From: Tim Wescott on
Jerry Avins wrote:
> Tim Wescott wrote:
>> Tauno Voipio wrote:
>>> On 27.2.10 4:02 , sven98de wrote:
>>>> Hi,
>>>> has somebody here implemented a RTTY (AFSK) decoder. Any suggestions
>>>> how to
>>>> proceed ?
>>>> I've found some technical abstracts about FSK demodulation in the WEB
>>>> like:
>>>> - frequency discriminator with IIR resonator
>>>> - frequency discriminator with FIR bandpass filter
>>>> - digital PLL
>>>>
>>>> Any idea how complex it is to implement one of these methods ?
>>>> What processor is sufficient for this, does a microcontroller the job ?
>>>
>>> Do you like to use it on real band conditions, or is this just
>>> an exercise.
>>>
>>> When there is QRM and QRN on the channel, a PLL is not a good
>>> idea, as it will pretty easily grab an interfering signal.
>>>
>>> The traditional TU (terminal unit) has two analog filters,
>>> one for mark and one for space. If you like to use it as a
>>> model, make a single pole-pair IIR resonator for both
>>> frequencies and compare the amplitudes.
>>
>> Or I/Q demodulate at both the mark and space frequency, do a running
>> average (a _real_ running average, not an exponential decay) for one
>> bit period, and choose the I/Q pair with the best magnitude. If you
>> deal with the synchronization issue, you can just sum-and-dump instead
>> of doing the running average.
>
> By "real running average" as opposed to "sum and dump", Do you mean
> forever and ever amen? The sensitivity would tank after a few bit times
> if you did that.

Boxcar filter, running average, y(t) = integral from (t - T0) to t dt, etc.

--
Tim Wescott
Control system and signal processing consulting
www.wescottdesign.com
From: Jerry Avins on
Tim Wescott wrote:
> Jerry Avins wrote:
>> Tim Wescott wrote:
>>> Tauno Voipio wrote:
>>>> On 27.2.10 4:02 , sven98de wrote:
>>>>> Hi,
>>>>> has somebody here implemented a RTTY (AFSK) decoder. Any
>>>>> suggestions how to
>>>>> proceed ?
>>>>> I've found some technical abstracts about FSK demodulation in the WEB
>>>>> like:
>>>>> - frequency discriminator with IIR resonator
>>>>> - frequency discriminator with FIR bandpass filter
>>>>> - digital PLL
>>>>>
>>>>> Any idea how complex it is to implement one of these methods ?
>>>>> What processor is sufficient for this, does a microcontroller the
>>>>> job ?
>>>>
>>>> Do you like to use it on real band conditions, or is this just
>>>> an exercise.
>>>>
>>>> When there is QRM and QRN on the channel, a PLL is not a good
>>>> idea, as it will pretty easily grab an interfering signal.
>>>>
>>>> The traditional TU (terminal unit) has two analog filters,
>>>> one for mark and one for space. If you like to use it as a
>>>> model, make a single pole-pair IIR resonator for both
>>>> frequencies and compare the amplitudes.
>>>
>>> Or I/Q demodulate at both the mark and space frequency, do a running
>>> average (a _real_ running average, not an exponential decay) for one
>>> bit period, and choose the I/Q pair with the best magnitude. If you
>>> deal with the synchronization issue, you can just sum-and-dump
>>> instead of doing the running average.
>>
>> By "real running average" as opposed to "sum and dump", Do you mean
>> forever and ever amen? The sensitivity would tank after a few bit
>> times if you did that.
>
> Boxcar filter, running average, y(t) = integral from (t - T0) to t dt, etc.

How dos a boxcar averager that is consulted only just before it is reset
differ from sum-and-dump, leaving aside its more complex implementation?
What am I missing?

Jerry
--
"I view the progress of science as ... the slow erosion of the tendency
to dichotomize." --Barbara Smuts, U. Mich.
�����������������������������������������������������������������������
From: Tim Wescott on
Jerry Avins wrote:
> Tim Wescott wrote:
>> Jerry Avins wrote:
>>> Tim Wescott wrote:
>>>> Tauno Voipio wrote:
>>>>> On 27.2.10 4:02 , sven98de wrote:
>>>>>> Hi,
>>>>>> has somebody here implemented a RTTY (AFSK) decoder. Any
>>>>>> suggestions how to
>>>>>> proceed ?
>>>>>> I've found some technical abstracts about FSK demodulation in the WEB
>>>>>> like:
>>>>>> - frequency discriminator with IIR resonator
>>>>>> - frequency discriminator with FIR bandpass filter
>>>>>> - digital PLL
>>>>>>
>>>>>> Any idea how complex it is to implement one of these methods ?
>>>>>> What processor is sufficient for this, does a microcontroller the
>>>>>> job ?
>>>>>
>>>>> Do you like to use it on real band conditions, or is this just
>>>>> an exercise.
>>>>>
>>>>> When there is QRM and QRN on the channel, a PLL is not a good
>>>>> idea, as it will pretty easily grab an interfering signal.
>>>>>
>>>>> The traditional TU (terminal unit) has two analog filters,
>>>>> one for mark and one for space. If you like to use it as a
>>>>> model, make a single pole-pair IIR resonator for both
>>>>> frequencies and compare the amplitudes.
>>>>
>>>> Or I/Q demodulate at both the mark and space frequency, do a running
>>>> average (a _real_ running average, not an exponential decay) for one
>>>> bit period, and choose the I/Q pair with the best magnitude. If you
>>>> deal with the synchronization issue, you can just sum-and-dump
>>>> instead of doing the running average.
>>>
>>> By "real running average" as opposed to "sum and dump", Do you mean
>>> forever and ever amen? The sensitivity would tank after a few bit
>>> times if you did that.
>>
>> Boxcar filter, running average, y(t) = integral from (t - T0) to t dt,
>> etc.
>
> How dos a boxcar averager that is consulted only just before it is reset
> differ from sum-and-dump, leaving aside its more complex implementation?

Not at all

> What am I missing?
>
Without synchronization to the data clock, sum-and-dump doesn't work.

--
Tim Wescott
Control system and signal processing consulting
www.wescottdesign.com
From: Jerry Avins on
Tim Wescott wrote:
> Jerry Avins wrote:
>> Tim Wescott wrote:
>>> Jerry Avins wrote:
>>>> Tim Wescott wrote:
>>>>> Tauno Voipio wrote:
>>>>>> On 27.2.10 4:02 , sven98de wrote:
>>>>>>> Hi,
>>>>>>> has somebody here implemented a RTTY (AFSK) decoder. Any
>>>>>>> suggestions how to
>>>>>>> proceed ?
>>>>>>> I've found some technical abstracts about FSK demodulation in the
>>>>>>> WEB
>>>>>>> like:
>>>>>>> - frequency discriminator with IIR resonator
>>>>>>> - frequency discriminator with FIR bandpass filter
>>>>>>> - digital PLL
>>>>>>>
>>>>>>> Any idea how complex it is to implement one of these methods ?
>>>>>>> What processor is sufficient for this, does a microcontroller the
>>>>>>> job ?
>>>>>>
>>>>>> Do you like to use it on real band conditions, or is this just
>>>>>> an exercise.
>>>>>>
>>>>>> When there is QRM and QRN on the channel, a PLL is not a good
>>>>>> idea, as it will pretty easily grab an interfering signal.
>>>>>>
>>>>>> The traditional TU (terminal unit) has two analog filters,
>>>>>> one for mark and one for space. If you like to use it as a
>>>>>> model, make a single pole-pair IIR resonator for both
>>>>>> frequencies and compare the amplitudes.
>>>>>
>>>>> Or I/Q demodulate at both the mark and space frequency, do a
>>>>> running average (a _real_ running average, not an exponential
>>>>> decay) for one bit period, and choose the I/Q pair with the best
>>>>> magnitude. If you deal with the synchronization issue, you can
>>>>> just sum-and-dump instead of doing the running average.
>>>>
>>>> By "real running average" as opposed to "sum and dump", Do you mean
>>>> forever and ever amen? The sensitivity would tank after a few bit
>>>> times if you did that.
>>>
>>> Boxcar filter, running average, y(t) = integral from (t - T0) to t
>>> dt, etc.
>>
>> How dos a boxcar averager that is consulted only just before it is
>> reset differ from sum-and-dump, leaving aside its more complex
>> implementation?
>
> Not at all
>
>> What am I missing?
>>
> Without synchronization to the data clock, sum-and-dump doesn't work.

Ah so! Thanks.

Jerry
--
"I view the progress of science as ... the slow erosion of the tendency
to dichotomize." --Barbara Smuts, U. Mich.
�����������������������������������������������������������������������
From: sven98de on
>On 27.2.10 4:02 , sven98de wrote:
>> Hi,
>> has somebody here implemented a RTTY (AFSK) decoder. Any suggestions how
to
>> proceed ?
>> I've found some technical abstracts about FSK demodulation in the WEB
>> like:
>> - frequency discriminator with IIR resonator
>> - frequency discriminator with FIR bandpass filter
>> - digital PLL
>>
>> Any idea how complex it is to implement one of these methods ?
>> What processor is sufficient for this, does a microcontroller the job ?
>
>Do you like to use it on real band conditions, or is this just
>an exercise.
>
>When there is QRM and QRN on the channel, a PLL is not a good
>idea, as it will pretty easily grab an interfering signal.
>
>The traditional TU (terminal unit) has two analog filters,
>one for mark and one for space. If you like to use it as a
>model, make a single pole-pair IIR resonator for both
>frequencies and compare the amplitudes.
>
>--
>
>Tauno Voipio, OH2UG
>

Yes I thought to use it on real band conditions, so there will be more or
less interferences.

Sven