Topic: ADI-2 DAC: Slow filter at higher sample rates

I am interested in using the Slow filter option on the ADI-2 DAC for its good impulse response, but the rolled off treble in the audible range and imaging down to pretty close to the audible spectrum when using 44.1 kHz sample rate is not what I want.

Is this something that can be cured by upsampling my music to say 176.4 kHz and then also using the DAC at this sample rate? The manual only shows frequency response for 44.1 kHz, so I don't know what happens at higher sample rates.

I am also a bit unsure about the lowest possible image frequency in this case. I think it would be 154.05 kHz (sample rate 176.4 kHz minus 22.05 kHz as the highest possible frequency in my source files), would the DAC even be able to output that?

Also, would impulse response somehow get worse again when upsampling?

2

Re: ADI-2 DAC: Slow filter at higher sample rates

Doco wrote:

I am interested in using the Slow filter option on the ADI-2 DAC for its good impulse response, but the rolled off treble in the audible range and imaging down to pretty close to the audible spectrum when using 44.1 kHz sample rate is not what I want.

I fully agree, but in the end the aliasing turns out to be no big issue in real-world...otherwise MQA wouldn't even exist...

And for the rolled-off treble there is a simple workaround: use the EQ to compensate. Here are two settings for Slow and NOS to linearize the treble response at 44.1 kHz sample rate:

Slow B5 G +4.0 F 16.5k Q 2.1 Shelf
NOS B5 G +3.0 F 14.3k Q 0.6 Shelf

As you can imagine with only 3 and 4 dB gain this is just a slight change, it linearizes NOS perfectly, and Slow good enough.

http://archiv.rme-audio.de/images/dafiltercurve_eqcomp_slow.png

http://archiv.rme-audio.de/images/dafiltercurve_eqcomp_nos.png

Now the point is that the Shelf filters have only small impact on the impulse response. Basically Slow stays Slow, and NOS stays NOS. See these two pictures, comparing original (red) with EQ'd (blue) result.

http://archiv.rme-audio.de/images/slow_eqcomp.png

http://archiv.rme-audio.de/images/nos_eqcomp.png

Doco wrote:

Is this something that can be cured by upsampling my music to say 176.4 kHz and then also using the DAC at this sample rate? The manual only shows frequency response for 44.1 kHz, so I don't know what happens at higher sample rates.

From single to quad speed (up to 192 kHz) the filters stay the same, they just double/quadruple in corner frequency (they move out of the hearing range), and the ringing is halve/quarter in time length.

Above 192 kHz the DAC always uses Slow, this is a fixed function.

But the point is a different one: whenever you upsample the data has to be filtered. This filter is exactly the same as when using the DAC itself in Single Speed, and can also be seen in the upsampled result. So you gain nothing.

Doco wrote:

I am also a bit unsure about the lowest possible image frequency in this case. I think it would be 154.05 kHz (sample rate 176.4 kHz minus 22.05 kHz as the highest possible frequency in my source files), would the DAC even be able to output that?

At 192 kHz there is no real argument against a steep filter, no matter if SD Sharp or Sharp. There will be no 96 kHz signal in your audio material that would make them ring. And even if then you are not able to hear this short, low level 96 kHz signal.

Regards
Matthias Carstens
RME

Re: ADI-2 DAC: Slow filter at higher sample rates

Thank you for the EQ settings. I will test them once my unit has finally arrived. I thought the treble roll off was important in getting the impulse response, but this looks very minor.

Am I correct to assume the settings for the Slow filter will also work for the SD Slow filter, given their frequency response is virtually identical up to 20 kHz?

4

Re: ADI-2 DAC: Slow filter at higher sample rates

Correct.

Regards
Matthias Carstens
RME

Re: ADI-2 DAC: Slow filter at higher sample rates

As you can imagine with only 3 and 4 dB gain this is just a slight change, it linearizes NOS perfectly, and Slow good enough.

Now the point is that the Shelf filters have only small impact on the impulse response. Basically Slow stays Slow, and NOS stays NOS. See these two pictures, comparing original (red) with EQ'd (blue) result.


So the small impact on the impulse response is true only for shelf filters (and not for Pick, High/Low cut) ?

6

Re: ADI-2 DAC: Slow filter at higher sample rates

Yes, peak filters usually cause more ringing than shelf filters when used in an application like this one, correcting the frequency response near the Nyquist border.

Regards
Matthias Carstens
RME

Re: ADI-2 DAC: Slow filter at higher sample rates

Thanks, good to know. So much things to learn. But I'm happy to learn smile !

8 (edited by dir 2018-05-04 19:05:53)

Re: ADI-2 DAC: Slow filter at higher sample rates

any EQ-recommendations for 48 kHz, 88 etc (in the NOS-mode)?
it will be interesting to EQ-correct input for recordings with slow filter

9

Re: ADI-2 DAC: Slow filter at higher sample rates

At 88 kHz there is no need to correct NOS anymore, IMHO. And the record EQ setting is exactly the same for Slow as shown above. For 48 kHz I will check.

Regards
Matthias Carstens
RME

10

Re: ADI-2 DAC: Slow filter at higher sample rates

http://i.piccy.info/i9/9b202991a0c2ed0793eb196a0fe46ac1/1525517191/19176/1239453/Spectrum.png
24/96 mode
it drops a little (around -1 dB) on the 24 kHz band

by the way, why RME calls super-slow filter of 4490 as NOS ? Actually NOS-DAC's are differrent story.

11

Re: ADI-2 DAC: Slow filter at higher sample rates

dir wrote:

by the way, why RME calls super-slow filter of 4490 as NOS ? Actually NOS-DAC's are differrent story.

Explained in the manual. Seeing the stairsteps and the massive aliasing effects it is bad enough to be called that (the reason that we made it available at all is that the oversampling is reduced, but not completely gone. Therefore the filtering is still there, weaker than with Delta-Sigma, but more than real NOS. It's an interesting approach, and the side-effects are tolerable in certain applications).

But the point is: the Pro is a Pro tool. Calling a filter 'SuperSlow' has no meaning for anyone. Imagine someone transfers precious recordings through an analog chain, uses the Pro (or DAC) as DAC, processes it, records it again at Hi-Res. Then a year later the engineer discovers that all transfers are completely messed up with high frequency noise, aliasing, and even the expected response in the higher treble lacks. Then that guy is fired and our reputation completely destroyed.

In HiFi people can do to the music whatever they like, it's their fun and usual has no serious consequences, right there at the end of the chain. In a studio environment R2R and NOS are an absolute no-go. So is 'SuperSlow'.

Regards
Matthias Carstens
RME

12

Re: ADI-2 DAC: Slow filter at higher sample rates

Here are examples for EQ settings to compensate Slow and NOS at 48 kHz sample rate:

Slow B5 G +2.5 F 16.3k Q 1.7 Shelf
NOS B5 G +3.5 F 16.5k Q 0.9 Shelf

As expected the higher frequency response of 48 kHz sample rate helps making the frequency response even more linear up to 20 kHz as with 44.1 kHz sample rate.

http://archiv.rme-audio.de/images/dafiltercurve_eqcomp_slow_48.png

http://archiv.rme-audio.de/images/dafiltercurve_eqcomp_nos_48.png

And here are the impulse responses for these EQ settings, comparing original (red) with EQ'd (blue) result.

http://archiv.rme-audio.de/images/slow_eqcomp_48.png

http://archiv.rme-audio.de/images/nos_eqcomp_48.png

Regards
Matthias Carstens
RME

13

Re: ADI-2 DAC: Slow filter at higher sample rates

Danke! This is really a very useful correction!
but what about the HD-signal? I must disable the EQ or will there be any presets for it, too?

14 (edited by HIFI 2018-05-17 06:59:38)

Re: ADI-2 DAC: Slow filter at higher sample rates

MC wrote:

Here are examples for EQ settings to compensate Slow and NOS at 48 kHz sample rate:

Slow B5 G +2.5 F 16.3k Q 1.7 Shelf
NOS B5 G +3.5 F 16.5k Q 0.9 Shelf

As expected the higher frequency response of 48 kHz sample rate helps making the frequency response even more linear up to 20 kHz as with 44.1 kHz sample rate.

https://archiv.rme-audio.de/images/dafiltercurve_eqcomp_slow_48.png

https://archiv.rme-audio.de/images/dafiltercurve_eqcomp_nos_48.png

And here are the impulse responses for these EQ settings, comparing original (red) with EQ'd (blue) result.

https://archiv.rme-audio.de/images/slow_eqcomp_48.png

https://archiv.rme-audio.de/images/nos_eqcomp_48.png


The simplest thing to set in the unit is setting the high and low frequencies. For Slow 10kHz to + 3.2db. For NOS 3.5kHz to + 2.5db. Get the same smooth frequency response up to 20 kHz.

15 (edited by PulsaR 2018-11-17 21:10:39)

Re: ADI-2 DAC: Slow filter at higher sample rates

Does the correction apply to "SD Slow" as well?

If not can we have correction data for SD Slow as well if it needs it?

Thanks

Re: ADI-2 DAC: Slow filter at higher sample rates

I think you have an error in your ADI-2 DAC user guide v1.4:-

DA Filter

Short Delay Sharp, Short Delay Slow, Sharp, NOS. The Digital to Analog Converter chip offers
several oversampling filters. Default is SD Sharp, offering the widest and most linear frequency
response and lowest latency. SD Slow causes a small drop in the higher frequency range.

Should it not be:-

Short Delay Sharp, Short Delay Slow, Sharp, Slow, NOS. (Add Slow)

17

Re: ADI-2 DAC: Slow filter at higher sample rates

Thanks for the note (I had received your mail to Support already). Yes, the word 'Slow' is missing.

SD Sharp / Sharp and SD Slow / Slow have the exact same frequency response, so the EQ settings work for both.

Regards
Matthias Carstens
RME

Re: ADI-2 DAC: Slow filter at higher sample rates

Does the DAC require any EQ'ing for 96kHz>?

P.S I love this DAC, it has replaced my Chord Hugo 2!

19

Re: ADI-2 DAC: Slow filter at higher sample rates

PulsaR wrote:

Does the DAC require any EQ'ing for 96kHz>?

The Slow and NOS filter do not need EQ at sample rates higher than 48 kHz, as the attenuation happens outside the hearing range then.

Regards
Matthias Carstens
RME

20 (edited by gioan 2019-01-27 11:01:06)

Re: ADI-2 DAC: Slow filter at higher sample rates

MC wrote:
PulsaR wrote:

Does the DAC require any EQ'ing for 96kHz>?

The Slow and NOS filter do not need EQ at sample rates higher than 48 kHz, as the attenuation happens outside the hearing range then.

Is this point valid also for upsampled 88/96/176 etc. from original 44/48khz recording? (I suppose yes, but not sure...)
Also would be interesting to see (if you have) frequency responses of slow/nos filters at 88/96/etc. khz

All this stuff by a pure "technical" curiosity, not audiophile mumbo jumbo...
I'm perfectly happy with "textbook" sharp filters at any samplerate :-)

Re: ADI-2 DAC: Slow filter at higher sample rates

gioan wrote:
MC wrote:
PulsaR wrote:

Does the DAC require any EQ'ing for 96kHz>?

The Slow and NOS filter do not need EQ at sample rates higher than 48 kHz, as the attenuation happens outside the hearing range then.

Is this point valid also for upsampled 88/96/176 etc. from original 44/48khz recording? (I suppose yes, but not sure...)
Also would be interesting to see (if you have) frequency responses of slow/nos filters at 88/96/etc. khz

All this stuff by a pure "technical" curiosity, not audiophile mumbo jumbo...
I'm perfectly happy with "textbook" sharp filters at any samplerate :-)

You should read post #2 wink

ADI-2 DAC (with stock PSU) - Neumann KH 310 A monitors - Cheap USB and XLR cables

22 (edited by gioan 2019-01-28 18:58:57)

Re: ADI-2 DAC: Slow filter at higher sample rates

N00b wrote:

You should read post #2 wink

This parts, I assume:

"From single to quad speed (up to 192 kHz) the filters stay the same, they just double/quadruple in corner frequency (they move out of the hearing range), and the ringing is halve/quarter in time lengt".

and

"whenever you upsample the data has to be filtered. This filter is exactly the same as when using the DAC itself in Single Speed, and can also be seen in the upsampled result. So you gain nothing".


The underlined parts of the two sentences appear inconsistent in my view.
If the filters double/quadruple in corners frequency (they move out of the hearing range) (and, I assume, the frequency roll-off move out of the hearing range), then is not exactly the same....
Shure, my lack of knowledge...

23 (edited by jiw 2019-01-28 21:22:43)

Re: ADI-2 DAC: Slow filter at higher sample rates

gioan wrote:
MC wrote:

The Slow and NOS filter do not need EQ at sample rates higher than 48 kHz, as the attenuation happens outside the hearing range then.

Is this point valid also for upsampled 88/96/176 etc. from original 44/48khz recording? (I suppose yes, but not sure...)

As MC says in post #2, the upsampling filter does exactly the same as the DAC for single speed data: remove content above the single speed Nyquist-frequency.

Thus, the impulse response and frequency response are dominated by those of the upsampling filter.

If you upsampled using a slow (shallow) filter, the HF attenuation is similar to that of the DAC's slow filter and you will need EQ to increase HF extension.

Re: ADI-2 DAC: Slow filter at higher sample rates

jiw wrote:

As MC says in post #2, the upsampling filter does exactly the same as the DAC for single speed data: remove content above the single speed Nyquist-frequency.

Thus, the impulse response and frequency response are dominated by those of the upsampling filter.

If you upsampled using a slow (shallow) filter, the HF attenuation is similar to that of the DAC's slow filter and you will need EQ to increase HF extension.

Ah ok...you (and MC of course) mean the interpolation low pass filter (at single speed Nyquist frequency) in the upsampling process.
Right, my lapse of memory.
Excuse me.

Re: ADI-2 DAC: Slow filter at higher sample rates

If i understand correctly.
1. When use files whith sample rate 44.1kHz output must be equalized for high frequency dropdown reason.
Valid for Slow or NOS filters.
Slow B5 G +4.0 F 16.5k Q 2.1 Shelf
NOS B5 G +3.0 F 14.3k Q 0.6 Shelf
2. When use files SR 48kHz
Slow B5 G +2.5 F 16.3k Q 1.7 Shelf
NOS B5 G +3.5 F 16.5k Q 0.9 Shelf

3. Streams higher 192kHz have only slow filter by default.
4. 88/96kHz not checked.

Question.
If frequency response have audible dropdown with 44.1/48 sampe rate streams why dont implement correction in dac firmware? Why user must be do it itself. And next question. How to view dynamic status of this corrections when switched from different files. Example. User done eq correction for 44.1 next switch to 48 done it for 48, next play 192 ets. Its not good for home users. This is good for profi whith one type of files used in one session on DAW. Ones time set and working.
Maybe add this corrections to dac firmware?

Re: ADI-2 DAC: Slow filter at higher sample rates

Hi.
Thank you all for interesting topic. I'm still learning this superbeast machine and very happy to have one. Please forgive me my question but in EQ settings - what "Q" stands for?  What value is that, can someone explain me please?
Any link to read about appreciated.

Re: ADI-2 DAC: Slow filter at higher sample rates

It explains itself if you watch it interactively at the PEQ.
You change whether influencing a broad or narrow range of frequency to be modified by the EQ setting.

BR Ramses - UFX III, 12Mic, XTC, ADI-2 Pro FS R BE, RayDAT, X10SRi-F, E5-1680v4, Win10Pro22H2, Cub13

Re: ADI-2 DAC: Slow filter at higher sample rates

Indeed smile   Now I know.  Thank you Ramses.

Re: ADI-2 DAC: Slow filter at higher sample rates

Q=Quality Factor.     The Lower the "Q" the Broader Range of Frequencies affected.  Likewise, a higher Q filter affects a narrower range.   Easily observed on the PEQ Screen. 

Curt

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

Re: ADI-2 DAC: Slow filter at higher sample rates

I am really out of my league here.  But what I have found is that I like the sound of NOS filter for some reason.  However I don't like the HF roll off.  So I tried my own PEQ filter in order to correct the roll off.  I then found the one posted here for NOS.  I will try that shortly.  But my lack of understanding leaves me a little confused.

I am trying to get CDs to sound good at 44.1kH sampling.  So the filter published here  will correct the HF roll off.  But what happens when I stream USB at 192kH?  What effect will the filter have.  Do I have to turn off filtering?

Also, when I turn off PEQ and look at the EQ window, it still shows the EQ curve even though I think I turned it off.  There is no indication in the EQ window that it is off.  How am I to understand this?

Also is there some free software available for modeling PEQ curves?

Thanks.

31

Re: ADI-2 DAC: Slow filter at higher sample rates

1. At 192 kHz sample rate you need to turn the filter off manually.

2. EQ disabled - curve shown in white instead of different colours.

Regards
Matthias Carstens
RME

Re: ADI-2 DAC: Slow filter at higher sample rates

Thanks.  I see that now.  Should I be concerned about the audibility of aliasing using the NOS filter?

Re: ADI-2 DAC: Slow filter at higher sample rates

I see that you said in a previous post that aliasing was not a real-world issue.  As for 192k, what happens to the frequency response when the NOS filter is in place?  Does the filter get replace with a slow filter above 192k or exactly at 192k?

34

Re: ADI-2 DAC: Slow filter at higher sample rates

1. You will see a small raise in upper treble.

2. Above 192 kHz.

Regards
Matthias Carstens
RME

35 (edited by orangejello 2019-08-03 19:23:24)

Re: ADI-2 DAC: Slow filter at higher sample rates

Thanks. 

BTW, I didn't find a mention in the documentation of the >192K filter always being "Slow".

36

Re: ADI-2 DAC: Slow filter at higher sample rates

Correct, I'll add  it.

Regards
Matthias Carstens
RME

37 (edited by zaurux 2019-08-04 14:47:39)

Re: ADI-2 DAC: Slow filter at higher sample rates

integra wrote:

...

3. Streams higher 192kHz have only slow filter by default.
4. 88/96kHz not checked.
...

If i use upsampling under Roon from 44.1 and 48 to 88/96, there is an effect with slow and NOS? Less treble like 44.1 and 88 ?

ROON on Lenovo > Allo USBridge > RME ADI-2 Dac > Audia Flight FL100  / Recital Audio Define HEFA

Re: ADI-2 DAC: Slow filter at higher sample rates

Wow, thanks MC! I just stumbled across this post. Changing the EQ with the slow filter has absolutely changed my perception of the RME DAC.

Up until now, I enjoyed the SD SLOW filter, mostly due to the fact it sounded most natural and organic to my ears. It also seemed to open up the soundstage and made things sound motr three dimensional. However, it often made a lot of music sound undefined and thus homogeneous as well. The SD SHARP filter gave music the required edge to help dynamics and perhaps had better "musicality" too but at the expense of the qualities I enjoyed with the SD SLOW.

I spent weeks thinking the perfect scenario for me would be a half way point between these two filters. After applying your EQ changes, I've found that sweet spot! I get all the smooth, airy sound of the SD SLOW with the positives of the SD SHARP combined. It's genuinely amazing to me how much this has changed my enjoyment of the RME DAC. I no longer get up between songs to change filters according to what song I'm about to play.

Re: ADI-2 DAC: Slow filter at higher sample rates

MC wrote:

Here are examples for EQ settings to compensate Slow and NOS at 48 kHz sample rate:

Slow B5 G +2.5 F 16.3k Q 1.7 Shelf
NOS B5 G +3.5 F 16.5k Q 0.9 Shelf

And for 44.1 kHz:
Slow B5 G +4.0 F 16.5k Q 2.1 Shelf
NOS B5 G +3.0 F 14.3k Q 0.6 Shelf

Hello Matthias,
thanks for this great information. This would be even good to have it in the manual!

My question:
Can those EQ presets be used with ADI-2 DAC FS (ESS), too? The known frequency response for the DAC 2 FS e.g. shows a decrease for NOS / 44.1 kHz beginning at 6 - 7 kHz - but correction starts at 14.3 kHz (for ADI 2) - wouldn't this be a little late for the DAC 2 FS (I don't know the linear frequency response of the ADI 2 DAC to be able to do an own comparison)?

40 (edited by KaiS 2023-04-23 21:17:00)

Re: ADI-2 DAC: Slow filter at higher sample rates

user317 wrote:

Can those EQ presets be used with ADI-2 DAC FS (ESS), too?

The filters, except “Sharp”, even if named similar, have a different shape in the ESS vs.the AKM DAC chips.
It’s visible in the graphs in the according manuals.

So, different EQ’s are needed.

Re: ADI-2 DAC: Slow filter at higher sample rates

Thanks Kai!

Hopefully Matthias can create those EQs for the ADI-2 DAC FS, too?

I'm wondering if there is any (online) application, which simulates creating EQs. Here, you can create filters, but the Q factor is ignored for the type higshelf. Does somebody know any other sites maybe?

Re: ADI-2 DAC: Slow filter at higher sample rates

Im very interested to experiment with compensation filter for 44.1 an 48 kHz for NOS and Slow with my ESS version ADI-2 DAC.

@Matthias: Would it be possible to share the settings like above especially for ESS?

Re: ADI-2 DAC: Slow filter at higher sample rates

MC wrote:
dir wrote:

by the way, why RME calls super-slow filter of 4490 as NOS ? Actually NOS-DAC's are differrent story.

In a studio environment R2R and NOS are an absolute no-go. So is 'SuperSlow'.

I find this comment to be a revelation.