Transmission delay after pfi0 trigger

[irad]

I am using a PXIe-5672 rfsg to generate a few waveforms in the script mode. The generation needs to be triggered by pulses coming through the PFI0 input. I observed a fairly large delay in signal generation, somewhere between 2 and 3 miliseconds, after the first trigger pulse. I am trying to find a way to reduce that delay to a few tens of microseconds.

The attached example illustrates the problem. There are three simple waveforms. Each one is generated at a different frequency and a different power level. The bandwidth is set to 80% of the IQ rate, which is a lot more than these simple waveforms need, but a more complex arbitrary waveform could need all the bandwidth available. Each waveform is generated when a pulse on PFI0 zero arrives and this is repeated five times. The pulse frequency is quite low, there is only one pulse per second. The pulse duration is about 10 microseconds. For each of the waveforms, there is a large delay in signal generation after the first pulse. Looking at an oscilloscope, I could see that the signal starts 2 to 3 miliseconds after the first pulse arrives. This is not the case for the other pulses though. The delay after the other four pulses is only around 100 micro seconds, which is good enough, although I would expect it to be even lower. 

I am not sure what is going on, but my best guess is that the generator is still tuning to a new frequency after the first pulse arrives, which is causing a delay in miliseconds. How can I get around this? 

[Tunde_S]

Hi irad,

The PXIe-5672 Vector Signal Generator is the combination of the NI 5442 waveform generator and the NI 5610 upconverter. The device that uses the trigger is the NI 5442. As a result, we’ll need to reference page 20 of the NI 5442 Specifications document for information about the waveform delay. As shown on page 20 of this specification document, we can reduce the waveform delay by increasing the sample clock rate (or IQ rate).

The KnowledgeBase article, found here, provides additional information about waveform delay.

Regards,

[irad]

Hi Tunde

Thanks for the link. From what I could see, the specification document provides a fairly good estimate of the actual delay. At 1 MS/s sampling rate, I saw a delay of about 34 microseconds. What I still don't  undertand is why the delay drops quite a lot after the first trigger. For example, with the following script

repeat 5
  wait until scriptTrigger0
  generate waveform
end repeat

the delay after the first trigger occurence is about 34 microseconds at 1 MS/s but it drops subsequently down to 3 or 4 microseconds. 

[dbell0971]

Hi Irad,

I tested this out today; I didn't have a full 5672, but I was able to test with a 5442 (mixer adds negligible delay).  For me, the delay was consistently at ~34 microseconds whether it was the first waveform or the not.  In your case, if I understand correctly, it is 3 microseconds from the trigger to the start of the upper sideband part of the waveform on the 2nd occurrence of scriptTrigger0.  Is that a correct understanding?  Is it possible to post an image/screencapture of your scope measurement?

My Measurement:

Also, if you're looking to decrease the delay further, you might try taking the IQ rate to an even higher rate.

Thanks,

David B

National Instruments

Applications Engineer

[irad]

Hi David, 

Yes, that is the correct understanding. The delay gets much shorter on the second trigger occurrence and all the other occurrences afterwards. Unfortunately, it's not easy to get an image out of the scope I am working with. I'll try to get another scope later and see if I can capture a trace in a file. 

Perhaps, there is something in the VI I attached in the first message that causes this behaviour. Anyway, it doesn't bother me because the delay gets shorter, not longer. I'll work with higher sampling rates to reduce the first delay. 

[dbell0971]

Hi irad,

Your VI looks pretty good, I'm not sure what could be causing this delay.  I'm glad to hear that the shorter delay doesn't bother you; if you decide you want to further investigate this, I'd recommend creating a service request at ni.com/support, we'd be happy to dig in with you more intensively.

Thanks and good luck,

David B.

National Instruments

Applications Engineer