Decoding FSK

>HardySpicer wrote:
>> On Sep 21, 10:21 am, Jon Mcleod <jonmcleod2...@yahoo.com> wrote:
>>> A Bell 202T modem uses FSK modulation (1200HZ, 2200HZ) and send data
at
>>> up to 1800 bits per second.
>>>
>>> I need to replace an "analog" version of this with a digital version,
>>> sampling the phone line with an A/D and "decoding" the 1's and 0's in
>>> firmware (ARM C).
>> Why? Analogue is geberally simpler and better for some applications.
>> Why re-invent the wheel?
>> 
>
>The old system (analog modem + Z80 cpu) has a bunch of obsolete parts. 
>They just want to redo it.  HOWEVER, that being said, the Arm7 in the 
>new design isnt going to be able to "do the math" in real time (the 
>angle function) so maybe you're on to something..

Is that because the ARM7 is already heavily loaded, because an ARM7 is far
more than you need for even the most sophisticated form of FSK
demodulation.

Steve

Rafael Deliano  <Rafael_DelianoENTFERNEN@t-online.de> wrote:

>Sorry its in german, but the pictures should
>be clear enough:
>http://www.embeddedforth.de/temp/fsk.pdf
>Its the usual delay & multiply version.

Seems to me such an approach is along the lines of a differential
demodulator, and as such loses 3 dB.   Or maybe it doesn't,
for some reason.

Have you (or anyone) implemented this and noted the performance?

Steve

On Sep 23, 12:22�am, spop...@speedymail.org (Steve Pope) wrote:
> Rafael Deliano &#4294967295;<Rafael_DelianoENTFER...@t-online.de> wrote:
>
> >Sorry its in german, but the pictures should
> >be clear enough:
> >http://www.embeddedforth.de/temp/fsk.pdf
> >Its the usual delay & multiply version.
>
> Seems to me such an approach is along the lines of a differential
> demodulator, and as such loses 3 dB. &#4294967295; Or maybe it doesn't,
> for some reason.
>
> Have you (or anyone) implemented this and noted the performance?
>
> Steve

Not everyone cares. The commonest use for Bell 202 these days is for
caller ID, and related analogue display services. The signal is so
clean with just one analogue line segment, that even at the copper
planning limit the receiver can get away with murder and be error
free. Saving a few cents, on the other hand, might be a key design
constraint.

Steve

 <steveu@coppice.org> wrote:

>On Sep 23, 12:22&#4294967295;am, spop...@speedymail.org (Steve Pope) wrote:

>> Seems to me such an approach is along the lines of a differential
>> demodulator, and as such loses 3 dB. &#4294967295; Or maybe it doesn't,
>> for some reason.

>> Have you (or anyone) implemented this and noted the performance?

>Not everyone cares. The commonest use for Bell 202 these days is for
>caller ID, and related analogue display services. The signal is so
>clean with just one analogue line segment, that even at the copper
>planning limit the receiver can get away with murder and be error
>free. Saving a few cents, on the other hand, might be a key design
>constraint.

I'm sure nobody cares, so long as it works.  OTOH the OP may
be facing a spec he needs to meet.  If not, it would not be too 
difficult to measure BER vs. SNR for the existing analog modem, 
so that one has a target.

(I understand some receiving modems of that vintage were implemented
as bandpass filters followed by level detectors -- a pretty
suboptimal modem design.)

I didn't realize caller ID was the likely application, thanks.

Steve

> http://www.embeddedforth.de/temp/fsk.pdf

Seems to me the site is not responding :(

-- 
Christen Fihl

> Seems to me the site is not responding :(
Sorry, i got an unplanned "server-upgrade"
that fouls it up for several days.

MfG  JRD

> Sorry, i got an unplanned "server-upgrade" that fouls it up for several 
> days.
Sorry to hear that.

Tried http://www.archive.org/web/web.php to see if they had copies :-)
But not easy

Looking forward for better days, good luck
Christen

>> Its the usual delay & multiply version. 
> Seems to me such an approach is along the lines of a differential
> demodulator, 
I would view it as an autocorrelator, but its usually not called
that in literature.

> Have you (or anyone) implemented this and noted the performance?
I have not yet come around to do it on a 68HC908 decode CLIP.
But it is the usual FSK-decoder for phone-line applications.
It seems less popular for radio.

Even the murky version where the multiplier is replaced by
a XOR-gate ( Bild 5 in my text ) was implemented in very early
300 baud modem-ICs.

Bell Labs had a look at 3 versions of digital FSK-decoders
and seemed to have liked it ( page 3 ):
http://www.embeddedforth.de/temp/fsk1.pdf
Note that they have 8kHz samplerate ( PCM-Codec )
and therefore problems to get the proper delay. They put a
fractional delay in to fix that. On a microprocessor one would
choose a more appropriate samplerate.

AMD did a coherent FSK ( Page 10 picture 7 ):
http://www.embeddedforth.de/temp/pskfsk..pdf

MfG  JRD

Works now, thanks

Christen Fihl

>>> Its the usual delay & multiply version. 
>> Seems to me such an approach is along the lines of a differential
>> demodulator, 
>I would view it as an autocorrelator, but its usually not called
>that in literature.
>
>> Have you (or anyone) implemented this and noted the performance?
>I have not yet come around to do it on a 68HC908 decode CLIP.
>But it is the usual FSK-decoder for phone-line applications.
>It seems less popular for radio.
>
>Even the murky version where the multiplier is replaced by
>a XOR-gate ( Bild 5 in my text ) was implemented in very early
>300 baud modem-ICs.
>
>Bell Labs had a look at 3 versions of digital FSK-decoders
>and seemed to have liked it ( page 3 ):
>http://www.embeddedforth.de/temp/fsk1.pdf
>Note that they have 8kHz samplerate ( PCM-Codec )
>and therefore problems to get the proper delay. They put a
>fractional delay in to fix that. On a microprocessor one would
>choose a more appropriate samplerate.
>
>AMD did a coherent FSK ( Page 10 picture 7 ):
>http://www.embeddedforth.de/temp/pskfsk..pdf
>
>MfG  JRD

AMD used coherent demodulation because they could do so at essentially no
cost. Their device already possessed the DSP crunching ability to do simple
QAM, so it needed only marginal extra resources to reprogram that DSP
hardware for a sophisticated method of FSK demodulation. Devices from the
same era which were only FSK demodulators generally used more basic
techniques, for cost reasons. Their goal was not highest performance, but a
competitive price. Many of these devices performed quite poorly, but were
priced to go.

Regards,
Steve