ANALYSIS: A Comparison of DSD Encoders & Decoders (KORG AudioGate, JRiver MC, Weiss Saracon)

Hello guys & gals, I've seen the question asked of comparing various DSD conversion programs on message boards over the years but have never seen someone try to "compare and contrast" with objective analysis. Let's at least give it a try here. I don't promise unequivocal answers, but hopefully a decent stab at it :-).

Remember that any conversion between DSD to PCM is a "lossy" process. Therefore, it is of course preferable to keep PCM sourced recordings in PCM and DSD likewise if possible. There will be some compromise in the accuracy each time conversion happens. Even though the bitrates for DSD64 and 24/96 PCM may be similar, the modulation technique used to represent the resultant sound wave is different (as per the above image). The question of course is how much difference and if it's quantifiable.

This question of conversion is important because as I have discussed before, many if not most DSD releases have gone through some kind of conversion for the flexibility and ease of editing in the PCM domain. The most blatant examples of conversion are the ones sourced from 44/48kHz material, but I'm sure many others are from 96/192kHz origin but they would not be easy to differentiate from a DSD original.

I. Procedure:

I do not have access to DSD recording gear but I can convert recorded PCM to DSD and back again to see what the conversion process does. For example, the PCM test signal from RightMark Audio Analyzer can be sent through the conversion process and we can see what happens to it to get an idea of the amount of degradation. For these tests, I chose to use the 24/96 test signal which I feel is a very reasonable hi-resolution specification exceeding DSD64 in a number of resolution domains. I know that 88kHz may be better as an integer multiple of the 2.8MHz sample rate but I figure these days in a high resolution studio, 24/96 is probably the standard and is common as high-resolution HDTracks and Blu-Ray audio releases.

Here then is the general procedure:
- Take the 24/96 RightMark test signal.
- Convert the PCM to DSD using the various encoders.
- Reconvert the DSD file back into 24/96 PCM using each program.
- Analyse effects of the 2-way conversion and differences between the programs.

I decided to use 3 commonly available conversion programs for this test - something free, something a consumer can afford, and finally the professional "standard" made available to me thanks to a friend who runs a studio. This will result in a total of 9 final 24/96 WAV files to "measure" with the RightMark software (3 PCM-to-DSD encoding x 3 DSD-to-PCM conversion). The 3 software programs used for conversion are:

1. KORG AudioGate 2.3.3. This software is available free. All it takes to run conversions is access to your Twitter account so the software tweets each time a conversion takes place. Small price to pay for the ability to do the conversion I suppose. I used the default DSD encoding (DSDIFF / Stereo Interleaved / 2.8MHz / 1-bit) and decoding to PCM (WAV / Stereo Interleaved / 96kHz / 24-bit) parameters. I noticed that AudioGate will apply a +6dB gain with the DSD to PCM conversion (-6dB DSD is equivalent to 0dBFS PCM, not uncommonly this standard is not followed and +6dB gain can result in clipping).

2. JRiver Media Center 19.0.117. I've used this program before to test the PCM-to-DSD conversion playback last year. You can also save the resultant PCM --> DSD and the converse DSD --> PCM conversion files as well. The DSD --> PCM conversion happens in 24/352.8 so I used the best resampler I have - iZotope RX 3 - to convert back to 24/96 for final analysis using a steep filter at 48kHz. There is also no +6dB gain applied so the default volume of the PCM output file is softer than with AudioGate and Saracon at default settings.

3. Weiss Saracon 01.61-27. The standard DSD <--> PCM conversion package used by a number of places like Channel Classics, many HDTracks releases, Pentatone... Again, I just used the default settings for conversion to DSD (dff, CRFB 8th Order, 0 gain, 2.8224MHz, Auto channel mode, Smart Interleave, Enable Stabilizer). Likewise the conversion back to PCM was with default settings (WAV, 24-bit fixed point, TPDF dither, 96.0kHz, +6 dB gain, Smart Interleave).

II. Result:

As usual, I'm going to present the data as summary charts to start. There are 3 DSD encoders and the same 3 can be used to decode DSD, so let's just present them organized by the encoder used. When I say something like "AudioGate then JRiver", I'm referring to the use of AudioGate as the DSD encoder, then using JRiver to do the conversion back to high-resolution PCM to be analyzed by RightMark (remember, for JRiver's case, I also used iZotope RX 3 to resample from 24/352 --> 24/96).

AudioGate as DSD encoder.

JRiver as DSD encoder.

Weiss Saracon as DSD encoder.

As you can see, the first column in each table is the 24/96 RightMark PCM test signal with no conversion done. These would be the ideal numbers if one could measure a perfect DAC/ADC setup or in this case the results of perfect conversion.

The rest of the columns reflect what happens to the 24/96 PCM test signal as it goes through the DSD conversion and decoding steps. Remember that RightMark is analyzing the audible 20Hz to 20kHz spectrum only. As you know, DSD64 conversion adds quite a lot of ultrasonic noise if left unfiltered and this would result in some poor noise levels and lower dynamic range if frequencies >20kHz were analysed.

Indeed, various amounts of distortion and imperfections can be seen. On the whole, it's far from bad though. At worst, the cumulative noise level is still down below -120dB and dynamic range >120dB with each of these encoder/decoder pairs.

Comparatively, you can see the free KORG AudioGate encoder table above seemed to have the worst results in terms of noise level irrespective of what other software was used to convert back to PCM. This is followed by JRiver and then Saracon puts out some very fine numbers.

There's a similar tendency when comparing the DSD-to-PCM decoder used. In general, the JRiver and Saracon DSD-to-PCM conversions (columns 3 & 4) resulted in better measurements of noise level, and dynamic range than AudioGate (column 2).

Let's now have a look at some individual graphs to see what's going on - here's using AudioGate to encode PCM-to-DSD:

Frequency Response

Notice the different software used to convert DSD back to PCM all have different low pass filters. As expected, PCM (white) is flat all the way to 48kHz. AudioGate (green) uses a very weak filter and is only attenuated by <1dB at 48kHz, followed by Saracon. JRiver at the default "Safe" setting has a steep 24kHz 48dB/octave slope applied as noted here (you can change this if you want up to 30kHz cutoff, 50kHz cutoff, or filter turned OFF).

Noise Level

There's deviance from the PCM noise floor using AudioGate DSD conversion as you can see. AudioGate is more noisy at converting PCM to DSD than the other programs (as will be evidenced later). The noise floor also isn't as smooth as the others (interesting notch at 10kHz and 20kHz).

In comparison, let's have a look at the JRiver PCM-to-DSD encoding:

Frequency Response

Noise Level

The frequency response curves are similar to the AudioGate DSD encoding representing the respective low-pass filter settings of the DSD-to-PCM converters. The main difference is with the noise level. As you can see, JRiver as DSD encoder is able to maintain a very clean noise floor essentially equivalent to 24-bit PCM until about 13kHz before rising - and this low noise floor is maintained by JRiver and Saracon when reconverting back to PCM. The AudioGate DSD-to-PCM conversion in comparison has a higher noise floor throughout the audible spectrum - perhaps a higher level of dithering is being applied?

Finally, let's look at Saracon used as the PCM-to-DSD encoder:

Frequency Response

Noise Level

A clean PCM-like noise floor all the way to 20kHz is achievable after going through Saracon DSD encoding but this quickly increases thereafter. Again, AudioGate conversion to PCM results in a higher noise floor which I speculate is due to stronger dithering.

III. Conclusion:

Since DSD <--> PCM isn't a straightforward process (like say resampling in PCM), as expected, at a "microscopic level", conversion software does make a difference in resolution.

What is much harder to quantify is audibility. Those frequency response, noise floor, distortion, crosstalk results are all below what I believe are human thresholds of audibility and overall there is minimal change to the 24/96 PCM original signal within the audible frequency range. Remember, the results I show here are with both conversion to DSD and back again to PCM, not just a single conversion step. Yet, I have seen commenters on-line insisting that the conversion results in audible deterioration in sound (even with just a single step like DSD --> PCM).

Looking at these 3 software programs, we can say with some certainty from an objective perspective that Saracon PCM-to-DSD transcoding maintains the lowest noise floor from 20Hz to 20kHz. JRiver is also very good in this respect, while AudioGate's results are less accurate but obviously still very good and of questionable audible significance given that the difference is still below the measured noise floor of all except maybe the very best DACs.

Of course, one has to pay big bucks for Saracon compared to the free AudioGate software!

As for DSD-to-PCM conversion, the main difference appears to be where each program has decided to put the low-pass filter to remove DSD's ultrasonic noise. Of the 3, JRiver has the most conservative low-pass filter at 24kHz (with small notable effect beginning around 20kHz) by default. Saracon allows a bit more to pass through up to around 30kHz, and AudioGate allows essentially everything to pass through up to 48kHz with 24/96 sampling. The only other difference seems to be a stronger dithering algorithm (I'm guessing here) with AudioGate such that the noise floor is marginally higher than the others. Again, we're looking at differences way way down in the noise floor so it really should not be an issue. I think the real question is where you think the low-pass filter should be set for DSD64 material (ie. at what point is recorded ultrasonic signal drowned out by noise and not worth keeping?)

From what I see here, I'm quite happy that Saracon is used in most commercial releases I've come across for DSD-to-PCM conversion. Within the 20Hz to 20kHz audible spectrum, it does appear to be the best even though I highly doubt one could go wrong with any of these. Just remember that the steep low-pass filter in Saracon means there's nothing above ~40kHz and therefore no point buying a Saracon DSD converted file above 96kHz (88kHz is all that's needed).

Over the years, I've listen to original DSD and compared to PCM conversions at 24/88 using Saracon and AudioGate output level matched as best I could (using the TEAC UD-501, never tried formal ABX or blinding). IMO, it's tough to assess since you can't instantaneously switch from DSD to PCM. The PCM converted files sound good to me and I would not hesitate to archive the DSD64 library as 24/88. Whatever difference has always been subtle at best (despite claims from the DSD faithful that somehow DSD sounds much better). I suppose it's possible that different DAC devices could also sound different depending on PCM or DSD input.

Has anyone out there done an ABX or other controlled listening test with DSD-to-PCM conversion? Would love to hear of your experience and preference... 

---------------

Rant of the week...
In the high-fidelity audio world we've often discussed the ills of severe dynamic range compression (DRC). I'm just going to go on my soapbox for a couple minutes and complain also about the ills of DRC for soundtracks these days... Notice how LOUD TV shows have become lately? A couple years ago, I tried watching NBC's Hannibal. Not only was the pacing terrible, meant for folks with ADHD, but the audio was so annoyingly grating that I could not tolerate more than 3 episodes. (I don't know if the series improved after those 3 episodes...)

More recently, I've become annoyed by the recent Cosmos: A Spacetime Odyssey hosted by Neil DeGrasse Tyson playing on Fox and National Geographic Channel. I mean... COME ON PEOPLE! This is a science program. This is a documentary (with some science fiction entertainment thrown in). WHY DOES IT HAVE TO BE SO LOUD? It's like there's no subtlety left... No opportunity to whisper... No opportunity to wonder... No opportunity to enjoy the eye-candy of some excellent CGI graphics without the blaring of some "majestic" soundtrack through many parts of the show. Aren't the ideas being presented supposed to be what it's all about? But yet at times, the narration gets muddled by the background audio.

While I can still enjoy Cosmos 2014 with my kids for the topical presentation, I'm left wondering how much better it could have been to allow the dialogue to take center stage and the background soundtrack to accentuate the emotional impact instead of being ridiculously front-and-center as if I'm supposed to watch this program on a tiny smartphone screen on the subway (maybe that's the target audience!). As usual, it's hard to know who to blame - is it the sound engineers working on this series behind the mixing console or the folks manning the TV station transmitting the signal running it through their compressor? Unfortunate.

I'll end with a quote from Carl Sagan. Certainly worth contemplating when reading comments posted on the Internet in general... (Not just as audiophiles.)

"We live in a society exquisitely dependent on science and technology, in which hardly anyone knows anything about science and technology." Carl Sagan (1989) [good article BTW]

I wonder what Mr. Sagan would think about the current state of affairs regarding the level of understanding of science in our society today. I suspect if he were still alive (he died in 1996), he'd be impressed by the access to information and interconnectedness we have these days through the Internet. That's not necessarily saying a lot though about the level of understanding.

Still a great read after all these years... Originally published 1980.

MEASUREMENTS: Nexus 7 to Audioengine D3 (A "Kinda Portable" Audiophile Playback)

Okay, for fun, I thought I'd grab a few measurements of something... Somewhat... "Portable" :-)

What you have here is my Nexus 7 tablet connected to an "on the go" cable (5" male microUSB to female standard USB, off eBay - pack of 3 for $10) --> Audioengine D3 DAC/amp --> Sennheiser HD800. Unfortunately, The D3 would not power up consistently when plugged into the Nexus 5 smartphone since that would have been even more portable! Looks like the D3 demanded more power than the Nexus 5's USB port could deliver.

I got USB Audio Player Pro software for the Android in order to get the USB DAC working. Unfortunately USB Audio Class 1/2 devices are not supported by Android by default. You can see the basic interface above (I happened to be playing some Bruno Mars Unorthodox Jukebox). It recognized the D3 without a hitch. It sounds the same to me connected to the HD800 like the other machines I tested the Audioengine D3 with last time so no need going into any subjective evaluation here.

I was more interested in whether the objective data showed any difference between this "mobile" set-up compared to the laptops / desktops.

RightMark Results:

Remember the clipping at 100% with the Audioengine D3. All the measurements are done with hardware volume attenuated to 92%.

16/44:

Frequency Response

Noise Level

THD

IMD

24/96:

Frequency Response

Noise Level

THD

IMD

The first column in the summary tables is the Nexus 7 + Audioengine D3. The second column is the ASUS Taichi laptop connected to the Audioengine D3. Finally the 3rd column is the Nexus 7 natively without using the external USB DAC.

As you can see, there is no substantial difference whether the ASUS laptop/ultrabook or Nexus 7 was used in either the 16/44 or 24/96 test case. The DAC determines the final audio output, not the source "transport" device.

The Audioengine D3 is substantially better as a DAC of course. As you can see, the Nexus 7 + D3 easily outclasses the native Nexus 7 audio output off the headphone jack. It has a flatter frequency response with lower noise floor measurable even at 16/44. The difference is more evident with a 24-bit audio signal... The 24/96 frequency response demonstrates that the native Nexus 7 in fact is incapable of 96kHz sample rate.

Jitter:

Slight difference between the 2 J-Test measurements. The noise floor seems a little bit higher on the whole with the Nexus 7 when I ran this test making the peaks such as the 16-bit modulation pattern less obvious. Also a little cleaner around the 24-bit 12kHz primary tone with the Nexus 7. The differences are down around the -120dB level and would not be audible IMO.

Conclusion:

Okay... So this set-up isn't exactly a portable device. But it does demonstrate that you can get excellent sound out of a small Android device with the USB DAC that is objectively equivalent to using a standard computer.

Truth be told, I don't need "audiophile quality" sound in a portable device. I can't remember the last time I listened to my smartphone or iPod somewhere quiet with expensive headphones. On-the-go, convenience trumps everything else IMO so there'd be no way I'd bother with full sized headphones. If I did bring full-sized cans around on a train/plane/subway, it certainly would not be the expensive high-resolution open design headphones!

Day to day, I have my Nexus 5 phone with me. There are a few albums saved on the phone as MP3 or FLAC when I want to listen. Otherwise there are countless apps to listen to Internet radio and music streaming services which is what I listen to most. The days of the isolated non-networked portable music player are long over for me and probably the vast majority of music lovers.

------------------

For fun I decided to jump on the Geek Out bandwagon for a spin and see how it goes.

Looks like we're starting to see some user reviews coming out now, the first more formal one I see being this one at Part-Time Audiophile. So far it looks encouraging. As expected there are a couple of comments about the heat production of the 1W model. No comment about how it handles DSD and it looks like those guys are Mac-centric, so not clear how it works out in the Windows world (ie. drivers). For what it is, I find purely subjective reviews interesting to read but I would prefer something a little more than the usual "drive by shooting" ;-).

After a bit of humming and hawing, I decided to go for a blue "Super Geek" ($250, 720mW, 3.4Vrms peak) model as the best compromise for my case. My rationale is simple... The amplitude difference between 720mW to 1W is only 1.4dB (as compared to 2dB between 450mW to 720mW). I'm still concerned about heat production since this is a class A design which sucks full power all the time... Assuming the enclosure design for heat dissipation is the same, the lower power model should drop the running temperature a few degrees. I can see myself using this as a line-level DAC if it measures really well and tap the headphone amp feature only on occasion (the Audioengine D3 appears lighter and smaller for travel). As a 'standard' USB DAC, this also means it'll likely be "on" all the time so I'd rather not have something too warm sitting on my table. Furthermore, with a laptop computer, an inefficient class A device means more power drainage. The $50 saved is trivial for this hobby and not really an issue.

Anyhow, once I get this, I'll let you know some results... It will be interesting to see how this compares to the TEAC UD-501 which has the same DAC chip and similar feature set (NOS, DXD, DSD, etc...). Hopefully it doesn't take too long to ship out - my understanding is that only the 1000mW "Super-Duper Geek" model is released so far.

Juergen mentioned having a look at the HpW-Works software package for jitter analysis. Might just do that although I remain unconvinced it makes any audible difference in 2014 especially with asynchronous USB. I have yet to see a good example of a decent modern piece of equipment where jitter can be shown as the culprit for impairing the sound quality.

Tonight's music:
Kodo - Mondo Head - Although I generally prefer the more traditional sound of Tsutsumi, this one not only sounds nice in stereo but fantastic in multichannel off the SACD!

Enjoy the tunes everyone!

MUSINGS: On Experts, Experience, and Opinions...

So last night, instead of going to bed early as I was supposed to, I decided to have a look at this interview of Allen Sides from Ocean Way Recording on TWiT.TV. As usual, Scott Wilkinson does a fantastic job with the interview and takes questions from the audience.

Obviously, Mr. Sides is a man of many years of experience and can speak authoritatively on MANY topics related to audio hardware, studio production, and historical anecdotes based on those years.

But some things bothered me. Around 16:00 there was talk about DSD: "somehow between making the recording and that SACD it doesn't sound quite as good as it should". Really? By 17:00, there's discussion about CD copies, different stampers sounding different, then an anecdote on Mariah Carey and how the pressed CD sounded bad compared to the reference. Okay... Maybe... How about someone ripping the disks and comparing the data integrity and talking about that? Given that Mariah was married to Tommy Mottola until 1998, this anecdote is now at least 16 years old so any hope of forensic assessment is long gone - does it still apply as a generalization these days?

Things get really bizarre by 26:00 - "I have never been able to even make a copy of a CD that sounds as good as the CD I started with... It always sounds worse". As you can see, Mr. Wilkinson was perplexed and commented that "it's almost like generational loss in analogue" that Mr. Sides is referring to. "Multiple degenerations"? Please...

I wished Mr. Wilkinson would have done a follow-up question like - "What if you bit-perfectly ripped that original CD to a computer - does it sound the same then?" "How about if you copy to a different hard drive, does it sound different?"

It's also clear that there are some limits to Mr. Sides' knowledge/experience which most of us in the hobby world would have no difficulty discussing. (34:30) Q: "What's your idea of FLAC encoding?" A: "I'm really not that familiar with it." Fair enough, a person cannot know everything.

Although I'm not that old at this point in my life, I have learned some things in my "travels" both personally and professionally. One which I hold dear is that no matter how much we can respect and trust the "experts" for their lived experience and knowledge, they (like us) are all just human. And as humans we all have idiosyncrasies and biases. In this case, I don't think it's much of a stretch for any of us who have spent hours on our audio systems, used EAC or dBpoweramp for ripping to ensure bit-perfect copies, to stand up with good confidence and tell Mr. Sides that he's just plain wrong about not being able to make a copy of a CD that sounds identical. The fact is that in more than 30 years of the existence of the CD format, there has been no evidence of this when variables are controlled for (eg. ensuring that bit perfect copies were achieved, the rater was blinded, etc.). If indeed "generational losses" were possible with digital, this would already be a well known fact and there'd be no uncertainty whatsoever! Moreover, if this were fact, it would change significantly how we deal with accuracy of our digital data (hey... how can I be sure that those numbers in my bank account are accurate?!). Experts can provide educated opinions, but ultimately they are just opinions and not necessarily fact. The same goes for his strange comment about SACD not sounding as good as it should between studio and the physical disk. What is more likely, that the digital data somehow mysteriously changed over time or that his own psychological expectations changed as the memory of the live studio event consolidated? (Assuming of course that the data wasn't altered by some mastering engineer along the way.)

In this world, there are many mysteries yet to be discovered and likely much we as a species will never know. But digital audio systems which are inventions of the human mind based on mathematical constructs and technologically engineered devices (like the CD) that ultimately changes the physical world (sound waves) were not produced by serendipity. I do not feel it's good enough that we should just shrug our shoulders and declare some experiences to have enduring truth like parts of this interview. That surrender to logic leads us into the realm of "anything is possible!" and ultimately the slippery slope on the path to "snake oil". This is especially significant in the impact on those already obsessive-compulsive and perfectionistic (ahem, like many audiophiles). Maintaining an objective approach hopefully allows a counterbalance to this. An opportunity to take a step back and question the things which seem to make no sense. An opportunity to explore reality with techniques and at times instruments of greater sensitivity than that which we are endowed with within this mortal shell. Although the human mind is the best "instrument" to perceive the beauty of music, accuracy of the reproduction chain is a different matter and can be detected by the use of objective techniques with obviously greater sensitivity.

It's fun listening to interviews like this and I certainly felt it was time well spent nonetheless.

MEASUREMENTS: Another Look - Audioengine D3 clipping at 100% volume.

The ability to interact with you guys over the last year or so running this blog has been vastly educational for me! Whether data gathering with the MP3 test last year or comments and suggestions with each post... In general I do try to keep up with comments if I can but after 80+ posts now, I apologize if there are some comments I've missed along the way.

The post today is thanks to the keen eye of "Solderdude Frans" and his comments & suggestions to the previous post on the Audioengine D3. As per Addendum 2 in that post, indeed there is indication that a 0dBFS signal is clipping with the D3. I had missed it due to the fact that I was looking at the square waveform without considering the possibility of clipping contributing to the shape observed. Also while calibrating for the RightMark tests, RightMark and E-MU were flashing red for clipping and I thought it was that the amplitude was too high for the E-MU 0404USB rather than considering the possibility that the clipping was actually from the D3 output itself.

So, as suggested by Frans, this is what a 0dBFS 1kHz sine wave looks like with no volume attenuation (ie. 100%):

Yup, the peaks are being clipped!

The point where the clipping ends is with the hardware volume control turned down to 92% (93% almost was good enough) in the Windows 8.1 control panel (software volume level in foobar stays at 100%):

I re-ran the RightMark measurements ignoring the warning about clipping to see what happens to the measurement results:

As expected, there's a deterioration to the measured THD and IMD results going from 92% to 100% with clipping - still low using the RightMark methodology but more than 10-fold worsening with the clipping in place.

Conclusion:
Interesting... Like the first sample AudioQuest Dragonfly reported in Stereophile awhile back, it looks like one needs to back off the Audioengine D3's volume setting a bit in order to avoid clipping. In the case of my sample, pulling down the Windows volume control to 92% did the trick.

Well, this fact obviously blemishes my impression of this little USB DAC as an accurate audio device. In practice, it's unlikely I will push the volumes to 100% but I wonder if this was by design. I find it hard to believe that the engineers did not check the clipping characteristics. Rather, my suspicion is that the engineers felt it was OK to allow the signal to clip a bit to increase the perceived amplification level. In a noisy environment, the extra amount of amplification of low level signal at the expense of distortion might be a reasonable trade-off that may not be too noticeable (possibly no problem at all if the music is soft and rarely hits 0dBFS) - of course, this would not be a "high fidelity" practice.

------------------

Seems like there are some ruffled feathers around the recent "Geek Out vs. others" measurements published by Light Harmonic. Good to see that the Geek Out is being designed with good measurements in mind; encouraging. I can see how other manufacturers can be a bit unhappy about all this... Although it's probably wise to have a 3rd party perform the tests rather than a direct LH release, I suppose if it gets the manufacturers thinking about better engineering and competition in this regard, that's a good thing. Curious that there was no frequency response measurement. I wonder if at 44kHz, the Geek Out does measure like a NOS DAC with some early roll-off in the high frequencies.

Ho! Larry somehow believes that "It's proven wrong to assume people could hear only below 20KHz" (see the comment section of that article). Ok. I suppose Ashihara's paper from 2006 could be used to argue this. But we're talking SPLs around 80 dB at 20kHz for most subjects as the threshold of hearing pure tones and there's quite a significant jump between 18 to 20kHz. Threshold for hearing pure tones isn't exactly evidence that this is beneficial for real music! Furthermore, if you look at the test subjects, they're aged 18 to 33 and the majority of these are young women! Good that hi-res can satisfy the golden-eared audiophile lady in our midst... Unfortunately it's not going to do much for the old boys I suspect ;-).

MEASUREMENTS: Audioengine D3 USB DAC / Headphone Amp.

In the last couple of years, we have seen a proliferation of small sized USB DACs. Devices small enough for laptop-sized portability aimed at the headphone user who wants a bit more power to drive better 'cans' and provide improved sonics than what's available through the laptop's phono jack.

I guess it must have begun with the AudioQuest Dragonfly back in 2012. Currently in the 2nd incarnation as version 1.2. Version 1.0 got quite a positive review in Stereophile among other magazines which has helped propel this class of device into audiophile acceptability (if not some mainstream approval?). With the Kickstarter success of the Geek (Out) last year, these devices continued to gather press.

To create a device like this can be difficult given the dependence on the USB port for power. I've certainly experienced first hand the noise pollution from plugging my TEAC DAC into the computer USB port and this being picked up by my Emotiva XSP-1 preamp's analogue passthrough. That was why I bought the USB-to-ethernet cable extender to provide some noise isolation that thankfully worked.

So, a few weeks ago as I was perusing the local computer store, I ran into a sale on the Audioengine D3 USB DAC. I figure, what the heck - at less than $180CAD, it'll give me something to play with and if it sounds good, maybe it'll accompany me on overseas trips with a good pair of headphones.

I'm not new to the Audioengine brand. I've been using their A2 powered speakers on my desktop for at least 2 years now and can certainly vouch for the build quality. As you can see from the picture above, the D3 package comes with the DAC itself which has 2 LEDs - white for power, and a blue one that lights up with 88 & 96kHz audio. It also comes with a good quality phono adaptor for 6.3mm(1/4")-to-3.5mm conversion. There's also a functional grey fabric pouch for the DAC - useful to prevent it from scratching other things due to the "metal injection molded" aluminum case which makes the device feel quite solid in the hand.

I made it a point to take a picture of the text on the side of the box as well. As you can see, Audioengine likes to drop some hints on what's inside... The DAC is the AKM4396 (same as my Squeezebox Transporter), and the audio amplifier is the LME49726 opamp. The 2Vrms output level is good as standard line level RCA also. Asynchronous USB communication rounds out the specs. About the only concern I had was the 10-ohm output impedance on the device (see Tech Specs). This suggests that the DAC would best be suited for 80+ohm headphones (see here for details).

One of the things I always check with headphone outputs is whether it's loud enough with my AKG Q701 as pictured above (the Audioengine A2 speaker can be seen also). Happily, it does indeed power the Q701 reasonably well. I can listen to dynamic music without needing to push the volume to 100% which I'll speak more about later.  Also in that picture you see the 2 LEDs lit up. The inner one is power, the outer slightly bluer one is for >48kHz audio.

One observation is that because of the aluminum case, this little USB DAC can get pretty warm after awhile. Not enough to burn fingers of course, but mildly uncomfortable if you put it in a shirt pocket for example while on the go (cools down quickly though). Make sure to use the little supplied pouch!

One last thing before we get to some measurements. This little DAC does not require a driver in Windows or with the Mac (alas, I do not have a Linux machine handy). There's no ASIO driver for that TI1020B asynchronous USB chip inside and audio will be sent out based on the Windows Mixer settings by default. This is also how you control the volume. Remember that if you don't want Windows to resample, make sure to go in and manually change the sample rate settings in the "Sound" panel. As with most of these Windows driverless mini-DACs, the maximum samplerate is 24/96 and no DSD support (like the Dragonfly - I assume the Geek Out will require drivers for DSD and >24/96). IMO this is not a problem. More important than DSD and 24/192 for me is that it can handle 88kHz natively for my DSD-to-PCM conversions (which this does).

There is one other way to get a bit-perfect stream without ASIO in Windows - use WASAPI (Windows Vista onwards). The foobar WASAPI component worked well for me and I used that for all the measurements with the Windows machines. WASAPI will do the samplerate switch automatically with this device.

I. Objective Measurements:

A. The Basics
As usual, let's start with getting some charts and numbers out there for the product so we can get a sense of what we can expect from the sound.

First, the digital oscilloscope displaying the 0dBFS 1kHz square wave (connected to ASUS Taichi Ultrabook):

That's some nice looking square waves! (See Addendum 2 - the shape here looks better due to clipping.) Essentially no overshoot and no ringing. Remarkably precise channel balance which handily beats the ASUS Essence One. Gotta say that I'm quite impressed by this graph since this is "just" a small USB DAC. A healthy 2.83V peak (which would correlate to a 2Vrms sine wave).

Impulse response (16/44) off Asus Taichi Ultrabook:

Standard symmetrical linear phase filter used. Typical of sharp roll-off filters as seen in a previous post. Absolute polarity is maintained.

-90.3dB 16-bit & 24-bit Signal (off ASUS Taichi Ultrabook):

16-bits

24-bits

Have a look at the "Waveform Peeping" page for some comparisons. Better-than-90dB dynamic range resolution. Enough resolution to demonstrate the -90.3dB square waveforms created by flipping the least significant bit of a 16-bit signal. It's not as clean/precise as my full-fledged desktop DACs like the Essence One or TEAC UD-501. The waveform appearance is reminiscent of the Squeezebox Touch but better in that there's no DC shift and the channel balance even at such low amplitude is better. Again, not bad at all for USB power off a laptop/ultrabook.

B. RightMark 6.3.0 (newest free version now with ASIO capability)

UPDATE: See the latest update demonstrating clippling at 100% volume with this device! So long as you keep volume setting down a few clicks (~92% in Windows), the following test results apply.

So far so good. It's time to grab some RightMark results to see how it fares in terms of distortion, dynamic range, noise level, etc.

Since this is a portable device which is dependent on the computer for power, I think one of the most important questions to answer is how well it functions between different machines. So far, I have not seen any reports looking at the variability between machines in the objective tests with these small USB DACs. With that goal in mind, I'll be providing results with a number of different computers.

Setup:
Test machine + AudioEngine D3 --> shielded phono-to-RCA cable --> E-MU 0404USB --> shielded USB --> Win 7 laptop running RightMark 6.3.0

Summary of the machines tested (the laptops were described here in more detail although OS was updated in some cases):

1. ASUS Taichi 21 Ultrabook (early 2013): Intel "Ivy Bridge" i5-3317U (1.7GHz dual), 4GB, Windows 8.1 x64, 128GB mSATA SSD, USB3 port, foobar + WASAPI component

2. Apple MacBook Pro, 17" (early 2008): Intel Core 2 Duo (2.6GHz dual), 6GB, OS X 10.8.2 "Mountain Lion", USB2 port, 240GB SATA SSD, Decibel 1.2.11 player

3. Apple MacBook Pro, 15" (mid 2009): Intel Core 2 Duo (2.26GHz dual), 8GB, OS X 10.9.2 "Mavericks", USB2 port, WD 640GB SATA HD, iTunes 11.1.5 (AIFF files)

4. HTPC: Intel "Haswell" Pentium G3220 (3GHz underclocked to 2.5GHz & undervolted, dual core), 400W Seasonic fanless power supply, 8GB, Windows 8.1 x64, USB3 port, 240GB SATA SSD, foobar + WASAPI component

5. Server: AMD "Trinity" A10-5800K APU (3.8GHz quad), 700W Antec power supply, 16GB, Windows Server 2012 R2, 128GB SSD + bunch of HD's inside! USB3 port, foobar + WASAPI component

6. Workstation: Intel "Ivy Bridge" i7-3770K (3.5GHz quad), 800W Antec power supply, 16GB, Windows 8.1, 240GB SSD + 2 HD's. USB3 port, foobar + WASAPI or + JPlay 5.2B (newest 6/12/2013 trial, Kernel Streaming/ULTRAstream/DirectLink).

Except where stated with the MacBook 15", all the audio files were encoded in FLAC. The D3 was connected either directly to the laptop or the desktop machine's motherboard USB connector (ie. no hub in between).

For those unfamiliar with computer hardware, the above may seem too technical. I just want to make sure I covered the variables thoroughly. Let's just say that this is a pretty decent range of machines from laptops to desktops and the CPUs range from relatively slow by today's standards (Pentium G3220) all the way to a reasonably fast Intel i7. Macs are a bit older as I've transitioned away from Apple in the last few years for work. I focused on USB3 ports where available as the standard these days. I'll throw in both OS X "Mountain Lion" and "Mavericks", Windows Server 2012 R2, as well as the JPlay software in one of the tests using the most extreme settings.

16/44:
Summary of this USB DAC connected to each computer:

Frequency Response

Noise Level

THD

IMD

As expected, these days, any advertised high-resolution DAC should have absolutely no issue with standard resolution 16/44 signals. We're basically looking at ideal and identical results here with this little USB dongle on all the computer hardware platforms. Clearly, from the frequency response, the AKM4396's standard "fast roll-off" filter is being used instead of the "slow roll-off" option for the chip (you can see the effect here with the Transporter).

24/96:

Now we move to high-resolution:

From the numbers it's looking really good across the board. No surprises with any of the machines and minimal inter-test variability.

Frequency Response

Noise Level

THD

IMD

Notice the use of JPlay for one of the measurements with the i7 Workstation setup. As with previous testing, there's no evidence of difference even with what I believe are very high settings - ULTRAstream / DirectLink. Also the MacBook Pro 15" used iTunes and AIFF uncompressed files rather than the standard FLAC with the others - again, no difference.

Here's the stereo crosstalk graph. Remember that this is highly dependent on the analogue interconnect cable used so doesn't really say much about the DAC per se. In this case, I'm using a short 3' length.

C. Jitter
Okay, to finish off, let's see if the J-Test shows any significant sidebands.

Some minor variation in noise between the machines (some of which I'm sure contributed by the E-MU 0404USB) but it's all rather low level and of no concern. Jitter modulation pattern easily visible in the 16-bit test. As usual, these J-Test graphs are meant for analysis of SPDIF interfaces. Asynchronous USB converters almost universally do not demonstrate issues. As I have said before, I doubt jitter is audible these days given how good even basic DACs are in this respect. I would seriously like to see evidence to suggest a need for stuff like "femto clocks" suggesting any difference that would be audible!

II. Comparisons:

One of the fun things about using the same test kit over the last while is that I've got a little database of objective results to make relative comparisons. Here's how the D3 stands compared to my other units (24/96 only, there's no point comparing 16/44):

The first 2 columns are the Audioengine D3 playing off my ASUS Taichi Ultrabook either on AC power or unplugged with the battery. Notice there is no difference. Some folks feel that running off battery power results in less noise; I certainly did not see that here.

A reminder - the better stereo crosstalk values for the D3 are because of shorter analogue interconnect cable (3' vs. 6').

A look at the graphs:

Frequency Response

Noise Level

THD

IMD

As you can see, the outlier here is the Squeezebox Touch with the "poorest" performance (realize though that the performance is still excellent!). The SB Touch rolls off the bass a little earlier than the others (-1dB by 20Hz) and with the highest noise level.

Although buried in the various overlaid graphs, the D3 noise level is remarkably low around 60Hz (mains fundamental frequency).

The Audioengine D3 fits at an intermediate level between the SB Touch and the rest of the high-performance DACs. The noise level is just a hair above the larger full-sized DACs. There's also more ultrasonic roll-off which I believe is inconsequential.

III. Subjective Evaluation:

I spent about 2 weeks listening with this headphone amp / DAC combination in the evenings while doing some other work. It sounds very good. Headphones used in the evaluation include my JVC HA-FXC51B IEM, Sony MDR-V6, AKG Q701, and Sennheiser HD800. With an output impedance of 10-ohms off the Audioengine D3, it's recommended to have >80-ohm headphones to maintain a relatively flat frequency response. Of the group, the worse impedance match would be the JVC IEM at 16-ohms. Despite this, the sound was quite nice if a bit bass shy with an accentuated mid-range that worked OK for some pop and rock recordings I was listening to (Donald Fagen's The Nightfly, Cat Stevens Tea For The Tillerman). I took the ultrabook, D3 DAC, and these IEMs around one weekend while my son was doing his sports lessons and it did not seem too cumbersome.

The sound was good with the 63-ohm Sony V6. These 'phones have plenty of bass and I had a chance to listen to a few vinyl rips. For example on Falco's Rock Me Amadeus (1985 Canadian Release) vinyl rip, you can hear the limitations of the digital sampling used in those days with an elevated background noise during the vocal overlay in the 1st half of the song. Daft Punk's Random Access Memories vinyl rip sounds very good as well with some surface noise reduction.

As I mentioned above, this little headphone amp can power the 63-ohm AKG Q701's reasonably well but I do push them up to ~80% with softer well recorded albums. For example, Rachel Podger's Guardian Angel (Channel Classics 24/96) sounds fantastic through these headphones. The open design of the AKG's are great for classical music transparency. I also find them very comfortable for longer listening sessions. Timbre and resonance of the violin in the recording space were well rendered... Enough resolution to easily hear the performer's occasional toe tap and breath sounds :-). As a side note, I do not recommend downloading the 24/192 versions off Channel Classics since the material was sourced from DSD64 and converted with Weiss Sararon - there's a steep lowpass filter in place and there's no audio above ~40kHz so 24/96 is more than enough. I also listened to the large choral arrangement of Mahler's Symphony No.8 ("Symphony of a Thousand") (Antoni Wit & Warsaw Philharmonic Orchestra, 2011 Blu-Ray rip). Comfortable listening level for this louder album was at ~65%. Lovely conveyance of the intermix of solo and choral voices; tons of detail to be unearthed in this recent Naxos recording of the 8th. [BTW, the multichannel mix is excellent as well!]

I spent a couple nights listening to the Sennheiser HD800 with this DAC. These are 300-ohm headphones so should be good impedance-wise and due to its sensitivity, can be driven to an uncomfortably loud volume (for me anyway). The thing I appreciate about the HD800 is that they're good at pretty much anything! The bass is tight without accentuation and the highs sound clean and dynamic. I keep wanting to use the term "precise" about the HD800. They can be very unforgiving of poor recordings; especially those with accentuated trebles (it's worth trying out the "Anaxilus mod"). First up was some Guns N' Roses - Appetite For Destruction (1997 MFSL release). Very good definition and detail even with loud and "busy" tracks like "Welcome To The Jungle". Not exactly high-fidelity but can still be used to judge quality of reproduction especially in teasing out the individual parts and making sure the sound doesn't become "muddy". Moving on to another memory of the 1980's, INXS' Kick (2002 Rhino remaster) sounded very good as well... I've always enjoyed the intro to "Devil Inside" with the synth-percussion and electric guitar as we get into Michael Hutchence's moderately reverbed vocals in this "classic" INXS lineup at their peak.

Moving on to the audiophile female vocalist genre... About 13 years ago, I was in Chengdu, China for some travelling when I came across a copy of Cai Qin's (蔡琴) "Golden Voice" (金片子 壹) album. This is one of the best recordings I have heard. If all CD's were recorded like this, I don't know who would even consider the need for "high-resolution". Needless to say, the rendition out through the Audioengine D3 and Sennheiser HD800 was fantastic. Simple instrumentation with the voice beautifully detailed consisting of classic Chinese songs originally written back in the early half of the 20th Century. I suspect many audiophiles would call this DAC/headphone combination "analytical". I just call it brutally honest; what I personally look for in true high-fidelity.

Finally, I hooked this little DAC using phono-to-RCA plugs to the Audioengine A2 powered speakers in the picture above to have a listen... My daughter really wanted to hear the Frozen soundtrack. Yup, sounds good - all that Disney high production value. Alas, although I really enjoy Idina Menzel's "Let It Go", I can never shake off the vision of Elphaba in my head! Overall, it sounds as good as my usual desktop DAC (ASUS Essence One) connected to the Audioengine A2 in terms of resolution and soundstage (within the limits of these small desktop speakers of course). So far I have not hooked this up to my main sound system downstairs (inconvenient re-routing of analogue cables). I suspect it will sound very good as well.

IV. Conclusions:

1. I was pleasantly surprised by the objective measurements! For something this small, the measured results are almost as good as the better DACs I've measured. All this driven off a USB port. Impressive.

2. Great to see that a USB DAC is capable of essentially identical performance off a number of different computer setups. The machines are of various age, make (Intel and AMD), both USB2 and USB3 ports used. Despite all the hoopla around fears of poor power quality or potential of electrical noise, this little device has demonstrated to me that these fears are unnecessary - or at least can be dealt with. It sounds clean, the noise floor is excellent. No evidence of jitter issues with the asynchronous USB interface.

3. Following from item 2, although previously demonstrated, again in these tests there were no significant differences in the sonic output with this DAC between computers running different OSes (Windows 8.1, Server 2012 R2, OS X "Mountain Lion", "Mavericks"), and playback software (foobar, JPlay, iTunes, Decibel). Likewise, I remain unconvinced that jitter is affected by anything other than the hardware interface itself. A good DAC and bit perfect is all that is necessary for quality playback as far as I can tell these days.

4. Specifically regarding the Audioengine D3, remember to keep the fabric pouch handy so as to avoid scratching other things packed with this stick due to the aluminum construction and sharp angles. Also, it does get warm after awhile. However, on the plus side, the construction is very solid! It might not have the color-changing Dragonfly, but the little blue LED is good enough to know when sampling rates have gone >48kHz. At 2.5" long (slightly longer than the Dragonfly), it does stick out from the computer a little bit but is generally unobtrusive.

5. Yes. Subjectively, the Audioengine D3 sounds very good and has enough power to drive my AKG Q701 - this is better than my TEAC UD-501. Most of my subjective headphone listening was done with the ASUS Taichi Ultrabook and desktop i7 workstation. I did not notice any difference in sound despite the different underlying computer hardware.

I think the proliferation of small high performance USB DAC / headphone amps is a good thing. I appreciate the recent Tom's Hardware article on computer audio suggesting that there's generally no need for anything better than the ubiquitous sound chip in the computer. That's probably also true. There's generally no need for better sound especially on-the-go where the ambient noise is high. However, "native" sonic capabilities of laptops can vary as I previously demonstrated. What a small device like this brings to the table are more power for demanding headphones and uniformly good sound quality off the USB port for whatever computer one owns. When I'm on the road with a good pair of headphones, this device will do nicely for listening in the quiet evening in a hotel...

Time will tell whether mini USB DACs like these will maintain much popularity however. For audiophiles it's not all that inconvenient to carry or plug in, but I'm not sure if this would continue to capture the interest of the mainstream market (assuming the interest in devices like the Dragonfly and Geek Out represents mainstream acceptance). In time, it's possible that laptops will on average achieve better sound quality which would negate the need for this class of device. We've already seen quality improvement with support of high sample rates and 24-bits. Remember the days of noisy audio where you can hear the electrical interference from spinning disk drives or busy CPUs? Or when everything was resampled to 48kHz? Thank goodness standards have improved.

I hope this review provides a good glimpse into how the Audioengine D3 measures and sounds. Certainly it has given me some perspective into this class of device as new ones come out. I 'hear' that the Geek Out is shipping in small quantities. I like that it has 2 outputs (one with <1-ohm impedance) and can deliver more power. Excitement is certainly there and things can get a little silly when this happens (as witnessed by this hilarious video). For the more powerful 1W model, I wonder what heat production and battery drain will be like with portable devices (supposedly Class A?). I'm most curious about how the crossfeed algorithm sounds. To date I have not found a headphone crossfeed DSP I've ever really liked. To call the feature "Awesomifier" I think could lead to the butt of many jokes if it actually doesn't end up "awesomifying" anything. Also, I saw on a video that it runs in non-oversampling mode which is certainly possible given the use of the PCM1795 DAC like I showed in these TEAC UD-501 measurements. I don't know if running this DAC in NOS mode (digital filters off) is wise for something like this. We should be seeing some reviews soon I hope...

OK, enough audio geekiness for now! It's Spring Break for the kids. Time for family R&R in the week ahead. :-)

Enjoy the tunes dear readers...

Addendum:
Around Christmas time I wrote a little opinion piece on the next gen video game machines and how I decided to just stick with the PC. Looks like indeed some of the new multiplatform games are probably best played on good ol' Windows... Wasted a few hours on this over the last few days :-)

Big push for Titanfall on the XBOX One but I think it's pretty clear that the Windows version has better visuals (true 1080P), better frame rates, and excellent gameplay even on a modestly powered machine.

Addendum 2:

Frans,

Managed to whip out the oscilloscope for a quick look...

-3dBFS 1kHz square wave, 44kHz

Slight Gibbs undershoot and overshoot seen, so clipping indeed likely contributing to the idealized appearance of the 0dBFS plot above. Nonetheless, still looks really clean and balanced compared to others...