General debate thread about audio measurements

[josh358]

If they're not ABX, they don't count, sorry and all that.

LOL

I will say that while I think ABX testing is limited (have you ever tried it? It's like handling a greased watermelon with a blindfold on), I'm a big believer in blind testing, for reasons that Floyd Toole has described. So I'm just working around practicalities as best I can. And even ABX testing has its role -- it can't tell you that something is inaudible as some believe, but it can demonstrate that something *is* audibly different, and generally, things can be ABX'd, including amplifiers, op amps, sampling rates, crossover notch distortion, and so forth. (Of course the Enid Lumley/Tice Clock stuff could never be, but, unfortunately, an ABX test can't demonstrate that.)

 

[Manbeard]

‘High frequency glare of multibit delta sigma dacs’!
Please save that shit for other forums.
Keith

Seriously. I'm sick of reading things like this on Audio *Science* Review.

 

[Shadrach]

(have you ever tried it? It's like handling a greased watermelon with a blindfold on),

I don't recall seeing any watermelons during the hundreds of ABX tests I did when I worked in avionics.
It's easy to forget that audio isn't restricted to a lot of obsessive recreational hi fi enthusiasts.
I think I'm going to leave you to your technical jargon at this point.............

 

[gvl]

Sorry if I violated some kind of forum taboo, I've enjoyed the measurements but I haven't contributed here before so don't know what they are.

That said I could amend this to the intentionally underdamped impulse response of the filters in the ESS chips, assuming that's what causes the issue (and I'm not sure that it is, although it certainly has an effect on the sound, which is why ESS does it even though they say it isn't technically correct).

As an engineer, I'm always curious about the relationship between measurements and subjective impressions, as the former are of little use without first establishing that correlation.

I have a Khadas Tone Board and a Topping DX7s, both are based on the same ESS chip. Both measured very good and similarly here. The former has the "glare" you're referring to, but the Topping is much easier to listen to. Whatever differences there are definitely not because of the ESS hardware.

 

[daftcombo]

I have a Khadas Tone Board and a Topping DX7s, both are based on the same ESS chip. Both measured very good and similarly here. The former has the "glare" you're referring to, but the Topping is much easier to listen to. Whatever differences there are definitely not because of the ESS hardware.

Really isn't there something in the measurements done here that can account for that on your system?

 

[josh358]

I have a Khadas Tone Board and a Topping DX7s, both are based on the same ESS chip. Both measured very good and similarly here. The former has the "glare" you're referring to, but the Topping is much easier to listen to. Whatever differences there are definitely not because of the ESS hardware.

I've read that some ESS-based DAC's don't have it. One possibility that comes to mind is that they're using different filters, either selecting them on the chip or their own, but I'm not sufficiently familiar with DAC design to know.

Here's what Keith L of Emotiva had to say about it. It jibes with my guess that it has something to do with the intentionally underdamped impulse response:

'The signal passing through a DAC is often altered in many subtle but significant ways... many of which involve timing.
However, most of those alterations are somewhat complex to measure, and VERY complex to interpret.
If you look at the analog output of several different DACs with a variety of input test signals, you will see significant differences.
Many manufacturers and magazine reviewers publish certain of these characteristics - like "impulse response" - and those measurements are in fact different for Sabre DACs than for most others.
Unfortunately, even when looking directly at the images of those signal characteristics, it isn't always obvious what a given difference represents in terms of SOUND.

'What is happening is that Sabre DACs have certain differences from most other DACs in terms of how they handle some sorts of impulse signals.
Because we're talking about subtle differences, and because no current DAC is perfect in this context, it may be difficult to suggest which ones are "more correct".
(Most of us here at Emotiva agree that the DACs we've chosen to use produce an output that we perceive as being more accurate to the original than Sabre DACs.)
However, if you compare oscilloscope images of certain test signals, the differences themselves are relatively easy to see.
And there are certain commonalities in how they are perceived by the majority of human listeners.

'When comparing a Sabre DAC to "an average DAC" from any of several other brands (with similar basic specs).....
- Some listeners fail to notice any significant difference
- Most listeners who notice a difference and LIKE Sabre DACs describe them as sounding "more detailed" or "more revealing"
- Most listeners who notice a difference and DISLIKE Sabre DACs describe them as sounding "etched" or "grainy" or "bright" or even "overly detailed"

'Regardless of which group you agree with, it seems somewhat obvious that the same general characteristics are being described by both.
Note that the surrounding circuitry has a major effect on how a DAC chip will sound... and different products that use the Sabre DAC chips seem to exhibit this characteristic sound to different degrees.
(However, it would be accurate to say that, in most situations where someone does notice a difference, this is the way they describe it.) '

http://emotivalounge.proboards.com/thread/52163/sabre-dac-glare-frequency-range

Interesting, no?

 

[Purité Audio]

I've read that some ESS-based DAC's don't have it. One possibility that comes to mind is that they're using different filters, either selecting them on the chip or their own, but I'm not sufficiently familiar with DAC design to know.

Here's what Keith L of Emotiva had to say about it. It jibes with my guess that it has something to do with the intentionally underdamped impulse response:

'The signal passing through a DAC is often altered in many subtle but significant ways... many of which involve timing.
However, most of those alterations are somewhat complex to measure, and VERY complex to interpret.
If you look at the analog output of several different DACs with a variety of input test signals, you will see significant differences.
Many manufacturers and magazine reviewers publish certain of these characteristics - like "impulse response" - and those measurements are in fact different for Sabre DACs than for most others.
Unfortunately, even when looking directly at the images of those signal characteristics, it isn't always obvious what a given difference represents in terms of SOUND.

'What is happening is that Sabre DACs have certain differences from most other DACs in terms of how they handle some sorts of impulse signals.
Because we're talking about subtle differences, and because no current DAC is perfect in this context, it may be difficult to suggest which ones are "more correct".
(Most of us here at Emotiva agree that the DACs we've chosen to use produce an output that we perceive as being more accurate to the original than Sabre DACs.)
However, if you compare oscilloscope images of certain test signals, the differences themselves are relatively easy to see.
And there are certain commonalities in how they are perceived by the majority of human listeners.

'When comparing a Sabre DAC to "an average DAC" from any of several other brands (with similar basic specs).....
- Some listeners fail to notice any significant difference
- Most listeners who notice a difference and LIKE Sabre DACs describe them as sounding "more detailed" or "more revealing"
- Most listeners who notice a difference and DISLIKE Sabre DACs describe them as sounding "etched" or "grainy" or "bright" or even "overly detailed"

'Regardless of which group you agree with, it seems somewhat obvious that the same general characteristics are being described by both.
Note that the surrounding circuitry has a major effect on how a DAC chip will sound... and different products that use the Sabre DAC chips seem to exhibit this characteristic sound to different degrees.
(However, it would be accurate to say that, in most situations where someone does notice a difference, this is the way they describe it.) '

http://emotivalounge.proboards.com/thread/52163/sabre-dac-glare-frequency-range

Interesting, no?

Just marketing pap.
Keith

 

[g29]

I have a Khadas Tone Board and a Topping DX7s, both are based on the same ESS chip. Both measured very good and similarly here. The former has the "glare" you're referring to, but the Topping is much easier to listen to. Whatever differences there are definitely not because of the ESS hardware.

From the Sabre DAC chip's spec sheet, it looks like the DAC maker has some sound customization options available to them.

"... Custom sound signature is supported via a fully programmable FIR filter with 7 presets. Residual distortion from suboptimal PCB components and layout can be minimized using ES9038Q2M’s unique THD compensation circuit, while chip-to-chip gain variation is minimized via a built-in auto gain calibration circuit. The ES9038Q2M SABRE32 Reference DAC sets the standard, SABRE SOUNDTM, for HD audio performance, typically consumes 40mW in normal operation mode (1.3mW in standby mode), and comes in an easy-to-use 30-QFN (3mm x 5mm)package. ... 7 presets or user programmable filters for custom sound signature. Bypassable oversampling filter. ... "

ES9038Q2M SABRE32 Reference DAC Spec Sheet

 

[josh358]

Just marketing pap.
Keith

No, I don't think so -- I read an interview online in which someone from ESS described the underdamped filters and admitted they were technically incorrect but they'd done it because it sounds so good. And overshoot is certainly easy to measure, and will have a sonic effect on transients. I'm just not sure that the overshoot corresponds to the "ESS glare" that you see mentioned so often. It would be interesting to hear from someone else who has worked with these chips.

 

[restorer-john]

But while the difference between good amplifiers is subtle, I've found that DAC's can sound dramatically different.

For me, it is the opposite. D/A converters rarely sound perceptibly different to me (unless one is a broken design) whereas amplifiers are more obvious in their differences.

In my opinion, it is the interactions with the loads (speakers) that cause them all to behave subtly different to one another. D/A converters and their attendant circuitry have a much easier life, with the following stages rarely offering anything but a completely benign load.

 

[RayDunzl]

It would be interesting to hear from someone else who has worked with these chips.

John Siau has posted here recently in the Benchmark threads.

 

[josh358]

I saw he's here. Maybe he'll chip in (sorry) and lend his expertise.

 

[josh358]

For me, it is the opposite. D/A converters rarely sound perceptibly different to me (unless one is a broken design) whereas amplifiers are more obvious in their differences.

In my opinion, it is the interactions with the loads (speakers) that cause them all to behave subtly different to one another. D/A converters and their attendant circuitry have a much easier life, with the following stages rarely offering anything but a completely benign load.

Interesting. Some of it may have to do with setup, e.g., my speakers (Magnepan Tympani IVA's) are a very non-reactive load and the only amps I've had here have low output impedance so wouldn't interact much anyway.

(I'd note though that this applies to "good" amplifiers operating in their linear range, I've had some cheap pro amps in here that sounded absolutely awful, and I didn't toy with anything extreme like SET amps or even with tubes.)

Meanwhile, I've spent the last week comparing DAC's and the differences were fascinating. At the most extreme, the difference between an ES9038 Pro DAC (the new Exasound e38 Mk II) and a Schiit Gungnir multibit B version was so obvious anyone would hear it the moment a track played. I also threw in the mix some other DAC's, an Exasound e28 Mk II (ES9018-based), a Lynx E22 card (which I believe is AKM based), my old Dragonfly Red (also an ES chip, forget which, which uses the minimum phase filter), and a few others. They all sounded different, though the biggest difference by far was between the delta-sigma DAC's and the Gungnir MB with its R2R chip and closed form filter. (When I listened to the closed form filter on HQPlayer and compared it to the linear and minimum phase filters, I could hear where some of the Gungnir's distinctive sound came from, but there was more to the difference than what the closed form filter did. And the converter chip of course is just part of the sonic signature of a DAC.)

 

[March Audio]

A manufacturer once told me that when reviewers review his equipment, they hear most of what he did. What I think differs is priorities -- the kind of music someone prefers, what they value most about sound. And also to some extent perspective, since no one can hear every piece of equipment under ideal conditions.

I've found personally that when it comes to subjective reviews, I have to calibrate the critics, and find those who share my tastes and priorities. But like the manufacturer I mentioned, I've found that they typically hear what I hear -- which is a good confirmation that I'm not imagining things if I read the review after I've listened myself -- but that they may have different sonic priorities or listen through very different systems. Once I'm familiar with a reviewer's preferences, I find the reviews helpful, albeit there's ultimately no substitute for hearing a piece of gear in good conditions.

I'd add that as an engineer, I've gone down the "objective" route several times over the years and been burned each time. The measurement suite in a magazine is too limited to fully encompass the sound of a product, the interpretation of the measurements too difficult. Just try listening to crossover notch distortion sometime, and compare its audibility to harmonic distortion. There's even a formula for the audibility of harmonic distortion -- as you might expect, the higher the harmonic the more audible it is -- making a single harmonic distortion figure next to meaningless.

Not that I can't tell something about the sound of equipment from the measurements -- but as I like to put it, where is the measurement for reproduction of depth? I actually know of many technical factors that influence that, including the timing of early reflections, the polar pattern, consistency of polar response ("power response"), baffle diffraction, group delay, stochastic timing in DAC's, amplitude response, even phase inverted third harmonic distortion. And there are no doubt other factors as well. So how are you going to infer that from measurements? You just can't, not completely, anyway.

Even where I know how something that is frequently measured, how do you translate the measurement into what you hear? Jitter, say -- jitter rejection is routinely measured, and I know what jitter sounds like. But I have no idea at what point I'd hear it. I'm sure that a DAC designer does, but I don't have that level of expertise.

I find measurements interesting because as an engineer I'm curious about how the correlate to what we hear and because they're important to design, and of course they can tell you something about the sound -- but ultimately, I know of no substitute for listening, despite all the vagaries and subjectivity.

Thing is you can ignore the measurement bit.

What you must do is compare blind and accurately level matched to get accurate assessments.

Every time I do this to audiophiles they can't hear the differences they otherwise claim.

 

[March Audio]

But personally, I prefer the Gumby, because it doesn't have the high frequency glare that multibit delta-sigma DAC's do. It sounds more analog and, more importantly, more like live music, on my system, anyway (but might not on less revealing speakers or on different kinds of music). So where's the measurement for that?

Where's the evidence that your subjective impression is accurate?

I have spent a long time subjectively listening to a soekris r2r dac against multiple different DS DAC's. There simply isn't the glare you mention. There are very subtle differences but no more than between any two dacs.

 

[March Audio]

I '

http://emotivalounge.proboards.com/thread/52163/sabre-dac-glare-frequency-range

Interesting, no?

Actually no because it has no supporting evidence. No controlled subjective testing that supports it.

 

[March Audio]

Try But while the difference between good amplifiers is subtle, I've found that DAC's can sound dramatically different. I mean, tone-control-level different, despite the fact that they're flat within a fraction of a dB. Comparing them has been a real ear opener.

That tells me that you you are not applying sufficient controls to your assessment.

 

[March Audio]

I really don't know. This time I'm annoyed to the point I actually plan to do a blind level matched test as the differences seem to be pretty obvious, but if I fail then it's my brain playing tricks on me.

I have performed a blind test between the TB and a Chord Mojo with 20 audiophiles. No one heard glare. A less controlled test was done with few people between it and a Seokris dac. No one heard glare.

 

[March Audio]

Seriously. I'm sick of reading things like this on Audio *Science* Review.

The science does not support its existence. Only audiophile anecdote.

 

[josh358]

That tells me that you you are not applying sufficient controls to your assessment

Perhaps, perhaps not. If a difference is subtle, and you want to be sure of it, I agree that a blind test is warranted. As you said, "What you must do is compare blind and accurately level matched to get accurate assessments."

That said, for my own purposes, I find that the problem of listening bias can often be circumvented by other means. For example, if I notice something, and then I learn that other listeners had the same impression independently, I tend to believe that what I heard was real. I always consider my subjective impressions provisional until I've received such confirmation -- at which point I consider them less provisional.

In this particular case, the thread from which I quoted an excerpt above begins with this:

'There seems to be some loose consensus that the Sabre-brand DAC chipsets have a "presence range" somewhere in the upper midrange to lower treble that sounds bright, despite measuring ruler-flat. Someone here on the Lounge brilliantly described it a "Like a photo in Photoshop where the contrast is turned up too much." KelthL seemed to think it was a deliberate artifact of the filter design to make the DAC sound "more present."

'That all said I concur that there is a fairly narrow band where the contrast sounds excessive (to me) on Sabre DAC products. My question today is "What, in your opinion, is that frequency range where the Sabre DACs sound too dynamic?"

'Now if you disagree with the premise, and like the Sabre DAC sound as is, then don't muddy the waters. But if you agree that Sabres sound bright, then what's your best guess as to the frequency span?'

http://emotivalounge.proboards.com/thread/52163/sabre-dac-glare-frequency-range

Exactly what I heard, and the effect was so pronounced anyone could have heard it. And I hadn't read this at the time I listened, so there's a pretty good parity check here -- good enough for me to lose interest in whether the DAC's differ, and become curious about why they do.

But if you wanted to eliminate any doubt, I agree that a blind test would be the way to go.

By the way, you took exception to Kal's impressions of the AHB2 and Parasound. As it happened, I *did* do a blind, level matched comparison between them (Parasound A21 -- Kal used an A31, which is similar) a couple of years ago when I bought my AHB2, and heard the same differences he did. And John Siau has ABX'd crossover distortion at very low levels; the Parasounds are high bias, but they'll still go into Class B above 10 watts into 8 ohms (John Curl says that you can improve their sound by biasing them until the heatsinks get hot!)

By the way, I don't think your experience with audiophiles is representative of what an experienced listener can hear, using the right program material and revealing loudspeakers.