[time-nuts] Simulating Oscillator Noise: Difficulties SimulatingFlicker FM Noise

[Bob Camp]

Hi

The thing that amazed me was that there were at least two of us in the room who *knew* the answer, and neither of them worked for HP ....

Bob


On Apr 23, 2010, at 2:16 AM, Magnus Danielson wrote:

> Hi Bob,
> 
> On 04/23/2010 04:10 AM, Bob Camp wrote:
>> Hi
>> 
>> Back in the late 70's HP was pushing their ADEV test setup to us. They hauled one in (yes indeed
>> more than one box) for a demo. It took them most of the morning to get it in from the parking lot
>> and set up / warmed up / running.
> 
> To much amusement I am sure.
> 
>> They ran us through a little presentation on how ADEV has two slopes (I *wish* I'd kept a copy...).
>> We brought out some sample oscillators, and ran them through. *Surprise* more than two slopes.
>> I then asked them "what do the other slopes mean?". No answer .... we'll get back to you ....
>> mumble mumble .... They never did get back to me.
> 
> Depending on the oscillator you expect to see white, 1/f and 1/f^3 phase modulation noise or white, 1/f^2 and 1/f^3 phase modulation noise. Looking at tabulated Allan variance:
> 
> http://en.wikipedia.org/wiki/Allan_variance#Power-law_noise
> (see also Tom's PDF)
> 
> We see that white and 1/f is weakly different such that they can be expected easilly be mistaken for having the same slope. The 1/f^3 noise has a flat Allan variance response. So, their oscillator during testing may simply have been of one kind and the oscillator you brought out of the other.
> 
> The Allan variance unability to separeate the white and 1/f noise triggered the development of the modified Allan variance, but that only happend after that incident.
> 
>> Needless to say they would have been a lot more likely to make the sale if they had one of your
>> one page charts.
> 
> A copy of Leesons paper (a two-page thing) in 1966 would have helped a lot.
> 
> Cheers,
> Magnus
> 
> _______________________________________________
> time-nuts mailing list -- time-nuts at febo.com
> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> and follow the instructions there.
>