What's a minimum pulse width achievable on counter output of X series cards?

[adat]

Hi, 

For one of our project there is a requirement to generate pulse train.

Core requirement is that we have to generate a Pulse (Torque Pulse) that follows anther pulse (Reference Pulse) & the delay of this torque pulse from reference pulse should be more precise means,

this delay should have an accuracy of ± 2ns (ns-nano seconds) & minimum pulse width requirement is 20ns . 

Our idea is to use counter to generate this pulse train.

I have gone through the possible solutions from NI & to achieve this accuracy thought to use NI PXIe 6356 Which is having 100MHz base clock for counters.

Since 100MHz=10ns will help me to achieve that much precise requirements of delay (multiples of 10ns will better give better accuracy).

Against to this I have some more queries which are mentioned below. 

·         Since I am not using analog output which has output rate of 3.33MS/s,

In this case, will there be any internal "overhead" in a system that will not allow the counter to have a base clock rate of 100MHz?

will it be possible to utilize 100% base clock of 100MHz for counter to generate pulse train?      

I called for support from NI & I got a reply that it is possible. So is it true?

·         What is the possible deviation/Noise in this clock period of 10ns?

 ·         What could be the delay in this clock period if we trigger the counter by using LabVIEW?

 ·         What are the triggering sources & accuracy available internally to trigger the counter?

 ·         How we can achieve better accuracy by using external triggering method? 

Your quick feedback on these queries will help us a lot. 

E-mail me to nilesh.adat@goodrich.com also. 

Thanks & Regards,

Nilesh

[John_P1 · Accepted solution]

Hi Nilesh,

There is no overhead that would prevent you from using the 100 MHz timebase--it's simply an oscillator that is on the 6356.  Analog Output is independent of the counter task.  To answer your other questions:

The base clock accuracy is specced at ±50 ppm (page 8 of the specs).  ±50 ppm of 10 ns is ±0.5 ps (±5e-13 seconds).

The amount of time between when a hardware-trigger is received and when the counter is updated is up to 1 timebase tick (10 ns).  If you are controlling this with software then the exact time that the trigger signal is generated is non-deterministic anyway so the 10 ns would be negligible.

There are quite a large number of signals you can use for triggering (both internal and external).  For the full list I would look at the device routes tab in Measurement and Automation Explorer.

Exactly what do you mean by accuracy?  The oscillator is specced at 50ppm -- if you have a more accurate reference clock you can PLL the oscillator to this for more accurate results.  External triggering doesn't affect the accuracy of the counter's timebase.

I'd like to keep the discussion on the forums so others might use this as a resource should they have similar questions in the future.  If you have a support contract and you want a more direct line of communication, you're more than welcome to open a phone or email service request at ni.com/ask.

If you'd like, you can provide more details about what exactly you need to do so we can give a recommendation on how to go about it.  The questions you are asking right now are pretty specific and I'm not sure how they necessarily relate to the big picture of the application as a whole.

Best Regards,

John

[adat]

Hi John,

Thank you very much for giving this usefull information.

Most of the doubts got cleared.

I will update you for any kind of issues if face it during functionality check.

Regards,

Nilesh

[Dennis_Knutson]

You should mark Johns' answer as the solution, not your thank you. You can change what you did by clicking on 'Options'.