Automatic receive bandwidth switch during transmit in digital modes

Don-KB6TSQ

I watched a video that
Steve Hicks made about noise.  You can find the video here: https://youtu.be/xXXj1Ko4ZXg?t=352  

In this video he mentions several noise sources:  Thermo, Gaussian, flicker, quantification, phase, additive, splatter, atmospheric, power line, spark, power supply and a few more I have forgotten.   In this talk he is only concerned with Thermo, Gaussian and Atmospheric noise.   He talks about  signal to digital conversion, FFT, dBm, bin size, noise floor and several other noise related topics.  This is a great technical talk on how our radios receive signals.  Being a practical person, my whole take away is that if I reduce my receive bandwidth during my digital mode qso (or any mode for that matter) signal strength would improve in relation to noise.

Most of the time I use a spot and pounce technique on FT8.  In order to see FT8 signals to decode,  I originally set the receive bandwidth to 2500-3000 Hz wide where stations are sending these signals.  When I spot a station I want to work, I double click on their line and my transmitter moves to that frequency and starts send my call and grid.  During this transmission, I use the RX box to change the receive bandwidth to about 100 Hz covering my transmitted signal.   When the station replies to my transmission, a -10 signal level initially is now +15 or higher most of the time.

So my new idea is this:  When in digital modes, could we add the ability to preset the receive bandwidth at the transmitted frequency.  It will need a toggle because you would not want this to happen while working stations on digital split within the original 2500-3000 Hz bandwidth.

KY6LA_Howard

Brilliant Idea

Never thought to try reducing bandwidth ... so I just tried it on a QSQ

Went from a -7 to a +9

Huge differenct

EduardoCarvalho

This is awesome. I will give it a shot later. I think working split may be more appropriate as the recommendation for FT8 is to not tx on the RX frequency.

Wayne VK4ACN

Ill have to try this. I usually have a 5khz bw. Maybe a macro on keypad woule automate it. Toggle Narrow or wide

Al K0VM

In theory for a noise limited signal, the effective band width for a digital program like FT8, is not the receiver bandwidth but rather the bandwidth of the decoder software..  The receiver bandwidth should have little effect unless there is distortion and other strong signals in the pass band..

AL, K0VM

Steven Linley

Yes, great idea, I do this all the time (I am used to switching filters, to quickly check for best reception with Superhetrodyne receivers) and a toggle would be very good. 
I have been using another slice to do this.

Don-KB6TSQ

I have added a video.  https://youtu.be/EgMUZnmjMrM

I do not talk in the video.  The first station I work starts as a -9 signal and by narrowing the bandwidth the station improves to +14.  The second station starts at -11 and is improved to +18 by narrowing the receive bandwidth.  The noise floor on the 3 kHz bandwidth is about s5, when the bandwidth is reduced to about 100 Hz.  The noise floor is lowered to about s1.

Al K0VM

Don,
  Here is an exchange that I had on WSJT group about FT8 SNR...
....

The users guide states that "Signal reports are specified as signal-to-noise ratio (S/N) in dB, using a standard reference noise bandwidth of 2500 Hz".  Does this assume that all receivers are using a 2500 hz bandwidth and the receive bandwidths that are not 2500 Hz will be in error or does WSJT-X attempt to ascertain the receiver bandwidth and normalize the SNR calculation to 2500 hz..  It appears to me that a 4 Khz RX bandwidth gives significantly worse SNR's than a 400 Hz  RX bandwidth.

AL, K0VM

Hi Al,

the answer to that is mode dependent. For modes other than FT8 I believe the noise is sampled in the modulation bandwidth (i.e. in the tone slots that are not currently containing a symbol tone). The measured noise is then extrapolated to a 2500Hz figure before calculating SNR.

Because FT8 is often received in a congested sub-band any attempt to measure noise in a narrow bandwidth is likely to be over estimated due to interfering signals. I believe the current strategy uses a baseline fitting algorithm that attempts to estimate the noise floor across the whole waterfall passband. This means that the waterfall passband needs to match the receiver's passband otherwise the regions outside of the receiver's passband will  unrealistically lower the noise measurement.

If you want to use a 400Hz receiver bandwidth with FT8 then you should correspondingly narrow the WSJT-X waterfall bandwidth.

73
Bill
G4WJS.

AL, K0VM

Don-KB6TSQ

Hi Al,

Thanks for your comments and sharing Bill's incites.  However, I am seeing 18 to 24 dB improvements at the receiver.  This is before and outside of the WSJT-X program manipulations.  I will try the 400 Hz bandwidth on the WSJT-X program.  I worked in research and love to experiment.

73 Don KB6TSQ

Al K0VM

I really doubt that narrowing up the RX filter improves the WSJT's ability to decode weak FT8 signals.  The internal software filters used by the decoder are on the order of 50hz or so.  On the other hand, the bandwidth that is used to calculate SNR is much wider and not directly determine  by the software.  Thus SNR is most accurately reported when the RX bandwidth matches the SNR bandwidth assumed by the software.

But that's what experiments are all about..
AL, K0VM

Mike-VA3MW