t0 value of simultaneously samples analog input

[ericson.15]

Hello,

I have an application that requires simultaneous sampling of accelerometer signals.  I use 4472 cards with a PXI-1042 DAQ system.  The examples I found to simultaneously sample are straightforward, so I'd like to assume all is well.  I see some things in my data that make me wonder (issues that may or may not crop up due to slightly non-simultaneous sampling), so I've gone back to look at the t0 time stamp of my waveforms to verify that the sampling has begun simultaneously.  I see an interesting phenomenon in my program that persists in this simple example.

I'm looking at the t0 values for two waveforms out to a lot of decimal places.  When my sampling frequency is a "round" number, like 5000, 10k, 20k, 25k,etc. - I always get a bunch of 9's in my two t0 values, every time, every while loop iteration.  See the pic below.  If I change to the sampling frequency to, say 5001Hz, or some arbitrary integer, all the 9's go away and the value of each decimal place is random.  It is good that the two t0 values are the same.  What I don't understand is all the 9's in the t0 when I use a 'round' number for sampling frequency.

This seems liek a funny phenomenon.  Can anyone explain?  How far out are the decimal places of my t0 time reliable?

[aNIta B]

Hi Tristan,

         I spoke to one of our R&D engineers about the behavior you were seeing.  The time stamp is simply the system time at which the waveform was brought in, so it's only going to be as accurate as your system clock (which I believe is a milisecond clock).  With the 9s behavior you're seeing:  there's not a good explanation to this behavior.  While that's a little odd, it shouldn't affect your results.  Let me know if you have other questions.  Thanks!

[ericson.15]

Thanks for the reply.  When you say that the time stamp is the system time, is that the time kept in my DAQ system, or my PC?

If this clock is only accurate to the milisecond, does that mean that "simultaneous" samples are only guarenteed to be simultaneous out to the milisecond?  Or is there some other mechanism that ensures simultaneous sampling more precisely than the time stamp value that I have access to?  (I hope that made sense.)  Thanks!

[A_Person]

The time stamp is based on the system time of the PC which, as mentioned before, is only accurate to the milliseconds.  This is only applied once the data is retreived from the card.

The timing source for sampling is on the DAQ card and is far more accurate than the PC clock.  You can find specific values in the specifications for your device.