Bonjour, I'd like to understand what is globally a bandwidth PLL. I know that a PLL uses in internal low pass filter (numerical). But, if we take a PLL as a black box, in relation with the input reference timing, is it again a low pass filter with just a high cutoff frequency, or a bandpass filter with both a low and high cutoff frequencies ? Thanks for some words about this. Best regards, Michelot

# PLL Basics

Michelot wrote:> Bonjour, > > I'd like to understand what is globally a bandwidth PLL. > > I know that a PLL uses in internal low pass filter (numerical). But, > if we take a PLL as a black box, in relation with the input reference > timing, is it again a low pass filter with just a high cutoff > frequency, or a bandpass filter with both a low and high cutoff > frequencies ?A PLL (that stands for "phase-locked loop") is not in itself a filter. It is a means for locking the phase of an oscillator (with or without constant offset) to a reference signal. Jerry -- Engineering is the art of making what you want from things you can get. �����������������������������������������������������������������������

Jerry Avins <jya@ieee.org> wrote in news:88gll.630$2O4.477@newsfe03.iad:> Michelot wrote: >> Bonjour, >> >> I'd like to understand what is globally a bandwidth PLL. >> >> I know that a PLL uses in internal low pass filter (numerical). But, >> if we take a PLL as a black box, in relation with the input reference >> timing, is it again a low pass filter with just a high cutoff >> frequency, or a bandpass filter with both a low and high cutoff >> frequencies ? > > A PLL (that stands for "phase-locked loop") is not in itself a filter. > It is a means for locking the phase of an oscillator (with or without > constant offset) to a reference signal. > > JerryIf you consider a phase perturbation on the input, a lowpass filtered version of that phase perturbation appears on the output. One can model the PLL as having a transfer function (phase out / phase in). The output centre frequency can be modelled as a linear phase ramp. Since linear ramps aren't very interesting (from control theory, we know that ramps can be tracked with zero error in a Type II system) we can ignore it, and just consider the variations from the ideal ramp. This enables one to analyse or simulate just the interesting stuff (noise, frequency domain performance, transient performance, etc.) at DC without having to worry about the carrier. This saves a huge amount of time in analysis or simulation. N.B. The two integrators we need for a Type II system come from the VCO (which acts like an ideal integrator: a constant input voltage results in a constantly ramping (phase) output) and the loop filter (which is often just a current source feeding an RC network). Regards, Allan (Currently listening to the Laughing Clowns)

On Fri, 13 Feb 2009 04:07:01 -0800, Michelot wrote:> Bonjour, > > I'd like to understand what is globally a bandwidth PLL. > > I know that a PLL uses in internal low pass filter (numerical). But, if > we take a PLL as a black box, in relation with the input reference > timing, is it again a low pass filter with just a high cutoff frequency, > or a bandpass filter with both a low and high cutoff frequencies ? > > Thanks for some words about this. > Best regards, > MichelotIf it weren't locked in phase, it wouldn't be a phase locked loop. If it weren't inherently low pass in phase, it wouldn't stay locked. Hence, if it can be called a PLL, it is low pass. -- http://www.wescottdesign.com

Allan Herriman wrote:> Jerry Avins <jya@ieee.org> wrote in news:88gll.630$2O4.477@newsfe03.iad: > >> Michelot wrote: >>> Bonjour, >>> >>> I'd like to understand what is globally a bandwidth PLL. >>> >>> I know that a PLL uses in internal low pass filter (numerical). But, >>> if we take a PLL as a black box, in relation with the input reference >>> timing, is it again a low pass filter with just a high cutoff >>> frequency, or a bandpass filter with both a low and high cutoff >>> frequencies ? >> A PLL (that stands for "phase-locked loop") is not in itself a filter. >> It is a means for locking the phase of an oscillator (with or without >> constant offset) to a reference signal. >> >> Jerry > > If you consider a phase perturbation on the input, a lowpass filtered > version of that phase perturbation appears on the output. > > One can model the PLL as having a transfer function (phase out / phase > in). > > The output centre frequency can be modelled as a linear phase ramp. > Since linear ramps aren't very interesting (from control theory, we know > that ramps can be tracked with zero error in a Type II system) we can > ignore it, and just consider the variations from the ideal ramp. This > enables one to analyse or simulate just the interesting stuff (noise, > frequency domain performance, transient performance, etc.) at DC without > having to worry about the carrier. This saves a huge amount of time in > analysis or simulation. > > N.B. The two integrators we need for a Type II system come from the VCO > (which acts like an ideal integrator: a constant input voltage results in > a constantly ramping (phase) output) and the loop filter (which is often > just a current source feeding an RC network).Of course filters are used in a PLL implementation. That doesn't mean a PLL is a filter, any more it is a VCO. Jerry -- Engineering is the art of making what you want from things you can get. �����������������������������������������������������������������������

On Fri, 13 Feb 2009 11:36:50 -0500, Jerry Avins <jya@ieee.org> wrote:>Of course filters are used in a PLL implementation. That doesn't mean a >PLL is a filter, any more it is a VCO.I'm curious as to how you make the distinction. What makes a PLL a non-filter? It has an H(s), it takes a noisy periodic signal and produces cleaned up, frequency shaped version of it. How is that qualitatively different from an RC filter?

Muzaffer Kal wrote:> On Fri, 13 Feb 2009 11:36:50 -0500, Jerry Avins <jya@ieee.org> wrote: > >> Of course filters are used in a PLL implementation. That doesn't mean a >> PLL is a filter, any more it is a VCO. > > I'm curious as to how you make the distinction. What makes a PLL a > non-filter? It has an H(s), it takes a noisy periodic signal and > produces cleaned up, frequency shaped version of it. How is that > qualitatively different from an RC filter?It's also a signal generator, no? The OP asked a a simple,question that (it seemed to me) embodied a misconception that I tried to dispel. To answer your question, a simple RC filter isn't bandpass. Jerry -- Engineering is the art of making what you want from things you can get. �����������������������������������������������������������������������

On Feb 14, 8:30�am, Jerry Avins <j...@ieee.org> wrote:> Muzaffer Kal wrote: > > On Fri, 13 Feb 2009 11:36:50 -0500, Jerry Avins <j...@ieee.org> wrote: > > >> Of course filters are used in a PLL implementation. That doesn't mean a > >> PLL is a filter, any more it is a VCO. > > > I'm curious as to how you make the distinction. What makes a PLL a > > non-filter? It has an H(s), it takes a noisy periodic signal and > > produces cleaned up, frequency shaped version of it. How is that > > qualitatively different from an RC filter? > > It's also a signal generator, no? The OP asked a a simple,question that > (it seemed to me) embodied a misconception that I tried to dispel. To > answer your question, a simple RC filter isn't bandpass. > > Jerry > -- > Engineering is the art of making what you want from things you can get. >����������������������������������������������������������������������� It's like a differentiator but for the phase. So you cannot think of the original signal as input if you want this type of analogy. Hardy

Bonsoir, Thanks very much Jerry, Allan, HardySpicer and Tim for your simple words. Sorry for my misconception, PLL is really not a filter. The word "bandwidth" (in Hz) is not only reserved to fiters, and there are certainly many other objects which have also a bandwidth, as PLL e.g.> Hence, if it can be called a PLL, it is low pass.So, it is low pass when it is locked, and when we make varied the input reference frequency. Is it correct ? If we haven't frequency multiplication or division, for the input reference frequency Fin = F0 we have Fout = k F0. If now, we have Fin = F0 + f(t), with f the frequency change according to the time, I think we would get Fout = kF0 + T(f) f(t), with T(f) a low pass transfert function. Is it correct like this? Best regards, Michelot

HardySpicer wrote:> On Feb 14, 8:30 am, Jerry Avins <j...@ieee.org> wrote: >> Muzaffer Kal wrote: >>> On Fri, 13 Feb 2009 11:36:50 -0500, Jerry Avins <j...@ieee.org> wrote: >>>> Of course filters are used in a PLL implementation. That doesn't mean a >>>> PLL is a filter, any more it is a VCO. >>> I'm curious as to how you make the distinction. What makes a PLL a >>> non-filter? It has an H(s), it takes a noisy periodic signal and >>> produces cleaned up, frequency shaped version of it. How is that >>> qualitatively different from an RC filter? >> It's also a signal generator, no? The OP asked a a simple,question that >> (it seemed to me) embodied a misconception that I tried to dispel. To >> answer your question, a simple RC filter isn't bandpass.> It's like a differentiator but for the phase. So you cannot think of > the original signal as input if you want this type of analogy.I don't understand your meaning here. Jerry -- Engineering is the art of making what you want from things you can get. �����������������������������������������������������������������������