Topband: CW Speaker - follow up

[David Raymond]

Greetings Topbanders. . .

My original post of February 7 has started a lot of discussion. . . some 
of it remarkably thoughtful and insightful.  The aggregate knowledge on 
this reflector is nothing short of amazing.

A quick recap. . .my original post detailed my experience with 
Adrian/VK2WF consistently copying my 160m CW signals consistently better 
than I copy his (see original post).  I'm operating the Flex remote via 
a Maestro using a nearby station.  Having had dozens of morning QSOs 
with VK2WF and compensating for the difference of ERP I was perplexed 
how Adrian was consistently copying my signal fairly well when I had a 
hard time even telling he was there (both of us have low noise levels).  
At some point a little later Adrian told me about his "CW speaker" and 
what a difference it made in literally pulling signals out of the noise. 
He mentioned that he could listen to his receiver with the CW speaker in 
the "Wide" (pass through) mode and my signal would be uncopyable, 
sometimes even undetectable.  He could then flip the switch to "Narrow" 
and my signal would be Q4 or better.  After suffering through many 
mornings with Adrian hearing me clearly better than I was hearing him I 
asked him if he would build me a CW speaker which he did.  To cut to the 
chase it arrived and I got it quickly hooked up to the Maestro using the 
"Line Out" audio.

I began listening and was immediately disappointed.  Clearly I wasn't 
seeing the "magic" that Adrian was.  I could listen on the CW speaker in 
the "Wide" mode and, having the Flex bandwidth set for 3 or 4 kHz and 
APF "off," I had normal receive audio as expected . . . more noise, 
close in signals causing QRM as usual. When I switched the speaker to 
"Narrow" with the receiver widened out the audio noise floor dropped 
significantly and already copyable CW signals were well above the very 
low audio noise floor (but S/N essentially the same copy as using the 
narrow filtering in the rig).  In this configuration I could not copy 
signals that were at or below the receiver noise floor.  I then returned 
the Flex to my usual CW setting of  250 Hz bandwidth and fair amount of 
APF dialed in (probably 50 or 75 Hz of audio bandwidth)  I could go from 
"Wide" to "Narrow" with _no_ apparent improvement in S/N or 
copyability.  I finally put the CW speaker on the bench testing with an 
audio oscillator with it showing an extremely sharp peak at the audio 
frequency of 610 Hz (my desired listening frequency) as expected with a 
very narrow response (maybe about 15 Hz).  The speaker was doing exactly 
what it was supposed to do.  I was stumped as to why no improvement on 
extremely weak signals that Adrian was seeing on his end.

In the meantime, another faithful Topbander had procured a CW speaker 
from N4IS to be used with his Elecraft K4D.  In conversing with him he 
confirmed he was having basically identical results with his CW speaker. 
. .no copy improvement signals beyond normal narrow filtering from the 
rig and nothing from the CW speaker once the signals dropped below the 
receiver noise floor.  After discussion and comparing notes further with 
Adrian we finally concluded that the difference seemed to be based on 
the fact that my receiver was an SDR while Adrian's was a home brew 
analog receiver (and rather elegant).  The other CW speaker user related 
an identical experience.  Based on the two CW speaker examples we 
concluded that the difference in copyability between Adrian and me (and 
the other user) occurred when incoming signal levels slipped below the 
SDR receiver noise floor making the signals no longer recoverable while 
signals below the noise floor on Adrian's analog receiver _were_ 
recoverable.

The response to my original post has been a bit overwhelming including 
many sent directly to me and some phone calls.  I will attempt to 
summarize what I've learned from the input I've received  as follows:

1.  In essence, it seems that once the signal slips below the SDR noise 
floor it is essentially "lost."  While it may exist in there somewhere 
it is unrecoverable with current SDR capability in amateur rigs.
2.  The loss of recoverable audio could be attributed to many things in 
the A-D/D-A conversion processes. . .sampling rate, dynamic range, phase 
noise, quantization errors, etc.
3.  All of this could be further compounded by the fact that I'm 
operating remote. . . adding in all kinds of unknown processes occurring 
in the circuitous digital path on the internet between my home (with the 
Maestro and CW speaker) and the remote site (with the Flex).
4.  At least two people suggested the use of a low noise preamp at the 
beverage to raise the overall composite RF signal level at the receiver 
available for processing  (acknowledging that S/N would remain the same) 
giving the A-D converter a better chance of recovering the weak signal 
bits from the low composite RF signal level (we have a 20 db low noise 
switchable preamp at the bev remote switching point made no change).
5.  I will take the CW speaker up to the remote site and hook it up 
directly to the Flex removing both the Maestro and the internet path 
from the equation to see if that makes any difference.

As I said previously, I could easily be missing something here along the 
way. . .I'm an old analog guy and far from a DSP expert.  If anyone has 
further ideas to share or suggestions for things to try please feel free 
to respond directly to me if you wish.  I will share anything 
significant I learn or any new developments here on the reflector.

73. . . Dave, W0FLS

Here's a photo of the VK2WF CW speaker: 
https://www.dropbox.com/scl/fi/kbddxg0lj1jb260bfr6je/2025-11-04-16.51.11.jpg?rlkey=cya2nnfnljbg3j98lju7bgyto&st=bdpq8qjl&dl=0