Login  Register

Re: Sensitivity - Was K4 Observations

Posted by Joe Subich, W4TV-4 on May 19, 2019; 6:15pm
URL: http://elecraft.85.s1.nabble.com/K4-Observations-tp7651899p7651934.html


There is one place that digital modes (like those by Joe Taylor and
associates) can improve the decoded SNR beyond simply reducing the
detection bandwidth.  If the modulation/encoding supports N states
but the encoding only uses M of those states, the decoding software
can make use of the "sparse constellation" to recognize states that
are impacted by noise and select the "closest" valid state.

This "coding gain" can improve the overall SNR beyond that provided
simply by the "matched" (or optimal) noise bandwidth.  However, with
all amateur modes (CW to FT8 & FT4) the majority of the SNR improvement
over SSB (or AM) is simply due to the use of optimal bandwidth to
reduce extraneous noise in the detector bandwidth.  Even with SSB,
properly tailoring the IF bandwidth will make several dB difference
in the detected SNR.  For example, a 2 KHz bandwidth (500 - 2500 Hz)
can provide significant improvement over a 2.8 KHz bandwidth (200 -
3000 Hz) under noisy conditions.

73,

    ... Joe, W4TV


On 2019-05-19 11:19 AM, Tom Azlin W7SUA wrote:

> More like a "feel good" detection SNR?
>
> I think fldigi uses a few bins either side of the signal to determine
> the noise in the SNR measurement. When I narrow my K3 IF bandwidth down
> to just the, say Olivia, bandwidth the SNR number climb up to 30 dB high
> as the filter cut the noise in the adjacent "noise" bins. If I use a
> 600-700 Hz filter or wider for a 500 Hz wide Olivia then the SNR
> measurements stay the same.
>
> So I have always thought along the lines of your two emails Joe. Plus
> long time ago I discovered how I could add FFTs up and a coherent signal
> would "climb" out of the random noise. So for a signal with considerable
> time per bin measurement you get that gain as well.
>
> So have always thought of the WSJT type negative numbers as bogus.
>
> 73, tom w7sua
>
> On 5/19/2019 7:18 AM, Joe Subich, W4TV wrote:
>> On 2019-05-19 9:50 AM, Wes wrote:
>>> FT8 reports negative SNRs number but we both know those are bogus.
>>
>> All of the modes that quote negative SNRs are doing so by using SNR
>> in a voice (2500 Hz) bandwidth *NOT* SNR in the detector bandwidth
>> (bandwidth of the final filter whether than be a narrow IF filter,
>> the "ear-brain" filter or a software [computation] filter).
>>
>> If one looks at the SNR thresholds of the various Joe Taylor "slow"
>> modes, 80% of the "negative" SNR can be attributed entirely to the
>> difference between the occupied bandwidth and the [excess] measurement
>> bandwidth.  The remainder can be attributed to software processing
>> algorithms that take advantage of the fact that noise is random while
>> the signal is not - in essence reporting using a "peak noise" level
>> while actually decoding against a "minimum noise" level (like copying
>> CW through static crashes - one looses a dit/dah during the crash but
>> fills that in from the context).
>>
>> 73,
>>
>>     ... Joe, W4TV
>>
>>
>> On 2019-05-19 9:50 AM, Wes wrote:
>>> I feel like I'm gonna be slappin' a tar baby by responding.
>>>
>>> Since we are discussion HF radios, I was assuming HF.  I realize
>>> JT65(-HF) and JT9 have been used on HF, but the QSOs are hardly
>>> random. If your computer clock is off, sorry, no QSO.  FT8 reports
>>> negative SNRs number but we both know those are bogus.
>>>
>>> Wes  N7WS
>>>
>>>
>>> On 5/19/2019 5:58 AM, Ed W0YK wrote:
>>>> JT65, JT9, FT8.
>>>>
>>>> 73,
>>>> Ed W0YK
>>>>
>>>> -------- Original message --------
>>>> From: Wes <[hidden email]>
>>>> Date: 5/19/19 07:49 (GMT-06:00)
>>>> To: [hidden email]
>>>> Subject: Re: [Elecraft] Sensitivity - Was K4 Observations
>>>>
>>>> What current modes hear below the noise level?
>>>>
>>>> Wes  N7WS
>>>
>>> ______________________________________________________________
>>> 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
> ______________________________________________________________
> 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