[SI-LIST] Re: Question on PLL for Data/Clock Recovery Apps

Larry,

I think you misread Ed's post.

Ed has it right on all points.  Phase is the integral of frequency, 
frequency, or angular velocity is the derivative of phase.

I believe that Ed was discussing the fact that the loop imposes a short 
term frequency error of opposite sign ( ave ) to the phase error in order 
to correct phase errors.  In the long term this forces the average 
frequency error to zero.  I believe that Ed tried to make clear that this 
frequency shift isn't a problem.  Rather, it's a necessary part of the 
solution of maintaining phase alignment.

On the cycle skipping issue, even with a PFD, Ed is again correct. If the 
loop bandwidth is insufficient for the incoming disturbance, even using a 
PFD, cycle skipping of one or more cycles will occur.  This is a good 
reason not be careful with, or just avoid VCO based oscillators with high 
division ratios:  power supply disturbances can lead to unwelcome surprises.

Regards,


Steve.


At 08:35 AM 12/11/2001 -0800, you wrote:

>You are getting some incorrect advice here. See below.
>
>Larry Miller
>
>-----Original Message-----
>From: Ed Miguel [mailto:emiguel@xxxxxxxxxx]
>Sent: Tuesday, December 11, 2001 7:53 AM
>To: si-list@xxxxxxxxxxxxx
>Subject: [SI-LIST] Re: Question on PLL for Data/Clock Recovery Apps
>
>
>Krishnan,
>I think you have to go back and look at the PLL as a PHASE locked loop and
>switch out of the frequency domain which is where your concerns for
>accuracy lie.   Look at the basic elements of a PLL, first there is a PHASE
>detector.  Any phase error will get translated through this which will
>include jitter, reference movement and VCO noise and movement.  This phase
>information will then get filtered, into average phase error, that will
>then be used to control the VCO.   Since the PLL is usually a PI
>(proportional/integral) closed loop servo system, the output VCO frequency
>will be adjusted but only enough to zero the phase error.   Remember that
>phase is an integral of frequency.
>
>----->>>Phase is the DERIVATIVE of frequency! Ed has it backwards.
>
>
>The response of the PLL is described
>in the gain of the loop as well as the zeta or damping factor. A PLL will
>typically not dwell out of frequency for a while but it will operate
>according to its loop transfer function.  If anything it will be slow to
>respond because of the filtering.   Of course this is true for the
>condition where the S/N is not so low that the signal is buried in noise
>(jitter) because under those conditions cycle skipping can occur.  It also
>assumes that the noise is gaussian.  If it is not then it will be treated
>as a phase transient and that will cause the PLL to follow the phase of
>transient (to zero it out) and in the process of doing that it may appear
>to shift in frequency for a while until it gets there.  Finally (I think)
>the PLL in acquisition is non-linear so there will be some frequency
>shifting taking place until it gets frequency locked at which point the
>phase error takes over the control.
>
>----->>> You have to achieve frequency lock before you can phase lock. If
>you are off frequency the PLL will cycle skip and is useless. Old phase
>detector designs did this and had a tendency to lock onto sub-harmonics of
>the reference signal (i.e., integral fractions of it like half). Any PLL
>phase detector that is worthwhile these days will keep its error output
>driving to eliminate the frequency error before it drops in the slot on
>phase.
>
>
>You raise some interesting questions and it is something that we all go
>through trying to visualize the operation of a PLL so you are not
>alone.  There are some good books on PLL's that you can search for under
>Phase Locked Loops or PLL's by Gardner or Wolaver or Egan or Encinas and
>there are some good Ap Notes out there on the Web.   In general I would
>say, look at the transfer function and note the response of the loop as a
>function of zeta, the damping factor.  This define how a PLL will typically
>respond.
>
>Ed Miguel
>Engineering Manager of TPG products
>Connor Winfield Corp.
>
>------------------------------------------------------------------
>To unsubscribe from si-list:
>si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field
>
>or to administer your membership from a web page, go to:
>http://www.freelists.org/webpage/si-list
>
>For help:
>si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field
>
>List archives are viewable at:
>                 http://www.freelists.org/archives/si-list
>or at our remote archives:
>                 http://groups.yahoo.com/group/si-list/messages
>Old (prior to June 6, 2001) list archives are viewable at:
>                 http://www.qsl.net/wb6tpu
>

------------------------------------------------------------------
To unsubscribe from si-list:
si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field

or to administer your membership from a web page, go to:
http://www.freelists.org/webpage/si-list

For help:
si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field

List archives are viewable at:     
                http://www.freelists.org/archives/si-list
or at our remote archives:
                http://groups.yahoo.com/group/si-list/messages 
Old (prior to June 6, 2001) list archives are viewable at:
                http://www.qsl.net/wb6tpu

Other related posts: