How to calculate ADC filter delay for USB 4431

[tmcarson]

I am making a continuous non regenerative AOAI measurement.  I need to compensate for both the ADC filter delay and the output filter delay

The output filter delay seems to be well documented in the NI DSA manual (link below)

http://www.ni.com/pdf/manuals/371235g.pdf

The ADC filter delay is harder to find the DSA manual says to look at the 443x spec.

The spec gives a "nominal" ADC filter delay of 38 samples.  I can not find a more specific definition (link below)

http://www.ni.com/pdf/manuals/372485c.pdf

I'm using labview 2010, on XP

attached is a sample vi.  it uses a producer consumer loop example I modified

I could manually find the total offset for different sample rate but i would prefer a way to calculate it.

Thanks

Timm C.

[dsb@NI · Accepted solution]

When you say 'calculate', do you mean automatically measure, compute by formula, or perform table lookup?

It is a good idea to measure the total (DAC + ADC + other) delay of the measurement device at each measurement sample rate used in your application. Best practice is to perform this measurement every time the system is powered up or restarted.

Recommend that you use VI provided by Sound and Vibration Toolkit to measure this delay and additional delays that can be introduced by the device under test:

<LabVIEW>\vi.lib\addons\Sound and Vibration\svt_Calibration\SVL Measure Propagation Delay (DAQmx).vi

If you do not have SVT/SVMS, you can recreate the delay measurement with LabVIEW Full functionality.

Start synchronize your AO and AI using the same implementation used in your measurement application.

Use a test signal such as a Gaussian monopulse and look for the trigger in both the generated signal as well as the acquired signal. The difference between the trigger locations is the delay.

Or use a test signal such as MLS or Chirp pattern and use cross correlation to measure delay.

But I like to use the SVT VI 😉

[tmcarson]

I would prefer to compute by formula.  I think I will need to settle for making a lookup table and restricting the sample rate choices. 

The end goal is to have a stand alone application that would not need to be measured each time.  My current system uses on sample rate with the delay fixed.  I'm trying to make something more flexible and robust.

I have measured the delay for a few combinations.  would it change for different DSAs of the same type?

Thanks for you insight

Timm C.

[dsb@NI]

The measured delay on a single device can change by up to one sample between restarts. We have observed that the delays of different devices are measured to be within one sample of each other.

If the measurement accuracy is sensitive to any change/difference less than a sample, then it is worth it to have a 'calibration' option in your application.

[tmcarson]

Thanks for the info. One or two samples is good enough for me.  do you know how this is handled in other systems made by companies like LMS or B&K.

Thanks again