SINAD Analysis.vi Bug

[Tims]

Hi All

I was wondering if anybody has noticed this bug and could tell me what I
am doing wrong or is it just a bug.

I have been using the SINAD analysis.vi and a NIDAQ card to measure
audio quality of radio receivers, and i found this strange phenomenon.
The SINAD analyser.vi works fine if the data from the NIDAQ card is fed
directly into it. But if a software filter is used between the SINAD
analyser and the NIDAQ card the measured SINAD drops sharply. This
clashes with theory in that the SINAD should improve since the amount of
noise in the system is reduced by the software filter.

I investigated. I found no problem with what the software filters were
doing in the frequency domain but I notice an impulse at the beginning
of the residue time
domain waveform out of the SINAD analysis function.
Maybe somebody could explain.

I have attached a demo VI created in LV 6.1. (doesn't require NIDAQ card
to demostrate this problem.

Tim

[Ben]

Hi Tims,

I took a look.
What impulse are you talking about?

Could post a screen shot with some sort of indication of this impulse that you are questioning?



Ben

[DSPGuy]

Tim,
Could the problem be the settling time of the filter? If you run your VI once with a filter topology other than None, then the SINAD is ~31dB. If you run continuously the SINAD is improved (~86dB for Elliptic) as compared to the no filter case after the first iteration.

[Tims]

Hi Ben,

Use my example VI and set the exported signal to residue, and then try
enabling and disabling the filters.

Tim

Ben wrote:

> Hi Tims,
>
> I took a look.
> What impulse are you talking about?
>
> Could post a screen shot with some sort of indication of this impulse
> that you are questioning?
>
> Ben

[Tims]

Hi DSPGuy,

The data I am measureing is not continous. I should have set the sine
generator reset to true..

Tim

DSPGuy wrote:

> Tim,
> Could the problem be the settling time of the filter? If you run your
> VI once with a filter topology other than None, then the SINAD is
> ~31dB. If you run continuously the SINAD is improved (~86dB for
> Elliptic) as compared to the no filter case after the first iteration.