Article: My Quest for a New DAC, Part 3 - Denafrips Terminator

[Geoff13]

Hello,

 

Thank you for a lovely and interesting review. So i also have a question. With the Terminator you don't have sibilance on the Sade recording. So would you say the Terminator smooths things out or how would you describe this.

 

I listen to a lot of vocal music and sibilance is pretty low in my set-up but if the Terminator gets rid of most sibilance than that's pretty amazing. I hear some sibilance on the Sade album with my NOS DAC. Would love it if that sibilance dissappeared.

 

Best regards

Geoff from Belgium

[austinpop]

1 hour ago, Geoff13 said:

Hello,

 

Thank you for a lovely and interesting review. So i also have a question. With the Terminator you don't have sibilance on the Sade recording. So would you say the Terminator smooths things out or how would you describe this.

 

I listen to a lot of vocal music and sibilance is pretty low in my set-up but if the Terminator gets rid of most sibilance than that's pretty amazing. I hear some sibilance on the Sade album with my NOS DAC. Would love it if that sibilance dissappeared.

 

Best regards

Geoff from Belgium

 

Hi Geoff,

 

In general, there are many malign forces that plague our audio systems. One of the guiding principles, lodestars if you will, we follow is to eliminate sibilance. In my experience, this is a system-wide effort, and tends to improve as you make improvements across the chain.

 

Ok, but that’s all motherhood and apple pie.

 

In my article, I used the word in a relative sense. In my friend’s system with the Yggy,  listening as is, I doubt anyone would consider the sound to be sibilant. It’s only in the comparative context, relative to the Terminator, that I felt the Terminator reduced sibilance slightly.

 

This is always the limitation of comparative language. You have to use a positive or negative phrase for the better (or worse) component, even though in isolation, there is nothing negative about it!

 

i wouldn’t go so far as to say the Terminator by itself can cure sibilance in a system. It could help, depending on what DAC it’s replacing.

 

I just want to reinforce that sibilance elimination is a systemwide effort.

[Ralf11]

So, I guess the next question is sibilance vis-a-vis digital glare...

[Account Closed]

Thanks for this great installment in "The Quest". I look forward to the final conclusion in episode (?). When you are done fooling around with computers, I would consider becoming a full time writer if I had your abilities.

 

I also would like to upgrade my DAC. I have a Mark Levinson 390s that has served me well for 13 years but it only does 44.1 or 48 via its SPDIF input. With true hi-res streaming coming online (Qobuz), I will need to upgrade soon. I may look into one of the new Ethernet direct options like the Merging or the MSB. I would really like to avoid the whole USB spaghetti and meatballs mess of boxes this time around. Hopefully there will be some lower priced options in that category going forward.

 

Looking forward to the next.

[austinpop]

18 hours ago, KDinsmore said:

When are you going to try an MSB? ?

 

Which one? Do they have one in this price range? I thought all their DACs were in the 5-digit USD price range!

[austinpop]

1 hour ago, Ralf11 said:

So, I guess the next question is sibilance vis-a-vis digital glare...

 

I can only talk about what these terms mean to me, and in that context, these are two different things.

 

Sibilance is a tendency where an audio system causes the "s" consonant in human voice (speech and/or song) to sound exaggerated, elongated, hissy, etc.

 

Digital glare is a more holistic impression of a system. Sometimes it's obvious harshness in the treble, or bright sounding, but sometimes it's just a subtle characteristic of the sound that is fatiguing after a while.

[Ralf11]

Yes, agreed.  I am wondering if there is overlap between the two - either in terms of the system and its design, or in terms of psycho-acoustics...

 

 

[Archimago]

Interesting discussion about sibilance guys...

 

Sure, it's likely a combination of factors and presumably the DAC is a factor here. But how much is contributed by the DAC and what is simply inherent in the musical recording? Maybe that's the way the Sade recording is supposed to sound (ie. "flawed" by design) and any DAC that smooths it out is doing the same thing for everything else! That would not necessarily be a good thing.

 

Here's a thought, can folks suggest maybe a handful of high quality recordings they think tend to cause sibilance on a poor DAC and presumably should not sound that way? Perhaps we can then look deeper into this concern and determine what a DAC could or should do to reproduce a "proper" sound.

[austinpop]

56 minutes ago, Archimago said:

Interesting discussion about sibilance guys...

 

Sure, it's likely a combination of factors and presumably the DAC is a factor here. But how much is contributed by the DAC and what is simply inherent in the musical recording? Maybe that's the way the Sade recording is supposed to sound (ie. "flawed" by design) and any DAC that smooths it out is doing the same thing for everything else! That would not necessarily be a good thing.

 

Here's a thought, can folks suggest maybe a handful of high quality recordings they think tend to cause sibilance on a poor DAC and presumably should not sound that way? Perhaps we can then look deeper into this concern and determine what a DAC could or should do to reproduce a "proper" sound.

 

Hi Archimago,

 

Thanks for raising this fascinating point. Yes, I agree this is a chicken and egg question. What's in the recording - inherently - vs. what is distortion? I have no expertise to make absolute statements here! Some observations, though:

1. It would certainly be fun to crowdsource here on CA a list of songs that both 
   • exhibit sibilance to some degree, and
   • for which sibilance has been found to improve with system improvements
2. Then for this list, we need some impeccable source who can represent the actual recording/artist, and make a clarification of "intent." Was that "distortion" intentional, or an artifact of the recording?
3. The above sounds pretty daunting, but perhaps possible.

Of course, let's also step back and reiterate that

1. sibilance is only one in many forms of distortion, and
2. improvements in sibilance (or more generally, distortion) are not just DAC-specific, but system-wide.

If this is a subject of interest (and it certainly is for me), it probably deserved a thread of its own.

[Ralf11]

I agree it is an interesting question (specifically as to sibilance, and more generally as to euphonic distortion - as an Audio Research tubey goodness pre-amp owner) and vote new thread

[Superdad]

On 9/13/2018 at 2:02 PM, austinpop said:

Sibilance is a tendency where an audio system causes the "s" consonant in human voice (speech and/or song) to sound exaggerated, elongated, hissy, etc.

 

 

 

[davide256]

On 9/14/2018 at 12:32 PM, austinpop said:

 

Hi Archimago,

 

Thanks for raising this fascinating point. Yes, I agree this is a chicken and egg question. What's in the recording - inherently - vs. what is distortion? I have no expertise to make absolute statements here! Some observations, though:

1. It would certainly be fun to crowdsource here on CA a list of songs that both 
   • exhibit sibilance to some degree, and
   • for which sibilance has been found to improve with system improvements
2. Then for this list, we need some impeccable source who can represent the actual recording/artist, and make a clarification of "intent." Was that "distortion" intentional, or an artifact of the recording?
3. The above sounds pretty daunting, but perhaps possible.

Of course, let's also step back and reiterate that

1. sibilance is only one in many forms of distortion, and
2. improvements in sibilance (or more generally, distortion) are not just DAC-specific, but system-wide.

If this is a subject of interest (and it certainly is for me), it probably deserved a thread of its own.

I nominate  recordings by "the Cranberries" as a test of whether a DAC can be excited into painful sibilance. Not the exaggerated" s" sound of a snake but rather the  inclusion of the affect of fingernail scraped across a chalkboard when traditional sibilance occurs.

[Ralf11]

which Cranberries CDs or tracks would you pick?

 

I am always looking to add to my test music list...

[austinpop]

Another good track for sibilance is Diamonds on the Soles of her Shoes from Paul Simon's Graceland.

[luisma]

As an owner of a Terminator for some time already, and I purchased when you initiated this quest with the 1st part, I tried different upsampling methods, DAC's OS, Roon DSP and HQPe with NAA and in my opinion and without trying to detriment Roon's or the DAC upsampling, HQP by far sounds the best. I know you already mentioned in your article the reasons for which  you weren't using HQP but since I tried all that I wanted to provide feedback.

Thank you for the article, very throughout and gives me the chance to try new things with the DAC.

 

[austinpop]

40 minutes ago, luisma said:

As an owner of a Terminator for some time already, and I purchased when you initiated this quest with the 1st part, I tried different upsampling methods, DAC's OS, Roon DSP and HQPe with NAA and in my opinion and without trying to detriment Roon's or the DAC upsampling, HQP by far sounds the best. I know you already mentioned in your article the reasons for which  you weren't using HQP but since I tried all that I wanted to provide feedback.

Thank you for the article, very throughout and gives me the chance to try new things with the DAC.

 

 

Thanks, that's great feedback. I know I can't test every use case, so it's good to hear from other CA'ers about aspects like this, that I couldn't test.

 

Can you be specific on what your upsampling strategy is with HQPlayer? Do you upsample everything to DXD, everything to DSD256, or do you do PCM-PCM, and DSD-DSD? What filters have you found to sound best?

[luisma]

Sure, without trying to hijack your column or review here it is.

 

This is what I have tried that it sounds good to me, please note that after some tones listening I came to the conclusion that after 16khz I can't hear anything, did some testing this week and maybe 16.5 or 17 khz tops but nothing more.

 

I have an Intel i7 NUC running Ubuntu (Jussi's kernel) with my own install of HQPE and Roon Server. Internet access is via wireless for streaming and wired unshielded ethernet connected directly to a fanless NAA core i3 running ubuntu minimal install by me as well running HQP NAA (networkaudiod) and Roon Bridge (not in active use). Both NUC and fanless PC are being fed by Uptone JS2 using the DC cables supplied by Alex C with the unit. From the NAA I use a custom made by Ghent Audio USB JSSG cable to connect the Denafrips Terminator which is powered by Audio Sensibility cable. The Terminator connects unbalanced XLR Wireworld Eclipse 7 to a Yamaha AS2100 SS AB integrated amp. Yamaha is connected via speaker cable  Anticables Level 3.1 to a pair of Klipsch P37F (Paladiums) and RCA interconnect to a Powersound Audio S3600 sealed sub.

 

My apologies but I'm still trying to accommodate the audiophile verbiage into my brain to describe things in an understandable way so I will use my own words here (Lehman's terminology) 

 

Trying Roon DSP Engine with Headroom Management -3 and Sample Rate PCM 384 and DSD 256 found the sound "hollow" / "not as full and coherent" as HQP. So at this point I'm discarding Roon's upsampling and proceeding to comment on HQP.

 

Trying HQPE, I read a lot about people preferring PCM over DSD, in my case is the opposite, for some reason it sounds better in my opinion listening to DSD, I tried DSD128 and DSD256 and I prefer the latter (as long as the filter runs on the core i7 with enough resources). When I was running PCM I found "closed form" with TPDF dithering to be my favorite filter.

With DSD I tried following some post "poli-sinc-mqa-mp" oversampling with DSD5 modulator and it sounds very good.

 

So at this point I do DSD mainly (my setting attached below)

 

Some last considerations:

 

1. The CPU taxing of PCM is less than DSD and will allow you to run other filters that might not be possible under DSD

 

2. I run into some issues reproducing music from Roon or HQP upsampling, after some tracks being played switching music or filters I got glitchy noises, slowdown of audio, high pitched  noise etc. Once I updated the Amanero firmware to CPLD_for_1080 and firmware 1099c (see below)  and added some packages to the linux install "sudo apt-get install libflac8 libasound2 libasound2-data libasound2-plugins" all the issues were gone and I can switch back and forth between filters with no issues. I am not sure if it was the firmware or the packages all I know it was very annoying and now it is gone for good. This issue was also present with Audirvana and J River so the firmware could be the main reason is gone.

 

3. Music for testing PCM and DSD

Supertramp - School

Caroline Campbell - Czardas, Skyfall

Caroline Campbell / William Joseph - Jar of Hearts and Piano Fantasy

Fat Larry's Band - Act like you Know

A-ha - Hunting High and Low (MTV unpluggeD)

Archie Shepp Quartet - true Ballads (The Thrill is Gone particularly is one of my favorites)

Of course Benny Goodman - Sing Sing Sing

 

4. I was using an Allo Bridge before but I am so happy I got it replaced by this Intel NAA, the CPU usage is 1% or less.

 

5. You may benefit with the use of a dedicated streamer and the I2S connections on the DAC (which may provide better clocking if I'm not mistaken than the USB connection)

 

6. Very important and I mentioned this on another thread, be sure Alvin lists the DAC on the papers as music equipment or something (he is actually already doing this) as customs may delay your order because the word "Terminator" doesn't click too well with customs on import documents  

 

I'm sorry if I overextended on my comments and description, this is about the Terminator and it is a great product.

 

This is just my opinion, I am not related in any commercial way (except that I'm a customer) with Denafrips, Vinshine Audio, HQPlayer, Roon or The Computer Audiophile. Many of you may have a different opinion or likes in music and sound and apologies if something I have stated is incorrect.

 

 

 

 

 

 

[firedog]

Roon Headroom setting: I have mine set at -0.5db, and that's enough to prevent 99% of clipping when up/resampling.  I think it sounds better than -3db.

[luisma]

6 hours ago, firedog said:

Roon Headroom setting: I have mine set at -0.5db, and that's enough to prevent 99% of clipping when up/resampling.  I think it sounds better than -3db.

Good tip, without headroom there is always clipping in my system, I will try -0.5 thank you 

[PeterG]

Thanks for another great review.  As I think I've said before, I love your head to head comparisons. At this level of performance, they are the only way to identify weaknesses.

[coot]

Regarding the name - did they realize using those letters they could call it "PAN-FRIEDS"?

[mansr]

5 minutes ago, coot said:

Regarding the name - did they realize using those letters they could call it "PAN-FRIEDS"?

More options:

• Fan's Pride
• Rap Fiends
• Fad Sniper
• Fair Spend

[davide256]

On 9/15/2018 at 4:15 PM, Ralf11 said:

which Cranberries CDs or tracks would you pick?

 

I am always looking to add to my test music list...

actually I think I just figured it out, looking at @firedog"s post (Thank you! )

 

 Icon CD compilation, track 3 "I can't be with you", track 1 "Dreams" 

Roses CD track 1 "Conduct"

 

these tracks normal or upsampled made my head hurt... but using headroom adjustment in Roon to reduce peak by -3 db reduced the migraine

 

[analogue]

3 hours ago, mansr said:

More options:

• Fan's Pride
• Rap Fiends
• Fad Sniper
• Fair Spend

Fena Drips (a new  coffee or tea  offering ,  based on fennel, avail at Whole Foods )

 

... or .. .

D ude,

E mergency

N eed

A ssistance ...

F uckin' audio

R eviewer

I s

P issin' me off 

S omethin' fierce

[AudioBang]

On 9/12/2018 at 12:42 PM, rickca said:

Excellent work, thank you!  Questions ...

 

Aren't you supposed to use NOS mode when using software upsampling?  Otherwise, aren't you hearing some interaction between the software DSP and the Denafrips OS mode?  Does ROON upsampling to 352.8/384 PCM bypass the Terminator OS mode filter?

It's really too bad you couldn't try PCM to DSD upsampling.  Does the Terminator OS mode do PCM to DSD conversion, or does it only upsample DSD input to DSD256?

 

It doesn't look like you didn't evaluate the Terminator with DSD input, is that right?  Isn't that supposed to be one of its strengths?

 

OK now I have to listen to the Apple announcements.

Was reading a 2013 post "Should I NOS-DAC ? where PeterST describes his DAC experiment for the ideal NOS DAC. He asserts that the misconception around NOS is not the NOS itself but the absence of the digital filter which is the source of ringing that plagues digital sound. From his experiments, to build the ideal "NOS" DAC, you would upsample the file before the DAC which takes all the aliasing distortion and raises it out of the audioband. From his  DAC test, you go from 30% harmonic distortion to .018% removing aliasing out of the audioband and you eliminate the filter ringing that is commensurate with the NOS design. 

 

I have an Antipodes server running Roon which will upsample a 44.1Khz sampled file to 8X [352.8Khz] or 16X OS [705.6Khz] . I'd love to see what that sounds like played back by a NOS filterless DAC to test PeterSTs assertion.  If I am following  his math correctly he asserts that the aliasing frequencies are determined by [for 352.8Khz OS input ]

 

 

1/2 sampling rate 352.8K =  176.4Khz + (176.4Khz - 20Khz) = 332.8Khz  

                                                   176.4Khz + (176.4Khz - 20hz) =   352.78Khz   

Audioband aliasing has been raised to between 332.8Khz and 352.78Khz

 

Now to filter these with a Jensen transformer if that will sufficiently reduce these levels before the amp. With a  -3dB  at 80Khz, -6dB at 160Khz, you would have -12dB at 320Khz using one transformer after the DAC .

Or perhaps one transformer between DAC and preamp and a second at the input to the amp for at least -24dB. The phase is virtually flat according to the Jensen Spec. I would like to know if the Holo Spring can accept OS inputs in NOS Mode  and what that would sound like as well. 

 

Rich M