cFP clock loose synchronization

[Graziano]

Hi!
   If you set cFP-2020 as a Time server client (i.e. it will be synchronized from a PC on the net) it will receive synch "frequently", not only on boot. I hope you can fortgive about synch problems.

   My problem was sligtly different, because I have no ethernet connection in my cFP (in final application, of course), while I need accurate datalogging, with timestamps.

   The synchronizations that happens on boot is cFP's operating system software clock synchronization, but, as I tested, you can retrieve "correct" time-stamps with FP polymorphic read.

   I hope I understood your question!!! Let me know!!!

   Bye!

graziano
  

[cs42]

Yes, you understood correctly.

So over all time should be more reliable if cFP is connected to a network and synchronizes to a "time server"... (even 1.8 minutes a month could be a problem in some circumstances, e.g. if you want to import your data in a database later).

Actually we didn't have a problem with this yet. All our cFP controllers are synchronised to the same pc which is synchronized to the net and all other PCs (the important ones with the databases e.g.) are getting their time from this one as well.

But hey, I'm curious.

Thanks for the info.

Cheers,
    Carsten

[Matthew_Williams]

Our testing (RT ~6.1.x) showed that using the built-in time server on the
Fieldpoint controller was unreliable.  We would intermittently get hugely
erroneous times, sometimes they would correct, sometimes not.

Based on this we elected to disable the built-in time server sync and
do the time update within our application.  Once we did that and
switched to using only the FP read timestamps as our time
reference everything was happy.

Matt

[Olli]

Hi,
yesterday I learned, that I cannot set a "normal" time server as a Fielpoint time server...it must be a PC running Fieldpoint software (again my RTFM problem ;-).
I need an absolute time reference in my measurement network, which is now the real time clock of a PXI. The PXI clock is synchronised via a network time server.
The fieldpoint is sending a software trigger via ethernet to all devices. The accuracy of the trigger synchronisation is less than one second. This is enough for our appplication, but the fielpoint real-time clock should not drift far away from the PXI's real-time clock. Can I find some specifications? I wonder how the Fieldpoint time service works? Does anybody have a link to this information?
Ciao,
Olli

[Matthew_Williams]

I don't know the details of how the built-in FP timeserver from NI works, but
we have used the attached to query NTP type timeservers from cFP controllers.

Matt

[Olli]

Hi Matt,
it looks useful your time protocol client VI. I just don't have the "RT Set Date and Time.vi". Anyhow, this post is very informativ so far.
Thank you all
Olli

[Matthew_Williams]

On my system that .vi is in RTUtility.llb, under C:\Program Files\National Instruments\LabVIEW 7.1\vi.lib\addons\rt\ETS\
I think that is installed with the FieldPoint addon stuff, rather than as part of LV proper.  It shows up under the
Real-Time Utilities on the block diagram palette.

Matt

[Olli]

Hi Matt,
OK i found this vi, but I cannot find the setdate.dll and settime.dll...it should be in the LV7.0 folder, but when I remember well, we have LV RT only from Version 7.2 on. Or maybe it's removed, because I always uninstall LV completly before installing a new version....at least I should keep the vi.libs. I will check this tomorrow, because someone's calling me home 😉
Have a nice Day or Night
Olli

[Matthew_Williams]

Did you find the .dll files?  The file from one of our 7.1 cFP nodes is
attached.  C:/ni-rt/startup/data/settime.dll in the Fieldpoint.

Matt

Message Edited by Matthew Williams on 01-30-2007 01:49 PM

Message Edited by Matthew Williams on 01-30-2007 01:50 PM

[Olli]

Hi Matt,
thanks for the reply and dlls. I found them in the folder Program Files\National Instruments\RT Images\LabVIEW 8.2.
Did I mention LV7.2 above? There I mixed something together with LV8 and LV7....
but anyhow your VI works now...maybe I will use it, when the other method will fail!
Thanks a lot!
Olli