Friday, 24 October 2014

MUSINGS: Articles of Disservice - Stereophile November 2014

Over the months, I have put up posts critical of magazine articles (like this one) but I must say that the November issue of Stereophile was an "impressive" read.

I just wanted to bring up a couple of notable articles that I found interesting but highly misguided. I found these articles disturbing because they perpetuate the status quo or express an opinion that lacks constructive merit and I think ultimately do a disservice to advancing the audio hobby.

First, I think it's worthy asking ourselves, what is the "mission" of the audiophile publications? I looked around but was not able to find a page describing a "mission statement" for Stereophile. I'm sure the purpose must include informing, and educating the readership around new products. Reviewing albums to consider. Cover trade shows to let us in on what's "around the corner". The objective measurements embedded in the reviews which I have commented on previously are useful. But at the the end of the day, is there a basic mission statement? You know something catchy like "the waging of war against the tyranny of inferior audio" (Audio Task Force) as quoted in the recent NY Audio Show report. Considering that audiophile magazines are "for profit" companies, I think it's all the more worthy of consideration; especially these days where ads and the relationship the magazines make with the industry has likely become the main source of revenue.

Steve Guttenberg's "As We See It" article titled Communication Breakdown touches on the supposed ills of dynamic range compression (see Loudness War). He starts off with a provocative statement: "Classical and jazz notwithstanding, an awful lot of new music is highly compressed, processed, and harsh, and it's about time we got used to it." He then talks about some "superstar producer" not liking his suggestion to have 2 mixes (crushed & non-crushed). Then he reminisces about childhood tinkerings with AM radio and how he likes the background noise slightly mistuned (hey I liked it slightly higher pitched when mistuned but can't say I liked the noise, just more as "tone control"). Then there's a little history lesson on distortion in rock & roll. Then a little something about analogue distortion vs. digital distortion. Then he basically says he has learned to enjoy the music "through the grit". So... I guess it's okay then to accept compressed and distorted music (including many jazz and soundtracks these days).


Well Steve - hell no. You've learned to tolerate the grit and enjoyed the music - I'm happy for you. You seriously don't think that most of us have clutched "to our chests our 180gm LP's of Dark Side Of The Moon and Aja and rejected all the new music" do you? I mean seriously, the Loudness War has been raging since the mid-1990's and I doubt many of us music lovers have not been able to explore "new" albums and bands for the last 20 years - a full generation! Talk about resurrecting and perpetuating a ridiculous straw man stereotype of the "old audiophile" (hilarious that the magazine front cover contains the artwork for Gaucho). Do you seriously think that many of us haven't moved on from Dark Side or AjaThe issue is not that we're not "used to it", the problem is many of us are sick of it because we know it can sound better.

Over the years, we have had tantalizing tastes of what good masterings could sound like with new music. Remember the "Unmastered" mix of Red Hot Chili Peppers' Californication (DR10 vs. DR4)? How about the Guitar Heroes III rip of Death Magnetic (DR12 vs. DR3)? Recently I was discussing with some folks about the Canadian Promo of Beck's Mutations (DR11 vs. DR7). How about the much improved Steve Hoffman vinyl remaster of Stadium Arcadium? Whatever people may think about using a simple algorithm like the DR Meter, there is no doubt when listening with a high-end system, nuances can be heard and listening fatigue is reduced tremendously with these alternate masterings. Audiophile reviewers often talk about "veils being lifted", well here is a clearly tangible one which the press could speak out about but instead we have articles like this nihilistic justification of the degradation of sound quality in Stereophile of all places!

Now before I get labeled as some kind of "distortion hater" for rock and pop, surely I am not. I accept an artist's decision to add distortion, noise, Protools plug-ins of all sorts; heck, Autotune is fine (better than raw talentless singing in some cases). Some albums are 'lo-fi' by design, I get it although it's not the kind of music I prefer. But certainly this does not mean we need to endure digital clipping distortions and flattening of dynamic depth across almost all genres, does it? When it clearly gets so bad that on a high-end sound system, the terrible distortions become so obvious, are we to just tolerate it and not complain? If all recordings sound poor, why even bother with expensive gear at all? New artists (and producers who make their music) need to understand that a poor sounding recording damages the credibility of the artist in the eyes of many. And there truly are limited opportunities to make a good first impression. Audiophiles may be a small part of the music listening public, but we can be quite vocal in "spreading the word" among family and friends, and I bet we buy more music than the vast majority of music listeners.

One example I can think of is perhaps the "lowest-fi" of all the albums I have - Iggy Pop did a "great" job with his 1997 remix/master of Raw Power (DR1!). Okay, so apparently he wanted it that way. But even there, I would argue that when Kevin Gray remastered the mix in 2012 for the vinyl release (DR10), the result was obviously superior - you could at least easily understand the lyrics. I don't recall Iggy launching any accusations that this less compressed mix somehow destroyed his artistic vision. Assuming the original master recording isn't too far gone, I hope to see a day when the dynamic peaks can be restored in the music of Arcade Fire, The War On Drugs, The Black Keys, or The Killers. (You can have a listen to the vinyl releases to have a preview of what these sound like with less compression!)

The article ends on this: "Over the long term, sure - maybe sonic realism will be the next big thing... In 2025." It's said in the financial trading world, "they don't ring a bell at the top" (or the bottom). Well, ding ding ding, Mr. Guttenberg. I sure hope this article marks the beginning of a shift towards musical realism. 2025 is only about 10 years from now - that's not really a terribly long time from now and a typical time frame for trend changes considering we've been enduring needless dynamic compression for 20 years. I suspect the next round of remasters will indeed be back to a more realistic sound because they just can't squeeze the dynamics any more! Furthermore, I would argue that for the high resolution digital audio movement to gain any traction, it will have to be married to a remastering renaissance with better dynamic range in mind in order to demonstrate a perceptible difference. (Generally, I consider buying a 24-bit album with DR<12 to be wasteful of money and storage space and will check out DR Database before considering any such purchase.)

Already, I'm encouraged to see the recent U2 album Songs Of Innocence (DR9) being better than previous efforts (No Line On The Horizon DR6, How To Dismantle An Atomic Bomb DR5). I was even more surprised by Train's recent work Bulletproof Picasso at DR11 (good job on the music and mastering boys; better than California 37 DR6, Save Me, San Francisco DR5)!

If I am correct, and we do see improved mastering and more sonic realism, it will certainly not be thanks to Mr. Guttenberg and Stereophile for their lack of advocating for true high fidelity in this issue. Gross "communication breakdown" indeed. Leadership gentlemen... Please find some courage to speak out with conviction.

Well, that's the software bit... How about the hardware side?

Consider Art Dudley's "Listening" column. He really should have subtitled this piece "Rage Against the Double Blind Test". In it he quotes from Malcolm Gladwell, takes liberties with comparing blind testing with the "Pepsi Challenge" (as if this is somehow highly relevant), seems to consider objectivism in general with disdain, and apparently has a phobia of guys in white lab coats, engineers and "Daddy-with-a-clipboard" (hmmm, father issues?). Anyhow, there really is too much in there to comment on completely so I invite the reader to have a gander.

"Daddy" (sans clipboard).

I hate to break it to you Mr. Dudley. Sit down so you don't hurt yourself... Those engineer guys (with clipboards) designed your sound system. Yes, they (at some point) figured out how to "cut" sound into that spinning vinyl disk. They used some fancy maths to figure out your tonearm to reduce distortion as it traced out the grooves. They studied electronics to design circuitry for that RIAA compensation curve. They figured out how to make speakers with low distortion and even put them together with appropriate crossovers. They considered the theorems involved in digital sound sampling and spent time researching fancy ways of encoding and error correcting that shiny disk. They figured out how to amplify a little signal with low overall distortion so it sounds decent at multiple watts. They figured out how to engineer computer programs to store, sort, decode, and transmit sonic data. They even were so thoughtful as to make something called the "remote control" so you could sit on your favourite listening chair and not waste energy getting up to change tracks if you desire! I almost forgot, since you love the Playstation 1 so much, I'm sure some engineer came up with that lovely plastic game controller too. Shocking, right!? How's it possible that anything that can convey artistic beauty come from measurements, graphs, charts, scientific principles, and yes, the occasional blind listening test (oh the horror!)?

Well, let's try to answer a few questions raised by Mr. Dudley in the article shall we?
"Are we disappointed when our favorite analog recordings are remastered from 44.1kHz files rather than from the original master tapes, because someone convinced the company that "that doesn't make any difference"?"
Yes. I would be disappointed I suppose if I were looking for a "pure" analog pressing as a matter of principle. That's not however to say that just because it's analog, it's good or has to be better than a 44kHz master. Many old 1980's LPs were derived from ~44kHz digital source recordings and sounded great (Dire Straits Brothers In Arms and Telegraph Road, Don Fagen The Nightfly come to mind). Many reissued LP's since have used well mastered 44kHz source and sound great. Let me ask you this... Would you honestly be able to tell where the source came from if the mastering engineer didn't reveal it to you? Over the years, other than with objective means, has any of the subjectivists been able to come up with a list of SACD's that look like they're sourced from 44/48kHz PCM by listening? If they haven't been able to do so with high resolution DSD audio, how plausible is it that vinyl listeners would be able to do so with the remixing and application of the RIAA EQ inherent in making LPs (not to mention distortions like surface noise)?
"Are we disappointed when an otherwise good electronics manufacturer lowers its manufacturing costs by switching from hand-wired circuits to PCB construction, because the company was persuaded that "that doesn't make a difference"?"
No. What makes hand-wired circuits "good" and PCB construction "bad"? Why would a company be "otherwise good" based on this construction criteria? Why do you engage in such black or white dichotomies? Over the years I've seen some really shoddy "hand crafted" construction so that's nothing special in my mind - made worse when it costs more. For complex circuits, I would consider good quality PCB construction superior in fact due to the likelihood of better precision if made by a reputable company. I also see nothing wrong with being able to repair a complex piece of electronics with replacing the PCB. Furthermore, if the manufacturer can lower costs and pass the savings down to the consumer, what's wrong with that? Please, give us a concrete example where switching from hand construction to PCB boards with essentially the same design resulted in a clearly diminished sound quality that was audible but not detectable by objective assessment (which is presumably what the engineer used to persuade).
"Are we disappointed when a manufacturer of classic loudspeakers begins making cabinets out of MDF instead of plywood, because an engineer convinced the company that "that doesn't make any difference"?"
Well, I can appreciate solid cabinetry and am happy to spend more on it if it's what I desire. But again, if the cost reduction is passed down to the customer, what's the problem? Are you again being black and white declaring that solid wood is definitely better? Are you mixing sturdiness, longevity, and aesthetics with "better sound quality"?

Of course, Mr. Dudley answers all these questions with this gem: "Yes, of course - and, in every case, we have the most single-minded, hardheaded objectivists to thank for lowering quality across the board." Pssst... Mr. Dudley, please do not muddle up material/physical/luxury "quality" from "sound quality". Has sound quality not generally improved over the years in both the low-end and high-end thanks to engineering efforts? I cannot help but feel that this man is angered by the very people and scientific know-how that has given him such pleasure over the years.

Then there's this beauty:
"The trouble is, many of the loudest people in the skeptic community, by their own admission, appear to be less interested in investigating seemingly anomalous events with fairness and an open mind than in shooting down everything that strikes them as "woo-woo"..."
Are you serious!? So, do you mean to say that an "open minded" subjective reviewer plugging in a pair of cables and declaring that the $500/ft pair sounds better than the $100/ft one after a bit of listening is "investigating". The term "investigate" requires some form of process of systematic inquiry by definition. Over the history of the subjective audiophile press, how many PWB rainbow foils, StopLight pens, expensive cables, and dubious "room treatments" have undergone systematic investigation? How about the recent Synergistic stuff like the Tranquility Base or the recently demo'ed Atmosphere? If anything, there is a resistance to scrutinize and investigate the most specious of claims by manufacturers - exactly the ones that need to be investigated. The whole point of blind testing is to remove potential confounding variables and an attempt here to discredit controlled test methods like blind listening tests is essentially to say one is really not interested in serious investigation into "anomalous events" or to try to separate verifiable fact from opinion.

To end off. Consider this quote:
"Perhaps it's our warm-and-fuzzy emotionalism that keeps those blinkered objectivists coming back again and again: We foolish, insecure record-lovers wish, in our hearts, for Daddy-with-a-clipboard to tell us what we ought to and ought not to buy - even though, in our brains, we know how thoroughly, obstructively mistaken they can be."
Well, there is one thing he is correct in. There's a sense of insecurity and fear in some audiophiles as embodied in articles like this. A desire to split what is good and what is bad yet oddly try to present the idea that it also doesn't matter to him (see the "Get off my lawn!" portion) but in an aggressive and divisive manner. I haven't seen a scientific looking bogeyman come out to push a product or tell anyone what to buy or not buy in ages! Rather, it's primarily the subjectivists who announce recommendations, sometimes conducting uncontrolled listening "tests" declaring the next best speaker / DAC / preamp / amplifier / cable / room treatment / etc. as worthy upgrades and in the process fueling insecurity. Articles such as this seem to be unable to dissociate between sound quality, aesthetics and material quality as if they are one and the same. If anything, the objective test results damper the hype in many product reviews and provides a point of reference on what is objective reality in terms of the quality of the sound itself. I can see how some manufacturers might not like this and find it inconvenient.

Just remember Mr. Dudley, many engineers over the decades made sound reproduction not only possible, but fantastic! That's a fact. Another fact is that objective analysis whether by blinded controlled listening tests or instrumentation can and is used to determine accuracy of the sonic reproduction, that's sonic fidelity. And I think many audiophiles would want high fidelity as a primary objective in this hobby.

[BTW: Alright, who has been using a robot avatar to cause grief to Mr. Dudley!?]

-----

I come back then to where I started. What ultimately is the "mission statement" or goal of an audiophile magazine like Stereophile? Is it anything like "the waging of war against the tyranny of inferior audio"? I actually hope it is... But do articles like these advance audio quality or foster reasonable discussion?

-----

PS: I'll be away for the next few weeks. Don't forget to participate in the LP Needle Drop Test :-). Enjoy the music...

Friday, 17 October 2014

TEST: Archimago's LP Needle Drop Blind Test.

Hello everyone. Welcome to another "blind" test! Unlike the previous High Bitrate MP3 Test last year and the more recent 24-Bit vs. 16-Bit Test, this one is much more subjective and essentially for fun :-). Not that previous tests weren't fun, but the results of this one is more for the experience of having tried (that's at least part of the fun of this hobby I hope!)...

This time, we're looking at vinyl "needle drops"; digitized output from turntable setups. You can download the test file here:
ftp://lptest.dyndns.org 
Login: LP 
Test: test
Download the file "Archimago's LP Test.zip" (~125MB).

Alternate download site (thanks again Ingemar):
www.privatebits.net/archimago/Archimago's LP Test.zip

Within the ZIP file, you will find 3 sample FLAC files - A, B, and C. Each is a high-resolution 24/96 audio recording of 2 minutes, 2 seconds duration. I trust this is a long enough sample to evaluate the sound quality.

Each file was created with the same LP; a 2012 180gm remaster of Paul Simon's "Graceland", specifically the last track on side 1 - "Diamonds On The Soles Of Her Shoes" (Matrix / Runout (Side A): 88691914721-A RE1 20315.1(3) STERLING RKS) so there's maximal inner groove distortion potentially. This LP is completely "virgin", purchased new about a month before I did these recordings and never been played before these samples. The LP looked clean and about as perfect (eg. no warps, dents, fill defects, scuffs, scratches...) as can be. I used an air blower to remove any obvious surface dust. It was NOT washed prior to playback since I did not want to inadvertently add anything (nor can I say I removed any potential deep embedded dust from the factory).

The recordings were done in the following order with the following systems:

1. Roksan TMS (first generation), SME309 magnesium tonearm, Ortofon Cadenza Black cartridge fed into a Whest PhonoStage.20 preamp. Audio output was through a pair of shielded 6' RCA cables plugged into the preamp (no balanced XLR on the Whest). Total cost for this setup should easily exceed $5000 on the used market.

Took a shot right after the vinyl needle drop was done...

2. My own stock Technics SL-1200 M3D turntable. Denon DL-110 cartridge fed into the Emotiva XSP-1 preamp (Gen1). This cartridge is a high-output moving coil (HOMC, 1.6mV) so the preamp was set at standard MM load and 47kohm load. Output recorded off 6' of MonoPrice balanced XLR cables. Total cost of this would be <$2000 considering the Technics was bought used. The Denon DL-110 costs about $150-$200 new shipped.

3. My own stock Technics SL-1200 M3D turntable. Shure M97xE cartridge fed into the same Emotive XSP-1 preamp (Gen1). This is a standard moving magnet cartridge with 4mV output. Again, 47kohm setting used on the preamp. Output recorded off 6' MonoPrice XLR cable. Slightly less expensive than (2) above with the Shure cartridge <$100. Again, total cost of this system would be <$2000.



Technics setup calibrated using Baerwald geometry and SME tonearm using the SME-supplied protractor (Stevenson?). Vertical Tracking Force optimized with digital scale (Ortofon = 2.3g, Denon = 1.8g, Shure = 1.25g dynamic stabilizer disengaged). Azimuth set to perpendicular by visualization using a mirror protractor. I tried to optimize the "stylus rake angle" to something like 92-degrees (I honestly believe it makes no sense to spend too much time or effort on this unless extremely misaligned; vinyl thickness differences and mild but common surface unevenness will easily affect this):

eg. Denon DL-110 stylus on vinyl. Photo taken with Nikon D800 using Tamron SP 90mm/2.8 macro lens at f8, manual focus on small tripod. Angles estimated & measured on screen with MB-Ruler.
Analogue-to-Digital conversion was done with the Creative E-MU 0404USB device at 24/96 using Audacity to record. I can confirm with A/B switching while playing the digital file off a Squeezebox Transporter and the Technics turntable playing at the same time that the sound is essentially identical when volume matched.

For consistency, the digital files were:
1. Trimmed to ~2'02" in length.
2. First second essentially silenced to provide the same "lead in".
3. Last 2 seconds faded out to silence.
4. All file volumes RMS normalized to -18.87dB. Note that there are a few clipped samples due to unanticipated dynamics of this recording but should not impede evaluation.
5. All samples compressed to FLAC lossless and tagged.
6. Randomized to Sample A / B / C.

Other than the above there was no other processing done to the files. Nothing like noise reduction or ClickRepair for example.

As you can see from the DR meter log file (foo_dr.txt from foobar2000 plug-in) in this ZIP file, the music has a good dynamic range of 14dB. (Excellent LP remaster!)

Your task:

1. Listen to the 3 tracks in native 24/96. Which sample did you think was the best? Which did you think was the worst? (It's OK to also feel there's no preference or even if you notice a difference, think that it would make no difference to musical enjoyment.)

2. How much difference did you hear? Although the cost differential is much more between the Roksan vs. Technics setups, suppose your system sounded like the "worst" sample, would you spend $1000 to upgrade it to sound like the "best" sample?

3. Go and fill out my simple survey. All 7 questions are mandatory and I will have to delete responses that have not been filled out correctly. Should take <5 minutes. Please also let me know which computer/DAC and what headphones/speakers used. Also whether you used something like the Foobar ABX tool for evaluation.
(http://freeonlinesurveys.com/s.asp?sid=9qk9tpzbxq8uwlk539831)

As usual, thanks to all who try this out! It's not often that one gets to hear turntable setups side-by-side and this was about as "controlled" as I could find a way to do this to get a taste of what disparate gear could sound like. As usual, although this test is more subjective than previous tests, make sure to listen and not just look at waveforms in an audio editor, also, I think it's better not to discuss one's results until after the test is complete so as not to influence others.

I'm going to run the survey until November 30th - I'll be overseas for a bit so in case there are issues, please leave a note in the comments section if you run into any difficulties. Have fun with this!

Best regards,
Archimago

-----------------------
Disclaimer: I believe this test conforms to the spirit of "fair use" for copyrighted material for the purpose of education and research. The author derives no financial benefit from conducting this survey.

Friday, 10 October 2014

MEASUREMENTS: Apple iPhone 4 & iPhone 6 audio output.

In terms of general look-n-feel, or usability of the product, there's nothing I can say here that has not been said about Apple's most popular devices - the iPhones. Though I'm not an Apple fanboy, my wife loves the Apple "ecosystem" and has been using an iPhone and Mac combination since the release of the iPhone 3G in June 2008.

There is no question that Apple produces some amazing devices focused on usability and pitched as lifestyle products. With enough financial resources, they can of course fund research and compete in the specifications arena as well. In the last few years they did fall behind on screen real-estate but it's good to see that with the iPhone 6, they're making headway in this area as well... This should really give the Android makers like Samsung some good competition in the Asian arena where logosyllabic writing systems predominate and a larger screen size is almost a must.

Since my wife hadn't upgraded her iPhone 4 in quite a while, out came the credit card for this thin "little" guy:



As you can see, this is the "gold" colored version of the regular 4.7" 64GB model (versus the 5.5" 'plus' model with the much-publicized bending tendency). I think the gold color looks nice - at least a little different from the usual silver or white and I'm sure this will sell well in Asia also. Not that it really matters much because if you use a case, the back will be covered anyhow (although the gold trim around the home button looks nice). Notice the camera lens protrusion. Some have commented that this looks bad. Indeed, it will prevent the unit from lying flat and it will "tip" somewhat. Again, it doesn't really matter if you use a back cover. I'd consider this a small cosmetic price to pay for better focusing mechanism, larger aperture, and optical image stabilization (the DxO folks rated the camera function very well).



As you can see, my almost-1-year-old LG/Nexus 5 phone on the right is a little larger with a 5" screen. But check out how thin the iPhone 6 (middle) is compared to the iPhone 4 (left) and Nexus 5. The iPhone 6 feels great in the hand. The rounded corners make it comfortable to hold and it's light but still feels solid. No, I did not try bending this thing :-).

One other thing to notice is that the headphone jack is now at the bottom of the phone. I find this less intuitive than the previous top-left placement.

In use, well, it's an iPhone :-). Runs all the usual apps, nice bright "Retina" screen (all modern 'premium' phones have excellent resolution these days anyhow), very snappy with the A8 1.38 GHz dual-core processor, 1GB of DDR3 RAM (modern Android phones have 2-3GB already). I'm curious what this M8 "motion coprocessor" will bring to the table in terms of future apps (funky 3-axis gyro, accelerometer, proximity sensor).

Let's Talk Sound...

It's interesting after all these years, I have seen few measurements of the sound quality out of these ubiquitous devices which I suspect has taken over the role of the iPod for music playback for many if not most Apple consumers. So without further ado, let's have a look at the output of both the iPhone 4 and 6 to compare and contrast what has happened in the objective measurements of these devices two "generations" apart.

Fist, let's look at the headphone out through the digital oscilloscope. Here's a 1kHz square wave played at full volume (0dBFS):

iPhone 4 - 1kHz square wave 0dBFS.

iPhone 6 - 1kHz square wave 0dBFS.

As you can see, the "square" wave tracings are very similar. Neither phones show any clipping (I confirmed with sine wave as well - not shown). The iPhone 6 is marginally "louder", putting out 1.4V versus the 1.3V from the iPhone 4. Channel balance is excellent on both machines.

Notice the "ringing" in the waveform with both phones... The reason why is readily apparent when we look at the impulse response:
iPhone 4 - 16/44 impulse response
iPhone 6 - 16/44 impulse response
They both maintain absolute phase but as you can see, both phones use a minimum phase filter with no pre-ringing. Interesting! Didn't know Apple has been doing this all these years...

Okay, let's now get to the usual RightMark 6.4.0 measurements.

Setup (the usual):
iPhone 4/6 [100% volume] --> shielded 3' phono-to-RCA cable --> E-MU 0404USB --> shielded USB --> Windows 7 PC

iPhone 4 firmware - iOS 7.1.2
iPhone 6 firmware - iOS 8.0.2

Screen brightness ~50%. These measurements were made with the iPhone 4 connected to my home WiFi router (no SIM card inside) and the iPhone 6 has my wife's SIM card inside and with cell phone and HSDPA data activated. I made no concessions for "better sound" since I don't believe people listen to these devices in a crippled fashion without data connection.

I used the latest ONKYO HF Player (1.2.1, $10 for the hi-res features and FLAC playback) to play the test files. No EQ or any other DSP process like upsampling applied for the test signals of course. I promised folks that I would try measuring the effect of various loads from the headphone output... Alas, I haven't found the time to get this together. Therefore, I'll just give you the results off the E-MU and will update with another post when I get some 30/100/300-ohm load measurements done.

16/44:
The summary result comparing the iPhones with Nexus 5, AudioEngine D3, Dragonfly 1.2, and TEAC UD-501 for a desktop "reference" DAC:


As you can see, the little iPhones hold their own in terms of 16-bit accuracy. As with most devices, 16-bit, 44kHz audio is not an issue these days and all competent devices would have no problem decoding this most basic bit depth and samplerate.

Slightly more high frequency roll-off with the iPhone 6. -0.5dB at 20kHz.

Noise level: All pretty close, Nexus 5 slightly noisier.

THD Graph

Stereo crosstalk: other than the TEAC using stereo RCA connectors, the others all using the same shielded 3' phono-to-RCA cable. The Dragonfly v1.2 has notably higher crosstalk.
24/44:
Okay, let us now go one step up into high-resolution territory. Can the iPhones manage better than 16-bits?


The answer is YES, the iPhone 6 is clearly capable of better noise floor and dynamic range when fed with 24-bit data. The measured performance is between 17-18 bits of dynamic range... Not bad for such a compact device and about on par with the AudioQuest Dragonfly 1.2 previously measured. Since I don't normally measure 24/44 with my other gear, I don't have comparisons in the table to other devices.

I measured the audio output using either the Onkyo HF player (FLAC files) or the Apple iTunes built-in "Music" app (AIFF). As you can see, the iPhone 4 functions as a 16-bit device in terms of noise floor and dynamic range performance when fed 24-bits. It benefits very slightly with 24-bit audio - at best 3dB improvement. Notice that there isn't any real difference between the Onkyo app and standard 'Music' app. Just remember to turn off any EQ feature to make sure it's bit-perfect (I noticed the iPhones Music app had set the EQ to "Classic" or something like that by default; I don't know if this was a setting my wife had previously set).

Some graphs:
Frequency Response: minimal difference.

iPhone 6 benefiting from the 24-bit data compared to the iPhone 4.

THD

Stereo crosstalk. Inter-test variation evident. (The -104.9dB iPhone 6 reading was atypically low; usually around -90-100dB.)
24/48:
Let us now raise the sample rate to 48kHz and see:

Click on the table to enlarge. 
Again, we see that there is essentially no difference between different music player apps on the iPhone (Onkyo HF Player vs. standard "Music" app linked to iTunes). This time I've included 24/48 results from the recently measured Microsoft Surface 3 laptop, the Squeezebox Touch, and a couple of USB DACs - the AudioQuest Dragonfly 1.2, and AudioEngine D3. Note that the Dragonfly and AudioEngine D3 were measured at 24/96 and I mainly wanted to demonstrate the noise floor performance. In terms of harmonic and intermodulation distortion, the iPhones perform well; the only "atypical" performer in terms of distortion is the Dragonfly.

In short, I am impressed by the low noise level and high dynamic range achieved by the iPhone 6's internal DAC! It's getting really close to the AudioEngine D3 in terms of low noise which is superb.

A few more graphs:
Frequency response: Squeezebox Touch seems to have more of a bass roll-off.

Noise floor. iPhone 4 unable to benefit significantly from 24-bit audio.

THD

Stereo crosstalk: Microsoft Surface 3 worst of the bunch here.
24/96:
The chart looks OK (using Onkyo HF Player with FLAC files):

But in reality, it's a no-go:

Neither iPhones are capable of native 96kHz samplerate and the data has been re-sampled down to either 44kHz or 48kHz.

Jitter:
The J-Test audio track was played off the iPhones with Onkyo HF Player for this measurement.



A bit more jitter noted with 24-bit data for the iPhone 4 (sidebands and wider "skirting"). Also notice the iPhone 4 has higher noise level so the jitter modulation pattern in the 16-bit test is not as evident compared to the iPhone 6.

Like essentially all decent hardware measured over the last while, it's hard to make a case for jitter being an audible issue given how low the distortion is; typically way below -100dB off the primary signal. The jitter spectra look the same whether I used the "Music" app or Onkyo HF Player; again, a reminder that software does not affect jitter performance as far as I can tell whether with these portable devices or with computer audio. Remember that this is even with either WiFi or HSDPA wireless data turned on.

Subjective:
iPhones are everywhere, go listen for yourself :-). Heck, bring in your favourite headphones and have a listen at the local Apple Store. Be that the case, I did of course listen to both the iPhone 4 and 6 with my headphones here at home. I really could not put a finger on any significant sonic difference between the two phones so spent most of the time listening to the iPhone 6.

What is most obvious using less efficient headphones like the AKG Q701 is that the headphone amp just isn't strong enough - no surprise there. 100% volume is merely 'loudish' with both the AKG and Sennheiser HD800. One effect of this is that bass frequencies requiring a bit more "oomph" just does not sound impressive using either headphone.



The iPhone 6 comes with the newer EarPod headphones with volume control. These sound better than the old earbuds from previous generations before the iPhone 5. They do feel more comfortable as well in the ear. But there's no denying that these don't sound that good with mediocre treble definition and muddy bass obvious within a few seconds after listening with the Sennheiser HD800.

I listened to both lossy 320kbps MP3 encoded with LAME and AIFF lossless music. The music sounds great using my Sennheiser HD800. Typical audiophile female vocals like Jane Monheit's Come Dream With Me sounded wonderfully detailed with vocal nuances intact. With the iPhone 6 plugged into my AudioEngine A2 speakers for "near-field" listening, the soundstage was excellent and voice well focused. The classic Miles Davis Kind Of Blue sounded nice and warm as it should - nothing was missing, including the elevated background noise and ambient sounds (is that someone clearing his throat 9 seconds into "So What"?). Mark Knopfler's Privateering sounds excellent with multi-layered strings, percussion and vocals on the title track. Loud tracks like Joe Satriani's "Crowd Chant" from Super Colossal (a dynamically compressed DR8 track) sounded fine with the phones capable to rendering details through the "mass" of vocals and playful guitar-voice interchange. More efficient headphones like the closed back Sony MDR-V6 or Audio-Technica ATH-M50 provide plenty of volume and as typical for headphones of this nature, the bass also takes on a more visceral property despite giving up a bit of detail compared to something like the HD800.

Summary:
Apple is without question the "800lb gorilla" of the music industry whether we're talking about music downloads or potentially the future of streaming audio. As a family of audio devices, the iPhone is arguably the most important music player in the world at this point in time. iPhones continue to be a point of entry in portable audio given the popularity. As I publish this post, I see the iPhone 6 is up for pre-order in China today; no doubt Apple will sell massive numbers there despite the higher price point compared to competitors.

While I cannot speak of the iPhone 5's audio performance, there has obviously been an improvement between the iPhone 4 to 6 with the ability to play 24-bit audio. The iPhone 6's DAC measures very well up to the maximum 48kHz samplerate. Furthermore, I was surprised that Apple has been using a minimum phase digital filter at least since the iPhone 4. Not that I really feel it makes much of any difference, but the iPhone can claim "no-pre-ringing" just like the Pono/Ayre folks might claim. Of course, with a typical minimum phase filter like we see here, there is quite a bit of post-ringing. The Ayre folks dampened that with a -6dB at 22kHz slow roll-off filter which IMO isn't necessarily a good thing if you want flat frequency response all the way to 20kHz.

As I mentioned at the start, I unfortunately have not had the time to start measuring using various simulated headphone loads to demonstrate frequency response anomalies especially with low impedance headphones. Ken Rockwell's review of the iPhone 5 puts the output impedance at 4.5 ohms which is excellent! Hopefully the iPhone 6 will be similar in this respect.

Given the measurements I'm seeing, the iPhone's DAC is excellent and can produce very accurate output. Given the small form factor and the need to balance power usage with other phone functions, headphone amplifier power is limited out of necessity. For sonic quality, therefore, the most important factor would be how well the headphone matches the amplifier. Assuming the iPhone 6 has an output impedance similar to the iPhone 5 around 5 ohms, a good ~40+ ohm set of high quality, high sensitivity headphones should provide excellent neutral sound. Not that lower impedance headphones would sound bad of course (most IEM's have low impedance for example), just potentially not as flat frequency response.

I must say that I am impressed by how smooth the iPhone 6 is in use speedwise and the thinner profile with softer curves definitely feels very comfortable in the hand compared to the more chunky iPhone 4. It reminds me of the curved iPhone 3G but with a more "premium" feeling metal case. My wife loves it already. Finally... There is one significant feature I wish the iPhone 6 had - inductive charging. Both my Nexus 5 and 7 have this feature and has wide compatibility with Qi chargers. For the better part of a year now, I have not had to plug the Nexus phone/tablet into anything at all. Even if the iPhone 6 were a couple of millimeters thicker, I think this feature would be worth it!

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

Epilogue:

Over the past year, there has been speculation about when Apple might go "high-res" with iTunes. Considering that iTunes does not even offer lossless downloads currently, I certainly would not hold my breath! Furthermore, with the iPhone 6 capable of 24-bit but not higher samplerates beyond 48kHz, the idea of 88kHz+ music doesn't even seem to be on the horizon.

Although I think lossless 16/44 iTunes would be a great idea, I believe Apple is smart not wasting their time in the high-resolution space for the masses. As I expressed previously, I believe "high resolution" audio will be a disappointment to most people after the novelty wears off; it just doesn't sound much better if at all given the same mastering. That's one issues. Another issue is that with devices like the iPod/iPhone/iPad, where there is limited storage space, you can already upload music in ALAC (I don't see why anyone should waste space with AIFF these days) if you really want. 24/96 is approximately 250% the size of the equivalent 16/44 lossless files, and 24/192 would be around 500% the size. Considering how little audible difference there is between high bitrate lossy (256-320kbps MP3 or AAC) and lossless already, it makes no sense to load up a portable device with all these  huge lossless high-resolution files when there are so many other things the storage space could be used for (eg. movies, documents, apps, photos, books/magazines, videos...). When you factor in that the typical listener is likely to be enjoying music in suboptimal conditions like the subway, bus, car, walking around the streets, or exercising - having audio files taking up so much space just doesn't make sense.

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

To all the Canadians out there. Have a wonderful Thanksgiving weekend! Enjoy some turkey and of course make sure to take in some lovely tunes...

Addendum (January 29, 2016):
Using a 1kHz signal, measured open vs. 20-ohm load, the iPhone 6 has an estimated output impedance of 3.2-ohms.

Friday, 3 October 2014

DIY / MUSINGS: Bi-Wired Canare 4S11 And Speaker Cable Discussion...

A couple weeks back I was tempted to make some of my own speaker cables out of bulk Canare 4S11 cables I had at my disposal. Note that with my Emotiva XPA-1L monoblock amplifiers situated close to the speakers themselves, I only needed 4' lengths of speaker cable. As such, it'd be ridiculous to claim that whatever speaker cable I use, there would be "huge" audible differences. I was more interested in upgrading the banana plugs to the 'locking' variant for a better connection and since I'm already constructing the cable, I might as well make a bi-wired set to have a listen and also because it's "cool" :-).

Without any live model willing to don my home made cables, I managed to find a reasonable substitute:

Apologies to this guy, and various ads like this over the years... I could get some heat shrink tubing to improve the cosmetics especially of the amplifier end, I suppose.
Okay then... So basically I have some "star quad" speaker cables now. Theoretically at least, the "star quad" cable configuration will reduce close electromagnetically-induced interference and this can be beneficial especially for low level signals - typically balanced microphone cables (here's Canare's "Star Quad Story"). This is not an issue with speaker cables carrying high level signals.

Canare and other manufacturers of these star quad speaker cables will also espouse the benefits of reduced radiation from the speaker cables (see this Canare PDF). I guess if one is running a pro set-up with very high power amplifiers, long lengths of speaker cable with low-level poorly shielded microphone wires nearby this could be an issue. As for the home set-up - highly unlikely of any benefit unless you do stuff like have long lengths of low level turntable phono cables lying around - obviously not "best practice"!

I found the Canare cable easy to work with, is quite flexible, appears to be constructed well ("Made In Japan") and this little DIY project took maybe 2-3 hours to put together and test. Most of the time was spent making sure I got the lengths correct. The Emotiva XPA-1 has quite a wide physical spread between binding posts - almost 1' apart. At about $1.50/ft, the cable is cheap and quite easily available at many cable suppliers on-line. The 4S11 cable consists of 4 conductors ("quad") under the sheath. Each multi-stranded conductor is 14AWG and a pair would be equivalent to 11AWG; as you can see, this makes it really easy to create a "bi-wire" configuration.

It has been many years since I've seen articles measuring speaker cables (many back in the early 2000's; perhaps controversies have settled somewhat?). Yes, speaker cables can be measured with sensitive equipment and generally, the idea is to keep the gauge/thickness adequate for the length in order to keep resistance low. Kudos to Audioholics for producing excellent no-nonsense articles on cables over the years. They have a nice article on this topic. Despite claims usually by manufacturers, good 12AWG copper zip cord speaker cables are essentially all one ever needs unless you're using >50 feet lengths (and 50 feet is being really conservative). Other parameters like the capacitance and inductance of the cable may affect the sound but we're only interested in audible frequencies up to 20kHz and with typically short runs of cable, these electrical parameters are unlikely to change sonic quality. It has been said that high feedback transistor and push-pull tube designs can oscillate at high frequencies (especially out-of-audio-band oscillations which may not be heard but can burn out tweeters and amplifiers) with high capacitance cables. I've heard legends of Phase Linear designs oscillating with the old Polk Cobra cables (low inductance, high capacitance Litz wire) from the 1970's for example. I suspect this only happens with 'exotic' cables and amplifiers of questionable design (otherwise we'd be hearing about smoking speakers and amps happening frequently)! Some manufacturers have dragged up phenomena like skin effect, but we're talking audio frequencies, not HF here and calculations have not suggested any significant issues.

There is clearly potential for sonic improvement with bi-amping especially with separate crossovers for speakers like with this article by Mitchco. But bi-wiring? I don't see how this improves anything other than potentially lowering resistance (thicker total wire gauge and double the connections to the speakers), or bypassing very poor speaker jumper plates. If anyone has seen analysis to suggest that electrically bi-wiring makes a difference, please comment!

One of my favourite cable review articles is for the classic Kimber Kable 8TC. Now this is the kind of review I wish I saw from the typical audiophile press. For comparison, here are some vital stats between the Canare 4S11, Kimber 8TC, Belden, and cheap zip cord looking at results where available between 20-20kHz (results from Audioholics or from manufacturer where measurements unavailable):

Canare 4S11 (11AWG - 2/4 conductors): (~$1.50/ft, I noticed some Ram Electronics 4S11 cable measurements here)
     DC Resistance: 2.6 mohm/ft
     Capacitance: 45 pF/ft
     Inductance: <0.12 uH/ft

Kimber 8TC (10AWG): (~$6.00/ft)
     DC Resistance: 2.19 mohm/ft
     Capacitance: 100 pF/ft
     Inductance: 0.037 uH/ft

Belden 10AWG: (~$0.75/ft bulk, as used in Blue Jeans cable's 5T00UP)
     DC Resistance: <2 mohm/ft
     Capacitance: 25 pF/ft
     Inductance: ~0.16 uH/ft

Zip cord (12AWG): ("Sound King" measured here, ~$0.50/ft)
     DC Resistance: <3.31 mohm/ft
     Capacitance: ~15 pF/ft
     Inductance: ~0.2 uH/ft

These are microscopic variations between very reasonable cables, would it be surprising if short runs of <50 feet or so results in no discernible difference?

Have a look at Audioholics' "Speaker Cable Faceoff 2" article for measurements of more expensive stuff... Here's one:

Cardas SE 9 (9.5AWG): (~$13/ft or so)
     DC Resistance: <3.38 mohm/ft
     Capacitance: <285 pF/ft!
     Inductance: <0.05 uH/ft

It's interesting how high the capacitance result is with the Cardas though, presumably by design. For the purpose of "high fidelity", I think most of us would agree that "the best cable is no cable". If this is true, then in principle shouldn't we be going for the lowest amount of these parameters? (Unless of course one is aiming for frequency coloration - the proverbial "tone control".)

Speaking of capacitance, let's go a little further, using my set-up as an example. One could do it by hand, but here's a quick calculator: Electro-Voice Cable Calculator. Plugging in the number for the Emotiva XPA-1L (250W, 8 ohm, 500 dampening factor), into the Paradigm Signature S8 speaker (8 ohms), with the 4S11 cable (let us be conservative and say 11AWG for 2 strands, 4 feet, 50 pF/ft), the roll-off frequency (-3dB) due to capacitance is at 99MHz; that is to say, capacitance just isn't going to be an audible issue. Play with the calculator using numbers from your own set-up. The fact is that with typical speaker cables, capacitance roll-off in the audible spectrum is not an issue until you're in the 1uF range (1,000,000 pF)! Of course, by the time you run say 2000 feet of cable to reach that level of capacitance, you'd be experiencing >5dB power loss (~60% less watts reaching your speakers), demonstrating that resistance as related to conductor gauge and length is much more important than capacitance in affecting signal integrity assuming the amplifier can handle such an extreme situation.

For those interested in the physics/calculations around inductance, have a look at this post with John Murphy's calculations to demonstrate how unlikely this is an issue as well.

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

Since it's impossible to do 'instantaneous' A/B testing with speaker cables accurately due to the time it takes to switch out the connections (unless I had a special switch box), I listened to my set-up with the Canare 4S11 bi-wired to one speaker, and my old 12G zip cord to the other to see if I can hear channel imbalance, tonal change, differences in level of details revealed. I tried and even got my wife and kids involved - no discernable anomaly. Mono music still sounds nicely centered and tonally balanced. Stereo soundstage is maintained; as an example, I love the old Ella Fitzgerald Pure Ella: Ella Sings Gershwin album that I got years ago. Ella still sounds like she's singing right in front of me in a private performance. Fancy stereo effects like QSound stereo widening and "surround" image still produced the 360-degree effect suggesting no significant untoward change in phase relationships between the speakers connected with different cables (remember, I'm only using 4'!).

Ultimately, it's fun doing these little projects... Especially at little cost. Although I felt that there was no audible benefit in this case (hey, I wish I did!), like I said, it was more so I could upgrade my banana plug connectors to the locking variety and having a set of bi-wired cables to demonstrate to friends if they ask and want to listen for themselves. You can easily find pre-built 4S11 cables on eBay or off Amazon from Ram Electronics. Of course, with only a little elbow grease and some time, I put this together with bi-wired speaker connectors for 1/2 the price. [2018 Addendum: I recently noted that Blue Jeans Cables is now selling 6' non-bi-wired and bi-wired speaker cables based on the Canare 4S11 at good prices with welded connections... Nice.]

(For more interesting reading: speaker wires and history from Roger Russell.)

Addendum: After about a week, I decided to reinsert the Paradigm Signature S8 (brass?) speaker jumper plates. The rationale is that I feel bi-wiring makes no difference and I didn't want to misplace the jumpers; so might as well leave them in place where they should be. This essentially makes the cables a set of 11AWG wires in star quad configuration with better cable-to-speaker contact since there are now two connectors. Electrons are smart little guys and can decide for themselves if they want to pass through that jumper plate :-).


Addendum 2: Train's new album Bulletproof Picasso just came out. I was having a listen to it and was impressed by how good the dynamics sounded. So I ripped the CD and discovered that it has a DR11 score - shocking for a modern pop album! Considering their previous efforts California 37 (2012) had a result of DR6 and Save Me, San Francisco (2009) of DR5, this was a very pleasant surprise... Could it be? Could it be that record producers are recognizing it's about time to terminate the loudness war? Could it be that they finally realize that distorted audio is bad for the music industry? Here's hoping the tide is indeed turning.