[hpsdr] calibration - noisefloor

[Joe Martin K5SO]

Georg,

You bring up an important issue that many of us (me included until  
quite recently!) do not understand very well with regard to HPSDR and  
how PowerSDR calibrates the Rx system.

First of all, HPSDR presents a special set of requirements with regard  
to the PowerSDR calbration routine, as I will explain momentarily.   
You perhaps have noticed that during the Level Calibration execution  
there are multiple switching events that take place between "Preamp  
ON" and "Preamp OFF" configurations.  This switching process brings us  
to the first challenge with respect to HPSDR.   Namely, HPSDR has no  
preamp!

Okay.  That being the case, what is done on Mercury is that it is  
configured such that when "Preamp ON" is selected by PowerSDR, Mercury  
connects directly to the BNC connector and when "Preamp OFF" is  
selected an attenuator (resistor) is switched in between the BNC  
connector and the normal input of the receiver.

The Calibration routine operates such that it takes a specified signal  
(freq and level are specified by you) that you supply to the BNC  
connector of Mercury and the routine "forces" that input signal level  
to show as the level that you told PowerSDR that it is, both when the  
"Preamp" is ON and when the "Preamp" is OFF.   That is, the system  
calibrates itself to show you the actual power level (in dBm) that is  
present at the input BNC connector to Mercury, regardless of whether  
you have selected "Preamp ON"  or "Preamp OFF".

Now, as you probably know, if you insert an attenuator in front of the  
first rf amplifier of an Rx system the noise figure (S/N) of the Rx  
system is degraded substantially, usually the noise figure is degraded  
by (almost) the same amount as the value of the attenuator in high  
performance systems.  The noise figure of an Rx system is normally  
dominated by the noise figure of the first amplifier of the system; if  
you degrade the noise figure of the first stage by X dB you also  
effectively degrade the noise figure of the entire system by  
approximately X dB.

When the HPSDR system "Preamp ON" is selected, that is the  
configuration for which you will obtain the lowest noise figure (best  
S/N) you will ever obtain as long as Mercury is the first rf stage in  
the system (i.e., no LNAs or transverters ahead of Mercury.

If your input signal to Mercury is, say, -73 dBm (sometimes defined as  
an S-9 signal level), and you have calibrated your system to that  
input, you will read -73 dBm on the panel meter (assuming the signal  
is positioned in the demodulation pass band of the filters selected)  
and the peak of the signal showing on the panadapter will be at -73  
dBm on the panadapter scale.

Now, if you select "Preamp OFF" you will STILL see -73 dBm displayed  
on the meter, and the signal on the panadapter will peak at -73 dBm on  
the scale but the S/N will be dramatically degraded (by roughly 20  
dB!).  That is, the noise floor will rise by about 20 dB when "Preamp  
OFF" is selected, because you have switched in an attenuator in "front  
"of the basic Mercury receiver.

In future versions of the diversity program I will most certainly   
change the label on the front panel control "Preamp" to say "ATTEN"  
instead so that this confusing issue will be less of a problem to  
someone trying to make sense of the noise floor behavior after running  
a level calibration.

I hope this explanation helps you understand better how PSDR  
calibrates the system and how, as a result of that calibration  
philosophy and also as a result of the hardware configuration of the  
preamp-less Mercury in HPSDR, it affects the noise floor that is  
displayed.

Once you understand this philosophy it makes logical sense and is a   
quite valid approach.  It's not, however, how I would've done it  
personally, hihi, but that's irrelevant.

73,  Joe K5SO




 1300232960.0