RTA settings question

[Dent]

In the help menus under Spectrum controls there is an example that shows:
Mode RTA 1/24 octave >37 Hz
The help menu tries to explain what the ">37 Hz" means but I just don't understand. As I move the mode up the list, i.e. RTA 1/12, RTA 1/6, RTA 1/3, etc., the > Hz numbers become smaller. What does this mean?

[brucek]

It's telling you that it's not an accurate true RTA display below that frequency indicated. It has to do with the FFT math. It is still an accurate level with respect to the pink noise used and the level received from the microphone, but it can't be called a true RTA representation below that frequency. Be sure to use periodic pink noise and the Hann windowing with the RTA.

brucek

[Dent]

Thanks for the reply. Now you mentioned using the Hann windowing with the RTA and periodic Pink Noise? The help menus specifically recommends using the Rectangular window with the periodic Pink Noise and RTA. Is there a specific reason you prefer Hann windowing instead?

[brucek]

Yeah, good point. The Hann recommendation came from my notes where there was a change in the beta version to modify the frequency restrictions on the signal generator.

The help files were then modified to add the message:

"The Hann window is well suited to most measurements, offering a good tradeoff between resolution and shoulder height. If very high dynamic range needs to be resolved (very small signals close to very large signals) use the 4-term or 7-term Blackman-Harris windows. If the spectral peak amplitudes must be accurately measured use the Flat Top window, this will provide amplitude accuracy of 0.01 dB regardless of where the tone being measured falls relative to the bins of the FFT. The other windows only show the spectral amplitude accurately if the tone is exactly on the centre of an FFT bin, if the tone falls between two bins the amplitude is lower, with the maximum error occurring exactly between two bins. This maximum error is 3.92dB for the Rectangular window, 1.42dB for Hann, 0.83dB for the 4-term Blackman-Harris and 0.4dB for the 7-term Blackman-Harris."

I started using it for everything, but you're right, if it's periodic you can use rectangular as it also says...

brucek

[JohnM]

It is essential to use the Rectangular window for accurate results with Periodic Noise. To see the difference this makes try connecting a loopback with the Spectrum or RTA, periodic noise and no averaging.

[brucek]

hehe, guess I better change my notes on that point.

[Dent]

brucek wrote:

It's telling you that it's not an accurate true RTA display below that frequency indicated. It has to do with the FFT math. It is still an accurate level with respect to the pink noise used and the level received from the microphone, but it can't be called a true RTA representation below that frequency. ...brucek

When I set things to the highest (most accurate) settings, i.e. Mode RTA 1/48 octave and FFT length 131072 it shows a ">39 Hz" in the "Mode" line so you are saying it is not an accurate display from 0 to 39 Hz? If I reduce the Mode setting to RTA 1/24 octave it then says ">19 Hz" which means it is only not an accurate display from 0 to 19 Hz now? Then dropping down one further to RTA 1/12 octave, the Hz setting now longer shows up so now the display is the most accurate all the way down to 0 Hz?

Why is it that decreasing the Mode to a "worse" setting gives a more accurate display?

[JohnM]

The display is accurate at all settings, but it has lower frequency resolution than an equivalent RTA below the frequency shown in the mode line. The RTA octave fraction and FFT length setting only change the frequency resolution of the display, not its accuracy.

Signals processed by FFT have a frequency resolution equal to the sample rate divided by the FFT length, for example a soundcard sampling at 48kHz and an FFT length of 65536 (64k) gives a frequency resolution of 48000/65536 = 0.732Hz. This is often caled the "bin" width of the FFT. An RTA has a resolution that varies with frequency as it is a fraction of an octave, i.e. a fraction of a doubling in frequency. The octave between 1kHz and 2kHz has 1000Hz whereas the octave between 10Hz and 20Hz has only 10Hz, so the width in Hz of 1/12th octave (for example) is much less at 10Hz than it is at 1kHz. The indication on the mode line tells you the frequency below which the resolution is determined by the FFT length rather than the RTA octave fraction, above that frequency the underlying FFT resolution is higher than the RTA setting requires, so as many FFT bins as correspond to the RTA octave fraction at each RTA frequency are added together to produce the RTA value. Below that frequency the individual FFT bins are used directly, scaled appropriately to show the correct level.