Timestamp PXI-5152 waveforms in GPS time

[galp]

Hello,

I am trying to collect waveforms with a PXI-5152 and correlate them with other data which is collected in GPS time.  I have a GPS receiver that can output a PPS pulse.  The PPS pulse is the only signal that the two systems can share.

NI support recommended the following.

Split my PPS signal and send one copy into a spare channel on the 5152. Prior to my data acquisition, I would collect one waveform of the PPS pulse.  The timestamp of this waveform would be used as my offset from the time counter on the 5152 to the GPS time (assuming I could figure out what second it was which I can)

The other PPS pulse would go to a 6682 time sync card.  I was told the card could sync to the PPS and output a 10MHz clock to the 5152.  

All measurements on the 5152 would then be converted to GPS time using the offset calculated earlier.

The problem I later found is that the 10MHz clock that is output by the  6682 is not disciplined by the PPS pulse.  It is freerunning.  Therefore, over time, there would be clock drift between the clock on the 5152 and my other data that is collected in GPS time.  This is a problem as I must have 2.5usec accuracy on the time of the data collected on the 5152 card in order to be able to correlate it with the other data set.

Now the question is can this design be salvaged.?  Is there a way to use the 6682?  It would seem that the 6682 should be able to compute number out 10MHz clock cycles between each PPS pulse. For example if it output 10,000,010 clock cycles between PPS, then there would be a 1 micro second drift for that second.   If I were able to read this error once per second then I could (in software) compute and update the error each second.  This would allow me to esentially eliminate the clock drift over the long term.

 If this is not possible then is there another card that can do something similar...I have been looking at the 6608 which may do something like this.

I appreciate any advice that is out there.  Thanks!

Paul

[Adam_U]

Hi Paul,

You're right, the 6682 out of the box cannot discipline CLK10 to one of the supported Time References. 

I think you have 2 high-level options. By leveraging a Trimble Thunderbolt's disciplined CLK10, you can create a system that samples synchronous to GPS. See this link for more information: http://zone.ni.com/devzone/cda/tut/p/id/6818 

If you instead want to correlate your free-running 5152 samples to GPS, you're dead on with the approach although this would likely need to be supported with new features from NI-Sync. We currently support timestamping a signal directly which for CLK10 would amount to far too much data. I'd e-mail nisync.beta@ni.com to discuss the possibility of this further.

Thanks for the detailed e-mail

-Adam