Forums

I-Q vs PLLs

Started by Unknown June 7, 2005
Which gives best performance, a Phase-locked-loop (say all digital -
software) or I-Q (ie using arctan and then differentiating)
demodulation.Reason I am asking is that a PLL is supposed to be the
best - is it better than just the pure number crunching.It woudl appear
to me that when you differentiate you introduce noise and the whole
point of a PLL is that it a servo which tracks rate of change of phase
and minimises noise.Also why is I-Q prefered rather than software PLL
in software radio? 


Thanks

sheepshaggerx@yahoo.co.uk wrote:
> Which gives best performance, a Phase-locked-loop (say all digital - > software) or I-Q (ie using arctan and then differentiating) > demodulation.Reason I am asking is that a PLL is supposed to be the > best - is it better than just the pure number crunching.It woudl appear > to me that when you differentiate you introduce noise and the whole > point of a PLL is that it a servo which tracks rate of change of phase > and minimises noise.Also why is I-Q prefered rather than software PLL > in software radio? > > > Thanks
I'm not a mind reader. What do you want to demodulate? Jerry -- Engineering is the art of making what you want from things you can get. �����������������������������������������������������������������������

Jerry Avins wrote:
> sheepshaggerx@yahoo.co.uk wrote: > > Which gives best performance, a Phase-locked-loop (say all digital - > > software) or I-Q (ie using arctan and then differentiating) > > demodulation.Reason I am asking is that a PLL is supposed to be the > > best - is it better than just the pure number crunching.It woudl appear > > to me that when you differentiate you introduce noise and the whole > > point of a PLL is that it a servo which tracks rate of change of phase > > and minimises noise.Also why is I-Q prefered rather than software PLL > > in software radio? > > > > > > Thanks > > I'm not a mind reader. What do you want to demodulate? >
I am talking about ordinary FM modulation.Lets say we are demodulating a sine wave. Thanks
On 7 Jun 2005 20:25:22 -0700, sheepshaggerx@yahoo.co.uk wrote:

>I am talking about ordinary FM modulation.Lets say we are demodulating >a sine wave. > >Thanks
There isn't much modulation in a sine wave...you may want to clarify further. Eric Jacobsen Minister of Algorithms, Intel Corp. My opinions may not be Intel's opinions. http://www.ericjacobsen.org
"Eric Jacobsen" <eric.jacobsen@ieee.org> wrote in message 
news:98ada1h80p0cahebroe28ec9415ealkhck@4ax.com...
> On 7 Jun 2005 20:25:22 -0700, sheepshaggerx@yahoo.co.uk wrote: > >>I am talking about ordinary FM modulation.Lets say we are demodulating >>a sine wave. >> >>Thanks > > There isn't much modulation in a sine wave...you may want to clarify > further.
Perhaps he means FSK. Leon
sheepshaggerx@yahoo.co.uk wrote:
> > Jerry Avins wrote: > >>sheepshaggerx@yahoo.co.uk wrote: >> >>>Which gives best performance, a Phase-locked-loop (say all digital - >>>software) or I-Q (ie using arctan and then differentiating) >>>demodulation.Reason I am asking is that a PLL is supposed to be the >>>best - is it better than just the pure number crunching.It woudl appear >>>to me that when you differentiate you introduce noise and the whole >>>point of a PLL is that it a servo which tracks rate of change of phase >>>and minimises noise.Also why is I-Q prefered rather than software PLL >>>in software radio? >>> >>> >>>Thanks >> >>I'm not a mind reader. What do you want to demodulate? >> > > > I am talking about ordinary FM modulation.Lets say we are demodulating > a sine wave. > > Thanks
Ah! So it's not AM. We're a step ahead now. Later, when I have more time. (PLLs have settling and sensitivity issues. If they follow too quickly, they are easily upset be noise.) Jerry -- Engineering is the art of making what you want from things you can get. &#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;
Hey Gumboots,

It really depends on what the signal looks like that you are trying to demod 
and what hardware and processing constraints are. (And what side of the bed 
you woke up on that morning) For instance, if you wanted to extract every 
possible bit of information out of a captured signal, and you had all the 
time in the world to do this, then I would use an I/Q arctan type approach. 
But if your processing power/time is limited and your demoding in real time, 
then I would use a software PLL. (roughly speaking)

You'll have to be very specific on your application if you want answers that 
are somewhat applicable. They both have many advantages and disadvantages.

Thomas

<sheepshaggerx@yahoo.co.uk> wrote in message 
news:1118196820.637569.107180@g47g2000cwa.googlegroups.com...
> Which gives best performance, a Phase-locked-loop (say all digital - > software) or I-Q (ie using arctan and then differentiating) > demodulation.Reason I am asking is that a PLL is supposed to be the > best - is it better than just the pure number crunching.It woudl appear > to me that when you differentiate you introduce noise and the whole > point of a PLL is that it a servo which tracks rate of change of phase > and minimises noise.Also why is I-Q prefered rather than software PLL > in software radio? > > > Thanks >
My application is in software radio so that the signal will have a
spectrum (random) anything up to say 15kHz for a normal FM stereo
radio. I use the sine wave as an example as we normally test systems on
sine waves and sweep them to do a freq response.

I am worried about the digital differentiation bit and noise. I though
the whole point of a PLL was that we went away from the old ideas of
differentiation (Foster Seeley).The PLL gave better performance. Well
here we are in the digital age and we go back to old ideas.Whats the
point in using digital demodulation if the SNR is poorer than analogue?

Thanks

sheepshaggerx@yahoo.co.uk wrote:
> My application is in software radio so that the signal will have a > spectrum (random) anything up to say 15kHz for a normal FM stereo > radio. I use the sine wave as an example as we normally test systems on > sine waves and sweep them to do a freq response.
Broadcast FM has a carrier deviation of 150 KHz in a channel 200 KHz wide.
> > I am worried about the digital differentiation bit and noise. I though > the whole point of a PLL was that we went away from the old ideas of > differentiation (Foster Seeley).The PLL gave better performance. Well > here we are in the digital age and we go back to old ideas. Whats the > point in using digital demodulation if the SNR is poorer than analogue?
If it were worse, it wouldn't be used. Theory be damned. Use whatever technique works best. What about the Foster-Seely discriminator or the Seely-Avins ratio detector involves differentiation? The _difference_ between the diode outputs is not _differentiation_. Jerry -- Engineering is the art of making what you want from things you can get. &#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;
Jerry Avins wrote:

> sheepshaggerx@yahoo.co.uk wrote: > >> My application is in software radio so that the signal will have a >> spectrum (random) anything up to say 15kHz for a normal FM stereo >> radio. I use the sine wave as an example as we normally test systems on >> sine waves and sweep them to do a freq response.
> Broadcast FM has a carrier deviation of 150 KHz in a channel 200 KHz wide.
I thought it was 75kHz, but it is plus or minus 75kHz so the bandwidth is (approximately) 150kHz. -- glen