[hpsdr] Janus rocks

[Michael White]

I made a series of on-air practical tests this weekend during the WPX CW 
contest.

Using my Softrock with the standard 7056 crystal controlled LO, I tuned 
40m looking for a weak cw signal in between all of the high power EU 
contest stations. and I eventually found a JA3 station at 7005.36, very 
weak but I could copy it well enough for a QSO.. JA is a very difficult 
polar path for G, and my modest fixed antenna does not fire in the right 
direction. The signal was very weak but stable enough over a long period.

I then switched the antenna over to my Ten Tec Orion II. The signal was 
just detectable,  but was below the noise level most of the time. I kept 
switching back and forth from one to the other, over period of an hour or 
more and I spent a lot of time optimising the filter settings on both. The 
optimum filter setting on the PowerSDR was around 200-300hz. On the Ten 
Tec the optimum SNR was found at 100hz (minimum setting) with the 
attenuator set to 6db . My Orion has the 300hz roofing filter fitted and 
the SNR improved significatly when the roofing filter kicked in.  Both 
radios exhibited excellent DSP filters with very little ringing even at 
such low bandwidths. 

I concluded that the SDR was better to copy the weak station than was the 
Orion, like for like under my live test conditions. At my urban QTH I get 
a fairly high noise floor which registered at about  -80dbm on PowerSDR 
with the pre-amp button set to "high".. I believe that with the Orion the 
wideband noise was smearing or leaking into the signal, wheras on PowerSDR 
the noise floor seemed to be less troublesome and the signal would carry 
to my ears more cleanly. 

I found the Binaural sound setting on the SDR better than mono, but it did 
not seem to improve the readability, it just sounds more comfortable in 
the headphones. Stereo sounds better than mono that is a fact, and 
Binaural gave me the same experience.

This weak signal (under extreme conditions) readability result came as a 
big surpirse to me. The direct conversion system really does compete with 
the best (I believe the Orion II is one of the best superhets around for 
CW contest work), and I believe the Janus 24 bit ADC design contributed 
significantly to this result, not forgetting the excellent Dttsp module in 
PowerSDR.

Using the 192khz sampling rate it was possible to tune from around 6960 
thru 7150 in software, with no apparent degredation in performance from 
end to end.

I also noticed that the Janus did not seem to generate any noticeable 
increase in the noise floor at the DC/local oscillator frequency. This 
allowed me to tune through the LO (7056) region without really being aware 
of where the LO was running. The low phase noise of the crystal oscillator 
will certainly be helping with this result, but other sound cards I have 
tested with the softrock hardware have exhibted clearly visible flicker 
noise at the low sub audio frequencies so upsetting performance in that 
region, not so with Janus.

The HPSDR rocks. Congratulations to the team. brilliant.!!.


Michael White
G3WOE