PREVIEW: Oppo Sonica DAC (Thanks Oppo for making it right... On the rationale for technical perfection.)

Usually, I don't post many "preview" type articles, preferring to give you guys "the goods" when I've had an adequate amount of time with a device (usually at least a month or so) including measurements and such. I'll discuss why I'm posting this earlier as we go along...

Well, this is what showed up at my door last week:

That's of course the recently released Oppo Sonica DAC. A "hi-fi" USB, S/PDIF, ethernet, WiFi, Bluetooth digital player with DLNA streaming capability, part of the Oppo Sonica "family" of audio devices. The only other device with this moniker currently being the Sonica Wi-Fi Speaker, capable of 24/192 streaming. Coming later this year apparently is the larger Sonica Grand.

Let's have a peek inside the box...

Intro:
As you might be aware, my last major DAC upgrade which I have been using as a "reference" of sorts for many of the measurements here has been the TEAC UD-501 since 2013. I still think it's a fantastic sounding DAC with great flexibility! It has adjustable filters including turning filtering completely off for "Non-OverSampling" (NOS) operation. It can decode up to DSD128 using DoP and native drivers. And the PCM input goes to 32/384kHz. Truth be told, I really don't need an upgrade but what the hey, it has been 3+ years and I was curious about the latest DAC chips... One of the current "kind of the hill" devices being the ESS Sabre32 ES9038Pro. As you can see from the link, there are all kinds of technical details to be impressed by: -122dB THD+N, -140dB dynamic range, multichannel (8) capable, various filters, customizable filter parameters, DSD512 capable, PCM 768kHz, etc...

Being value-oriented, the Oppo Sonica DAC (~US$800) caught my eyes as an interesting device worth evaluating as one of the newest devices. Here she is outside of the box.

Included: DAC, instruction manual, generic stock IEC power cable.

Simple front controls with On/Off switch, whitish-colored OLED screen, left control knob with stop points and push-button selector, right continuous volume control knob. Other than the single On/Off switch, there is no remote controller. The Sonica smartphone/tablet app does allow putting the device into an "off" state but not turn it "on" (you will have to manually push the button). It weighs about 10 lbs, feels sturdy in the hand, and it's got a nice brush metal all black aluminum exterior. The form factor is "long" on length and narrower than my TEAC UD-501.

Setup with the left control knob allows you to go through a few selections like whether you activate the digital volume attenuation for the outputs. The OLED can be set to full brightness, an intermediate "Dim" setting, and an off setting where it will turn on for about 5 seconds when samplerates change or it detects a new signal. There's also selection for pairing the Bluetooth and network connections. There is currently no ability to change the digital filters used.

Looking at the back of the device, we see the "business" end with the various connectors:

As you can see, this device was "Manufactured December 2016".

Good selection of inputs - USB-B, S/PDIF coaxial and TosLink, ethernet, and interestingly the AUX analogue RCA input for devices like a phono pre-amp output. There is WiFi which was easy to set up with the Oppo Sonica app available on Android and iDevices. There's also a Bluetooth 4.1 setting to pair with your mobile devices for audio streaming as well. Convenient. Both front and rear USB-A 2.0 connector for USB sticks and presumably hard drives (not tried yet). A ground lug if needed as well (likely to help with analogue AUX input noise if present) and +12V triggers for synchronized turning equipment on/off (since it lacks a remote, I suppose this is one way to control with other devices in your system). Detachable IEC power cable - as usual, I just used the included stock cable.

The outputs consist of a single pair of RCA single-ended and XLR balanced connectors. Notice that it does not have any headphone output which might be significant for some folks. During playback both RCA and XLR are active at the same time so you could connect one of them to a headphone amp if desired. Connectors appear to be of high quality and gold plated.

Installation was not hard. The Sonica Android app worked well in searching and finding the WiFi network. Also the Sonica app can be used to update the firmware, currently the official firmware is: Sonica-33-0422 / MCU-09-0306 / USB-0110. I also installed the Windows 10 driver 3.34.0. The firmware update was done over WiFi and did not take long (of course, better to use ethernet for updates just due to signal stability, but my WiFi is strong).

Subjective Impressions:
Given that I just received the device last week, I haven't spent too much time listening yet but have put in a few hours (no, I'm no believer in "break-in" requiring tens or even hundreds of hours!)... No question it sounds excellent. Even before putting the device to any kind of objective measurements, it's clear that this thing has an excellent noise floor especially connected over the XLR cables to the rest of my system. The device itself produces no hum or buzz; the internal toroidal linear power supply was silent even with my ears against the device.

I put it through its paces with a few older and newer albums. This DAC can deliver the goods in terms of detail retrieval. For example, playback of the soundtrack from Whiplash (2014, DR10) was excellent. Tracks with percussion are always good to hear the "speed" of sonic reproduction and transients. You certainly get this in the album right from the first track with the "snare liftoff". The album has a collection of brief segments from the excellent movie, soundtrack moments, and jazz band pieces. A full-range system capable of low bass recommended.

While we're on the topic of soundtracks, for some good 'ol pop fun from the 70's, check out the Guardians of the Galaxy: Awesome Mix Vol. 2 (2017, DR9) "mixtape". With a selection from ELO, George Harrison, Looking Glass, Cat Stevens, Fleetwood Mac, Sweet, Jay & The Americans... It really is about having fun and not worrying about audiophile minutiae. While it is a bit on the dynamically compressed side given that these are songs from back in the day, it's not extreme at least. Unfortunately, a song like "Brandy (You're A Fine Girl)" sounded a bit too "shouty" at parts. Oh BTW, there's David Hasselhoff doing a campy disco tune called "Guardians Inferno" - I dare someone to request this one at an audio show :-).

Going back into my oldies collection, I dragged out Santana's Abraxas (MFSL UDCD from 2008, DR11) which I had not listened to in awhile. Nice tonality from this classic rock album of 1970. The wind chimes on "Singing Winds, Crying Beasts" sound lovely through this DAC; lots of detail and spatial cues creates a nice "scattering" as if the chimes are placed all through the room. The iconic Santana guitar solos throughout the album sounded great placed front-and-center in the mix. Admittedly, although this mastering sounds good, there's no denying the limitations of the source recording with notable surface noise and a bit of the proverbial "veil" in the sound compared to modern digital recordings. For me, a good DAC should be able to faithfully render the album contents warts and all.

Finally, for some genuine "high resolution" listening, Rachel Podger & Brecon Baroque's Vivaldi: L'Estro Armonico (Channel Classics 24/96, 2016, DR13) was a treat. Twelve concertos that convey "musical rapture" with intensity, texture, pacing and most importantly energy. The various string instruments are beautifully overlaid and the DAC has no difficulty negotiating the complexity of the intermingling of instruments. I love the contrast with the slower tempo, more deliberate pacing of "Concerto No. 2 in G minor - Larghetto". There's a beautiful natural ambiance and decay of the notes in the recording space captured in the recording. As usual, well done Channel Classics!

I'll continue to listen of course. The subjective side sounds good... But let me get to why I'm publishing a preview this week.

Oppo... What are you doing?
I often start my measurements of any product with a look at the digital filters using a 16/44 impulse response and construct the "Digital Filter Composite" (DFC) graph based on Juergen Reis' work around the same time. It's all quite easy to do and literally takes a few minutes to grab the results. Usually, while I'm at it, I'll also have a look at the jitter measurements. Already, I have a basic idea of how this device functions objectively and can tell you that in many ways the performance is very impressive.

But before I get to the other stuff in a proper, full write up in the weeks ahead, I wanted to bring forth a couple of unexpected results.

The impulse response for the 44kHz filter they're using looks like this:

Looks like Oppo decided to go with a minimum phase steep digital filter setting... Fair enough. But here is the DFC:

Whoa Oppo, what is this!? The digital filter graph is obviously highly anomalous for a high-fidelity DAC. It's demonstrating some characteristics of intersample overload even with a -4dBFS signal. I have not seen this happen before in other DACs up to this point; even with inexpensive devices. As you can see there are also various ultrasonic distortions present especially what looks like a rather strong "reflection" of the 19 & 20kHz signal up around 63kHz for some reason.

Then there are the jitter test results:

That's unexpected for a high quality asynchronous USB DAC these days. We're seeing jitter sidebands in both the 16 and 24-bit versions of the J-Test. I also see the same kind of jitter anomaly in the 24/96 version of the 24-bit J-Test (ignore the 14kHz noise spike):

Ahem, Oppo... Considering the amazing performance of the BDP-105 posted recently which demonstrated excellent digital filter performance (although some intersample overload still present) and essentially perfect jitter test results, this is rather seriously underwhelming. Surely the ES9038Pro can objectively perform with much better jitter rejection that this! Even though the anomaly isn't necessarily all that audible, in principle, a new device released in late 2016/early 2017 in this price range should not have these issues after all these years of advancement in asynchronous USB playback.

My suspicion is that Oppo is not utilizing the standard filters for the ESS DAC chip and instead is doing some kind of low quality upsampling itself. Who knows, maybe it's borrowing the algorithm used in the Sonica Speaker which incorporates room compensation DSP? Very odd I must say. I presumed that this can be fully improved with a firmware update.

If they insist in using this type of filter and these measurements are actually "defective by design", then I would respectfully suggest that they incorporate the ability for us to change digital filter settings. They have already advertised this feature with the just released Oppo UDP-205 4K/UHD player based on the online manual. Considering the Sonica DAC's target audiophile audience willing to spend close to US$1K, I would have thought this feature would be a given and more of a benefit than for the video crowd! I would be much more happy with a standard, steep, linear phase filter like what the Oppo BDP-105 had than this current implementation.

With that, I sent a support request to Oppo on their website. Within an hour, I got a response and within a day, an updated E-mail with the pictures above was "sent to engineering" to be replicated in-house...

Oppo responds...
Within 5 days of the initial E-mail, Oppo made good on its investigation into this issue. They sent me a copy of their beta firmware (Sonica-33-0505B), and the machine was upgraded with a front USB stick:

With a 16/44 impulse signal, the filter is still minimum phase:

But look what has changed - here's the DFC:

Still a little bit of overload in the 0dBFS signal but I suspect that's inherent in the ESS DAC filters (and typical of essentially all DAC chips these days). And jitter?

Again, ignore the 14kHz spike there... It's an aberration from the ADC.

Ahhhhh... That's more like it!

Along with the new beta firmware, they sent me a package of 14 FFT measurements. Here are a few of them that I thought was particularly interesting comparing the Sonica DAC vs. Oppo BDP-105 jitter using Audio Precision gear:

Notice how little the jitter difference is comparing the ES9018 (BDP-105) with the ES9038Pro (Sonica DAC)! We basically see a tiny bit more low-level jitter at the base of the primary signal with the ES9018.

Thank you Oppo. Although preferably it would be better not having a customer bring the issue up, I am impressed by the quick customer service and engineering response including providing the beta firmware build and objective confirmatory testing! This is certainly the mark of professionalism with any good technology company. My faith has been restored in the Sonica DAC and ES9038Pro. Of course, it would still be nice to have the ability to select the various default ESS filters through the menu since many devices including the UDP-205 even has it!

We'll talk more about the Sonica DAC with more measurements in the next while.

Here's a question to the pure subjectivists out there... Assuming Oppo did not at some point issue a firmware update that addressed the jitter (let's not even mention the digital filter anomaly), do you think "Golden Eared" subjectivist reviewers would have noticed? I would very much doubt it despite all the lip service non-technical hardware reviewers give to the importance of jitter (and claims of being able to hear "immense" differences like this fellow).

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

For the record, even if Oppo did not come through with the new firmware, I can still improve the performance of the jitter and DFC results by doing my own upsampling to 768kHz. Something like this...

Those results were from the DAC before the beta firmware was installed. We'll talk more about how to achieve the above results another time - it's really quite easy and you don't need expensive stuff...

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

Archimago, why do you care about technical perfection?
Finally, I thought I'd just address the question above for a bit. As you can see, I've said that the DAC subjectively sounds great already, but yet I'm obviously quite critical about the amount of jitter and the unusual "digital filter composite" graph.

The answer is simple. I believe that hardware should perform to an ideal technical standard from which I can take control to personally tweak the sound to my standards. IMO, philosophically, not only do audiophiles strive for "high fidelity", but I believe the ultimate goal of the audiophile is to be able to subjectively tweak the sound we want to hear to our own specifications, not just be a passive recipient of what we buy. Intuitively I think we all know this to some degree... We customize/optimize the system with speaker placement, maybe add our own acoustic room treatments, and in the old days when more equipment had them available, we played with our tone controls and EQ's. Of course some questionable manufacturers want us to believe tiny shirt-button sized "transducers", weird devices that neutralize "RF", and kilobuck cables will do the "tuning" job to capitalize on these customization cravings of passionate audiophiles.

I've often seen writers ascribe a special reverence to various "rock star" hardware designers and we somehow are supposed to trust the opinions of various (frequently aging and aged) audiophile reviewers with supposed "Golden Ears". Yes, I can respect some amazing hardware designers especially those who put their energies into the physics side such as speaker design. But what I truly respect in the end is how close the gear comes to the ideal standard with low noise floor, excellent dynamic range, excellent frequency response, time domain accuracy, and freedom from unwanted distortions.  It really doesn't matter who does the job and whether I have faith in what they hear. It is the unnamed and often unheralded engineering teams that are the "heroes" these days. For today's article, I appreciate the work of the unnamed Oppo hardware/software engineer(s) who put that beta firmware together. These properties (low noise, excellent frequency response, etc...) are already quite obtainable in DACs - which is why getting the technical measurements right with a modern high-fidelity device like the Sonica DAC is important to me.

When a DAC can perform up to ideal standards, it essentially becomes like a very high quality TV screen or monitor upon which the audiophile can not just project the music he/she enjoys with the quality of a transparent window into the music, but also apply whatever fine-tuning of the sound he/she wants to make. Complex DSP like digital room correction can be applied and even changes like amplitude attenuation will still be rendered with great detail because the device has excellent dynamic range to draw from. Time domain nuances will not be "blurred" by the device's inherent jitter. And in the case of an oversampling filter, to be free from intersample overload means that we won't prematurely run into ugly distortions as the signal approaches 0dBFS when antialiasing is applied. Arguably, we can already say that with reputable DACs, even with small anomalies, transparency has long been achieved due to the limitations of human hearing.

In sum, high fidelity hardware not only allows us to extract all the detail in our music, but also affords the freedom to go beyond the "stock" sound with retained accuracy of reproduction. The realm of customization is the ultimate expression of subjective preference! We can become the designer of the sound we're after and take into account variables which equipment designers cannot because they did not design for our rooms, do not have our speakers and downstream gear (like amplifiers), do not have our ears, and cannot know our subjective preferences.

Have a great week everyone... Hope you're all enjoying the music!

Addendum (May 12, 2017):
For those looking for the beta firmware, check this post for link and instructions.

Addendum (May 23, 2017):
Official firmware with the improved jitter and digital filter performance now up (33-0511).

Addendum (June 10, 2017):
Measurements published now.