# Resonator Tolerance

Discussion in 'Electronic Design' started by [email protected], Apr 24, 2006.

1. ### Guest

Hi,

I have some questions on Resonator Tolerance, and PPM.

On an engineering specs, it's mentioned that the initial tolerance is
0.50%. How do you convert to PPM?

How do you calculate the timing (microseconds) when the following are
given?
- 50ppm
- 16MHz

Thanks,

2. ### Tim WescottGuest

PPM means "parts per million", i.e. 1PPM means 1/1000000.

% means "per cent", or "parts per hundred" in english. 1% means 1/100.

I will _not_ do the math for you.

--

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

3. ### Guest

i know ppm and %

but i can't figure out why 50ppm = 0.5%

to my calculation, 50 ppm = 0.005%

and 0.5% = 5000ppm

my friend just said, simple divide/multiply by 100, i just don't get it,

4. ### Guest

that's only the first part,

How do you calculate the timing (microseconds) when the following are
given for a given duration?
- 50ppm
- 16MHz

5. ### Guest

A 16MHz waveform repeats every 62.5nsec. If there is a 50ppm tolerance
on the frequency, there will be a 50ppm tolerance on the period, which
is 3.125psec.

This sort of question should be posted on sci.electronics.basics, or -
better - dealt with by your instructor, who is paid to sort out this
sort of elementary problem.

6. ### Phil AllisonGuest

** Shame you got crap replies from others here.

** That equates to 5,000 ppm.

Or +/- 80kHz in a 16 MHz device.

** The " 50ppm " figure describes the *temperature stability* of the
resonator.

Very likely + 50 ppm change *per degree C * !!!

The timing tolerance is dominated by the initial tolerance of +/- 0.5 %.

........ Phil

7. ### Guest

Hi Bill,

Thanks, for the explination, I have tried to figure out by myself, but
it didn't work.
And no, I don't have an instructor.
the tolerance be?

8. ### Guest

Thanks Phil~

So if the 50ppm refers to changes of the tolerance by the degree, then
i think i would look at the initial tolerance,

thanks~

btw, how do i know what is the tolerence in seconds for a given signal?

9. ### Guest

A 16MHz waveform repeats every 62.5nsec. If there is a 50ppm tolerance
on the frequency, there will be a 50ppm tolerance on the period, which
is 3.125psec.

or should i say that, there an timing tolerance of 3.125psec on the
waveforms that i measured?

10. ### Rene TschaggelarGuest

If the timing tolerance is 50ppm, meaning a 20000th, then the time
measurement can be off for 1 second every 20000 seconds.

Rene

11. ### Phil AllisonGuest

** There is no "given signal".

You are MASSIVELY fucking confused.

........ Phil

12. ### Winfield HillGuest

wrote...
If you have a recent vintage digital scope, it may be able to
make quasi-accurate measurements for you, but an analog scope
most likely cannot get you much past 0.5%, you'll be measuring
the scope not the component. Go grab a frequency counter.

13. ### Pooh BearGuest

wrote:

This is done using mathematics ! Multiplication and division - that kind of
thing you know.

Graham

It isn't.

Graham

15. ### Guest

Hi Graham,

Thanks for the clarification. I thought I was missing something in
maths.

16. ### Guest

Hi,

Let me paint the picture again.

I am measuring the "waveform" from GoLogic. The waveform doesn't not
tally with the values being flash into the device that is used to
produced the waveforms. It devaites by a few microsecs.

I was given that the device resonator's initial tolerance is 0.5%.

But 0.5% of what? Of one message frame? Of one OnTime? OffTime? Or
OnOffTime?

So, how do I make use of the 0.5% tolerance to calculate the range of
timing (in terms of +/- XXX microsecs)

17. ### Spehro PefhanyGuest

All of the above, modulo jitter in the oscillator etc. It's all
proportional, so percentages stay the same.
Suppose the time would be YYY microseconds (say the division
ratio isn't exact so perhaps not exactly equal to what it would
ideally be), the range of timing is YYY*0.995 to YYY*1.005,
ignoring any jitter. The range in error is the difference between
the ideal value and the above range of values.

With asynch serial RS-232 type communications the error in the bit
position of the last bit is what you normally need to worry about
(because it's the worst case situation), and it's about 10 times the
error in the timing. So with a 0.5% resonator and perfect division
ratio you'd have about a 5% error in the middle of the last bit time
when the sampling takes place (assuming no jitter, perfect edges, and
no error at all in the other side).

Best regards,
Spehro Pefhany

18. ### Guest

Hi Spehro,

Thanks for the explination.
I'm using the USB version of GoLogic.

btw, you mentioned that the timing would be YYY +/- 0.5%. So if my
OnTime is 100us, so it should range from 95us to 105us. But I always
thought the 0.5% should be for the whole frame or the whole message
transmitted.

19. ### Phil AllisonGuest

** You really are a complete goose.

Even basic arithmetic is beyond you.

......... Phil

20. ### Guest

Hi Spehro,

You mentioned timing, but timing of the what?.....
- onTime
- offTime
- OnOff Time
- frame?.....

btw, my friends speculate that the initail tolerance that is spec in
the datasheet (0.5%) is for one OnOff Time.

and said that the 50 PPM is refer to 50 pulses in 1 million, will out
of out the specs......

I'm not sure what to take...