5665 RFSA record timestamp

[jrybka]

Hello,

  I have one 5665 RF Signal Analyzer and one 6682H timing module in PXI chassis. 6682H maintains its timebase from GPS. I am doing continuous IQ acquisition at 25M IQ rate, but I need accurate timestamps from 6682H. wfmInfo from 5665 is not in absolute fashion, and is derived from (non-disciplined) OCXO onboard RFSA oscillator block.

Is there any possibility to get accurate timestamp for e.g. each 512MB block of acquired IQ data ?

Is it possible to use multiple record fetch ? Will the whole acquisiton be continuous (this is the fact I didn't find anywhere in documentation - if multiple records are sampled without any delay between records, and that there are not missing even a single sample) ? Can each individual record trigger a timestamp on 6682H ?

Thank you for any reply.

Jakub

[adamcseh]

Dear Jakub,

one way of obtaining accurate timestamps of the RFSA acquisition is to synchronize the start of the acquisition (trigger) with the 6682. The device is able to generate a trigger signal with known timestamp. You can use this signal to trigger the IQ sampling of the analyzer. An example snippet is attached.

The actual delay between generation and triggering will be affected by the propagation delay on the PXI trigger lines which is in the order of 10 nanoseconds. If your requirements are tighter than this, you can use the PXI Start trigger lines with matched and precisely known lengths. (This requires the 6682 instead of the 6682H.)

I hope this helps you.

Best Regards,

[jrybka]

Dear Adam,

  thank you for your reply. I know I can start RFSA acquisition based on 6682H timestamp trigger. What I need is exact opposite. Read timestamp from 6682H based on *some* signal from RFSA, which tells "now I am starting next block of sampled data". I am sampling IQ continuously, not in blocks. And I need to mark every n-th block with timestamp.

I think there is a possible solution: Synchronize 6682H and RFSA oscillator block (and thus 5622 digitizer) from external GPS disciplined OCXO, and start RFSA continuous acquisiton based on timestamp. That way, any absolute timestamp in sampled signal would be relative to trigger timestamp, and due to disciplined 5622 OSP timebase the calculated time *should* be accurate for some time. I can re-start acquisiton, at best, once a day.

Is there somewhere any better explanation on synchronizing 6682H and 5665 than in http://www.ni.com/white-paper/9734/en ? Do you think that this solution could work with good precision of timestamp ?

Thank you.

Best Regards,

Jakub Rybka

[adamcseh]

Dear Jakub,

thanks for the additional details, it helped to get a better picture of what you are trying to do.

I think your proposed solution can work fine. If you synchronize all possible clocks to the GPS disciplined one and start the continuous acquisition with a known timestamp, than the samples will follow each other according to the sample clock, which has a known frequency and is synchronized to your absolute reference. This means that the timestamp of every sample could be easily calculated.

Good luck with it!

Best Regards,