Well, my last monday morning mail ( having my tongue tiwsted a little for
Shakespeare's language ) I want to say the truth there is no need for an
owen to achieve extremely stable clocks ( near 25C ! )
The clock ethalon I have used is the german radio broadcasting which
covers all romanian teritory and a standard ( 24 hour with seconds)
commercial clock.
And again, thermostating a cuartz oscillator can gave worst results mostly
because:
 Any cuartz crystal can be defined as a series of an Re ( equivalent
resistance ) and Xe ( equivalent reactance ) no matter if they have serial
or parallel configuration ( AT or X axis cutting ). Imagine the graphic
showing the dependence of Xe versus frequency. Any crystal quartz have
two points in which the crystal is pure resistive. These points are
defined as resonant frequency ( fr ) and atiresonant frequency ( fa ).
Any parallel resonant quartz will oscillate somewhere between these two
frequencies and depends of capacitive load on the cuartz.
As an example the standard ( AT axis cutted ) 32768 quartz have fr at
about 32760 and the fa at about 32790. Let say we've got a stable
oscillation at exactly 32768Hz at 25C. Thermostating at higher temperature
( usual at 50...60C ) will shift the oscillating frequency because of
strong variation of Xe with temperature. If the Xe/frequency
characteristic have a high variation at thermostated temperature the effect
will be a high frequency deviation for a very small temperature deviation.
This is happening all the time when insufficient datasheet are comming
with the crystal and your oven is not perfect...
 Staying on piclist from about 2 years I heard a lot of funny explanation
about why is needed the series resistor on PI microcontroller oscillating
network. The funniest I ever heard ( and my mouth is now open up to my
ears ) is that some small quartz can be damaged if this resistor is not
used.
AGREE, ONLY IF YOU HAVE USE A 2KG HAMMER ! and smash the litte quartz with
him...
 The PI network ( capacitor to gnd, cuartz oscillator, capacitor to gnd )
can be used only with an 180 degree phase shift amplifier. The most common
amplifier on microcontroller is an inverter gate. The phase shift
generated by PI network is extremely sensitive at frequency deviation,
this is the input condition for a stable global oscillation ( with cuartz
crystal ) To have a stable oscillation the equivalent resistance Re must
be zero. Case which is rarely found in practice.
The serial resitance is used to modify a little the phase shift and the
load on the crystal, but only that load created by input/output of the
oscillator amplifier, nothing more.
Some time this help, some time create problems, and this depends by cuartz
structure and phase shift stability of the whole network.
I hope you have understood me now.
For romanian readers of this list, there is an old good book treating this
subject: "CMOS integrated circuits" by Iulian Ardelean and others at
Editura Tehnica, Bucuresti, 1986

Best from Transilvania,

Vasile


On Mon, 18 Feb 2002, Bob Ammerman wrote:

> > BTW, the best accuracy I've got with bresenham methode was +/- 2
> > sec/month. Could be better ?
> >
> > best regards, Vasile
> >
>
> This represents an error of about 0.8 ppm. This is pretty amazing, given
> that crystals are generally rated an order of magnitude or 2 worse than
> 1ppm. (or are you using an oven, etc).
>
> Bob Ammerman
> RAm Systems
>
> --
> http://www.piclist.com hint: The PICList is archived three different
> ways.  See http://www.piclist.com/#archives for details.
>
>

--
http://www.piclist.com#nomail Going offline? Don't AutoReply us!
email listserv@mitvma.mit.edu with SET PICList DIGEST in the body