NI PCI-7813R GPS 1pps/10Mhz synchronization

[Kirco]

Hi,

We are considering to buy for our satellite laser system one PCI-7813R board. The crucial in whether to buy or not this board is GPS synchronization since there are other instruments working and all have to be synchronized to GPS PPS and 10MHz signal. Is it possible to FPGA programm or there is some other possibility to use the 1PPS (or IRIG-B is also an option) and a 10MHZ signal for time synchronization of the board counter? It is pretty important to us and thanks in advance to your replies.

Kirco 

[JoshuaP]

The best GPS/IRIG option that NI has to offer is the PXI-6682.  It allows you to synchronize an entire PXI chassis to GPS/IRIG/1588 and offers the ability generate triggers and clocks at specified times that stay locked to the sync reference or timestamp events that are occuring on trigger lines or the external PFI lines.  It's one down side is that it doesn't allow you to lock the 10 Mhz clock to GPS.  In order to accomplish that I would suggest using the PXI-6682 in combination with the Trimble Thunderbold.  I've seen other customers applications that used this combination work really well.  The Thurderbold provides the synchronized 10 MHz clock to the 6682 on it's clock in terminal and the 6682 does all the rest.  If you don't need the 10 MHz clock disciplined to GPS you can use the 6682 by it's self, which can generate clocks from DC to 1.5 MHz.

You can then add other PXI modules (like the R-Series you mentioned) to the chassis and the 6682 can provide the synchronized signals to them or timestamps events from them.

Trimble Thunderbolt

-Josh

[Kirco]

Thank you for the reply, actually the board you suggest is kinda GPS receiver itself.We already have one Motorola 12+ timing receiver and all other instruments are synchronized to this. I was just wondering if somehow one could reset the counter of the device mentioned in my post every second according to the PPS sygnal etc.... maybe there is some fpga routine who can do that... any suggestions?

[Veikko]

Hi Kirco,

Have you checked this:

How Do I Synchronize the FPGA Clock in my R-Series Intelligent I/O Board to my PXI clock?
http://digital.ni.com/public.nsf/allkb/53306F0FDA486B298625738A007B130F?OpenDocument

I think it possible synchronize with this to 1PPS and FPGA. I think you should contact to NI on this one to get the exact information.

Veikko

[Kirco]

Hi Veikko,

thank you for your thread. Actually, it is possible as you pointed out in your link. However, it assumes that you already have one time system in the PXI chasis as I pointed out. It however asumes that you already have this PXI clock, what we do not have. Doing that implies also buying the pxi chasis etc, the clock, installing everything etc. Also, synchronizing this fpga to the PXI clock may introduce some inconsistencies to other instruments synchronized to our GPS receiver and 10MHz hydrogen maser signal. Well, the problematics is actually simple. We simply want to put this FPGA card into one PCI slot of our industry  PC. We furthermore provide 1 PPS from our GPS precision timig receiver and 10Mhz signal from the hydrogen maser for synchronization of this card. If there is some FPGA routine allowing this synchronization would be great. Our instruments are all synchronized to these signals and that would be the best possoble solution without big changes. So, the synchronization is performed with these 2 signals. On our Industry comp are also other cards synchronized to these pps and 10Mhz signal. Of course your link would work, but in that case the abovementioned modifications are necessary and possible bias to other instruments synchronized to abovementioned sources. So, anyone knows of a way to synchronize this FPGA card to external 1PPS and 10Mhz signals? I am pretty sure it could be possible just the matter is how. I am not really LAbview FPGA expert but my feeling is telling me it could be possible. Our old ISA card was synchronized to these signals only once before the start of the session very simple. When 1 PPS arrives, the internal card counter was reset to start from 0 and the number of day seconds was red from our GPS receiver via rs232 to keep the absolute time. The software we use is DOS/ industrial fortran (internal memory addressing possible)  for real time computing. Not really modern, but effective and fast so we keep that so. Thanks again for your post and hope someone will come with a solution. 

[wiebe@CARYA]

Kirco,

I'm doing something very similar. Perhaps our solution might be usefull (the priniple works).

Let me explain my situation. I have to send messages based on a timestamp. This timestamp is absolute time, so simply synchronizing the clock is not sufficient. To accomplisch the task, I created a single cycle timed loop that counts FPGA ticks. This counter value is put in an indicator, and is read in parallel loops with a local variable (a global would do too).

This counter can be reset to the actual time when it gets a 1 PPS pulse. The mechanism I use is fairly simple: The host PC reads the NMEA string, and passes the time to the FPGA as a U64. The U64 represents ticks since 1-1-1904. If the FPGA receives a 1 PPS pulse, it starts a counter. If the time is changed and the timer is smaller then .5 sec. it updates the time to the NMEA time + the counter value.

This mechanism provides me a globally readably timer that is synchronised every 1 PPS pulse it receives. And it's cheap and simple.

But this is where I'm having troubles... The 1 pps pulse synchronises the timer every second (at best, but it's availability is a requirement). The FPGA clock has a deviation of about 800 ticks within this second! This comes down to 20 us, and we have to be 1u accurate! So we have to use an external clock, but this clock doesn't have to be synchronised with the GPS, it just needs to be accurate (< 1 PPM, but .1 PPM would be better). We found how to tell the FPGA to synchronise with the PXI clock, but the deviation is exactly the same. We also used an external clock provided to the PXI backframe, but the results are still exactly the same! It doesn't seem the be working at all.

Using LV8.5, and a PXI-7813R. I'll post this question in a seperate thread, but if someone post an answer here, I wouldn't mind.

Regards,

Wiebe.

[Kirco]

Hi,

for the time being we dropped the idea of using 7813. It simply is not as accurate as we need the time accuracy (our application demands for 5ns absolute time accuracy). Anyway, the absolute time synchronization we had in mind was to use 10Mhz signal comming from the GPS receiver as counter base and use 1 PPS only once for initial time synchronization. Anyway, the problem has been now solved with other FPGA which supports more accurate clock. Hope it helps

Kirco

[wiebe@CARYA]

The 7813R should be able to run on 5ns clock (200 MHz), according to spec? Not much margin for error though.

Regards,

Wiebe.

[alex.]

Hi Kirco,

We are looking for a solution for synchronize time base and start DAQ in 2 different site with error<10ns.

May you give any suggestion on it?

Thanks,

Turbot

turbot_yu@hotmail.com

[wiebe@CARYA]

Hi,

What do you mean by 2 different sites? Are those sites far apart? If not, it will be really hard, and honestly, I have no idea if it can be done. I think a 1 PPS pulse from a gnss receiver has more error then 10 ns...

Regards,

Wiebe.