Hi fi news measurements


#1

Now that the review has been out for a while, I wonder if Ted can comment on the noise and distortion measurements.

i wonder if the distortion they measured in the low frequencies has anything to the bass issues some are hearing.


#2
crabby said

Now that the review has been out for a while, I wonder if Ted can comment on the noise and distortion measurements.

i wonder if the distortion they measured in the low frequencies has anything to the bass issues some are hearing.

Howdy

If you look at the Stereophile review of the Playback Designs MPS-5, a DAC that’s well respected and one that I like the sound of and I personally wouldn’t mind owning, you’ll see similar noise measurements. The review is at http://www.stereophile.com/hirezplayers/playback_designs_mps-5_sacdcd_player/index.html, and the measurements are at http://www.stereophile.com/how-is-ted-coding-the-fpgaent/playback-designs-mps-5-sacdcd-player-measurements

It’s both Playback Designs’ and our contention that some of the measurement parameters that are traditionally used for PCM based DACs have less correlation to what we hear with one bit DSD based DACs than they do with PCM based DACs. We’d hope that more appropriate measurements are discovered as more one bit DSD DACs arrive on the scene.

Having a black background is very important to a good resulting sound and if the measured noise were a serious detriment then I’d expect that the first place it would show up would be as hiss and a not black background.

In more detail: The noise isn’t unexpected, but what matters is that it’s uncolored and the resolution of the DAC is much better than the noise floor. Unlike PCM dacs with active output stages or multibit sigma delta dacs with active output stages the passive one bit architecture we use is an inherently noisy process. Further reducing the noise generated by the high speed switching from one rail to the other would be cost prohibitive and as long as people hear a black background even when listening at high volumes the noise is low enough. In particular that noise won’t affect apparent frequency responses or more specifically bass perception.

I can’t say if the distortion measurements correlate with a few people’s complains about the bass. The rising distortion with rising levels in the low bass usually lead to a fuller bass sound which is the opposite of what some are reporting.

-Ted


#3

This reminds me of the early days of Wadia which allowed a little high frequency roll-off in exchange for phase coherency. The sound was strikingly better, but the objectivist clan had fits.

Subsequently, a number of CDPs provided user selectable filters. This way, one could choose better sound or better measurements - whichever one preferred.


#4
Elk said This reminds me of the early days of Wadia which allowed a little high frequency roll-off in exchange for phase coherency. The sound was strikingly better, but the objectivist clan had fits.

Subsequently, a number of CDPs provided user selectable filters. This way, one could choose better sound or better measurements - whichever one preferred.


I like that some Ayre equipment has a setting for listening and another for measurement :)

#5

Excellent!


#6

Hilarious.


#7

Damn! Why does my system sound like crap? OH…I left it on the measurements setting! blush_gif

T


#8

While I believe measurements have great validity, I do question the reliance on frequency response measurements based on a swept sine wave.

A sinewave is an elegant simple thing. But it bares no relationship to the complex waveforms of music.


#9
Elk said While I believe measurements have great validity, I do question the reliance on frequency response measurements based on a swept sine wave.

A sinewave is an elegant simple thing. But it bares no relationship to the complex waveforms of music.


I surely don’t mind tests but as you say a swept sine doesn’t resemble anything we care to listen to. Also I believe that phase response is very important, especially for transients, and few report more than an impulse response or a step response, which, tho they have that information in them are hard to read/understand for most of us. I’d like to see things like group delay for example. These things make less sense for PCM where most of the time there aren’t analog filters or IIR filters: the result would merely be “minimum phase” or “linear phase” (I can tell you which I’d gravitate to.)

For a PCM/DSD device like the DS which have chains of possible filters (or at least more than a few filters) a FR and impulse responces or group delay plots for each of the formats (like 44.1, 88.2, 176.4, 352.8, as well as DSD and double rate DSD) might be useful. Often we see the FR for 44.1, 88.2 and 96kHz, but less often the whole kit and caboodle.

The other problem is that sine wave tests assume a linear system. But we know that there are non-linear things going on, THD, IMD, etc. Sigma delta modulators are necessarily that linear either, some testing sensitive to the artifacts of SDM’s might help separate the wheat from the chaff in DSD dacs.

I guess I should say that I didn’t pay money for the complete Miller Audio Research tests of the DSD so I don’t know exactly what they did measure.


#10

Ted

If you come visit I can probably get us a day visit to the NRC.

Here is another Paul who went there often.

http://www.hifiplus.com/articles/psbs-paul-barton-takes-playbackhi-fi-on-a-guided-tour-of-canadas-nrc-acoustics-labspart-1/

The findings of Toole’s program were in one sense startling and at the same time remarkably useful: When the brand name, size, type and price of speakers were concealed from listeners, persons with normal hearing agreed on which speakers sounded pleasing and accurate --“musical” if you will-- and which ones were inaccurate and downright unpleasant. In the course of these tests, the notion of “golden-eared” listeners was largely dismissed. As long as listeners had a few hours of training on what to listen for (“fat” emphasized bass, strident harsh treble, muffled midrange, narrow “boxy” colorations and the like), they ranked good speakers and bad speakers the same way. And what was even more intriguing: If a speaker’s on-axis (in front) and off-axis frequency-response measurements could be kept as similar as possible, especially within a 15-degree “listening window” and especially over the midrange, the speaker would score highly in blind listening tests. While this is an oversimplification of decades of research at the NRC and by individual designers at Axiom and at other firms, it has proven to be extraordinarily predictive. Although no two speakers designed according to the NRC mantra ever sound exactly alike, there is nevertheless a remarkable congruence in what might be called “the Canadian sound,” and that is one of openness, transparency, “linearity” (smoothness), and fidelity.”

http://www.axiomaudio.com/NRC#


#11

Ted,

I hooked up the Direct Stream’s XLR outputs to a Ward Beck XTM4 VU meter http://ward-beck.com/products/metering_monitoring/xtm4 without music playing (silence).

It measured “noise” at a level of -38 dBu on the left channel and -40 dBu on the right channel.

An Audio Aero DAC hooked up to the same meter measures -80 dBu.

However, a Neutrik Minilyzer hooked up to each DAC measures about -80dBu “noise” on both the Audio Aero and Direct Stream.

I wonder what you think the Ward Beck meter is detecting?


#12
st50maint said Ted,

I hooked up the Direct Stream’s XLR outputs to a Ward Beck XTM4 VU meter http://ward-beck.com/products/metering_monitoring/xtm4 without music playing (silence).

It measured “noise” at a level of -38 dBu on the left channel and -40 dBu on the right channel.

An Audio Aero DAC hooked up to the same meter measures -80 dBu.

However, a Neutrik Minilyzer hooked up to each DAC measures about -80dBu “noise” on both the Audio Aero and Direct Stream.

I wonder what you think the Ward Beck meter is detecting?

It depends how wide of a bandwidth it measures over. For example a SACD bit stream will measure at less than -120dB over the audio band but perhaps -40dB over 0-100kHz. In a slightly broader sense different frequency weightings could come into it too.

#13
tpauline said Damn! Why does my system sound like crap? OH...I left it on the measurements setting! blush_gif

T


While I never want to make light of someone else’s work this idea of a measurement setting and a listen setting goes beyond anything reasonable and escapes me totally. The idea behind measuring is to attempt to quantify what we hear when music runs through a certain piece of gear. What am I missing here?

#14
Paul McGowan said
tpauline said Damn! Why does my system sound like crap? OH...I left it on the measurements setting! blush_gif

T

While I never want to make light of someone else’s work this idea of a measurement setting and a listen setting goes beyond anything reasonable and escapes me totally. The idea behind measuring is to attempt to quantify what we hear when music runs through a certain piece of gear. What am I missing here?


I think it’s an elegant answer to the good measurements not correlating with good sound problem. He’s showing that he knows how to make something that measures good and he knows how to make something that sounds good. I’m sure he doesn’t mind poking fun at people who think that the standard measurements correlate well with good sound.

#15

Hmmm. Ok, I can buy that. Thanks!