• Registered: 2008-01-01
  • Posts: 3,025

Topic: 25 Hz bar on Spectrum Analyzer too high?

I wonder why the 25 Hz bar of the Spectrum Analyzer (both Digicheck 4.5 and 5.0 prerelease) does always show a higher input/output db-level than the other bands? Is this a reliable measurement or a bug? If it is a reliable measurement how do I have to interprete it?

Here are some examples:

Input noise-floor (stereo) of inputs 7+8

http://img145.imageshack.us/img145/9585/digicheckinputys3.png

Output 7+8 White Noise (Realtime Analyzer)

http://img142.imageshack.us/img142/9962/digicheckwhiteew8.png

Output 7+8 Pink Noise (Realtime Analyzer)

http://img99.imageshack.us/img99/4531/digicheckpinkot0.png

[b]Output 7+8 Bron Noise (Realtime Analyzer)[b]

http://img365.imageshack.us/img365/1146/digicheckbrownwb2.png


As a comparison Brown Noise from another Noise Generator software (Sound Generator):

http://img149.imageshack.us/img149/3566/digicheckbrown2tw0.png


And finally a sine-tone at 48 kHz (Realtime Analyzer):

http://img210.imageshack.us/img210/9485/digichecksine48kpy0.png

2 Reply by MC

  • MC
  • MC
  • Administrator
  • Offline
  • From: Germany/Thailand
  • Registered: 2007-04-06
  • Posts: 25,582

Re: 25 Hz bar on Spectrum Analyzer too high?

Explained in the online help - no bandpass, DC included.

Regards
Matthias Carstens
RME

MC's Website

  • Registered: 2008-01-01
  • Posts: 3,025

Re: 25 Hz bar on Spectrum Analyzer too high?

Thanks for the quick reply! smile

So this is the DC offset of the converters? What use is the 25 Hz band readings then? And why does it read alot higher with the brown noise created by one software than by another if the other bands (including the neighboring 31.5 Hz band) are nearly the same?

Sorry if this sounds like a stupid question, but I lack the theoretical background to interprete the 25 Hz readings.

4 Reply by MC

  • MC
  • MC
  • Administrator
  • Offline
  • From: Germany/Thailand
  • Registered: 2007-04-06
  • Posts: 25,582

Re: 25 Hz bar on Spectrum Analyzer too high?

First this band is not a bandpass (as the others), but a low pass, so it covers a wider range and thus shows more energy when given the same signal. We added this feature to detect low frequency rumble (when mastering long play records) and DC problems easily. That said it seems the one software indeed produces a high amount of DC, or has a wrong filter architecture around zero to 10 Hz. The display also shows DC errors of older converters very good (like the ones done with early DAT recorders).

As the area below 25 Hz does no include useful information sonically (ideally this one would be empty), having a low pass instead of a bandpass makes a lot of sense (that our competitors did not yet notice...grin...).

Regards
Matthias Carstens
RME

MC's Website

5 Reply by MC

  • MC
  • MC
  • Administrator
  • Offline
  • From: Germany/Thailand
  • Registered: 2007-04-06
  • Posts: 25,582

Re: 25 Hz bar on Spectrum Analyzer too high?

MC wrote:

As the area below 25 Hz does no include useful information sonically (ideally this one would be empty)

Maybe I should add some information here. A good record does not contain any low frequency junk, which just wastes energy and steals power. Therefore in mastering a subsonic filter is used very often. Especially as these days everything is done completely within the computer, but programmers from various (even professional) software just code without thinking about the results of what they do. This can not only be seen in the oversampling topic (plug-ins violating the Nyquist border with illegal calculations, producing absurd wave forms), but also with your software example, where the signal should not be as wrong as it is. Be it EQ, LFOs or whatever, the lower limit is 0 Hz for them, which is as wrong as not taking care of the half sample rate border.

Regards
Matthias Carstens
RME

MC's Website

  • Registered: 2008-01-01
  • Posts: 3,025

Re: 25 Hz bar on Spectrum Analyzer too high?

Thanks again for the good explanation. I did read the part of the manual/help that explains this, but frankly I didn't understand its implications. Now all is clear and I like the idea of the lowpass filter on the 25 Hz "band".

I'm currently trying several free software signal-generators in order to measure my room as a basis for some acoustic treatment. Too bad that Digicheck doesn't offer signal-generation.

Concerning my results and how Digicheck displays them I plan to use Realtime Analyzer. It's an analyzing software similiar to Digicheck that offers its signal-generation options free of charge. It's documented frequency range starts at 10 Hz, which fits your explanation. My guess is that the other software "Sound Generator" starts at 0 Hz and thus produces a larger spike in the 25 Hz "band" accordingly. That's a pitty, because this software is the only one allowing 24-bit playback.

I assume that the 25 Hz spike of the Input measurement is due to the higher energy of the accumulated input-noise spectrum too?

PS: My suggestion/wish for Digicheck would be to include a signal-generator. wink

7 Reply by Timur Born (edited by Timur 2008-06-03 03:30:42)

  • Registered: 2008-01-01
  • Posts: 3,025

Re: 25 Hz bar on Spectrum Analyzer too high?

One last question: Is the highest band using a bandpass or a highpass filter? I ask because I noticed that the highest band also is a bit higher in all of my measurements and when sending pink noise through it at 44.1 kHz sample-rate there is a 1 dB raise at 20 kHz that shifts to 40 kHz when using 96 kHz sample-rate.

Digicheck is a great tool by the way, it helps me to understand what is going on in my system and room.

  • Registered: 2007-11-20
  • Posts: 16

Re: 25 Hz bar on Spectrum Analyzer too high?

hey MC it seems like you have an idea of what is programmed correctly and incorrectly. Could you recommend a high pass / low cut filter plugin?