[Elecraft] LP Pan versus P3 Panadapters

[David Woolley (E.L)]

Sorry, but the only way of handling this well is proper interleaved style.

Edward R. Cole wrote:
> 
> The sensitivity of the display is tied to the sampling rate (FFT/BLK 
> size on the control panel of the SDR-IQ).  With maximum sampling of 
> 262,144 I get a bin size of 0.42 Hz.  This also referred to as the RBW 

I believe that one of the early articles was talking about professional 
systems that actually run interleaved FFTs with frequency offsets, so 
several bins overlap and are averaged.

I suspect it is true that all amateur radio power spectrum estimation is 
done with a single FFT.

> (resolution band width).  The narrower the FFT routine in the sw is, the 
> more sensitive the display is.  It is analogous of using narrow filters 
> to hear very weak signals.  But one cannot copy CW using bandwidths 
> under 25-Hz due to filter "ringing" effect, so the panadapter display 

The real reason is that the sidebands of fast CW extend beyond 25Hz. 
The first order sidebands of 12 wpm morse are at +/- 5Hz, so, with 25Hz 
bandwidth you have lost most of the signal (as against the carrier) at 
more than 30 wpm.

> can be several times more sensitive than what one hears, depending on 
> what the sw permits.

What you are comparing with the pan adapter is the ability to hear the 
carrier, not the modulation.  The effective bandwidth of the ears is, I 
believe, less than 25Hz at typical sidetone frequencies, so whilst you 
may not perform as well at detecting the carrier as the FFT, you will 
perform much better than the ability to recover the modulation.
> 
> I usually run my SDR-IQ at 65,536 or 32,768 samples/sec with RBW of 1.7 
> or 3.4 Hz.  WSJT sw that is used for eme has a RBW of 4.3 Hz (I believe) 
> and that will display signals 10-dB weaker than can be heard.  WSJT has 
> a max span of about 4-KHz.

There is non-coherent integration going on here as well.  You get a 10 
log10 (n) improvement in SNR of the carrier by averaging n samples.  As 
far as I know the EME reflection process spreads the spectrum and 
prevents coherent integration of even very slow morse. 
http://www.proaxis.com/~boblark/wksig1.htm seems to suggest something 
around 7Hz/GHz of spreading.
>