Crystal oscillator frequency stability?

[[email protected]]

Hi guys,

I'm using a cheap, can-type clock oscillator in a project as a
frequency standard. It's 10 MHz. Here's the specs:

http://www.abracon.com/Oscillators/ach.pdf

What I don't understand is that in the literature they say that the
frequency stability is only +/-100 ppm. If I'm figuring right, that
means that the clock can vary per second by as much as +/-1000 clock
ticks. But when I connect the output to a frequency counter it reads
10,000,060 Hz and only fluctuates maybe one or two ticks over time,
which is orders of magnitude better than what they claim.

Did I just get lucky, or am I misunderstanding the specs? Is this an
average error over a batch of many parts? Are they taking the entire
temperature range into account? What gives?

Thanks.

Sid

 

[colin]

Hi guys,

I'm using a cheap, can-type clock oscillator in a project as a
frequency standard. It's 10 MHz. Here's the specs:

http://www.abracon.com/Oscillators/ach.pdf

What I don't understand is that in the literature they say that the
frequency stability is only +/-100 ppm. If I'm figuring right, that
means that the clock can vary per second by as much as +/-1000 clock
ticks. But when I connect the output to a frequency counter it reads
10,000,060 Hz and only fluctuates maybe one or two ticks over time,
which is orders of magnitude better than what they claim.

Did I just get lucky, or am I misunderstanding the specs? Is this an
average error over a batch of many parts? Are they taking the entire
temperature range into account? What gives?

youl find it varies with many things :-
temperature at a certain rate,
supply voltage at a certain rate
and ageing over long time at a certain maximum rate,
plus the initial accuracy,
all this in total will probably be less than +-100ppm.
(and also acceleration too ! but this probably isnt included)
so dont expect to see any such wild variations in the space of a second or
two.

Colin =^.^=

 

[Phil Allison]

** Groper alert !

I'm using a cheap, can-type clock oscillator in a project as a
frequency standard. It's 10 MHz. Here's the specs:

http://www.abracon.com/Oscillators/ach.pdf

What I don't understand is that in the literature they say that the
frequency stability is only +/-100 ppm. If I'm figuring right, that
means that the clock can vary per second by as much as +/-1000 clock
ticks. But when I connect the output to a frequency counter it reads
10,000,060 Hz and only fluctuates maybe one or two ticks over time,
which is orders of magnitude better than what they claim.

Did I just get lucky, or am I misunderstanding the specs? Is this an
average error over a batch of many parts? Are they taking the entire
temperature range into account? What gives?

** That last option is the one.

Crystal stability is normally quoted over the specified operating temp
range.

So, +/- 100 ppm over the range from 0 to 70 C equates to roughly +/- 3
ppm per degree C. This is an average figure and it may well be that it is
much better at a particular temperature - like around 20C.

Don't forget the initial accuracy figure too and the fact you can usually
tweak a crystal onto its marked frequency with a little capacitive loading.

It follows, that if you can keep the temp of the crystal stable, its
frequency is very stable.



........ Phil

 

[John Larkin]

Hi guys,

I'm using a cheap, can-type clock oscillator in a project as a
frequency standard. It's 10 MHz. Here's the specs:

http://www.abracon.com/Oscillators/ach.pdf

What I don't understand is that in the literature they say that the
frequency stability is only +/-100 ppm. If I'm figuring right, that
means that the clock can vary per second by as much as +/-1000 clock
ticks. But when I connect the output to a frequency counter it reads
10,000,060 Hz and only fluctuates maybe one or two ticks over time,
which is orders of magnitude better than what they claim.

Did I just get lucky, or am I misunderstanding the specs? Is this an
average error over a batch of many parts? Are they taking the entire
temperature range into account? What gives?

Thanks.

Sid

That's typical for a cheap can oscillator; they're usually within 10
ppm of dead-on. But they only guarantee +-100 PPM.

But from that initial value, expect a few PPM drift per year and a
good fraction of a PPM per degree C (much less if you're lucky) and
some more power supply voltage influence. But short-term, stability
should be better than 0.1 PPM, even for a cheap rock.

John

 

[mpm]

On Feb 18, 8:33?pm, [email protected] wrote:
But when I connect the output to a frequency counter it reads

10,000,060 Hz...

Do not forget, your counter also has a timebase.
You must consider the drift and age of that timebase (probably crystal
derived?) on your quoted measurement. It could be that the crystal is
dead-on, and the instrument is 60 Hz "off".

In reality, it is highly likely that they are both "off", at least a
little...

Convert these numbers to percentages and you will discover that for
most circuits, the accuracy is plenty sufficient.

-mpm