[Elecraft] K3: DSP NR for Data A mode?

2012-07-20 Thread Bill Frantz
Is it possible to implement Noise Reduction in the DSP for Data 
A mode? Or is it a really stupid idea?

The PSK software I use -- cocoaModem for the Mac -- has a noise 
reduction mode which works reasonably well when it is receiving 
a broad spectrum signal -- 2 KHz or more. It does not work well 
at all when I narrow the DSP bandwidth to eliminate strong 
nearby signals. This situation leads to the annoying situation 
where I can copy a weak station when the bandwidth is broad, but 
not when I have narrowed it to eliminate strong nearby stations. 
The strong nearby stations overload my SignaLink AtoD converter 
which keeps me from copying the station. It seems to me the DSP 
is in a position to apply noise reduction before the narrow band 
filter, if that is the best way to perform noise reduction.

Cheers - Bill, AE6JV

-
Bill Frantz| Airline peanut bag: Produced  | Periwinkle
(408)356-8506  | in a facility that processes   | 16345 
Englewood Ave
www.pwpconsult.com | peanuts and other nuts. - Duh | Los Gatos, 
CA 95032

__
Elecraft mailing list
Home: http://mailman.qth.net/mailman/listinfo/elecraft
Help: http://mailman.qth.net/mmfaq.htm
Post: mailto:Elecraft@mailman.qth.net

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


Re: [Elecraft] K3: DSP NR for Data A mode?

2012-07-20 Thread N5GE

Have you tried using a narrower filter in the K3 or using the K3 NR?

Amateur Radio Operator N5GE
ARRL Lifetime Member
QCWA Lifetime Member

On Fri, 20 Jul 2012 09:03:20 -0700, Bill Frantz fra...@pwpconsult.com wrote:

Is it possible to implement Noise Reduction in the DSP for Data 
A mode? Or is it a really stupid idea?

The PSK software I use -- cocoaModem for the Mac -- has a noise 
reduction mode which works reasonably well when it is receiving 
a broad spectrum signal -- 2 KHz or more. It does not work well 
at all when I narrow the DSP bandwidth to eliminate strong 
nearby signals. This situation leads to the annoying situation 
where I can copy a weak station when the bandwidth is broad, but 
not when I have narrowed it to eliminate strong nearby stations. 
The strong nearby stations overload my SignaLink AtoD converter 
which keeps me from copying the station. It seems to me the DSP 
is in a position to apply noise reduction before the narrow band 
filter, if that is the best way to perform noise reduction.

Cheers - Bill, AE6JV

-
Bill Frantz| Airline peanut bag: Produced  | Periwinkle
(408)356-8506  | in a facility that processes   | 16345 
Englewood Ave
www.pwpconsult.com | peanuts and other nuts. - Duh | Los Gatos, 
CA 95032

__
Elecraft mailing list
Home: http://mailman.qth.net/mailman/listinfo/elecraft
Help: http://mailman.qth.net/mmfaq.htm
Post: mailto:Elecraft@mailman.qth.net

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:Elecraft@mailman.qth.net

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


Re: [Elecraft] K3: DSP NR for Data A mode?

2012-07-20 Thread Bill Frantz
Note that NR is not available in DATA A mode -- K3 manual p25: 
NR turns noise reduction on. It doesn’t apply to DATA or FM 
modes, or with AGC turned off.

Ray identified why I would like the K3 NR to work in DATA A when 
he said, NR algorithms -need- a broad chunk of spectrum. If 
the DSP NR could work on a broad spectrum before filtering it 
would have enough data to separate signal from noise. I'm not 
trying to get NR to deal with strong nearby signals, but when I 
narrow the bandwidth to get rid of them, my post-narrowing NR no 
longer works so I'm looking for pre-narrowing NR.

I have used the notch filter, but it takes time to deploy, and 
can't deal with two separate loud signals. By experience, 
narrowing the DSP bandwidth is operationally easier and more effective.

I suppose I should bite the bullet and get a 24 bit USB sound 
card. Does anyone have any recommendations?

[It hurts to spend money for a 24 bit A/D converter which is 
going to convert analog the radio has produced using a 24 bit 
D/A converter. It's just unaesthetic.]

Cheers - Bill, AE6JV

On 7/20/12 at 10:43, li...@n5ge.com (N5GE) wrote:

Have you tried using a narrower filter in the K3 or using the K3 NR?

On 7/20/12 at 11:11, raysil...@verizon.net (Ray Sills) wrote:

Well, the way I see it Noise Reduction won't help you get 
rid of strong nearby (in frequency and/or QTH) signals.  That 
would be the job of a notch filter.  In fact, I think NR 
algorithms -need- a broad chunk of spectrum, so as to be able 
to identify what is signal and what is noise.  If you 
narrow the DSP bandwidth, the software many not have enough 
noise to work with.

If those strong signals are overloading the A/D converter, you 
would be better off lowering the level fed to the converter.  I 
think the SignaLink uses a 16 bit converter, and this is where 
a 24 bit converter would work much better, since it has 24 dB 
more dynamic range (before clipping).

---
Bill Frantz| The only thing we have to   | Periwinkle
(408)356-8506  | fear is fear itself. - FDR  | 16345 
Englewood Ave
www.pwpconsult.com | Inaugural address, 3/4/1933  | Los Gatos, 
CA 95032

__
Elecraft mailing list
Home: http://mailman.qth.net/mailman/listinfo/elecraft
Help: http://mailman.qth.net/mmfaq.htm
Post: mailto:Elecraft@mailman.qth.net

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

Re: [Elecraft] K3: DSP NR for Data A mode?

2012-07-20 Thread Don Wilhelm
Bill,

I have not found 24 bit soundcards to be a requirement for data modes.  
16 bit soundcards will do just fine.

Yes, NR is not available in Data modes because it distorts the signal 
and would result in reduced copyability.  The object is to produce as 
clean a signal as possible to the soundcard.

What you observed is a true consequence of how most noise reduction 
algorithms work.  They assume that the noise is distributed evenly 
across the bandwidth they are receiving, and configure a filter to 
reduce that noise.  If you narrow the bandwidth, that noise begins to 
appear (to the mathematics) more and more like a signal that should not 
be interfered with.

I would suggest that the best solution is to allow the data mode 
software to sort things out to the best of its ability.  If you reduce 
the signal width sufficiently to home in to the desired signal, then I 
would expect the data software to be able to decode properly (at its 
limits of rejection) using a wide bandwidth, and also be able to decode 
quite well with a very narrow bandwidth, but at intermediate widths, the 
datamode application software will not have enough bandwidth to properly 
discern what is noise and which is a proper signal.

The noise reduction in many data mode applications is actually quite 
good.  So my suggestion is to allow that data mode application to use 
its noise reduction capability (which requires a wide bandwidth), but if 
you are struggling to receive a signal, narrow the RX bandwidth to the 
narrowest possible and see if the QSO can continue.

73,
Don W3FPR


On 7/20/2012 5:57 PM, Bill Frantz wrote:
 Note that NR is not available in DATA A mode -- K3 manual p25:
 NR turns noise reduction on. It doesn’t apply to DATA or FM
 modes, or with AGC turned off.

 Ray identified why I would like the K3 NR to work in DATA A when
 he said, NR algorithms -need- a broad chunk of spectrum. If
 the DSP NR could work on a broad spectrum before filtering it
 would have enough data to separate signal from noise. I'm not
 trying to get NR to deal with strong nearby signals, but when I
 narrow the bandwidth to get rid of them, my post-narrowing NR no
 longer works so I'm looking for pre-narrowing NR.

 I have used the notch filter, but it takes time to deploy, and
 can't deal with two separate loud signals. By experience,
 narrowing the DSP bandwidth is operationally easier and more effective.

 I suppose I should bite the bullet and get a 24 bit USB sound
 card. Does anyone have any recommendations?

 [It hurts to spend money for a 24 bit A/D converter which is
 going to convert analog the radio has produced using a 24 bit
 D/A converter. It's just unaesthetic.]

 Cheers - Bill, AE6JV

 On 7/20/12 at 10:43, li...@n5ge.com (N5GE) wrote:

 Have you tried using a narrower filter in the K3 or using the K3 NR?
 On 7/20/12 at 11:11, raysil...@verizon.net (Ray Sills) wrote:

 Well, the way I see it Noise Reduction won't help you get
 rid of strong nearby (in frequency and/or QTH) signals.  That
 would be the job of a notch filter.  In fact, I think NR
 algorithms -need- a broad chunk of spectrum, so as to be able
 to identify what is signal and what is noise.  If you
 narrow the DSP bandwidth, the software many not have enough
 noise to work with.

 If those strong signals are overloading the A/D converter, you
 would be better off lowering the level fed to the converter.  I
 think the SignaLink uses a 16 bit converter, and this is where
 a 24 bit converter would work much better, since it has 24 dB
 more dynamic range (before clipping).
 ---
 Bill Frantz| The only thing we have to   | Periwinkle
 (408)356-8506  | fear is fear itself. - FDR  | 16345
 Englewood Ave
 www.pwpconsult.com | Inaugural address, 3/4/1933  | Los Gatos,
 CA 95032

 __
 Elecraft mailing list
 Home: http://mailman.qth.net/mailman/listinfo/elecraft
 Help: http://mailman.qth.net/mmfaq.htm
 Post: mailto:Elecraft@mailman.qth.net

 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:Elecraft@mailman.qth.net

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