Spectrafoo

[edemunari]

Sorry for wasting your time but is there anyone who should explain me why if i connect the Optical output of Digi interface (192 i/o) to the optical input of Mac (2x 2.8 GHz Quad-core, Mac OSx 10.5.5) and setting a mono channel with the digidesign signal generator to -20 Pink Noise i got in the spectrogram c.a. -42 in the power scale (shouldn't get the same vaues, i.e. -20)? At the same time the K14 meter says -6 and so (adding to the +14 full scale limit) correctly -20 as the signal generator and the peak meter of protools say.

BTW also spectra do the same so maybe i'm missing something or never understand the way i have to meter this...

http://duc.digidesign.com/images/smilies/mad.gif

Thanks in advance, all the best, Eriberto

[Andre Knecht]

When measuring absolute levels you should use a sine wave, because its level is constant. Wide [pseudo] random spectral signals, such as pink noise, typically contain peaks that exceed their nominal level (especially in the lower octaves).

IHTH.

[Serge Perron]

if i connect the Optical output of Digi interface (192 i/o) to the optical input of Mac (2x 2.8 GHz Quad-core, Mac OSx 10.5.5) and setting a mono channel with the digidesign signal generator to -20 Pink Noise i got in the spectrogram c.a. -42 in the power scale (shouldn't get the same vaues, i.e. -20)? At the same time the K14 meter says -6 and so (adding to the +14 full scale limit) correctly -20 as the signal generator and the peak meter of protools say.

Eriberto,

Do you mean that the "spectrograph" instrument reads -42 when you send it -20 dBFS pink noise from the Pro Tools signal generator? I have duplicated those results when I set the spectrograph to a resolution of 1/12 octave. What this means is that the power in each of those 1/12 octave bands is about -42 dB, but does not represent the total power.

I have a SpectraFoo level meter seeing the same pink noise signal. The meter values change depending on the scale chosen and whether you check the AES Std. RMS Ref. box in the level meter control pop up menu. Looking at the numerical level values in the pop up menu and using the K14 scale with the AES Std. RMS, the pink noise is at -6 dB peak RMS, as you observed. Using the SpectraFoo scale, in which 0 dBFS is the reference, the pink noise reads -20 dBu.

As Andre stated, you should use a sine wave to measure absolute levels. When the signal generator is set to a 1 KHz sine at -20 dBFS, my spectrograph reads -18.29 dB at 1 KHz, much closer to the expected value. But the numerical meter readings are the same for 1 KHz as they are for the pink noise: -6 dB on the K14 scale and -20 dBu on the SpectraFoo scale. But you'll notice that the 1 KHz steady state tone reading is much more stable than the pink noise.

Serge

[edemunari]

Originally Posted by Serge Perron

Eriberto,

Do you mean that the "spectrograph" instrument reads -42 when you send it -20 dBFS pink noise from the Pro Tools signal generator? I have duplicated those results when I set the spectrograph to a resolution of 1/12 octave. What this means is that the power in each of those 1/12 octave bands is about -42 dB, but does not represent the total power.

Serge

Yes the problem is not in the level meter, even if the sine wave is more stable all the waveforms confirm the -6 / -20 reading depending on the scale choosed.
What i'm trying to understand is strictly related to spectrograph, but not for only the 1/12 resolution. What you noticed happens with all the resolution choosed... more or less (anyway withouth reaching the -20)
The spectrograph reads always c.a. -40 dB. Because this happens with Spectre and also with roger nichols inspector xl, that is Rtas, i'm wondering what i'm doing wrong or maybe can't uderstand expecially because i can't get any math getting -42 as results of the pink noise metering.
Thanks in advance, all the best, Eriberto

[Serge Perron]

The spectrograph reads always c.a. -40 dB. Because this happens with Spectre and also with roger nichols inspector xl, that is Rtas, i'm wondering what i'm doing wrong or maybe can't uderstand expecially because i can't get any math getting -42 as results of the pink noise metering.

Spectrum analyzers divide a signal into frequency bands. The greater the number of bands, the more the signal is divided and the lower the energy level in each band will be. The -42 dB you are seeing is probably the energy level in one frequency band only. If the energy in each band are combined, you will get back the original level of the pink noise.

Since pink noise has equal energy per octave, I believe the math to determine the spectrograph reading in each band is: dBFS = 10log(1/number of bands) - 20.

To test this equation, if there is only one band: 10log(1/1) - 20 = -20dBFS, which is the reference level.

If there are 2 bands: 10log(1/2) - 20 = -23 dB, so there is 3 dB less in each band.

For a real example using the SpectraFoo 1/12 octave spectrograph, the number of bands is 173 (I actually counted them). So the math is 10log(1/173) - 20 = -42.4 dBFS. That's about what you get in each band when you send pink noise @ -20 dBFS from the Pro Tools signal generator plug in.

As an analogy, if you send pink noise @ -20dBFS to a graphic EQ with all the sliders at unity gain, then send the EQ output to a meter, it will read -20 dBFS. Now if you fully attenuate all sliders except one, the meter reading will be much lower. That is what is happening with the spectrograph levels.

Serge

[edemunari]

WOW!!! Serge!!!
Really clear and easy thanks a lot!!!

Eriberto