I try to assemble PXI DSA system with precise time synchronization (better than 1 microsecond). In according to reference design (see http://zone.ni.com/devzone/cda/tut/p/id/6818#toc0) I planned to use:
NI PXI-PCI8366 MXI-Express Kit with Fiber Optic Cable
NI PXI-4462, 24 Bit, 204.8 kS/s
NI PXI-6682 GPS, IRIG-B, IEEE 1588 Sync and Time Module with TCXO
In this design the Trimble Thunderbolt E GPS receiver is used for disciplined 10MHz clock.
Is it possible to use the NI PXI-6653 instead Trimble GPS? Can I discipline PXI-6653 10 MHz oscillator to GPS signal from PXI-6682 board?
Or , may be, I must use NI PXIe-6674T and NI PXI-6682H?
Thanks for the question. The document you referenced says, "The Trimble Thunderbolt E (www.trimble.com) is a GPS receiver that provides a 10MHz GPS disciplined clock. This is the receiver used to validate this architecture." This means that the Trimble is basically used for benchmarking to show the accuracy of the 6682 module. For your application, you would not need a Trimble GPS at all, and in that diagram, you would only need to plug in an antenna to the PFI line in the 6682.
Basically, what we're trying to show here is that the 6682 can be used in synchronization tasks across multiple chassis spanning large distances by using GPS timing. Clock drift can occur between absolute GPS time and the timing card in your system, but everything in your system will still be synchronized to each other. This drift is also corrected by the 6682 through that antenna to absolute GPS timing.
Please let me know if you have further questions!
Thank you for answer!
As I understand, if I use 6682 module, I can start measurement with submicrosecond accuracy. But 10 MHz reference clock has 10 Hz accuracy, i.e. if my record time is the 10000 seconds I can obtain 0.01 second time discrepancy from the ideal (or GPS) time at the end of record. For compensation this error I must use more stable 10 MHz reference clock. Fore example, I can discipline this clock by the GPS pps signal using PLL circuit.
Is my conclusion true?
I'm not sure if this will help answer your question, but the 6682 is able to create a trigger with around 100 nanoseconds of jitter. However, its 10 MHz clock that it provides is free running and not syncronized to GPS. The TCXO that is used on the 6682 has 1ppm of accuracy, meaning that it may be off by as much as 10 Hz. Note that this doesn't effect the 6682 ability to generate a trigger within 100 nanoseconds of GPS or its ability to timestamp events to +/- 100 nanoseconds if that is what you desire.
In the application that you reference the usage of the Trimble Thunderbolt was to also align the 10 MHz clock to GPS.
I just want to clarify a couple of points:
+ the PXI-6682 cannot discipline its TCXO to GPS (or any other sync source). The TCXO is always free-running.
+ If you want to discipline PXI_Clk10, you have 2 options:
(1) use a PXI-6682 in Slot 2 and configure it to drive PXI_Clk10_In with its ClkIn input. Connect a 10MHz disciplined-clock source to the PXI-6682's ClkIn input. That source can be Trimble's Thunderbolt, or any other device that can provide a disciplined clock. This is what is described in the DevZone article you referenced (http://zone.ni.com/devzone/cda/tut/p/id/6818#toc0)
(2) The other option is to use a PXI-6653 for PXI or PXIe-6674 for PXIe installed in the chassis' timing slot, and the PXI-6682 installed in any other PXI slot. The use our Clk10 disciplining software to make the board in the timing slot discipline its OCXO to the PXI-6682's GPS signals. That is described here: http://www.ni.com/pdf/manuals/373176b.pdf
So, to answer your question, yes, it is definitely possible to use a PXI-6653 instead of the Trimble Thunderbolt to discipline Clk10 to GPS.
Hope this helps.
I try to order this system and have some discussion with technicians about time accuracy of this one. We have discrepancy conserning time accuracy for long time record.
The model task is:
The system is: 1036 + 4462 + 6682 + 6653. 10 MHz clock of the 6653 board disciplined by the GPS signal from the 6682.
We start quasi-continuous record using 6682 triger (for example pps signal from GPS). Sample rate is 204.8 kS/s.
The question is:
1. What is the time accuracy of the START TIME?
2. Let's suppose, that record will have 204.8*10^3*10^5 samples (i.e. record length is about 27 h). What time accuracy of the last sample will be?
3. What time accuracy will be in the middle of the record?
With GPS you will find it very hard to find absolute specs, since there are many factors that can affect the performance and accuracy. But we do have empirical resutls of tests we have done using the clock displining setup you are using.
Empirically, we have seen that systems using the clock disciplining mechanism stay within 100ns of GPS time at all times, as long as the signals from the GPS satellites are not blocked from the antenna. Therefore, you can typically expect your start time, a sample in the middle of your acquistion and the end of the acquistion to have accuracy of 100ns.
Hope this helps.
Thank you for answer. That's what I thought. But technicians say me, that I can obtain "GPS accuracy" only for the first sample. And if I want the same accuracy for the middle and end of record, I must use Trimble GPS with 10 MHz oscillator...
If you do not discipline the oscillator, then that would be true, you would only get the 100ns accuracy for the start trigger. Which is why they are saying you would need the Trimble 10MHz disciplined oscillator.
So, in other words, you can get the 100ns accuracy throughout the duration of your acquisition by having a disciplined oscillator, and that disciplined oscillator can come from either the external Trimble box, or from the PXI-6653 in combination with the PXI-6682 as described in the document linked a few posts above (http://www.ni.com/pdf/manuals/373176b.pdf).
I am struggling with similar issues that are mentioned above. I would like to get some guidelines from people who work a little on timing and synchronization aspects in LabVIEW. I know right now how to configure the DSAs using DAQmx and how to trigger DSAs using niSync however it is of great unknown how to route the signals between PXI-6682 and PXI-6653 and how to send the timebase from PXI-6653 to DSAs via backplane. I have PXI-4495 as well as 4472 in two different chassis. I also started a separated topic here since this seems to be quite complex issue.
Thanks for any help in advance.