovery usually follows Equalizer.
Demapping / Symbol Decoding is in the end.

> 2- what is Timing and Carrier Recovery (If you have web site to my
> answer please write it)

Don't remember any site offhand - but lots of modem
papers/books/tutorials out there.

> 3-does all these phases can be implemented by software?
Absolutely. 

> 4- IS AGC(autmatic Gain Control) necessary or not?
Yes.

> 5- I have Descrambler and demodulator for V.27ter, Can You trace it to
> see it is correct or not because I have some Problems in fax decoding.

What kind of problems ? 

Sachin

scngupta@yahoo.com (Sachin Gupta) wrote in message news:<d2308724.0311150726.642223f@posting.google.com>...
> > > > And what are the main phases for decoding V.27ter faxes???(i.e:
> > > > demodulation descrambling....?) and what else.
> > > 
> > > - Demodulation
> > > - Equalizer !
> > > - Timing and Carrier Recovery
> > > - Demapping/Symbol Decoding
> > > - Descrambling
> > > 
> > > Thats basically it. Hope this answers your other question as well. 
> > > 
> > > Sachin
> >
> > Thank You Sachin for your help.
> > And I have  questions about your answer:
> > 
> > 1- Is it ordered correctly or not (i.e. Demodulation then Equalizer
> > then Timing and Carrier Recovery then Demapping/Symbol Decoding then
> > Descrambling) .
> 
> Depends on your design. Carrier Recovery usually follows Equalizer.
> Demapping / Symbol Decoding is in the end.
> 
> > 2- what is Timing and Carrier Recovery (If you have web site to my
> > answer please write it)
> 
> Don't remember any site offhand - but lots of modem
> papers/books/tutorials out there.
> 
> > 3-does all these phases can be implemented by software?
> Absolutely. 
> 
> > 4- IS AGC(autmatic Gain Control) necessary or not?
> Yes.
> 
> > 5- I have Descrambler and demodulator for V.27ter, Can You trace it to
> > see it is correct or not because I have some Problems in fax decoding.
> 
> What kind of problems ? 
> 
> Sachin

Thank you Sachin 

The problem is :
    
   I have implement a demodulator (DQPSK for 2400 b/s) and a
descrambler (1+x6+x7)and decoder (MH) only.
 And then I sent a fax using MH for encoding to test the demodulator,
the descrambler and the decoder but when I display the fax only (30%
of the fax is ok) but the remaining parts are garbage. So, what is the
problem? is it necessary to implement the equalizer or there is some
bugs in the demodulator or descrambler???

And if you have to see the code to check I'll post it here
And If is the equalizer necessary, How can I program It (Links to web
site about equalizer implementation because the V.27ter is not clear)


Thank You Sachin

amoor_netx@yahoo.com (amoorS) wrote in message > > Sachin
> Thank you Sachin 
> 
> The problem is :
>     
>    I have implement a demodulator (DQPSK for 2400 b/s) and a
> descrambler (1+x6+x7)and decoder (MH) only.
>  And then I sent a fax using MH for encoding to test the demodulator,
> the descrambler and the decoder but when I display the fax only (30%
> of the fax is ok) but the remaining parts are garbage. So, what is the
> problem? is it necessary to implement the equalizer or there is some
> bugs in the demodulator or descrambler???
> 
> And if you have to see the code to check I'll post it here
> And If is the equalizer necessary, How can I program It (Links to web
> site about equalizer implementation because the V.27ter is not clear)
> 
> 
> Thank You Sachin

Your descrambler is wrong. It should not be just 1 + x^6 + x^7. There
are some fiddle factors you need to deal with for long repeating
sequences. This is specified by a rather confusing logic diagram in
the V.27ter spec. I went round in circles a few times recently until I
got it right. Just using 1 + x^6 + x^7 will give about the error rate
you described.

Regards,
Steve

steveu@coppice.org (Steve Underwood) wrote in message news:<80bddbd7.0311160437.6c3e1c8e@posting.google.com>...
> amoor_netx@yahoo.com (amoorS) wrote in message > > Sachin
> > Thank you Sachin 
> > 
> > The problem is :
> >     
> >    I have implement a demodulator (DQPSK for 2400 b/s) and a
> > descrambler (1+x6+x7)and decoder (MH) only.
> >  And then I sent a fax using MH for encoding to test the demodulator,
> > the descrambler and the decoder but when I display the fax only (30%
> > of the fax is ok) but the remaining parts are garbage. So, what is the
> > problem? is it necessary to implement the equalizer or there is some
> > bugs in the demodulator or descrambler???

What's MH ? 
I have a couple of questions to ask:

- Have you plotted your constellation diagram at the receiver side,
does it look something like the ones for DPSK ?

- Equalizer is usually necessary, but depending on what your channel
conditions are - you might not need it. Are you sending data on
telephone line ?

- Remember V.27ter mapping and demapping/symbol detection is
differential PSK. The information is modulated in phase changes and
not absolute phases.

> > And if you have to see the code to check I'll post it here
> > And If is the equalizer necessary, How can I program It (Links to web
> > site about equalizer implementation because the V.27ter is not clear)

Take a look at TI's application note on channel equalization.
Equalizer implementation is something which will take some time to
study. Have you studied Least Mean Squares (LMS) Equalization. If not
you can study a standard Digital Communications text book to
understand it or any advanced text on Signal Processing.

> Your descrambler is wrong. It should not be just 1 + x^6 + x^7. There
> are some fiddle factors you need to deal with for long repeating
> sequences. This is specified by a rather confusing logic diagram in
> the V.27ter spec. I went round in circles a few times recently until I
> got it right. Just using 1 + x^6 + x^7 will give about the error rate
> you described.

Ah ! The infamous V.27ter scrambler - I struggled to understand it for
quite some time till a comp.dsp veteran helped me.

Figure I-I/V.27ter is especially helpful (you can try to implement
whats happening inside it), the text on the other hand is very
unhelpful. I hardly remember how I implemented it. But you should
definitely study it for some time and try and implement it on your
own. It is definitely worth the effort.

Sachin

scngupta@yahoo.com (Sachin Gupta) wrote in message [...]
> What's MH ? 

Modified Huffman coding. Its the simplest 1-D compression scheme
specified for FAXes in the T.6 spec.

[...]
> Ah ! The infamous V.27ter scrambler - I struggled to understand it for
> quite some time till a comp.dsp veteran helped me.
> 
> Figure I-I/V.27ter is especially helpful (you can try to implement
> whats happening inside it), the text on the other hand is very
> unhelpful. I hardly remember how I implemented it. But you should
> definitely study it for some time and try and implement it on your
> own. It is definitely worth the effort.

Isn't it sad when the only ways to figure out what the spec means are
so round about. My solution was to sent FAXes, and play around with
the scrambler until they came out right :-( Decrypting the scrambler
requirement took me longer than implementing the rest of the modem.

The text doesn't even try to be helpful. It just refers you to the
rather woolly diagram. The diagram doesn't show the whole timing
generation, and relies on some flaky design. That is why I ended up
groping in the dark for the intended solution. Once you get something
that works, its functionality doesn't seem entirely meaningful.

Regards,
Steve

steveu@coppice.org (Steve Underwood) wrote in message news:<80bddbd7.0311181715.20986950@posting.google.com>...
> scngupta@yahoo.com (Sachin Gupta) wrote in message [...]
> > What's MH ? 
> 
> Modified Huffman coding. Its the simplest 1-D compression scheme
> specified for FAXes in the T.6 spec.

Thanks. Never worked on that part. 

<snip>

> Isn't it sad when the only ways to figure out what the spec means are
> so round about. My solution was to sent FAXes, and play around with
> the scrambler until they came out right :-( Decrypting the scrambler
> requirement took me longer than implementing the rest of the modem.

Remember the spec is dated 1970's. I was born in late 1970's. May be
that was the best that they could do at that time, but could have
modified it later on. Later specs (V.29 and V.17) didn't repeat that
same mistake. Probably nobody thought that the effort was worth it - I
mean ITU-T resolution etc. Also, if you think what will you change to
that specification. I won't dare to propose to change the Text - I
understand the figure (with flip-flops, gates) and counters and
implemented that, I still don't understand the text and how to change
it. Even if we did - which company will fund the effort for an ITU
resolution ? Mine won't.
 
> The text doesn't even try to be helpful. It just refers you to the
> rather woolly diagram. The diagram doesn't show the whole timing
> generation, and relies on some flaky design. That is why I ended up
> groping in the dark for the intended solution. Once you get something
> that works, its functionality doesn't seem entirely meaningful.

It is what that guy called "Character Building Exercise". I did it at
his suggestion and prodding and am glad that I did it. Undertsanding
it is so round-about that you end up building a lot of character
instead. Helps a lot in your career :-)

Sachin

scngupta@yahoo.com (Sachin Gupta) wrote in message 
> 
> Remember the spec is dated 1970's. I was born in late 1970's. May be
> that was the best that they could do at that time, but could have
> modified it later on. Later specs (V.29 and V.17) didn't repeat that
> same mistake. Probably nobody thought that the effort was worth it - I
> mean ITU-T resolution etc. Also, if you think what will you change to
> that specification. I won't dare to propose to change the Text - I
> understand the figure (with flip-flops, gates) and counters and
> implemented that, I still don't understand the text and how to change
> it. Even if we did - which company will fund the effort for an ITU
> resolution ? Mine won't.

Of what relevance is the date? I was working in the late 1970s, and if
I had written a spec like that it would have rightly been thrown back
in my face, with vociferous complaints. Actually the V.27ter spec.
appears to be newer than the V.29 spec (V.27ter is somewhat later than
V.27). V.29 is of late 70s vintage, and appears to have had no updates
since first publication. The grotty point in the V.29 spec. is it
doesn't fully define the transmit side filter. V.27ter says RRC, but
V.29 is open ended.

Regards,
Steve