Synchronization of five PCI-6602

[ayumisano]

Dear all,

I would like to know whether it is possible to synchronize five PCI-6602 and how. (I was told that it's impossible to synchroize more than 4 cards, by an NI staff) My aim is to synchronize a NI 6509 and five PCI-6602.

Moreover, I would like to know the delay time of taking data and finishing taking data of a software type digital I/O card.

Thank you very much for your attention!

Ayumi

[peteski]

Dear all,

I would like to know whether it is possible to synchronize five PCI-6602 and how. (I was told that it's impossible to synchroize more than 4 cards, by an NI staff) My aim is to synchronize a NI 6509 and five PCI-6602.

Moreover, I would like to know the delay time of taking data and finishing taking data of a software type digital I/O card.

Thank you very much for your attention!

Ayumi

Ayumi,

Without even an order-of-magnitude estimate of the tolerance to which you need to synchronize, it is impossible to provide much advice. It's a completely different matter if you have to synchronize things to a tolerance of say 12.5 ns (which is the smallest time period potentially resolvable by the counter/timers on the PCI-6602) vs. synchronizing things to a tolerance of 1 second. And all the orders of magnitude in-between imply different trade-offs.

It is also important to provide information as to whether you need counter/timers or DIOs synchronized, and if you are using DIOs, how many are inputs, how many are outputs, and if there are outputs which need to react to the inputs.

-Pete Liiva

[ayumisano]

Ayumi,

Without even an order-of-magnitude estimate of the tolerance to which you need to synchronize, it is impossible to provide much advice. It's a completely different matter if you have to synchronize things to a tolerance of say 12.5 ns (which is the smallest time period potentially resolvable by the counter/timers on the PCI-6602) vs. synchronizing things to a tolerance of 1 second. And all the orders of magnitude in-between imply different trade-offs.

It is also important to provide information as to whether you need counter/timers or DIOs synchronized, and if you are using DIOs, how many are inputs, how many are outputs, and if there are outputs which need to react to the inputs.

-Pete Liiva

Dear Pete Liiva,

Thank you for your reply!

I am using the digital input of PCI-6602 and NI 6509 only, there are at least 20 inputs for each card. (five PCI-6602 and one NI 6509) Actually the average event rate is not that high, it's only 1 event per second, but of course the faster the better. If the event rate is one event per second, how long would the delay be acceptable?

Thank you in advance~

Ayumi

[peteski]

Dear Pete Liiva,

Thank you for your reply!

I am using the digital input of PCI-6602 and NI 6509 only, there are at least 20 inputs for each card. (five PCI-6602 and one NI 6509) Actually the average event rate is not that high, it's only 1 event per second, but of course the faster the better. If the event rate is one event per second, how long would the delay be acceptable?

Thank you in advance~

Ayumi

Ayumi,

At a rate of one event a second, synchronicity should be quite easy to do through software. If you start getting to a rate of one event per millisecond or faster, much greater care may be needed. Next question is do you need to trigger off one of the digital inputs, or do you want to asynchronously sample the digital lines with respect to their events?

-Pete Liiva

[laksa]

You can synchronize 5 PCI-6602 by using a special cable which NI sells called RTSI cable. If you look at the PCI-6602 card, there is a little connector at the top. This is called the RTSI connector. I am not sure which NI guy you talked to and if he says it's not possible he must be an inexperience NI engineer. You can check the RTSI cables from this website:

http://sine.ni.com/nips/cds/view/p/lang/en/nid/13605

This cable allows you to share the clock between a number of PCI-6602 cards. So, all of them can be synchronized with each other. This is true synchronization as the RTSI allows you route the clock signals from one card to be shared among the rest of the cards.

-laksa

[peteski]

laksa,

Yes, I agree with you that the RTSI cable will allow for synchronization of signals between the PCI-6602 cards, but it won't help with the PCI-6509, as it does not have a RTSI cable interface (or at least it appears not to on the NI Website <- click for link).

If synchronization needs only to be within an order of magnitude or two of 1 second, the RTSI cable is not really necessary, all the synchronization can be done through software and careful application of either "Get Date/Time In Seconds" or "Tick Count (ms)" calls. HOWEVER, if millisecond or better synchronization is required, it will be critical to get these cards running on the same clock, whether it be an internally or an externally supplied clock pulse.

Also, if greater than 1 millisecond synchronization is needed on a large multitude of Digital Inputs, the PCI-6602 may not be the best choice. It seems to me that the card is designed as a multi counter/timer card first and as a DIO card second. And though the counter/timers could be used as extremely well synchronized Digital Inputs, in this mode you would only get 8 channels per card.

Mind you, sometimes the best tool for the job is the tool you have, as opposed to the tool you can't have, depending on your budget, time, and management resources

-Pete Liiva

[Mike Ashe]

Yes, I agree with you that the RTSI cable will allow for synchronization of signals between the PCI-6602 cards, but it won't help with the PCI-6509, as it does not have a RTSI cable interface

Well, you could slightly modify a RTSI cable to give you a wire out from the RTSI trigger line you are using, then route that into one of the inputs of the 6509 card, which has input change notificaton. Probably good enough for sync to the order of a few milliseconds, maybe better. Certainly fast enough for what you're trying to do.