From: Chuck Simmons
Organization: You jest.
X-Mailer: Mozilla 4.61 [en] (X11; U; Linux 2.0.33 i586)
Subject: Re: Phase locked loops and confusion.
References: <3DDED492.2B9873B0@webaccess.net> <3DDF16D8.90072F5@webaccess.net> <3DDF229B.B85180DD@ieee.org>
Date: Sat, 23 Nov 2002 12:41:21 GMT
NNTP-Posting-Date: Sat, 23 Nov 2002 04:41:21 PST
Kevin Aylward wrote:
> analog wrote:
> > Chuck Simmons, John Larkin wrote in part:
> >>> Yeah, the charge pump adds another integrator to the loop. I don't
> >>> know why the charge pump thing is so popular;
> It makes the charge and discharge current to the capacitor equal and
> independent of capacitor voltage. If the capacitor was charged from a
> voltage source through a resistor the current would vary. For instance,
> suppose that the input frequency was such that 1V was the steady state
> voltage on the cap, and that the PD output was 0 to 5 volts. The current
> would switch from 4/R to 1/R. If the input frequency now changed such
> that the steady states voltage was 2V, the current would switch from 3/R
> to 2/R. Now, what would be the result of this on the system?
> most of them have
> >>> nasty deadbands. A charge pump pd doesn't inherently drive the
> >>> loop in the right direction when the freq error is large; it takes
> >>> a bit of additional logic (like in the 4046) to do that.
> Actually, I was a bit misleading on my dead band post, now I have woken
> up a bit. The dead band, is of course, the result of the whole phase
> detector/charge pump arrangement. I thought you were implying some sort
> of x-over distortion type effect, which wouldn't really make sense at
> all. So, if the phase detector puts out pulses two narrow they won't
> make it through the charge pump to the filter cap, due to limited BW of
> the pump. The fix for this, is to force the phase detector to generate
> equal narrow +/- pulses when the input frequencies/phases are matched.
Yes, we are aware of that one. Translating this into digital equivalents
was not to difficult. I haven't seen the logic but it simulates well.
The point that came up that I mentioned is that of deciding the phase
detector structure and the type of filter needed seems confusing to some
because of the charge pump and cap being popular but not very usual in
some of the papers. Actually, we have to have an integrator in the loop
but we have choices on how it is introduced (how we add up phase error
and introduce the sum into the output). The structure of the loop filter
is pretty much dictated by that.
Actually, after one of the engineers printed the paper he had read, the
design cleared up rapidly after I pointed out that I was counting one
more integrator than he was and that I had no argument with the paper.
The additional integrator simply required a greatly different filter
... The times have been,
That, when the brains were out,
the man would die. ... Macbeth
Chuck Simmons firstname.lastname@example.org