[hpsdr] SDR DF

[Joe Martin K5SO]

Hi Frank, 

Yes, don't worry, I'm not planning to try to implement any Doppler-baased method for HF DF!   

There actually are several serious drawbacks to the Doppler approach, not the least of which are that the magnitude of the Doppler effect is directly proportional to the frequency of operation and the introduction of additional switching noise that you have brought to focus.  If you look carefully at the mentioned video you'll see that the actual demonstration is done in the 70 cm band  (416 MHz, as I recall) and the angular accuracy demonstrated even at that high frequency is not particularly good.  It would be much worse at HF, of course.

I certainly agree with you that multiple receivers using instantaneous phase difference measurements is much preferred over Dopller at HF.  In fact, there are special algorithms available these days that can yield better than 1 degree of accuracy (e.g., the so-called MUSIC algorithm for DF) if you have the processing power available for it.  But even if one doesn't use such state-of-the-art algorithms a straight-forward modification of the early Watson-Watt algorithm with our high performance receivers and "Adcock-pair" antennas should yield a great result for HF direction finding with as few as three receivers/antennas.  The two antenna pairs in the three-antenna case would be Ant1-Ant2 and Ant1-Ant3, preferablly with the three antennas not in a single line.

I, for one, am still working hard to find a suitable method of powering up multiple receivers so that they have the same phase each time they power up.  That one detail seems to be the only major hurdle left in actually implementing DF with our multiple-Mercury HPSDR rig on HF.   

Frankly, I'm surprised that I have as yet been unable to solve the startup phase problem.  Maybe someone has a new suggestion as to how we might do it.  I've considered Phil's latest novel suggestion about perhaps using a common CORDIC for all three receivers but that seems to require being able to pass the 16-bit ADC values at 122.88 MHz sampling rate from the non-master receivers, in parallel, to the master receiver where the common CORDIC would reside.  That high-speed data transfer seems to me to be problematic to implement and would seem to me to be not possible or advisable to do via the Atlas bus even if we decided to commandeer the necessary Atlas bus lines to do it.  

It continues to seem to me that there must be a method we can devise to force the Mercs to start up identically; unfortunately the method so far has eluded me.  I'm still trying, though, even though you don't hear much about it from me, hihi.  At this point I would be willing to try to implement even a quite complex method to accomplish it, as I may have now tried most of the simple reset-based approaches with no success.  The problem is important enought to warrant even more work, I think; not only for DF but for other phase-critical applications.  

Thanks, Frank, for the comments and thoughts on this.  I agree with your conclusions and suggestions.   We should be able to make this work!  

73,  Joe K5SO


On Mar 11, 2012, at 6:50 PM, FRANCIS CARCIA wrote:

> ***** High Performance Software Defined Radio Discussion List *****
> 
> Hi All,
> I'm just catching up on mail after a week in the windy city full of road toll boths. My goodness all they need is trolls.
> I have 1 question on the method shown in the video. I have found MUXes generate a transient when they change channels. This is caused by the digital control bleeding into the analog. Does anyone know how the user blanks this transient? A 3.3 or 5 volt digital address control put needs many dB of isolation
> The faster you rotate the more noise you generate. This isn't a problem with strong local signals but when you listen to weak stations this transient becomes a problem raising the noise floor. I tried to build the ultimate hardware noise blanker years ago so know the switch problem well. So Joe I think it might be better with multiple receivers if you could solve the timing problem. Frank WA1GFZ

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openhpsdr.org/pipermail/hpsdr-openhpsdr.org/attachments/20120311/fa71b2fe/attachment-0004.htm>