Very excited by what I am reading in this group! I have a specific test requirement that I'm hoping the cDAQs might work with but it isn't so straightforward.
We have a system based on cRIOs (9082 and 9025). This has requirements for time synchronisation as it aggregates data on specific clock intervals (for IEC 61000-4-30). To test this, we want to generate a pattern that is aligned with the clock times.
We already have a test setup using DAQmx and have been unable to properly complete these tests but TSN may change that. I know with the cDAQs we can trigger a start time based on TSN but is there a way we can align the cDAQ clocks to NTP which can act as a common source for both the cDAQ generating the test pattern and the cRIOs under test?
Thanks for the question. I bit of technical info about TSN cDAQs 9185/9 before I go directly to your question. NI TSN cDAQs 9185/9 use 802.1AS protocol to keep their clocks synchronized with each other or other TSN capable devices such as cRIO 9035/9 (sync), cRIO 904x, IC 317x or FieldDAQ. When you connect any of the above devices to each other or over an 802.1AS switch they will automatically synchronize their internal clocks to each other.
When you connect chain of TSN cDAQs to Windows host PC they only stay synchronized to each other but not to Windows. The start time you provide from a non-tsn host (Like Windows PC) gets ‘translated’ to 802.1AS time and cDAQs start the acquisition. There is NO synchronization between cDAQ and Windows host after that.
Even though NTP and 802.1AS are synchronization protocols they are vastly different from each other. The 802.1AS synchronization accuracy is on order of magnitude better (nanoseconds) compare to NTP (in seconds to microseconds) range. I believe that for this reason no one attempted to do this. At least with the current NI offerings that would not be possible, there is no way to align/bridge cDAQ 802.1AS clocks or even other 802.1AS device’s clocks with NTP clock.
I hope this helps, let me know if you have more questions.
A further clarification. With TSN you can do most of what you are asking for:
The only part that will not work is aligning to anther signal coming from NTP.
As mentioned NTP is not precise enough to serve as the master clock source.
If you don't care about "absolute" time (ie all the devices are tightly aligned to each other but not to a global or wall clock time) then you don't need to do anything, this will work with your proposed set-up (assuming you use a cRIO controller that supports TSN - All the 904x controllers support TSN).
If you need the devices to be aligned with other timed elements across the world then the best option is to get a GPS receiver and use that to master the network clock. NI offers this capability in PXI and there are other devices on the market which can also do this. If you decide to go this route let us know because we may also need to change some details of the PTP profile used for time synchronization.
Thanks for the responses. That more or less confirms my thoughts.
Unfortunately the cRIOs are running an application which includes their own time synchronisation over NTP or GPS so we can't use 1588 or 802.11AS on them directly.
I have some quotes for external master clocks so I think that is the best solution then, depends on whether the customer thinks it is worth paying for them or not!
What about syncing a cRIO 9046 to the GPS time from a GPS module (NI-9467)?