pxi4461 latency from starttrigger signal to actual output

[graduate student 44]

i use a pxi4461 to output an analog signal. i've connected this analog output from the pxi4461 to one of the input channels on a pxi4472 card. on the same chassis, i'm having a pxi7831R card timestamp the 'Dev0/AO/StartTrigger' signal on the pxi4461 card. i'm also timestamping the pxi4472's 'AI Convert' signal using the same clock.

whenever an analog output is spit out from the pxi4461, i get two timestamps, 1) the time when 'StartTrigger' signal on pxi4461 produces an impulse, and 2) the time of an impulse from the pxi4472's 'AI Convert' signal whose corresponding sample point records the start of the analog output fed to it. somehow, i always see a latency of about 1msec between these two timestamps. i'm
anxious to find out the source of this discrepancy and would appreciate it if anyone has any specs on the following:

1) for the pxi4461, the latency between when 'startTrigger' produces an impulse and when the analog output is actually outputed?
2) for the pxi4472, the latency between when a signal reaches the input channel and when it is A/D sampled (so that an 'AI Convert' signal produces a pulse)?

thanks!

[JedRetherford]

graduate student,

The behavior that you describe with your PXI-4461 and PXI-4472 is probably not due to latency, instead it is related to the onboard filtering that is part of the DSA boards. This behavior is described in the manuals for the devices:

PXI-4461 (p.2-8 - p.2-21)
http://www.ni.com/pdf/manuals/373088a.pdf

PXI-4472 (p.3-4 - p.3-10)
http://www.ni.com/pdf/manuals/322940c.pdf

While you will not be able to remove the delays that these filters cause, the delays are deterministic and you should be able to easily compensate for them.

Best of luck with your application.

Regards,
Jed R.
Applications Engineer
National Instruments

[graduate student 44]

thanks Jed! that makes sense.

i find the documentation rather confusing. this is probably mainly due to my lack of understanding of sampling and filtering in general. i'm sampling at 32k/s. the delay on pxi4472 shouldn't change for different input, right? is this delay dependent on the filter coefficient? if yes, where do i find it and what will be the easiest way to calculate the theorectical delay on the 4472 filters? thansk!

[graduate student 44]

hi Jed, i read through more of the specs for 4472. it mentions a "Delay through ADC anti-aliasing filter=38.8 sample periods", so does that mean if i'm sampling at 32k/s, the delay (time between when a sample point is spit out by the A/D converter and when it corresponding input reaches the 4472 channel) is 38.8/32000sec=1.2125msec?

[Anonymous]

hi Jed, i read through more of the specs for 4472. it mentions a
"Delay through ADC anti-aliasing filter=38.8 sample periods", so does
that mean if i'm sampling at 32k/s, the delay (time between when a
sample point is spit out by the A/D converter and when it
corresponding input reaches the 4472 channel) is
38.8/32000sec=1.2125msec?

[JedRetherford]

graduate student,

Yes, your math looks correct. I would expect that this is the property of the card that is causing the problems that your are experiencing. Once you compensate for the delay, you should have no problems using your hardware.

Regards,
Jed R.
Applications Engineer
National Instruments