PXI-6608: Typical accuracy & stability for a 10 Hertz pulse-train output ?

[scooby]

I need to generate a very accurate and stable
10 Hertz continuous pulse-train for a test system.
What typical frequency accuracy & stability
can expect to achieve from the PXI-6608 for
a generated 10 Hertz continuous pulse-train.
I have read the specs ... just want to make
certain I am interpreting them correctly.

Thanks!

[Bill B]

Hello,

Thank you for contacting National Instruments.

The base clock accuracy of your 10MHz Oven-Controlled Crystal Oscillator (OCXO) on your PXI-6608 is 75ppb. Since the frequency of this signal is 10MHz, the accuracy would then be (75/1,000,000,000)*10MHz = +/-0.75Hz. This base frequency is divided by 1M to produce your 10Hz output signal. Therefore, (10,000,000.75)/(1,000,000) = 10.00000075Hz or (9,999,999.25)/1,000,000 = 9.99999925Hz.

Regards,
Bill B
Applications Engineer
National Instruments

[scooby]

That is what I hope you would say.

So then in terms of frequency stability ...
the 6608 is spec'd out at <= 0.00045 ppm/day or
(0.00045/1000000)*10MHz = 0.0045 Hz/day in 10MHz and
(0.00045/1000000)*10Hz = 0.0045 uHz/day in 10Hz.

So I can expect the drift in my 10 Hz pulse train
to be no greater than 0.0045 microHertz per day
for a properly calibrated board?

Thanks very much !

[acanivell]

Hello there!

Though this was long time ago already I need confirmation on the same kind of calculations.

(However, in my case, I am measuring frequency, not reading, also with the 6608)

I thought the stability of 6608 OCXO was 75ppb per year (or 45ppb as stated in the OCXO notes). So therefore 11.25ppb/day. Not 0.45ppb/day.

In any case, if the stability of the OCXO is 75ppb :

- Stability on the 10Mhz clock: 10,000,000*75/1,000,000,000 = 0.75Hz

The question is, if you measure a 10Hz signal with this device, will you have a 0.75Hz accuracy? Or would you have the accuracy of the clock that you use to MEASURE (using a 100Khz timebase):

- Stability on time base: 100,000*75/100,000,0000 = 0.0075Hz

So apart from my measurement error (according to http://zone.ni.com/devzone/cda/tut/p/id/3619) I have to add up this error like this:

TOTAL ERROR = MEASUREMENT ERROR + CLOCK ERROR = X + 0.0075hZ

Being XX the calculated error made by the frequency estimation algorithm.

[acanivell]

Maybe some of the conclusions about derived clocks here might be of your interest:

http://forums.ni.com/ni/board/message?board.id=40&message.id=7564&jump=true#M7564

Regards