Topic: ADI-2 DAC FS Spectral Analyzer issue?

I'm working on my speakers placement to get best frequency response.

I played testing signals on USB input (white noise, pink noise, sine frequency 1000 Hz) and at random realized that ADI-2 DAC FS Spectral Analyzer page shows different pictures as expected. Can someone repeat my testing or comment my results?
I tried different setup (testing samples from various web pages, Win10 - foobar with/without ASIO, MacOS, iPhone) - always got same results. Note: Spectral Analyzer in Foobar works as expected for my testing signals.

Is this the issue with Spectral Analyzer only? Or deffective unit? 

White noise - in theory power density should be flat.
Pink noise - power density should fall off at 10 dB/decade
Sine 1000 Hz - Just 1000 Hz bar should be visible.

See screenshots here:
https://trozsypa.rajce.idnes.cz/ADI-2_DAC/

2

Re: ADI-2 DAC FS Spectral Analyzer issue?

White noise is flat with FFT. Pink noise is flat with bandpass filters - as used in the ADI-2.

Regards
Matthias Carstens
RME

3 (edited by ning 2020-01-18 19:54:34)

Re: ADI-2 DAC FS Spectral Analyzer issue?

Oh I noticed that long time ago. MC, could you add a setting option to disable the filter? Sometimes (such as in measurements) I would like to see a non filtered result so that I can quickly know what’s going on.

4

Re: ADI-2 DAC FS Spectral Analyzer issue?

Disabling the filters would disable the Analyser...that request makes no sense..

Regards
Matthias Carstens
RME

5 (edited by ning 2020-01-19 06:06:46)

Re: ADI-2 DAC FS Spectral Analyzer issue?

I mean for example if there is a 1khz tone just show an 1khz bar. Just the most straightforward fft algorithm will do.

(Perhaps you applied integral smoothing?)

6

Re: ADI-2 DAC FS Spectral Analyzer issue?

We don't use FFT.

Regards
Matthias Carstens
RME

Re: ADI-2 DAC FS Spectral Analyzer issue?

I didn't know FFT is not used for Analyzer. Next time I must read the user's guide - bandpass filters are mentioned in Analyzer section.
Requires FFT too much computing power and hardware was not designed this way? Or reason why bandpass filter is better than FFT?

Anyway, picture of "Sine 1000 Hz" signal is clear to me now. Neighboring "channels" don't have ideal selectivity. Hence bandpass filter partially leaks signal through.
But I still not understand issue with white noise signal. Let's assume that all bandpass filters have same selectivity and gain. On input of each filter is same energy (white noise spectrum is flat - it contains any frequency). So, output of each filter should be same and picture should be flat. 
I'm not expert for simulation of bandpass filter in digital domain :-(
MC, can you comment which of my assumption is wrong?

8 (edited by ning 2020-01-19 17:27:01)

Re: ADI-2 DAC FS Spectral Analyzer issue?

> Requires FFT too much computing power and hardware was not designed this way?

I guess so. FFT requires a buffer of lot of samples, and it's not easy to run in parallel (it's a sequential model). The spectrum is probably implemented in a FPGA, which enable the advertised 30-band filter running in parallel.

> Or reason why bandpass filter is better than FFT?

I guess it's because most other gadgets use bandpass filter too, and that's a tradition inherited from analogue devices.  band filters are easy to be implemented in analogue circuits, so were common in old devices.

9 (edited by mark2748 2020-01-20 21:38:33)

Re: ADI-2 DAC FS Spectral Analyzer issue?

Re flat spectrum display for pink noise:  The RME ADI-2 DAC spectrum analyzer displays one-third octave bands.  That is, the analysis filter bandwidths are nonlinear.  The increase in bandwidth with center frequency just compensates the 1/f frequency spectrum of pink noise, so the net result is a flat line display.  Octave Bands are commonly used for audio analysis because they closely approximate human hearing.  See for example the Wikipedia articles on "Pink Noise" and "Octave Band".  It would help if RME mentioned this in their user manual.

10 (edited by Curt962 2020-01-21 03:41:36)

Re: ADI-2 DAC FS Spectral Analyzer issue?

Trozspa,

Come on over to our thread "Recommended Reading for EQ..." 

https://forum.rme-audio.de/viewtopic.php?id=29041

We'd be delighted to offer some assistance!    The RME is Dead-Bolt Linear, and yet its display is never referenced in our tests.  It's your Analyzer Tool (REW, etc)that tells the Tale! (Not the RME Display)  It's the Speaker/Room interface that is sorely lacking in most cases.  We can help with that, and have happy participants to prove it.

Our Track Record is good!!   smile

Curt

Vintage 2018 ADI-2 DAC. "Classic AKM4490 Edition"
Cables:  Red, and White Ones.
Speakers:  Yes

Re: ADI-2 DAC FS Spectral Analyzer issue?

Thank you mark2748, I understand functionality of spectrum analyzer much more now.

mark2748 wrote:

Re flat spectrum display for pink noise:  The RME ADI-2 DAC spectrum analyzer displays one-third octave bands.  That is, the analysis filter bandwidths are nonlinear.  The increase in bandwidth with center frequency just compensates the 1/f frequency spectrum of pink noise, so the net result is a flat line display.  Octave Bands are commonly used for audio analysis because they closely approximate human hearing.  See for example the Wikipedia articles on "Pink Noise" and "Octave Band".  It would help if RME mentioned this in their user manual.

Re: ADI-2 DAC FS Spectral Analyzer issue?

Like Curt I don't quite understand what you are trying to do. The panel shows the level of the input at various frequencies. However, this depends on the music and not on the room. For the room's signature, you will have to measure in-room response, for example with REW. Such measurements use frequency sweeps and not real music.

RME ADI-2 DAC fs, Quad 606-2 power amp, Quad 2805 speakers, B&W PV1d sub with Antimode 8033 Cinema

Re: ADI-2 DAC FS Spectral Analyzer issue?

Topic of this post is spectrum analyzer functionality only, not the room's response on testing signal.
I was not aware that the way how spectrum analyzer is designed and implemented may results to different representation for same input signal (FFT vs. bandpass filters).

willem wrote:

Like Curt I don't quite understand what you are trying to do. The panel shows the level of the input at various frequencies. However, this depends on the music and not on the room. For the room's signature, you will have to measure in-room response, for example with REW. Such measurements use frequency sweeps and not real music.