RME ADI-2/4 Pro SE Filter Response Measurements

[johny_2000]

Here are my own measurements of the RME ADI-2/4 Pro SE filter response. I believe they closely match the RME manual and MatLAB simulation results. Sample rate 192kHz

SD Sharp:

SD Slow:

Sharp:

Slow:

SD LD:

Brickwall:

NOS:

 

[ppataki]

Just purely by the looks, isn't NOS the best? I see the least ringing there

 

[ZolaIII]

Interesting enough only SD slow has a good proper impuls response and even it pre rings a little bit so I have to ask how you tested? SD sharp doesn't pre ring at all but it doesn't settle down even in 2x time, rest pre rings (except NOS) and negative side is bad.

 

[morillon]

mesureaments very very easy to do in multitones (free) from pkane too ;-)
but to associate the answer in FR...

 

[staticV3]

Just purely by the looks, isn't NOS the best? I see the least ringing there

Yes, NOS has the best possible impulse response.

The problems start when you try to play anything else with it.

 

[ppataki]

The problems start when you try to play anything else with it.

Hi @staticV3
What do you mean by this?

 

[morillon]

observe the FR with differents filters

 

[staticV3]

Hi @staticV3
What do you mean by this?

See Archimago's excellent articles on NOS DACs:

NOS vs. Digital Filtering DACs: Exploring filtering turned off, implications, fidelity and subjective audibility. (Recent BorderPatrol DAC chatter...)

A blog for audiophiles about more objective topics. Measurements of audio gear. Reasonable, realistic, no snakeoil assessment of sound, and equipment.

archimago.blogspot.com

REVIEW & MEASUREMENTS: Cayin RU6 R-2R USB Dongle DAC/Headphone Amp. Considering "Intentional Imperfections". (Topping D90SE -60dBFS input graphs.)

A blog for audiophiles about more objective topics. Measurements of audio gear. Reasonable, realistic, no snakeoil assessment of sound, and equipment.

archimago.blogspot.com

DEMO: Listening to the Cayin RU6 R-2R noise and distortions with headphone load. Confirming subjective impressions. And Jim Austin / Stereophile's "Thoughts On Reviewing".

A blog for audiophiles about more objective topics. Measurements of audio gear. Reasonable, realistic, no snakeoil assessment of sound, and equipment.

archimago.blogspot.com

MEASUREMENTS: Digital Filters and Impulse Response... (TEAC UD-501)

A blog for audiophiles about more objective topics. Measurements of audio gear. Reasonable, realistic, no snakeoil assessment of sound, and equipment.

archimago.blogspot.com

 

[Mnyb]

The ringings are just a response to the special test signal who could not “ legally“ occur in properly bandwidth limited signals trough sampling .

IE it won’t happen with music ( normally but let’s makes this simple ) , it’s a case of wrongly interpret test signals by lay people

Each of these filters has a theoretical response and the measurements tells you the actual filters are close to theoretical nothing more .
One should not at all interpret these curves as if they represent some kind of transient response with music.

 

[johny_2000]

Interesting enough only SD slow has a good proper impuls response and even it pre rings a little bit so I have to ask how you tested? SD sharp doesn't pre ring at all but it doesn't settle down even in 2x time, rest pre rings (except NOS) and negative side is bad.

Thank you. Everything was done in REW with a loopback connection of the device. I also had similar concerns about SD filters.

 

[johny_2000]

observe the FR with differents filters

This is what I'm going to do next. You just inspired me

 

[staticV3]

This is what I'm going to do next. You just inspired me

Best method is to play random white noise on the DAC at 44.1 or 48kHz, then sample that noise at 96kHz ADC sample rate or higher, and use extremely long averages (1k+).

FFT size can be small, 32k is more than enough in most cases.

You then get graphs like this:

https://media.discordapp.net/attach...976e58a0ee37588fb18709c83a7a67bc503e99d1a46a&

 

[Sokel]

Best method is to play random white noise on the DAC at 44.1 or 48kHz, then sample that noise at 96kHz ADC sample rate or higher, and use extremely long averages (1k+).

FFT size can be small, 32k is more than enough in most cases.

You then get graphs like this:

https://media.discordapp.net/attach...976e58a0ee37588fb18709c83a7a67bc503e99d1a46a&

Confirm this,and it makes nice visuals too:

 

[johny_2000]

Best method is to play random white noise on the DAC at 44.1 or 48kHz, then sample that noise at 96kHz ADC sample rate or higher, and use extremely long averages (1k+).

This method appears to require the use of two devices for playback and recording since different sample rates are used.

 

[staticV3]

This method appears to require the use of two devices for playback and recording since different sample rates are used.

It's the only way to isolate the DAC filter.

If you run DAC and ADC at the same rate, then you'll measure the sum of DAC filter + ADC filter instead.

 

[Rja4000]

Best method is to play random white noise on the DAC at 44.1 or 48kHz, then sample that noise at 96kHz ADC sample rate or higher, and use extremely long averages (1k+).

I use a slightly different method.

Of course, 2 devices are needed, and the ADC has to run at higher sampling rate.
But I keep both DAC and ADC in sync, using SPDIF link between them.

Then I'm able to use a higher sampling frequency on the ADC as long as it's a multiple of the DAC frequency.
And I can then use method like periodic noise from REW or FFR (Hpw) to measure response instantly.

It's not always working, but when it works it's fast and neat.

 

[staticV3]

I use a slightly different method.

Of course, 2 devices are needed, and the ADC has to run at higher sampling rate.
But I keep both DAC and ADC in sync, using SPDIF link between them.

Then I'm able to use a higher sampling frequency on the ADC as long as it's a multiple of the DAC frequency.
And I can then use method like periodic noise from REW or FFR (Hpw) to measure response instantly.

It's not always working, but when it works it's fast and neat.

Whenever you get to use windowless/rectangular window FFT, it just scratches an itch. So satisfying!

 

[Sokel]

Whenever you get to use windowless/rectangular window FFT, it just scratches an itch. So satisfying!

This is also fun to watch and pretty accurate:

 

[MC_RME]

This method appears to require the use of two devices for playback and recording since different sample rates are used.

The first post requires the same. This is the reason your measurements don't look like those in the data sheet or my measurements in the manual. The ADC must run on highest sample rate to fully capture the DA signal without affecting it itself.

 

[johny_2000]

The first post requires the same. This is the reason your measurements don't look like those in the data sheet or my measurements in the manual. The ADC must run on highest sample rate to fully capture the DA signal without affecting it itself.

Good point.
Which ADC should I use for these measurements?
E1DA Cosmos / APX555B / APx516B or another RMA ADC?