labview/daqmx frequency precision

[AntLee29]

I plan on outputting a low-frequency sawtooth signal using daqmx with my PCI-6259 to a precision of 14 decimal places.  Does Labview and this daq card support this level of precision?  If not, how close can it get?  The exact frequency I'd like to use is 0.00152587890625 Hz.  Thanks.

Message Edited by AntLee29 on 08-13-2007 03:40 PM

[Kevin_Price]

Ok, I checked out your reciprocal and you want a period of exactly 655.36 seconds.  If, for example, you set an output rate of 100 Hz, then you could define your sawtooth to have exactly 2^16 = 65536 values.  I *think* that's also exactly the bit depth of the D/A converter for the analog outputs.

So far, so good... sort of.  You say you want "14 digit precision" on the output frequency -- are you really sure about that?  That implies that you need the output frequency to be stable to within 1 part in 10^14!!!   Well, the onboard crystal oscillator won't be nearly that good.  The spec sheet I looked at rates it at 50 parts per million, and that's only at constant 25 C temperature.  So you can only get something like 5 digit precision.

On the other hand, since the *period* of the sawtooth is supposed to be 655.36 seconds, then perhaps 5 digit precision will be ok after all?

Why exactly do you say you need "14 digit precision"?  Compared to what standard?  For what purpose?  In what way(s) will 5-digit precision be a problem for your app?

-Kevin P.

[AntLee29]

Hi Kevin.  It turns out getting all 14 digits is not required, but it's simply best if I can get as close as possible.  When I said 14 digit precision, that was with respect to the input frequency value for the daqmx functions.  The hardware is not yet set up so I am unable to test it.  You are correct that I want a period of 655.36 seconds, which is why I plan to wire 0.00152587890625 Hz at the input nodes.  A 5-digit precision will be ok, but are you saying the 5-digit precision means the closest frequency I can use is 0.00153 Hz?  Or does it mean the period can get within 655.3600x seconds?  Both will be ok if that's the best I can get, but some clarification would be nice.  Thanks for your input!

[Kevin_Price]

My remark about "5 digit precision" is based on the spec sheet which rates the oscillator at 50 parts per million, or 0.005%.  This would correspond to a period of about 655.36 +/- 0.03.  So it'll be the 5th significant digit in the nominal period that may vary.  Similar for frequency.  It'll likewise be the 5th significant digit that'll be in question.

-Kevin P.