SVFA Power Spectrum - Averaging issue

[Oleks]

Hello all,

I am using SVFA Power Spectrum to process a signal from a pressure trasnducer. All options associated with this block seems to work fine with the exeption of averaging parameters. I used Linear/One Shot mode and set number oaf averages to 20. The prioblem is once it reaches 20, it keeps going further. I added a boolean indicator to check if it processed the required number of averages, it did in fact. So now I am puzzled. Restart averagting seems to work as intended though.

Any thoughts?

[WillPwr]

Hello, Oleks!

Did you make sure to set your "weighting mode" to linear (0)? It will default to exponential if not, which could be causing the behavior you're seeing.

[Oleks]

As far as I can tell, It is in the linear mode (see the picture attached). That is what confuses me.

[Tori_W]

Oleks:

My name is Tori in Applications Engineering at NI and I've been looking into this a bit more. I think you have stumbled upon an important issue. It seems the Peak Hold type of averaging is a bit different from the RMS and Vector averaging types that seem to provide the correct behavior by restarting averaging at the number specified - in this case 20. I can't quite tell at this point what the problem is and need to figure out if Peak Hold should have the same functionality as RMS and Vector or not. If it does, this may be a bug. If it does not, then there needs to be documentation explaining what is going on here. I will likely get an opinion from R&D and will keep you posted. In the meantime, can you use RMS or Vector instead? Do you need another kind of workaround? Also, please send me your VI for reference. Have a nice evening.

Tori W.

[Tori_W · Accepted solution]

Oleks:

From the Sound and Vibration Help File for Time Averaging Modes, we can see that this is expected behavior:

"In peak-hold averaging, the largest measured sound pressure level value of all previous values is computed and returned until a new value exceeds the current maximum. The new value becomes the new maximum value and is the value returned until a new value exceeds it. Peak hold actually is not a true form of averaging because successive measurements are not mathematically averaged. However, as with other averaging processes, peak-hold averaging combines the results of several measurements into one final measurement. As with exponential averaging, the averaging process continues indefinitely."

So because Peak Hold does not average in a strict sense, it does not restart averaging at the number of averages you specified like RMS and Vector averaging that are a true average.

Tori W.

[Oleks]

Thank you Tori,

That makes perfect sense now!

Oleksandr