[hpsdr] Infinite precision, phased arrays,and us

Thu Sep 7 19:30:49 PDT 2006

Hi Nyall

Thanks for taking the time. I wasn't having trouble with the concept of 
beamforming - I was focused very narrowly on the concept of using I/Q to 
achieve the phase shifts.

Cheers
Murray
VK6HL

At 05:13 PM 7/09/2006, Nyall Davies wrote:
>I am having difficulty in knowing whether you are not understanding what I 
>do or whether I do not understand your problem.
>I apologise if I am stating the obvious.
>It is easier to understand the Tx path but the same reasoning applies on Rx
>
>On the grounds that a diagram is worth a 1000 words I have attached a 
>crude diagram (as a .gif)
>In Fig 1 I have two antenna elements fed in phase - these add broadside on 
>and produce a figure of eight polar diag
>Fig 2 I feed them out of phase ie phase in one path180 degrees - they add 
>end on now and cancel broadside.
>By adjusting the phase through 0 to 180 degrees we have swung the beam 90 
>degrees
>Fig 3 shows the situation for an intermediate position. We can add as many 
>elements as we like as long as we continue to add more and more phase 
>shift in successive ones. we can have any distance between them as long as 
>we take it into account in the phase.
>
>We should note that the aperture of the antenna has gone down and thus its 
>gain.
>
>The question is "How do we achieve the phase change?"
>
>Assuming we produce baseband digital information at some low clock rate we 
>would naturally up the sampling rate with a CIC filter.
>ie we would have a number of stages of differentiator, the a sampling rate 
>changer and then a number of stages of integrator. This naturally does 
>interpolation.
>I have not examinined this in depth but I presume we simply put a series 
>of identical samples into the integrator and the mere fact that it is an 
>integrator produces the interpolated samples. So we have a baseband output 
>at whatever sampling rate we want. I presume we would go up to the 
>sampling rate of the Rx A/D say 100 MHz
>
>How do we get the required phase change? - the easiest way conceptually is 
>to now frequency shift with a mixer (digital)- so let us split the path 
>into two mixers and  adjust the phase of the LO samples as desired. 0 to 
>+- 180 degrees for +- 90 beam swing. That demonstrates the accuracy 
>required. This will produce the right phase at whatever RF we produce as 
>phase is conserved through a mixing process.
>
>We could have started the two channels at baseband but we would have to 
>shift the phase across the whole audio band. We are all familiar with the 
>complex networks used to do this in the past but would now be mucking 
>about with Hilbert filters or the such like.
>A few moments with a piece of paper multiplying  a+jb by 1+j0 or 0+j1 or 
>sqrt(2)+j sqrt(2) etc will demonstrate that phase shifting need only a 
>multiple ( for a single frequency)
>
>The two outputs would feed separate Tx modules. While this is quite 
>appropriate to Radar phased arrays with distributed PAs it is less so to 
>amateur radio.
>
>In a Radar array the problem of fill time arises. As a RF pulse hits the 
>array it will reach one element of the array first. At that stage we have 
>a very broad beam (Isotropic) as the other elements receive RF the beam it 
>more and more forms. The same will happen at the back end of the pulse. It 
>could prevent us using very narrow pulse in a big phased.
>
>
>I had not seen the idea of beam forming as the way to go for HF radio but 
>as we are always trying to resolve signals in QRM the inclusion of one or 
>more null steerers in the Rx path would be very valuable - we take a lot 
>of care to make our filtering good in the frequency domain but ignore the 
>spacial domain that is just as valuable to us.
>
>
>I hope this helps a bit
>
>Nyall G8IBR
>
>
>----- Original Message ----- From: "Murray Lang" 
><murray.lang at metoceanengineers.com>
>To: <hpsdr at hpsdr.org>
>Sent: Thursday, September 07, 2006 6:57 AM
>Subject: Re: [hpsdr] Infinite precision, phased arrays,and us
>
>
>>***** High Performance Software Defined Radio Discussion List *****
>>
>>Thanks again Bob,
>>
>>Just to take a step backwards, this all came about from discussion in
>>another thread a while back where the topic of beamforming arose. At one
>>point I flippantly discounted controlling the RF phase shifts via I/Q but
>>was pulled up and reminded that with I/Q you can do anything. OK, you can
>>do phase modulation so why not a steady phase shift? I was uncomfortable
>>with it but couldn't put my finger on exactly why so I let it go. Now I'm
>>grabbing it back again.
>>
>>Yes, I knew that their would be more than one QSD/E with synchronised
>>(English) clocks. I also understood that the HPSDR dudes were already
>>catering for synchronised clocks - it's not new. I'm just trying to
>>understand why the concept of achieving phase shifts via cunning
>>manipulation of IQ is not sitting quietly in my head.
>>
>>It's one thing to shift the phase to accurately represent a modulating
>>audio signal, but to have such fine control as to be accurate to within a
>>fraction of a wavelength of the RF carrier seems to me a lot to ask of a
>>sound card. Consider a phase modulated signal applied to such a phase
>>shifted carrier. I/Q would have both shifts encoded into it: a component
>>accurate at AF and a component accurate at RF.
>>
>>Murray
>>VK6HL
>>
>>
>>
>>
>>
>>At 12:33 PM 7/09/2006, Robert McGwier wrote:
>>>***** High Performance Software Defined Radio Discussion List *****
>>>
>>>Baaa.  You are doing it justice.  Believe me.  There are 100 hundred
>>>people reading this that wanted to ask the questions and didn't.
>>>
>>>As I said in the previous,  for a phased array in the HPSDR system,
>>>we are not likely to use only one QSE for transmit or QSD for receive.
>>>  We are not likely to use a single sound card.   You will need one QSD
>>>and one sound card pair for EACH ELEMENT in the phased array and the
>>>QSD oscillator<<S>>  must all be the same source just as the sampling
>>>clock on the A/D's must all be derived from the same clock.   But they
>>>are separate devices, one per element in our narrow band phased array.
>>>   If we chose to use only one QSD and sound card, then the combing of
>>>the elements must be done before it and there must be a phase shifter
>>>in each antenna element.
>>>
>>>I think you are getting there.  You are asking the right questions to
>>>understand the system.
>>>
>>>Bob
>>>N4HY
>>>
>>>
>>>
>>>At 09:41 AM 7/09/2006, Murray Lang wrote:
>>> > ***** High Performance Software Defined Radio Discussion List *****
>>> >
>>> > What about the situation where QSE is the only "mixing" stage and I/Q
>>>is from a sound card? Where does frequency response come into it? You've
>>>only talked about word size. You seem to be saying that phase is a DC
>>>component of I/Q.
>>>
>>>Sorry Bob, I realise that I haven't done justice to what you've
>>>posted. I can see that the commutative nature of mixing helps but I'm
>>>having trouble relating that to the situation I've identified above.
>>>The only oscillator is the 4xoutput carrier and it's being modulated
>>>by a sound card.
>>>
>>>Murray
>>>VK6HL
>>>_______________________________________________
>>>HPSDR Discussion List
>>>To post msg: hpsdr at hpsdr.org
>>>Subscription help: http://lists.hpsdr.org/listinfo.cgi/hpsdr-hpsdr.org
>>>HPSDR web page: http://hpsdr.org
>>>Archives: http://lists.hpsdr.org/pipermail/hpsdr-hpsdr.org/
>>
>>
>>_______________________________________________
>>HPSDR Discussion List
>>To post msg: hpsdr at hpsdr.org
>>Subscription help: http://lists.hpsdr.org/listinfo.cgi/hpsdr-hpsdr.org
>>HPSDR web page: http://hpsdr.org
>>Archives: http://lists.hpsdr.org/pipermail/hpsdr-hpsdr.org/
>>
>
>

 1157682649.0