[time-nuts] Need schoolin PPM vs E to the umth?

[Bruce Griffiths]

Chuck

Chuck Harris wrote:
> n3izn at aol.com wrote:
>   
>> So can any one bring it down a few clicks and explain the accuracy thing to this balding hippie?
>>
>> I'm leaning towards ham radio applications. GPSDO are rated in one way like 1 X 10 to something. Ham radios are
>> coming with "High Stability" oscillators that are rated in ? PPM.
>>
>> In laymen's terms what do those numbers mean and how do they relate to each other?
>>     
>
> They convey the same information as an accuracy percentage, only they
> allow for an easier to understand number for higher stability sources.
>
> For example, an oscillator that is accurate to 1%, is accurate to
> one part in 100, or expressed in scientific notation, 1 part per 10E2.
>
>   
Wrong, it should read: 1 part in 1E2.
1E2 is actually shorthand for 1x10^2 .

For example see:
 http://en.wikipedia.org/wiki/Scientific_notation

This is the way the E notation is implemented on HP scientific calculators.

> Most people on this group consider that pretty (well awfully, actually)
> crude.
>
> As you get to more and more stable frequency sources, you might hear
> of 1 part per million, or 1 PPM, or 1 part in 10E6.  In percentages,
> that would be 0.0001%.
>
>   
1 PPM is 1 part in 1E6 .
You are just perpetuating a misinterpretation that was relatively common
in some old OCXO datasheets.
> And as you become a time-nut, you start to talk of parts in 10E12, which
> in percentages would be 0.0000000001%... Which is just too cumbersome
> to use.  You have to count the zeros every time you look at the number,
> and you still aren't sure what it means.
>
> Any specification that tells of the accuracy of a frequency standard
> needs some additional information... notably, a time frame over which
> the accuracy is valid.  So you will hear things like 1PPM/year, or
> 1 part in 10E12 per second.
>
>   
1 part in 10E12 per second is actually 1 part in 10^11 per second.
> Some information about environmental conditions would also be useful.
>
> Also, implicit in an accuracy specification is +/- which creates an
> error band for the error to be contained within.
>
> When one becomes a true time-nut, one starts to worry about trends in
> stability issues, and at that point, statistical analysis becomes
> important, and you start to talk of things like Allen-Variances.
>
> -Chuck Harris
>   
>> Hope this doesn't start to much controversy on the board.
>>
>> 73 Chris
>>
>>
>>
>>     

Bruce