NOS vs. Digital Filtering DACs: Exploring filtering turned off, implications, fidelity and subjective audibility. (Recent BorderPatrol DAC chatter...)

Hmmm... Non-OverSampled waveforms - "accurate", "high fidelity"?

The waveforms above were captured with my RME ADI-2 Pro FS ADC at 384kHz. As you can see, on the left side, we see the waveform from the TEAC UD-501 DAC with the digital filter turned off (Non-OverSampling - NOS mode). In the middle, we have the TEAC's "Sharp" oversampling filter engaged. And on the right, we see the same waveforms captured from an old 16-bit Philips TDA1543 x 4 DAC with chips in parallel (supposedly improves noise level), a multi-bit NOS chip, without any oversampling.

Here's the TDA1543 x 4 DAC board. Typically fed with a 12V DC power supply.

Furthermore, we see that these are supposed to be -6dBFS sine waves at 1kHz, 5kHz, 10kHz, and 16kHz. Clearly, the "Sharp" digital filtered tracing shows us that the reconstruction algorithm does a good job in recreating the intended waveform compared to the NOS tracings with visual inspection. While the NOS output from the TEAC and TDA1543 look similar, there are differences we can make out in terms of the output from the TEAC appearing to have more overshoot at the transitions.

Something else to notice is that as the frequency increases, with fewer samples to define the sinusoidal wave (since we only have 44.1kHz sample points), notice that many of the frequency peaks do not reach the full amplitude of the sine wave with the NOS output. This correlates with the dip in high frequency amplitude response which we can easily measure:

Notice in the graph above, I've included both the RME ADI-2 Pro FS output in NOS mode as well as the old TDA1543x4 DAC compared to the TEAC UD-501 with "Sharp" digital filtering turned on. This shows us that regardless of old (TDA1543) or modern (RME ADI-2 Pro FS with AKM DAC) device used, the moment you turn off the digital filter, one creates this dip in the treble response which in the measurements above are down to about -3dB by 20kHz.

For fun, I can use the "Extremus" filter setting with my Raspberry Pi 3 B+ Touch to oversample the 44.1kHz signal to 88.2kHz (this DAC can handle up to 96kHz) and create a smoother looking sine wave:

Notice that the upsampled waveform is still rather jagged looking. This is because the TDA1543 is only a 16-bit DAC, plus we're only upsampling to 88.2kHz here. What is obvious is that now all the sine wave peaks are at close to full amplitude at 10kHz at least. And this improvement of frequency response can be confirmed with measurements:

The SoX upsampled 16/44.1 test signal to 88.2kHz now has a flatter frequency response but it's still obviously not perfect. Less than -1dB dip into 20kHz compared to -3dB without filtering.

Finally, remember that those jaggy, squarish NOS waves imply that there are high frequency anomalies. Here's a look at the FFT analysis up to 192kHz (384kHz sampling rate using RME ADI-2 Pro FS) of what is simply the 10kHz sine wave through the TEAC UD-501 with filter engaged, turned off, the TDA1543 DAC in NOS and when sent a signal from the Raspberry Pi 3 with SoX upsampling to 88.2kHz.

Obviously, the old TDA1543 device (bottom row) is much noisier than the TEAC DAC (based on much newer TI/Burr-Brown PCM1795 chips). We see with the TEAC in NOS mode the ultrasonic imaging distortions, the result of ignoring the proper reconstruction based on sampling theorem, compared to the typical "Sharp" digital filter available to most modern DACs. As for the TDA1543, notice the improvement with less high frequency distortion when we use even a 2x oversampling from 44.1kHz to 88.2kHz for this 16-bit DAC.

As fate would have it, I had started writing this article when I was reminded this week that Stereophile reviewed and measured the BorderPatrol DAC SE back in August of this year. Imagine, in 2018, a 'new' DAC designed around the geriatric Philips TDA1543 DAC chip released back in 1991 with fancy copper case and supposedly great power supply going for somewhere around US$1000-2000. Evidently despite the power supply and fancy chassis, the device measures rather poorly to say the least - go figure... The owner of the company even claimed "I was aiming to make an affordable DAC that sounds refined, clean, relaxed, fluid, colorful and human and free from the artificial ‘hi-fi’ sound that characterizes so many budget designs"! Since when did jagged edged reproduction of digital become anything but "artificial"!? Since when is this "refined" as claimed by the Stereophile reviewer? And how in the world does that price tag represent a "budget design" or "a very reasonable price" when clearly that jagged simple sine wave output can be surpassed by almost any decent modern DAC at US$200 or less?

I see in the last few years that there are other "high end" DACs based on the TDA1543 around. For example, there's this 16-chip Computer Audio Design 1543 MKII DAC proclaimed to be the "world's finest" at a mere £7,250.

Stereophile's "As We See It" penned by Jon Iverson for this month's issue brought out the BorderPatrol DAC again (I noticed some similarities in the philosophy expressed with my previous post here). In the article, Iverson quotes Jon Atkinson as saying "If this is a 'great' DAC, I'll have to hang up my measurements." While I may not agree with some of Atkinson's opinions, obviously I do not want to see the loss of measurements in Stereophile; if anything there should be more emphasis on the importance of measurements especially for controversial stuff like cables, questionable DACs like this, and digital tweaks. Over the years, I have opined that it is the objective measurements in Stereophile that actually make the magazine worthwhile reading as an archive for information rather than just subjective opinion.

IMO, obviously Mr. Atkinson has no reason to worry about his findings regarding this not-great DAC. When measurements clearly show that there are all kinds of issues with a device - channel imbalance, linearity anomalies, high-frequency noise, clearly poor low-level performance, high jitter, and terrible ultrasonic characteristics inherent in NOS devices - how can anyone who desires "high fidelity" audio argue that these are "good" (forget "great") characteristics?

(BTW, notice Figure 8 on the measurements page and the similarity with my capture of the -90dB signal from back in 2013 of the TDA1543 x 4 DAC I have. Despite the expense and fancy power supply, it certainly doesn't look like the low-level waveform has improved any!)

Apparently the subjective reviewer was not able to hear these anomalies. The question is should the subjective reviewer have been able to? Should subjective reviewers have the hearing acuity to appreciate the channel imbalance found? Should the reviewer have been able to tell that this DAC inverted polarity (a "golden ear" claim of significance)? Or that the DAC allowed all kinds of nasty high frequency noise to hit the amp and speakers which could actually have added distortions in the audible frequencies?

I do not want to single out this subjective reviewer because the fact is that this applies to all subjective-only reviewers regardless of expectations of them having "golden ears". When faced with the objective realities of the device, the reviewer and we as the audience need to come to terms with obvious discrepancies especially in clear situations like with these NOS DACs. Let's remind ourselves that audiophile reviewers are indeed human with idiosyncrasies and biases (of course!).

Let's summarize... IMO...

1. NOS DACs are not "accurate" or "high fidelity" when asked to reproduce typical 44.1kHz samplerate material. Clearly they have reduced frequency response and create all kinds of ultrasonic distortions. Personally, the only time I ever listen to my DACs in NOS mode are if I'm upsampling to 192+kHz using good software that does the filtering for me or if I listen to native hi-res files 96+kHz (maybe with piCorePlayer with SoX upsampling, JRiver with upsampling, or HQPlayer).

2. Regarding the aphorism that gets trotted out by pure subjectivists frequently: "If it measures good and sounds bad—it's bad. If it measures bad and sounds good, you've measured the wrong thing." (D. R. von Recklinghausen) Supposedly this is the unofficial "motto" of Stereophile that originated from before the early 1990's. At least the rest of that article reads reasonably well.

While this might be true in some circumstances especially in the old days, who ever said this was universally applicable? After decades, with the maturity of sound technology, assuming the collection of measurements was done appropriately, an alternate explanation is just as likely: "If it measures bad and sounds good, maybe your hearing isn't as good as you think." I think it would be hard to argue against this perspective when a subjective writer waxes poetic about stair-stepped squarish waves coming out of an old NOS DAC as if there is some special, non "digital-sounding" quality. From my perspective, these squarish waves are as "digital-sounding" as it gets! Another example of this might be how forgiving our ears/mind are to the effects of jitter; objectively, it takes quite a lot of timing irregularity before most people would be able to put their finger on an audible problem. Yet think of all the times various reviewers have claimed that cables of all things affect jitter significantly or manufacturers seem to think femtoseconds are audible...

When the technology is advanced enough to exceed perceptual thresholds, this is exactly what one would expect. The human perceptual system has been surpassed in many domains already. When this happens and there is a desire to differentiate devices, creative imagination becomes a greater factor and the mind projects descriptions of the devices fed by psychological expectations and biases than objective reality.

3. As I've said in the past, it's okay to be a "euphonophile". Some distortions could be subjectively perceived as "better". Examples might be the "BBC dip" or the "Harman Listener Target Curve" where non-flat frequency responses are psychoacoustically preferred. Likewise, higher even harmonic distortion with certain tube amps might sound better to some than an ultra-clean solid state sound. Our job as audiophiles is to recognize this. In my opinion, there is such a thing as an "objective euphonophile". A person who is honest enough with him/herself to say that he/she likes certain types of distortions in the sound, can objectively point to that which they desire, because it "sounds good" to them. This also means that the person accepts that he is not aiming for "high fidelity" sound. Rather, his pursuit is of a certain "coloration" to the sound that is possibly idiosyncratic rather than universal or ideal - by definition, this pursuit is not faithful to what was on the recording.

As such, a NOS DAC connoisseur could still say "I love the sound of old NOS DACs because I think the higher noise floor and ultrasonic frequencies through my amplifier and speakers result in a subjective 'lusciousness' that I desire even though it wasn't on the original recording". If a reviewer said something like that, I think many would applaud the honesty and congratulate the person for the level of insight and implied understanding of the technology. No need to waffle on right/left brain ambivalence, claim authority based on some kind of "experience", or flowery language to suggest some questionable redeeming quality (as we see in the Stereophile review of the BorderPatrol).

This same principle applies to those who love vinyl and turntables. One can appreciate the limitations to the technology. Absolutely, acknowledge benefits like album art and the physicality of vinyl. At times I like the sound from my LPs as well. But please don't keep insisting that LPs are "high resolution" media that surpasses true hi-res digital though! Let's be honest about the "fake news" in the audiophile world...

Audiophiles. Know thyself?!

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

Just want to let everyone know of a cool new tool by Yamamoto2002, our man in Japan :-) - WWShowUsbDeviceTree.

With this program, you can inspect a bunch of USB parameters of devices hooked up to your computer which of course includes USB DACs. Things like USB Audio Class compatibility, compatible bit-depth, power requirements... Here's what my ASUS Xonar Essence One looks like:

Yeah, totally geeky. Have fun if you're into this kind of thing. :-)

I was over at an audiophile friend's place the other evening. He put on the DSD128 version of Jacintha's Fire & Rain. Sounds good with nice dynamic range (~DR13) if you're into the audiophile female vocals genre and James Taylor cover songs :-|. When examining the DSD128 files, I noticed the noise floor was higher than expected for a high resolution production.

The Bohemian Rhapsody (DR8) soundtrack with its mix of remastered tracks and live recordings was certainly fun listening to that evening! And if you're interested in some older-sounding new pop music, be sure to check out Steve Perry's Traces (2018, DR6) - especially if you're a Journey fan back in the day :-).

Enjoy the music folks. Happy Thanksgiving weekend to the American friends.