crystal precision

P

Prospero

Guest
I'm a bit confuse about crystals precision and stability.

Is it possible to tune a 4Mz crystal with a simple trimmer with 1Hz
precision?
Crystal specs usually state +-30ppm stability but I assume this is an
untrimmed crystal.
If anyone can shade some light about this, I would be gratefull!

--- Propero ---
 
Precision and stability are two separate things.

And precision usually refers to the ability to measure something to within a
defined accuracy. So for instance you could measure the crystal frequency
to a precision of 1 Hz.

The accuracy of the crystal is a measure of how close it is to some defined
value, under certain measuring conditions. So a crystal might measure
40000125 Hz ( measured to a precision of 1 Hz ), and be called a 4000000 Hz
crystal, to an accuracy of better than 40 ppm.

Stability refers to how the frequency might change with environmental
conditions, like time, temperature, humidity, magnetic field, etc, etc. So
you have to know what your environment is, and what changes to expect which
are likely to affect the crystal. So you might have a crystal which changes
less than 30 ppm over the temp range of 10-40 degrees Celsius, for example.

Pulling a crystal with a trimmer ( normally a small capacitor in parallel,
which lowers the frequency ) can only be done over a very limited range, and
only lower in frequency. So you might be able to pull a 40000125 Hz crystal
down to 4000000 Hz, but not a 39999954 Hz one up to 4000000 at all.
So buying any 4000000 Hz crystal, ( which has a 50% chance of being lower
than the nominal ) is not a good option, if you want to trim it. Also the
further you pull the frequency, the more you affect the stability, since
also the change in capacitance of the trimmer with the environment has to be
taken into account.

Hope this helps.

--
Regards,

Adrian Jansen
J & K MicroSystems
Microcomputer solutions for industrial control
"Prospero" <prospero@invalid.net> wrote in message
news:30Bjc.128816$2V6.1429656@wagner.videotron.net...
I'm a bit confuse about crystals precision and stability.

Is it possible to tune a 4Mz crystal with a simple trimmer with 1Hz
precision?
Crystal specs usually state +-30ppm stability but I assume this is an
untrimmed crystal.
If anyone can shade some light about this, I would be gratefull!

--- Propero ---
Click to expand...
 
On Tue, 27 Apr 2004 18:19:28 -0400, "Prospero" <prospero@invalid.net>
wrote:

I'm a bit confuse about crystals precision and stability.

Is it possible to tune a 4Mz crystal with a simple trimmer with 1Hz
precision?
Click to expand...
For a short while. A cheap crystal oscillator will drift some decent
fraction of a PPM per degree C, and several PPM per year, and will
have additional powersupply, shock, and g-force (orientation)
sensitivity. The trimmer cap will need to have good settability, too,
maybe a piston cap. Cheap trimmers cover their capacitance range in
maybe 1/3 of a turn.

1 Hz, 0.25 PPM, is probably not dependable for a simple XO on a
longterm basis, maybe not for an hour.

John
 
Hi Prospero,

To get even close to that precision for more than a few minutes would
require that the whole circuit resides in a temperature regulated
housing. Why not synchronize it to WWV or something like that? Their
signals at 10MHz and 15MHz are strong in most of the US if you live
there. Other countries have similar services but maybe not on such
convenient frequencies.

Regards, Joerg
http://www.analogconsultants.com
 
Thanks for your answers!

I may be better off tuning to a cesium clock somewhere :)

--- Prspero ---

"Prospero" <prospero@invalid.net> a écrit dans le message de
news:30Bjc.128816$2V6.1429656@wagner.videotron.net...
I'm a bit confuse about crystals precision and stability.

Is it possible to tune a 4Mz crystal with a simple trimmer with 1Hz
precision?
Crystal specs usually state +-30ppm stability but I assume this is an
untrimmed crystal.
If anyone can shade some light about this, I would be gratefull!

--- Propero ---
Click to expand...
 
"Prospero" <prospero@invalid.net> wrote in message
news:30Bjc.128816$2V6.1429656@wagner.videotron.net...
I'm a bit confuse about crystals precision and stability.

Is it possible to tune a 4Mz crystal with a simple trimmer with 1Hz
precision?
Crystal specs usually state +-30ppm stability but I assume this is an
untrimmed crystal.
If anyone can shade some light about this, I would be gratefull!

--- Propero ---



Crystals are marked for the frequency they will oscillate at when presented
Click to expand...
with a particular load capacitance. 30 PF is a common value. Assume the
crystal was perfect when it left the factory; then it will oscillate at
4.000000 MHz if it sees 30 PF. Make the capacitor bigger, and the frequency
goes down. Make the capacitor smaller, and the frequency goes up. AT cut 4
MHz crystals will pull hundreds of Hz.

Having trimmed a crystal to frequency is not the same as having it stay
there. There are aging and temperature effects. An oscillator with 1 part
per million stability is considered precision and has to be temperature
compensated. You are looking for 0.25ppm.

There should be lots of information at crystal manufacturer's web sites.
Also look up TCXOs, but don't be surprised at the $$$.

Tam
 
Have a look at some crystal references. A google search for "quartz
crystal resonators" will turn up a bunch of stuff.
http://www.rakon.com/generated/1-25/technicle_article_files/frame.htm
is probably reasonable.

Generally there will be a calibration accuracy and some specification
of temperature stability. There may also be a spec on long term
drift, though that's usually reserved for high-accuracy crystals.
There are other important specs as well: one is the load capacitance
for the calibrated frequency. If your capacitance trimming can adjust
the load capacitance both above and below the calibration value, then
you can adjust the frequency below and above the nominal value, and
likely you'll be able to hit the 4MHz as close as you have patience to
adjust it; and then it will promptly drift off as the temperature
changes and the crystal ages and some other things. If you control
the temperature and pick a temperature near a zero-tempco point for
the crystal, it's probably reasonable with a good oscillator circuit
to hold within 1 part per million for fairly long periods, probably
many days. Without the temperature control, expect much larger
variation: see the S-curve of freq versus temperature which is
characteristic of AT-cut crystals (almost certainly what you have at
4MHz).

Cheers,
Tom




"Prospero" <prospero@invalid.net> wrote in message news:<30Bjc.128816$2V6.1429656@wagner.videotron.net>...
I'm a bit confuse about crystals precision and stability.

Is it possible to tune a 4Mz crystal with a simple trimmer with 1Hz
precision?
Crystal specs usually state +-30ppm stability but I assume this is an
untrimmed crystal.
If anyone can shade some light about this, I would be gratefull!

--- Propero ---
Click to expand...
 
If you are after a good time base, consider the GPS signals. Search on
Google for Totally Accurate Clock.

--
Regards,

Adrian Jansen
J & K MicroSystems
Microcomputer solutions for industrial control
"Prospero" <prospero@invalid.net> wrote in message
news:_WBjc.65627$Tu2.1353688@weber.videotron.net...
Thanks for your answers!

I may be better off tuning to a cesium clock somewhere :)

--- Prspero ---

"Prospero" <prospero@invalid.net> a écrit dans le message de
news:30Bjc.128816$2V6.1429656@wagner.videotron.net...
I'm a bit confuse about crystals precision and stability.

Is it possible to tune a 4Mz crystal with a simple trimmer with 1Hz
precision?
Crystal specs usually state +-30ppm stability but I assume this is an
untrimmed crystal.
If anyone can shade some light about this, I would be gratefull!

--- Propero ---
Click to expand...
 
Prospero wrote:
Thanks for your answers!

I may be better off tuning to a cesium clock somewhere :)
Click to expand...
if you have lots of dollars, a rubidium clock is
cheaper, but still darn good.

And much cheaper is a temperature stabilized crystal

and still cheaper is a temperature compensated crystal oscillator.
These can be small enough to fit onto a circuit board.

If I wanted to build a pretty good oscillator for hobby
use, I think I'd get a cheap Casio watch (occasionally
sold for $5) and try to pick of the oscillator signal
with a jFet or maybe a pickup coil.


--
local optimization seldom leads to global optimization

my e-mail address is: <my first name> <my last name> AT mmm DOT com
 
In sci.electronics.design, Roy McCammon <rbmccammon@mmm.com> wrote:

Prospero wrote:
Thanks for your answers!

I may be better off tuning to a cesium clock somewhere :)

if you have lots of dollars, a rubidium clock is
cheaper, but still darn good.

And much cheaper is a temperature stabilized crystal

and still cheaper is a temperature compensated crystal oscillator.
These can be small enough to fit onto a circuit board.

If I wanted to build a pretty good oscillator for hobby
use, I think I'd get a cheap Casio watch (occasionally
sold for $5) and try to pick of the oscillator signal
with a jFet or maybe a pickup coil.
Click to expand...
For better-than-average crystal frequency stability, there are
temperature-compensated crystal oscillators such as the Dallas/Maxim
DS32KHz termperature-compensated oscillator, TCXO's from this company
(carried by digikey):

http://www.ecsxtal.com

There's some general info on the Technical Guides page, and looking
at their products shows the range of accuracy and stability available.
The best shows 2.5ppm (the DS32KHZ is about the same), which is 10 Hz
at 4MHz. A crystal oven can do a little better, but if you want much
better it will take one of those atomic oscillator/clocks or a
WWV/WWVH/GPS receiver.
 
Hello Ben,

The "atomic clock" version won't work for most places in the US. The
signal from NIST in Boulder, CO, is so weak that it can really be only
used at night at places like here in California. Our clocks look for that
signal around midnight, synchronize and then go stand-alone. They don't
synchronize again until the next night. Good for time keeping but not for
a lab oscillator.

WWV and WWVH are much better. But not all over the world.

Regards, Joerg
http://www.analogconsultants.com
 
In sci.electronics.design, Joerg
<notthisjoergsch@removethispacbell.net> wrote:

Hello Ben,

The "atomic clock" version won't work for most places in the US.
Click to expand...
Actually by "atomic clock" I meant an actual Cesium oscilator and
divider, generating and outputting a stable frequency standard, rather
than a receiver. It's more expensive than a receiver, but doesn't rely
on radio propagation.
I was annoyed when I started seeing "atomic clocks" that were just
cheap quartz clocks with WWVH receivers in them. It's odd how the
terminology got changed when those were marketed.

Anyway, here's a REAL fully self-contained atomic wristwatch, and
was just what I was thinking of by "atomic clock." It should put the
OP on the fast track to getting his 4MHz with well within 1 Hz
accuracy:

http://www.leapsecond.com/pages/atomic-bill/index.htm

The
signal from NIST in Boulder, CO, is so weak that it can really be only
used at night at places like here in California. Our clocks look for that
signal around midnight, synchronize and then go stand-alone. They don't
synchronize again until the next night. Good for time keeping but not for
a lab oscillator.

WWV and WWVH are much better. But not all over the world.

Regards, Joerg
http://www.analogconsultants.com
Click to expand...
-----
http://mindspring.com/~benbradley
 
Anyway, here's a REAL fully self-contained atomic wristwatch, and
was just what I was thinking of by "atomic clock." It should put the
OP on the fast track to getting his 4MHz with well within 1 Hz
accuracy:

http://www.leapsecond.com/pages/atomic-bill/index.htm
Click to expand...

But.. No alarm function!
 
On Thu, 29 Apr 2004 21:08:15 -0400, Ben Bradley
<ben_nospam_bradley@mindspring.example.com> wrote:



Actually by "atomic clock" I meant an actual Cesium oscilator and
divider, generating and outputting a stable frequency standard, rather
than a receiver. It's more expensive than a receiver, but doesn't rely
on radio propagation.
Click to expand...
You can find unused Rubidium oscillators on ebay now and then, $300 or
so. They output 10 MHz. They're not legally primary standards like a
Cesium is, but are seldom off more than a part in 1e9 or so.

John
 
"Ben Bradley" <ben_nospam_bradley@mindspring.example.com> wrote in message
news:07tv801pl32u8p9mjevi9mbtbt5i1nuiv9@4ax.com...
In sci.electronics.design, Roy McCammon <rbmccammon@mmm.com> wrote:

Prospero wrote:
Thanks for your answers!

I may be better off tuning to a cesium clock somewhere :)

if you have lots of dollars, a rubidium clock is
cheaper, but still darn good.

And much cheaper is a temperature stabilized crystal

and still cheaper is a temperature compensated crystal oscillator.
These can be small enough to fit onto a circuit board.

If I wanted to build a pretty good oscillator for hobby
use, I think I'd get a cheap Casio watch (occasionally
sold for $5) and try to pick of the oscillator signal
with a jFet or maybe a pickup coil.

For better-than-average crystal frequency stability, there are
temperature-compensated crystal oscillators such as the Dallas/Maxim
DS32KHz termperature-compensated oscillator, TCXO's from this company
(carried by digikey):

http://www.ecsxtal.com

There's some general info on the Technical Guides page, and looking
at their products shows the range of accuracy and stability available.
The best shows 2.5ppm (the DS32KHZ is about the same), which is 10 Hz
at 4MHz. A crystal oven can do a little better, but if you want much
better it will take one of those atomic oscillator/clocks or a
WWV/WWVH/GPS receiver.
Click to expand...
An ovenized oven works even better - place the ENTIRE oven-stabilised XTAL
oscillator inside another temperature-controlled environment. This runs the
crystal oven controller at a very constant temperature, greatly improving
the temperature stability of the XTAL oscillator. Do some web browsing
(double oven crystal oscillator) and see what you can find - 5E-11 over
0-50C range. A TCXO is more like 2.5ppm, OCXO about 1E-8. Voila.
 
You must log in or register to reply here.

Welcome to EDABoard.com

Sponsor

Back
Top