RTTY FILter

[wa9fvp]

Sometimes in the evenings I like to copy WLO broadcasting news and weather using RTTY and SITOR mode B the frequency is 8.475.225 LSB.  I decided to use the K3's data mode and take advantage of the dual data filters.

I made this observation while using my Universal "M-8000 Communications Terminal" with the Butterfly filter installed. The Butterfly filter is a perfect filter for demodulating RTTY.  It uses a combination of bandpass and notch to generate a coherent butterfly pattern.  The M8000 is connected to an X/Y oscilloscope.  I use the M8000 Decoder to copy RTTY and SITOR and it's connected to my K3's line out.  The photos below were taken while I was copying 45/170 RTTY.

When I copy WLO using LSB (the width control is set to 450Hz and with an I.F. offset of 2.25 I get a perfect butterfly pattern.  When I switch to DATA mode using the dual filters the Butterfly pattern is distorted with non coherent phase shifts and in the noisy conditions the copy degrades.
 

[wa9fvp]

Here's are the actual Butterfly filter outputs using an Agilent DSOX3024 digital storage scope connected to the mark/space (or X/Y) outputs on the M8000.   The first image is the actual DSOX3024 screen shot  with the K3 in USB mode on 20 meters, copying RTTY 45/170.   The second photo is the same setup except the K3 is in FSK/DATA mode.
 

[Jack Smith-6]

Looks like the FSK/DATA mode filter may be ringing - eyeballing the
filter envelope looks like it has some dips.  A bit more bandwidth might
help - and the other time I've seen this is when there's a tuning error,
such that the two bandpass filters are not exactly centered on the
mark/space frequencies. That observation is not with respect to the K3,
but relates to observations I made back in the days of Model 15 page
printers with analog filter demodulators.


Jack K8ZOA

On 9/3/2011 11:23 AM, wa9fvp wrote:

> Here's are the actual Butterfly filter outputs using an Agilent DSOX3024
> digital storage scope connected to the mark/space (or X/Y) outputs on the
> M8000.   The first image is the actual DSOX3024 screen shot  with the K3 in
> USB mode on 20 meters, copying RTTY 45/170.   The second photo is the same
> setup except the K3 is in FSK/DATA mode.
>
> http://elecraft.365791.n2.nabble.com/file/n6756776/rtty_usb_mode.bmp
> http://elecraft.365791.n2.nabble.com/file/n6756776/rtty_data_mode1.bmp
>
>
> -----
> Jack WA9FVP
> Willco Electronics
> --
> View this message in context: http://elecraft.365791.n2.nabble.com/RTTY-FILter-tp6756724p6756776.html
> Sent from the Elecraft mailing list archive at Nabble.com.
> ______________________________________________________________
> Elecraft mailing list
> Home: http://mailman.qth.net/mailman/listinfo/elecraft
> Help: http://mailman.qth.net/mmfaq.htm
> Post: mailto:[hidden email]
>
> This list hosted by: http://www.qsl.net
> Please help support this email list: http://www.qsl.net/donate.html
>

______________________________________________________________
Elecraft mailing list
Home: http://mailman.qth.net/mailman/listinfo/elecraft
Help: http://mailman.qth.net/mmfaq.htm
Post: mailto:[hidden email]

This list hosted by: http://www.qsl.net
Please help support this email list: http://www.qsl.net/donate.html

[wa9fvp]

The butterfly filters are tuned exactly to the tone frequency.  When the cross pattern is vertical, it's spot on 2125Hz, horizontal is exactly 2295Hz.  I changed the DATA Mode to AFSK using a single filter and that works the best.  I found that the width control is optimized between 500 and 450 Hz.   Less than 450, the non-coherent phase shifts are introduced, as you mentioned "is caused by filter ringing".

[wa9fvp]

The next two photos shows the phase relationships as the tone transitions from mark to space.  The butterfly filter maintains a perfect out of phase relationship using the AFSK, single filter, data mode.  With the dual filter, the phase relationship is sometimes in phase and then transistions to an out of phase condition. The phase non-coherence can cause a distortion in the data symbol and increase the bit error rate.

The bottom line is.  If your using an external RTTY box like a PK-232, KAM or even RTTY decode software,  use the AFSK (single filter) and not the FSK (dual filter).

[Joe Subich, W4TV-4]

> I found that the width control is optimized between 500 and 450 Hz.
> Less than 450, the non-coherent phase shifts are introduced, as you
> mentioned "is caused by filter ringing".

What roofing filter are you using?  If it is an Elecraft 5-pole filter
is the the offset properly set?

Are you using the K3's Dual PB filter?

Remember, the K3's DSP filters have very steep skirts.  The selectivity
in most other rigs is entirely related to the skirts of the IF filters;
the filters in other "modern" DSP based are much "softer" and tend to
emulated the 1;8:1 - 2.2:1 shape factor (depending on bandwidth) of the
crystal filters.  "Ringing" tends to be directly related to the nature
of the filter in the area of the "knee" and once the bandwidth is less
than 450 Hz the knee is impacting the keying sidebands.

73,

    ... Joe, W4TV


On 9/3/2011 12:32 PM, wa9fvp wrote:

> The butterfly filters are tuned exactly to the tone frequency.  When the
> cross pattern is vertical, it's spot on 2125Hz, horizontal is exactly
> 2295Hz.  I changed the DATA Mode to AFSK using a single filter and that
> works the best.  I found that the width control is optimized between 500 and
> 450 Hz.   Less than 450, the non-coherent phase shifts are introduced, as
> you mentioned "is caused by filter ringing".
>
> -----
> Jack WA9FVP
> Willco Electronics
> --
> View this message in context: http://elecraft.365791.n2.nabble.com/RTTY-FILter-tp6756724p6756903.html
> Sent from the Elecraft mailing list archive at Nabble.com.
> ______________________________________________________________
> Elecraft mailing list
> Home: http://mailman.qth.net/mailman/listinfo/elecraft
> Help: http://mailman.qth.net/mmfaq.htm
> Post: mailto:[hidden email]
>
> This list hosted by: http://www.qsl.net
> Please help support this email list: http://www.qsl.net/donate.html
>

______________________________________________________________
Elecraft mailing list
Home: http://mailman.qth.net/mailman/listinfo/elecraft
Help: http://mailman.qth.net/mmfaq.htm
Post: mailto:[hidden email]

This list hosted by: http://www.qsl.net
Please help support this email list: http://www.qsl.net/donate.html

[wa9fvp]

My K3 defaults to FL3 which is the stock 2.7/5 filter.  There is a 400 Hz filter in FL2 but when it's selected, the phase distortion kicks in just a little.   You have to under the concept of the Butterfly filter.  Both mark and space filters are coherent filters and phase distortion in the roofing filters can effect the butterfly if it's too narrow.  FL3 (2.7/5) is selected but as I dial down the width control, the DSP narrows the 15KHz IF.  The DSP filter isn't a problem when AFSK is selected.  It's the dual filter that's the problem.

[Joe Subich, W4TV-4]

> The Butterfly filter doesn’t introduce phase shifts or impulse
> response. It’s normalized for 170 Hz shift and for the most part,
> the filter bandwith is set by the Universal M8000 firmware.

Every filter has differential phase shift and and group delay (impulse
response) characteristics.  Cascading filters causes an interaction
between individual phase responses of the filters.

Have you measured the response of the Elecraft Dual PB filter without
the Universal Butterfly filter in line?  That is the only way you can
provide an clean evaluation of the Elecraft Dual PB filter.

> As I said “The bottom line is, If your using an external decoder, the
> AFSK single filter works the best”.

First, it's not the "AFSK single filter" as the filtering is identical
in both the AFSK A and FSK D data sub-modes.  Secondly, that is not
always the case ... if one uses the "AA6YQ FIR 512" profile in MMTTY
(another "butterfly" type filter) results are somewhat better with a
wider DSP setting and no Dual PB filter in the K3.  However, if one
uses a different "profile" (e.g., standard RTTY) or one of the optional
detectors (FFT or PLL) using a narrow (e.g., 400 Hz) DSP setting with
the Dual PB filter can provide better results that a simple wide DSP
setting.

None of that is surprising given the nature of cascaded narrow filters

73,

    ... Joe, W4TV


On 9/4/2011 2:23 PM, Radio Repair wrote:

> Hi Joe,
>
> The Butterfly filter doesn’t introduce phase shifts or impulse response.
> It’s normalized for 170 Hz shift and for the most part, the filter bandwith
> is set by the Universal M8000 firmware.   There are 6 bandwith settings
> from -3 to +3  that I can change on the M8000 front panel that will allow
> me to optimise the bandwidth.  If the filter is set too narrow the butterfly
> display will show the impulse response but the default setting “NOM” is
> optimal for all modes, shifts and speeds.  As I said “The bottom line is, If
> your using an external decoder, the AFSK single filter works the best”.
>
> I don’t know how the K3 handles the dual filtering internal to the DSP but
> I assume it’s optimized to the internal DATA decoder.   More than likely it’s
> OK to use the dual filter when your copying and sending RTTY on the K3
> without the use of an external decoder.
>
>
>

______________________________________________________________
Elecraft mailing list
Home: http://mailman.qth.net/mailman/listinfo/elecraft
Help: http://mailman.qth.net/mmfaq.htm
Post: mailto:[hidden email]

This list hosted by: http://www.qsl.net
Please help support this email list: http://www.qsl.net/donate.html