MEASUREMENTS: Oppo BDP-105 RCA, XLR, HDMI, and other stuff for the record (like MQA CDs?!)...

Recently, I was looking at my directory of audio measurement data and realized that I forgot to put up the measurements I had collected on the Oppo BDP-105 Blu-Ray player that I borrowed from a friend in late 2016. I believe the BDP-105 has been discontinued in anticipation of the next generation flagship. The last version of the 105 was the BDP-105D which added DarbeeVision DSP video processing. This would likely not affect the audio side which is what I explore in this article and I'll obviously leave it to the videophiles as to whether this processing is beneficial or not.

As you probably are aware, this device caused quite the stir in the audio and home theater communities back in 2013 when it was first released. For a reasonable price as a universal disk player, it was capable not only of the usual CD/DVD/(3D) Blu-Ray playback but also SACD, DVD-A, and even HDCD decoding. All with flexible digital input choices - USB, TosLink, coaxial, and HDMI. For analogue output, it has single-ended RCA (stereo and multichannel) and balanced XLR analogue (stereo only) connectors. Already in 2013, Oppo had gained a reputation as a good brand with their previous models, but I think this device really put them fully on the map for audiophiles and videophiles alike looking for a reasonably priced (~US$1200 at the time) device that neither sacrificed features nor quality.

For audiophiles, the high-resolution audio output is handled by the pair of ESS Sabre32 Reference ES9018 DAC chips inside. According to the BDP-105 Wiki for those curious: "In the BDP-105's stereo board design, the 4 pairs of DACs in the ESS9018 DAC are allocated as: 1 pair for the RCA outputs, 1 pair for the XLR outputs, and 2 pairs stacked for the headphone amplifier." The other ESS chip is used for the multichannel RCA output.

Years ago, I published some results for this player (along with a DSD followup) and showed that it is indeed a very competent high-resolution device. That was back in 2013 when I was just starting this blog. Over time, I have incorporated a few more measurement and I thought it would be fun to revisit the device. This time, with my newer ADC, have a look at the digital filter in greater detail, explore the difference between the use of RCA and balanced XLR cables, and see if digital input with the USB-B interface differs from HDMI, and the S/PDIF variants... I suspect many of us have heard the BDP-105 over the years, so this might provide a nice opportunity for objective-subjective cross-correlation.

As you can see from the back of the device, this thing is very versatile with quite a number of input and outputs:

To begin the evaluation, the device was updated to the latest Oppo firmware - "Main Version BDP-10X-83-0715" released August 12, 2016. Firmwares are found here. I also installed the newest ASIO driver for Windows 10 - version 3.26 - found on the same page.

Measurement chain is my usual set-up consisting of:

Source (Microsoft Surface 3 i5 / Raspberry Pi 3) --> 6' generic shielded USB [ASIO in Windows] or HDMI [WASAPI in Windows] --> Oppo BDP-105 --> either generic 6' RCA or XLR --> Focusrite Forte ADC --> 6' USB cable --> Windows 10 measurement laptop

The output I'm measuring is the stereo RCA or XLR (left side of the rear panel). Not the multichannel outputs which in the picture above was what I had plugged the RCA cable into at the time to check if both the stereo connectors and multichannel RCA were active at the same time (they were). The power cable is a generic "hospital grade" cable (thicker gauge than the stock cord, not that it makes any difference). The device is plugged into my Belkin PureAV PF60 power conditioner.

As you can see, I made measurements with both a typical Windows laptop like the Surface 3 as well as with a standard Raspberry Pi 3 (not the more fancy touchscreen version described a couple weeks back).

Notice that the audio driver interface for USB is the latest Oppo ASIO driver in Windows, and for the HDMI interface, I'll use WASAPI through foobar2000. Default ALSA drivers in Linux for the Raspberry Pi 3 using piCorePlayer 3.02.

Part I: Digital Oscilloscope, Digital Filter, Impulse Response 

Here's the 1kHz 0dBFS 16/44 square wave output from the RCA connector on a digital oscilloscope:

So far looking good. Peak voltage of 3.22V correlates with 2.3V rms. Channel balance is excellent. The small, symmetrical Gibbs phenomenon at the edges of the square wave suggest that a linear phase filter is being employed.

And here is the impulse response of said linear phase anti-imaging filter:

Based on the pre- and post-ringing of the impulse response, we can surmise that this is a typical steep filter. Polarity is maintained. We can show how the filter acts using what I've called over the last couple years the "Digital Filter Composite" similar to Jürgen Reis' discussions in Stereophile:

Digital Filter Composite - Oppo BDP-105. Note that the 37kHz spike in the graph is from the Focusrite Forte device as previously documented, not the Oppo.

If you've followed the measurements posted on this blog in the past, normally I just measure the wideband white noise with "0dBFS peaks". On this one, I added the "-4dBFS peaks" version. The reason is that Stereophile typically shows only the "-4dBFS" version in their measurements. I believe that we really should aim to have devices where the digital filter is immune to overload (intersample overs), which is what the "0dBFS" tracing shows is happening with the Oppo BDP-105. In fact most "off the shelf" DAC filters will overload. One of the few times this has not happened is with the Light Harmonic Geek Out V2 using the newer ESS SABRE 9018AQ2M low power chip (can't tell whether this is a function of the DAC chip or the firmware applying some attenuation). Remember that with "loudness war" mastered music with low DR where volume is pushed way up there, it sure would be nice to have digital filters not add any more distortion (not that these masterings would be "audiophile approved" in any way, but since we're perfectionists, we might as well aim for perfection)!

Part II: RightMark Tests

Alright then... Time for some audio test signal comparisons.

16/44.1:
As usual, we start with the "standard resolution" CD-quality summary comparing a few devices over the years of various sorts from desktop DACs like the TEAC UD-501, to an ethernet streamer like the Squeezebox Transporter, to low-power USB DAC like the Geek Out V2, and to a mobile player like the PonoPlayer:

Numerically, you can see that 16/44 presents no challenge to any decent device. A 16-bit noise floor is no big deal these days. The far left column is the result using the Pi 3 piCorePlayer streamer sending a USB signal to the Oppo, the second column is the data from the Surface 3 sending the digital data through the HDMI interface.

Here are the graphs showing that some devices like the Geek Out V2 has a little more crosstalk, the PonoPlayer has high frequency roll-off. Otherwise, as is typical of different DACs, upper frequency response >10kHz varies a little between devices depending on slight differences like timing and filter parameters:

24/96:
Where measurements typically gets more interesting these days is of course when we start getting more demanding; like at the 24/96 bit depth and sampling rate. Summary of the same devices:

Okay, overall the numerical scores are not bad across the board. We see some differences between devices as expected. Looking at the BDP-105 scores, you see an interesting change in frequency response between the USB and HDMI inputs. When using HDMI, the analogue output is not as "ruler flat" and this can be seen in the composite graphs:

Notice that the "Surface HDMI - BDP-105 RCA" frequency response curve is similar to the purple Geek Out V2 tracing and not as flat as the Pi 3 USB plot. This is actually a real finding - the Surface 3 connected to the USB port follows what the Pi 3 shows, and likewise this pattern is also seen with XLR output (see later). Nothing to worry about since this is highly unlikely to be an audible difference (<0.5dB roll-off at 20kHz) but it does show the sensitivity of objective measurements.

Notice also that the PonoPlayer by 24/96 does not show strong high-frequency roll-off compared to the 44.1kHz data above since the higher sample rate is not affected by the customized minimum phase upsampling filter that Ayre utilizes.

24/192:
Finally, here's how the Oppo BDP-105 performs at an even higher sample rate. The Squeezebox Transporter is incapable of 192kHz so doesn't appear here.

As above, we see that the HDMI interface rolls off the highs a little earlier than with the USB interface. Again, this is quite evident in the graph:

Part III: RCA vs. XLR output

So, you might be wondering then if the XLR outputs are indeed objectively better than the RCA output. Here's the comparison:

Yup. What's rather obvious here is that balanced XLR output significantly improves the noise floor and in all cases except with the HDMI interface, consistently shows improved distortion characteristics. See the color labeling above which I hope helps simplify things. Notice the last 2 columns are with the S/PDIF input using my old CM6631A asynchronous USB to TosLink/coaxial digital converter measured back in 2013. Check out the composite graphs (I'll use the simplified labeling since whether I used a Surface Windows 10 computer or Raspberry Pi 3 made no difference):

Let's talk about this...

First we see a slight difference in the frequency response among the different interfaces. I've zoomed into the results (top left) to show that basically, the asynchronous USB interface results in flatter response whether using XLR or RCA output compared to HDMI and the S/PDIF interfaces.

As expected, noise floor is significantly lower with the XLR output than RCA. But there's something not-quite-normal going on with the HDMI-XLR output (light blue)! We're still looking at -130dB, but it's noisy comparatively with a 60Hz hum plus harmonics. This noise anomaly does not seem to affect USB or the S/PDIF interfaces.

Stereo crosstalk, as expected, higher with the RCA output.

Finally, THD+N sweep shows low distortion. However, the most notable difference is again the HDMI-XLR combination where distortion is higher than the others probably a reflection of the noise floor anomaly.

Part IV: Jitter

Here are the Dunn J-Test plots using the XLR output to achieve the lowest noise floor. Notice the small difference in J-Test results between asynchronous USB, HDMI, S/PDIF coaxial, and S/PDIF TosLink:

Looking at the graphs above, it's quite clear that the results from the BDP-105 are just excellent no matter which digital interface is used. However, I think it's quite clear that the HDMI 1.3 graphs are worse with slight skirting at the base of the primary frequency in the 24-bit test, plus slightly more sideband amplitude on the 16-bit plot.

I seriously doubt jitter would be audible given these excellent results in any event!

Part V: CONCLUSIONS

Although now discontinued, I'm guessing that the Oppo BDP-105 remains in the homes and minds of many audiophiles; surprisingly even with very recent positive "reviews" online at Dagogo (June 2016) and The Absolute Sound (March 2017!). How nice that subjective reviewers can appreciate devices with clearly excellent objective performance :-). Subjectively, I agree with these reviews. The BDP-105 is a fantastic sounding machine. The sound quality is clean, it renders the full spectrum precisely, is capable of very accurate channel balance with stable instrument placement in the soundstage. This is what I call "high fidelity"; a device that doesn't add or subtract to the source that is being provided for it to convert.

Albums produced with audiophile sensibilities and care like Manfred Honeck & Pittsburgh Symphony's Richard Strauss: Tone Poems SACD (Reference Recordings, 2014, DR15) sounded wonderfully dynamic with glorious orchestral nuances. To be expected, low dynamic range music like most modern pop sound flat and unengaging (won't name names here...). This is the price to pay for high-fidelity - a device that is honest will both allow us the opportunity to experience the magnificence of high quality productions and just as obviously present the grossest of deficiencies.

Let me now add a few specific conclusions based on the objective performance:

1. The digital filter does overload with my wide bandwidth white noise signal containing 0dBFS peaks. Ideally, I would love to see the day when all DAC chips have overload protection in the filter design.

2. For high resolution stereo, clearly the USB 2.0 interface is the best. Everything from frequency response, to overall noise floor, to distortion results, the USB measurements were excellent. I had no issues with using the Raspberry Pi 3 USB output and the ASIO driver in Windows 10 worked without a hitch.

This might be a surprise for some audiophiles. Those who hang around certain message boards probably have been exposed to various sentiments about USB being "noisy" or as a "general computer interface" is somehow poor for audio applications. Nonsense. As much as some folks complain, the objective findings are there for anyone who cares to look and understand that truly in more ways than not, "bits are bits" after all especially when dealing with an asynchronous digital interface. Whatever complaints are made about timing issues like jitter, those issues have been solved quite a while ago!

Of course there exists a small group of companies catering to the lingering concerns about USB audio. Notice that these companies almost never produce objective findings to show their equipment makes even a whiff of difference. I do believe it is important for consumers to remember that claims are not the same as facts, and that the onus of evidence needs to be placed on companies to explain their claims and of course the "professional" audio press should do a little of their own investigative reporting if they concur with marketing material.

3. The HDMI interface performed the worst. All relative of course, but compared to the USB interface, HDMI acted rather... Shall I say... "Idiosyncratically". Don't get me wrong, it doesn't sound bad by any means and it is the only interface with the BDP-105 for lossless multichannel sound (alas I did not measure sound quality through the multichannel RCA outputs). But for hi-res stereo, I can say that the frequency response was not as flat as it could be (not that important), there was a strange low level hum when measured with the XLR outputs, and distortion results were not as clean. Of course I am just testing a single machine and I suppose there could have been something wrong with this unit's HDMI board causing the very slight anomaly.

4. The S/PDIF coaxial and TosLink interfaces performed very well with the Oppo. I was impressed by the excellent low jitter performance! Not as good as the USB interface but better overall than the HDMI results. Often, TosLink can be found to have more jitter than coaxial in many devices. This was not the case in my testing with the Oppo BDP-105.

If you're wondering, "quad rate" PCM 176.4kHz and 192kHz were not supported by the coaxial and TosLink inputs on this device.

5. Jitter. What's there to say? With good modern devices like this (remember this machine was released in 2013), no concerns that the small amount detected with the HDMI interface and S/PDIF would be audible. It is rather strange the obsession with jitter in the press and among truly extreme audiophiles nitpicking about what amounts to picoseconds in essentially any decent digital device these days.

Honestly folks, to be consistent, if anyone feels timing irregularities causing jitter is a problem, then they really should avoid vinyl playback altogether. Likewise, if the audiophile press truly thinks that what amounts to tens of picoseconds or even hundreds of picoseconds of timing irregularity is that big a deal to differentiate the sound quality of DACs, then they really should not ever consider LP's as anything close to "high fidelity" based on this time domain basis.

6. XLR is superior to RCA. No surprise. Balanced analogue outputs and cables have superior noise rejection; easily 7dB lower noise level measured with the Focusrite Forte.

I think it's important to remember this because when you see reviews of all kinds of devices from expensive power supplies, to USB cables, to supposedly better computer streamers, to software supposedly capable of "lowering noise", to power conditioners, etc... see whether the reviewer is using a DAC capable of balanced output or simply connected with RCA cables. If the reviewer is using RCA output and suggesting that one spend hundreds if not thousands of dollars worth of upgrades, perhaps it's wise to suggest that he/she upgrade to devices with balanced circuitry first before recommending what would most likely be highly questionable tweaks.

Finally, I just want to put in a word about aftermarket mods. I have actually heard and measured another Oppo BDP-105 which was sent to a well-known modding company adding vacuum tubes to the analogue output stage, upgraded connectors, fancy "audiophile" components, improved power supply, etc... These mods are almost always highly regarded by the typical audiophile press as making the sound "better". They're also often very expensive (the cost of the mod often higher than the stock device). Let's just say that the analogue output of the device I tested changed from what I show above to something with less than 16-bits of dynamic range after modding... :-(

Yes, the modded vacuum tube BDP-105 monstrosity had a "warmth" and "smoothness" to it. But it was clearly technically compromised with an obvious coloration and to my ears just sounded too "relaxed" and "mellow". This is obviously not my preferred philosophy behind "high fidelity" or general spirit of what an audiophile strives for in trying to find the "absolute sound". I can respect "euphonophiles" for spending money to seek out this kind of sound (sure, if it makes an owner happy... have at it...) but ultimately I hope they realize that the device is no longer capable of "high-definition audio" and money spent on 24-bit music would be completely wasted unless there's evidence that the mod actually maintains/improves resolution!

Looking to the future, since late 2016, the next generation of Oppo players are now out beginning with the Oppo UDP-203 supporting UHD Blu-Rays for 4K playback. Later this year, the UDP-205 will be released. For those interested in audio only, check out the Sonica DAC with the latest ESS Sabre ES9038Pro, the successor to the ES9018 in this BDP-105. But seriously folks, given the remarkable performance of the BDP-105, at this level of performance, whatever gains in analogue output precision will be tiny - good luck in a controlled A/B comparison!

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

In other news, I hate to drag it up again but I see that now there's some kind of move to issue MQA-encoded CD's. Mon Dieu! That's getting rather desperate isn't it, MQA/Meridian?

As discussed a number of times in this blog, there's really no magic to MQA (see here, here). We can appreciate what it's doing in decompression of the least significant bits in an encoded 24/44 or 48 PCM bitstream and see the lossy ultrasonic reconstruction. We can even think about the digital filter it uses and model it as discussed here. No folks, there's nothing apparently all that special about the time domain performance as much as people like J.V. Serinus continues to try defending on a post like this recently parroting the "adspeak".**

Sure, with a 24-bit file, there are inefficiencies in the noise floor that can be exploited and used with some potential benefits if one believes there is value in ultrasonic frequencies (not my position of course based on last week's comments).

But think about this... What inefficiency is there in 16-bit, 44.1kHz CD resolution to take advantage of? Seriously, do we really want an encoding scheme to mess even with the 16th bit of standard CD resolution assuming this is an audiophile-level recording originally produced with something like 24-bit resolution and dithered down to 16-bits already?

Although we can't say for sure what this is about, we can imagine what it likely is up to! I wouldn't be surprised if all it is is some data stuck in the 16th bit to tell the MQA DAC that "Yup, this is MQA!" and just upsample using the "custom" MQA digital filter parameters with no actual unfolding of ultrasonic material (because there's just not enough data to decode anything worthwhile). My suspicion is that basically this is what a "renderer" like a low-powered AudioQuest Dragonfly would do (when outfitted with MQA firmware). The difference being that with the MQA-CD, it would just be playing with 44kHz data rather than the 88/96kHz software unfold like from TIDAL. Reminds me of old HDCD-encoded disks that light up the HDCD indicator but do not utilize potential benefits like Peak Extend. I wouldn't be surprised at all if HDCD actually had more potential to influence sound quality.

Like I said, it seems desperate reaching this point where MQA would stoop down to CD-level bitrates. I don't know about you, but I have not noticed many people proclaiming that MQA decoded TIDAL streams sound "revolutionary" in any way. It's not exactly setting the world on fire at 24/48, so why not dilute down to 16-bits as an attempt at expansion for product licensing or to keep MQA in the news cycle, right?

Sure, we can wait and hear how these MQA CD's sound. But because there is no magic to any of this, it's not hard to imagine the potential - or in this case, likely lack of potential by inference. Rather than as a move worthy of celebration by MQA proponents, I think this is more a sign of MQA's "last stand". Such is the natural history of hype such as Pono not long ago. As I had expressed years ago, there is little potential in "high resolution audio" to make a big audible difference. We're certainly not going to hear much of any difference with the music most people listen to; this is what Pono's failure taught us. Pono had more celebrity power behind it plus the novelty behind the idea of hi-res for the masses. MQA in comparison has neither the celebrity presence nor novelty (except really as a type of compression technique).

Carry on...

Have a great week ahead everyone. Enjoy the music!

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

** As an aside, notice in 1988 with a recently posted review of the DCM Time Frame TF1000 speakers, John Atkinson wrote this:

In any case, the question of whether absolute time integrity of the signal is subjectively a good thing is moot: no-one has yet shown that loudspeakers with time-coherent behavior always sound better; in fact, there are speakers with quite time-smeared behavior which perform well in the main area where time-coherent designs are supposed to be preeminent—that of stereo imaging. Perhaps only when all the other common problems are fixed will time coherence serve to distinguish the subjective sheep from the objectively identical goats.

I disagree with that statement to a large extent because these days we can use DSP to improve time domain performance. I have heard and measured the objective improvements. But if people like Atkinson can say the above about speakers - essentially thinking that time-coherent behaviour is not that important - then why make a big deal about MQA's claimed time domain performance in already highly accurate digital conversion? Seriously, even if MQA is capable of improving the time domain performance which I find doubtful, are improvements with speakers not potentially much more significant on orders of magnitude!? Looks to me like magazine writers say whatever they want - let reason and idealistic concepts of fidelity be damned. Let's be honest, they say what their buddies in the Industry and advertisers want them to say; and the buddy du jour just happens to be MQA/Meridian/Stuart.

Interestingly in that article, Atkinson points at Peter Aczel and The Audio Critic which is a strongly objective-leaning publication. As humans, we all have psychological biases... Does it surprise me that Stereophile wasn't as impressed with a speaker like this as much as perhaps Aczel might have been?

Speaking of Aczel, for those who have not read it, have a look at his article "What I have learned after six decades in audio" from 2015. I can't seem to reach the page on the official site, but the text is available on Reddit here. Like it or not, there is much wisdom in those words IMO. What a shame that much of his writings seem to be unavailable on the web!