Message-ID: <3A69FEB9.BAABE575@bellatlantic.net>
Date: Sat, 20 Jan 2001 16:10:17 -0500
From: "Andre' Kesteloot" <akestelo@bellatlantic.net>
MIME-Version: 1.0
To: rsgb_lf_group@blacksheep.org
Subject: Re: LF: PLL inaccuracy
References: <000f01c0783a$c6049800$11e27ad5@default>
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 8bit
Precedence: bulk
Reply-To: rsgb_lf_group@blacksheep.org
Sender: <majordom@post.thorcom.com>

Alan Melia wrote:

> Hi Andre' that is not my appreciation of how off air standards work. You do
> not try to get a lock up in 64 microseconds (one frame). In this case the
> PLL is not trying to compensate for poor short term stability. The
> short-term stability of the VCO must be excellent. My Droitwich standard can
> take up to 20 minutes to reach lock after switch-on. The VCO is a very
> stable VXO which still have reasonable performance when the locking signal
> disappears. I think you must consider trading time for accuracy. A good PLL
> should always have the best VCO you can get to reduce the incidental FM. So
> over 10 minutes of nudging it in the right direction (not compensating for
> its short term stability) it is easy to see you can get 1000 times (at
> least) better performance. As you say there will always be a slight error,
> but it can be made vanishingly small.
>
> Cheers de Alan G3NYK
> Alan.Melia@btinternet.com

I am not sure I can agree with this understanding. The 
_time_  required to
reach lock is indeed a function of the time constant in the
feedback loop, but
the total amount of possible drift is not.  The way a PLL
works  _is_  because
it drifts ever so slightly in either direction, thereby
creating a DC component
that is then used to bring the VCO back in the other
direction.
If we insert a divide-by-64 between the VCO and the PLL
comparator, then the
error in the comparator (phase difference) must be
multiplied by 64.

For simplicity's sake, let's assume that we have a phase
comparator that needs a
1 Hz difference between its two inputs before it creates a
DC correction signal 
large enough to control the VCO (not a very good phase
comparator, assuredly, 
but just for demonstration purposes).

The second phase comparator's input is the TV line
frequency.
The VCO operates at 1,000,000 Hz and has a divide-by-64
inserted between its
output and the phase comparator.
If the VCO is exactly at 1MHz, then the output of the
divider is 15,625Hz, but
the VCO can drift all the way to 1,000,064 Hz before the
phase comparator will
see a 1 Hz difference, and start reacting.

73
Andre' N4ICK