SVT ITU-R 468-4 at 48kHz sample rate

[Pappion]

ITU-R 468-4 (aka CCIR 1k Weigthed) is only available in fixed rate VI which does not cover 48kHz sample rate..

Any work around for this?

I may also need 8k, 16k, 32k, 44.1k and 48k.

BTW: A-weigthing is avialable in Weighting (Arbitary Rate) pallete.

LV2014, LV2012, Sound Vibration Toolkit.

P.S. It be nice to replace some of the high priced R&S stuff.

 Just what is the Idle Channel Noise on a 4461 (1v dBFs)? I'm kinda put off by BNC with one signalinput  on shield. Would feel better about Twin BNC. Real or Not?

[Roohcifer]

Hi Pappion,

The ITU-R 468-4 fixed rate filter does cover 44.1k and 48k.  Please see this document for a full list of supported sample rates:  

http://zone.ni.com/reference/en-XX/help/372416B-01/svtconcepts/apply_wfilters/

The ITU-R 468-4 filter passes frequencies well above 10kHz as per this curve:

http://en.wikipedia.org/wiki/ITU-R_468_noise_weighting

As a result, I don't really understand the motivation for using sample rates as low as 8k or 16k (or, for that matter, 20k, even though that is the minimum supported sample rate for the filter as per the document at the top of this post).  You won't be capturing frequencies that are relevant to the filter if your sample rates are this low.

The SFDR of the 4461 is listed for the various gain ranges on p. 6 of the spec doc:

http://www.ni.com/pdf/manuals/373770j.pdf

I'm not sure I understand the final question, but each BNC connection on the 4461 is differential, where the coaxial pin is + and the shell is -.

Let me know if this answers you questions!

Regards,

[Pappion]

Roohcifer,

I'm writing this from home, as it is now crunch time at work. I don't know when I can get back to it.

At 48kHz the fixed rate ITU-R 468-4 gave an error. The error explantion did not include all of the frequencies your document lists.

Lying to it by changing dt 1/50k worked, so it's not the data.

Paycheck is Motivation for 8k, 16k, and 32k. It's a moblie phone thing.

Voice doesn't need much bandwidth. Sometimes power consumption is the more important consideration.

"The SFDR of the 4461 is listed for the various gain ranges on p. 6 of the spec doc:"

Yes that all looks great, but SFDR is not a number I can present.

It is useful with an undefined BW vs. SNR which needs a defined BW.

It's not the AES17 Idle Channel Noise (ICN) with CCIR 1k (or 2k) weighted filter, BW 20-20kHz and AES17 20k upper band filter.

It's not the AES17 Dynamic Range (DR) test which is just the above with -60 dBFs 997Hz sine present, notched out later for Vrms measurement.

(It is harder to cheat as a DAC must be actively clocked during the DR version of ICN test.)

ICN and DR results prefferably expressed as dBV (dBFS 1Vrms), or Vrms.

[Roohcifer]

Hi Pappion,

I tested the code below, and it worked when I configured the "Simulate Signal" express VI's sampling rate to 48k, 50k, and any other frequency listed in the document.  All other frequencies result in an error. 

If your data is not sampled precisely at 48k, there will be an error.  To check the true sample period of your data, wire the brown waveform wire containing your data into a "Get Waveform Atributes" block, change the "Y" component to "dt", and put an indicator on the "dt" terminal.  Right click on the front panel "dt" indicator, click on "Display Format", and change "digits" to 17.  When you run your code, the indicator should read 2.0833333333333333E-5.  If not, then your data was not sampled precisely at 48k.

To resample your data in software, you can use the "Resample Waveforms" VI:

http://zone.ni.com/reference/en-XX/help/371361J-01/lvwave/resample_waveforms_cont/

I tested it the above VI, and the resulting filtering returns no errors when you set an appropriate dt.  You can also use this VI to upsample your 8k, 16k, and 32k data, as there is no way to get the ITU-R filter VI to accept other data rates without recoding the algorithm.  The only other option I can think of would be sampling your data at an accepted rate, and then applying a software LP filter to achieve the bandwidth you desire.  Either way, the VI can only accept the frequencies listed in that document.

The only specs we have for the 4461 are those published in the spec sheet I linked previously.  For obvious reasons, we cannot test our cards in every conceivable industry's standards.  As such, we provide what specs we can.  Should the specs in that sheet not be sufficient, I would reach out to your local NI sales representative and ask whether they have a loaner card you could test to verify the specs yourself.  That said, there are typical AI Idle Channel Noise specs in Vrms  and typical DR for a 1kHz input tone at -60 dBFS for variouw BWs in the spec sheet, though there does not appear to be any specifid filter.  Again, it would be impractical to test this card under every inductry standard spec, so the data sheet presents general, seemingly unfiltered specs (except where indicated otherwise) that have some amount of meaning to every industry.

[Pappion]

I will check the dt for accuracy. Thank-you.

Still it would be nice to have a full package, ie. multi-rate CCIR (ITU-468-4).

Since this hardware and software is sold as a Sound package, some concessions to AES17 seems in order.

It's frustrating, your parked on 3rd base, when a home run is just a step away.

Maybe you all should hire an audio engineer to polish things up a bit.

[Pappion]

It does work at 48kHz, I had to fix the dt rounding error. (dt reciprocal, round off, and reciprocal again).

The Error Expanation does not include 48kHz as a vaild sample rate. (attached).

Last year I played with a PXI-5922 24 bit digitizer card.

I guess it's similar to PXI-4461 but the input circuitry is O'scope oriented, not Audio oriented.

Still, I could get excellent 99dB SINAD performance in a small voltage range. (attached).

Too bad you all wasted months of time putting together a demo. I had to cancel and buy R&S UPVs.