Enjoying new filters and SAS
Wanted to mention that I find the new spectral filter to be very powerful.
I also find the SAS or pseudo-stereo Synchronous AM very useful. It seems the upper and lower bands take turns as to which one is better at any given moment. If combined electrically or in software, I suspect there is cancelling that happens. When using SAS mode, the bands just move around left to right and it seems I can't usually hear the signal better than the other SAM modes. I think the brain can engage it's signal processing and pull out better information.
Thanks, John!
Ken
I also find the SAS or pseudo-stereo Synchronous AM very useful. It seems the upper and lower bands take turns as to which one is better at any given moment. If combined electrically or in software, I suspect there is cancelling that happens. When using SAS mode, the bands just move around left to right and it seems I can't usually hear the signal better than the other SAM modes. I think the brain can engage it's signal processing and pull out better information.
Thanks, John!
Ken
Comments
Mr John!!! You are a star!!! 😂
I came back to the Kiwi forum after six-plus months away, and what do I find? You have resurrected a 70-year-old forgotten/ignored technique I've been trying (unsuccessfully) to promote for over a year. And longed to experience since reading about it in my teens, 60 years ago!
Your "SAS", or something apparently equivalent. Coherent lower/upper sidebands demodulated and presented to left/right ears.
This was often referred to as "binaural" in the literature, but that's a term hijacked for two other different audio techniques in the last 20 years or so. Now I think of it as "sideband binaural".
For me, the brilliance of SAS is primarily in spreading all the (non-coherent) noise across the sound stage, leaving the wanted channel centred and allowing the brain to focus via the "cocktail party effect". Adjacent channel interference tends to come from the sides as well.
A more subtle benefit is often found for co-channel stations with small carrier offsets from nominal. The tiniest spatial separation is enough to improve aural discrimination so much more effectively than with plain AM or even SAM. Sometimes even much weaker stations can be resolved.
In my opinion this is an indispensable tool for serious DXing. I hope many other listeners are going to explore and enjoy its capabilities - and the word gets around to other potential customers.
If you've researched the topic you will know that KiwiSDR is unique. Binaural sideband AFAIK has never been implemented in analogue radios. And NO OTHER SDR program offers this feature. The couple that provide a "BIN" or pseudo-stereo function do something completely different, completely missing the potential achieved by Kiwi. There MAY have been just one SDR going down the same track, but that looks like a dead donkey now.
I'm very curious to find out how you implemented this (I did it with Hilbert) and have a couple of ideas arising from my experiments. Don't know if you feel like chatting?
Thanks to Ken for his post that directed me here, and to you for having the inspiration and making the effort to provide it.
All the best
Paul
Paul. Thanks, but all the SAM/SAS stuff was shamelessly stolen verbatim from Warren Pratt's fantastic (open source) wdsp package as found in the Teensy-ConvolutionSDR project: github.com/DD4WH/Teensy-ConvolutionSDR
See this forum post: http://forum.kiwisdr.com/discussion/1990/v1-390-cross-compilation-sam-demodulators
Yes, the code (https://github.com/jks-prv/Beagle_SDR_GPS/blob/master/rx/wdsp/SAM_demod.cpp) has a Hilbert in it. Not that I fully understand what a Hilbert is or does (I read the Wikipedia page and my eyes glaze over). But that doesn't stop me from hacking at stuff until I fail or maybe get something to work. I tried a dozen different FFT-based things to do channel nulling without success before I remembered the SAS code and got that to work (Christoph has since given me another FFT scheme to try).
I find there is also a pseudo-stereo effect when you listen to IQ mode in headphones. It's fun to close your eyes and listen to a CW contest in this mode. It's not quite a full stereo experience but your brain does a remarkable job of making it seem so. No doubt due to the orthogonal phase difference presented to the ears.
Oho, John, maybe Teensy added that after I looked in on it a long time back, or I just didn't understand what it was doing. Can't remember.
Re: I-Q "pseudo-stereo", that's something various people provide (under various names), but if you listen carefully it has the same "disembodied" character as wrongly-phased stereo - you can't get a solid focus on individual components. It's a waste of time, except as a special effect.
SAS or "sideband binaural" "does the business" because the 90 degree phase shift (Hilbert) in one channel as it were compensates for the 90 degree quadrature of Q. I'll call the result Q*.
Then summing I+Q* gives LSB and I-Q* gives USB, according to the standard (now almost never used) "phasing method" of sideband selection.
With this, sending phase-coherent LSB to L and USB to R we're really motoring because (under perfect propagation and DSP) the phase-locked station is 100% centred in your head. Noise (non-coherent) is spread across the sound stage (unlike with plain SAM). Adjacent channel muck is off to the side. Effective SNR is greatly enhanced, maybe by 6 dB.
Next, the Luxembourg effect spoils this, up to a point, with the wanted station wandering around a bit. But your brain still tracks it OK.
Next again, a co-channel station can often be distinguished by slight offset in position or (maybe most often) by wandering around - so the brain gets a bit of help here as well.
The Spanish or UK Single Frequency Networks (SFNs) on MW are a very interesting special case. A jumble with AM or SAM can sometimes be clarified a little with SAS.
And lastly, something you don't provide is free vernier tuning to override the PLL. A range of +/- 10 to 50 Hz would cover most MW carrier offsets and, with a resolution of about 0.1 Hz, allows you to "tune in" individual offset stations. That brings them one by one centre stage (albeit with slow wandering) and can emphasise weaker stations in the presence of significantly stronger ones.
That really does work. Using Gnu Radio Companion on IQ recordings, I've resolved both a US plus a Canadian station on one channel underneath a much stronger local station. It's painstaking work, but that's DXing for you.
I'm not saying this is necessarily the best technique for all purposes - for example the recent SDR# channel cancelling is a stunning piece of work. But it's a simple "upgrade" for any SDR software and has never been generally exploited.
One last point. For general listening SAS has no advantage over SAM, and may even be inferior. But it is a specialist tool that performs miracles (well not quite). And its value cannot be appreciated from only a casual glance - you have to listen and get used to what it's telling you. And yes, some listeners may feel uncomfortable, a bit disorientated.
Sorry if I've bored you. I'd hate for this to be a "button" that no-one understands and just gets passed over.
"maybe Teensy added that after I looked in on it a long time back"
Oh dear, just checked back and it was implemented by end-2017. Just goes to show how easily you can miss something significant. Also curious that the Teensy Convolution team hasn't made more of a song about it!
Another BTW: ExpertSDR v.2* is the only interface I've come across to provide tuning in 0.1 Hz steps (sure, the RX calibration probably won't be that good, but it's the relative change that counts here). Couple that with a hi-res FFT to distinguish carriers about 1 Hz apart, to make a beautiful binaural resolving tool.
Just sayin'.
* Seems to be gone in the forthcoming v.3?
Using SAS mode listening to some active CW portions of the hambands is indeed quite interesting.
To really make the binaural effect shine and translating radio frequencies to L/R audio signals what would be helpful is an option to inhibit the carrier search for such cases.
Wishing all of you a much better year 2021,
Ben
Hi, Ben. Seconding your suggestion. Bears some relationship to what I wrote above:
"And lastly, something you don't provide is free vernier tuning to override the PLL. A range of +/- 10 to 50 Hz would cover most MW carrier offsets and, with a resolution of about 0.1 Hz, allows you to "tune in" individual offset stations. That brings them one by one centre stage (albeit with slow wandering) and can emphasise weaker stations in the presence of significantly stronger ones."