Objective Reviews & Commentary - An Engineer's Perspective

December 21, 2011

Winter Solstice

snowflake by andrew magillHAPPY HOLIDAYS! Alas, unlike last week’s 7500 word MacBook Air review, there’s no big article today. The plan was to release more details of the Objective Desktop Amp (ODA) by the first day of winter but I’m still waiting on a few key hurdles (mostly related to the ODAC) first. So the days will be getting longer before that happens. For now, here are a few updates on assorted topics. (photo: Andrew Magill)

INTERESTING BENCHMARK ARTICLES: As I’ve shared many times, I have immense respect for John Siau at Benchmark Media—makers of the DAC1. I've had the pleasure of discussing audio in person with him and he’s artfully navigated (much more so than me!) the objective/subjective line in audio. He genuinely understands what matters, what doesn't, and why. John’s talents are not only apparent in his products but, unlike most commercial designers, he's also written several excellent papers, presented at Audio Engineering Society conferences, and provided detailed measurements of Benchmark gear. From my perspective, he’s equal parts academic, engineer, audiophile, and enviable entrepreneur. If you’re at all into understanding why measurements matter, you might want to check out two of his most recent articles. I do recognize he’s not completely unbiased but I believe most of what he has to say in the papers is relatively objective and easily verified:

COMING ATTRACTIONS: Here are some articles to look forward to in the coming weeks:

  • FiiO E10 DAC Review – With any luck this will be up by next Wednesday.
  • ODA/ODAC Update – As mentioned above I hope to make some good progress on the ODA and ODAC and should have a more details to publish in the first few weeks of January.
  • Audiophile Beliefs Update – Somewhat in the vein of the Benchmark articles above, I still receive a lot of questions about various audiophile beliefs. They include balanced audio, bit depth, USB vs S/PDIF, output impedance, high resolution formats, jitter, tube amps, single-ended (SE) amps, and correlating measurements with blind listening tests. I won’t try to tackle all those topics in a single article, but I would like to elaborate on at least a few of them sooner rather than later. For some of these the best I can do is shine a bit more light on a fuzzy topic, for others there are some widespread myths that can be put to rest.
  • Clip Zip Review – My poor Clip Zip hasn’t even been unboxed yet. I’m really curious how it compares to my trust Clip+. If the battery has any charge left when I unbox it at least I’ll know it can be safely stored for many months. ;)
  • USB Pro Audio Interfaces - I’ve been testing various interfaces ranging in price from around $100 to $400. All, theoretically at least, are capable of 24 bit operation over USB. In such devices most of the money goes into the recording hardware, but I thought it would be interesting to see how the DACs in these devices measure up for playback. It’s rare to see them tested for playback-only performance. Most results available are from RMAA, and on top of RMAA’s usual limitations, they’re typically loopback tests of the combined A/D and D/A performance. The dScope allows breaking out out just the playback performance and measuring it in ways RMAA cannot.

HAPPY HOLIDAYS TO ALL: I hope everyone gets to relax with some good music and enjoy at least a few days away from work, school, etc. I need to go see how the elves are coming along with those hand carved ODA enclosures….


  1. NwAvGuy,

    I would like to wish you and your family happy holidays, and a much deserved rest from a great blog that has brought pleasure to many over 2011 (but don't stay way too long!!) :)


  2. Every bit of this sounds interesting to me. I'm surprised you're alredy chipping away at the pro audio interfaces. That topic is very appealing to me, but every single item mentioned is great.

  3. Merry Christmas to you too! Will be very interested to read your extended thoughts on jitter.

  4. I look forward to reading these posts in the new year. Happy holidays.

  5. Thanks for all of this! I found early on that you can find pro-gear that sounds better for much cheaper than home audio gear - especially in the active monitor category (though you do trade in looks/waf quite often!).

    I recently purchased the Mackie BlackJack Onyx (£120) after seeing the review on proaudiostar and have loved it! i have two other similarly priced headamp/DACs and the Mackie is better in every way (though this is purely subjective).

    All that to say, I look forward to your pro usb audio interface article!

    The review that converted me is here:

    Thanks again - James

    1. I just got my Blackjack today. I'll use it for vinyl digitizing and it seems fine. Although a phono stage makes for a noisy source to test stuff with :-). I like the separate pots for monitor in, monitor out and phones.

      Once again, the Guy is right that pro gear doesn't have low output impedance on the phone outputs. I played a 1 kHz sine wave and dialed the volume to 2 V open. Then I loaded one channel with 10 Ohms and the voltage dropped to about 0.5 V, giving an output impedance of about 30 Ohms. Not great, but far away from the desired <1 Ohm and it could be worse.

    2. Just repeated the test with 1 Ohm open and I got almost exactly 0.25 V into 10 Ohms. So same thing, 30 Ohms . I was worried that 2 V into 10 Ohms would already clip.

  6. > Audiophile Beliefs Update
    This is a really good idea - "Mythbusters" of audiophile world!

  7. Thank you very much for your articles. For me they are the highlight of the week. Your work is the biggest contribution ot Head-fi ever. It is nice to see that your Blog is gaoining followers in a fast speed! Happy Hollidays to you to!

  8. Happy holidays to you too! Don't stress yourself.

  9. Audiophile beliefs update - I can't wait for this!

    Have a great break, guy.

  10. I wish a good 2012. For me, best blog of 2011.

  11. NWAVGuy: Happy Holidays to you and your family!

    I must say that for me this is the best blog of 2011. I always have an open tab with this:)

  12. Darn you and your hype building articles. ;)

    Happy Holidays.

  13. I'm looking forward to those articles you've got planned.

    Don't work too hard burn yourself out though, take a break if you need to. It would be a shame to lose a great resource like this.

    Merry Christmas, happy Festivus, Kwanzaa, Hanukkah, New Years, and all that jazz...

  14. NwAvGuy,
    Merry Christmas to you and yours! I wish you the very best! May you be richly blessed in all you set your mind to do. I has been very good to "meet" you here and have the opportunity to learn from your knowledge and experience. The hobby is BETTER for you being in it. Looking forward to all that the new year brings. Thank you.

  15. Keep up with the objective , technical reviews.
    It's always good to see a FR chart , instead of comments like "sonic bliss" , "ear-gasm".
    Thank you very much.Have a nice holiday

  16. Merry Christmas! I'm really looking forward to your E10 review!

  17. Happy holidays, NWAVGUY!

    Speaking of solstice - here's an animated graphic I did for wikipedia that shows the Sun's analemma - we're at the 'bottom' of the 'eight' today...



  18. any chance that we could add silver cables to those audiophile beliefs? I'm not sure how you could test it... But, I hear no difference. I'm curious how you feel about all these fancy cables that are supposed to improve sound quality.

  19. Thanks to everyone for all the holiday wishes and other encouragement! To Chris, you need to use Silver Bells with silver cables to hear the amazing benefits. The audio currents through the silver in the cables cause the silver in the bells to ring in perfect harmony with the high frequency content of the music adding new levels of air and depth to the sound. For best results the bells should point towards the North Pole. You can find the bells at OverpricedAudioSnakeOil.com in three different quality levels depending on your budget.

    Seriously, unless line-level (interconnect) audio cables are intentionally designed to mess up audio signals there are no significant measurable differences in their audio performance in normal use regardless of what expensive conductor and insulators are used or how they're constructed. You can measure differences at RF frequencies and in other ways that have nothing to do with their audio performance (a trick used by some manufactures trying to show their cables are objectively better). But, by all the traditional measures of audio quality, a $5 cable will generally measure the same as a $500 cable.

    Speaker cables do differ in their resistance (mostly related to the thickness of the wire) and other other properties (mostly related to their construction). Some exotic cables are even more likely to cause amplifier instability.

    The resistance, especially for longer runs, can cause audible and measurable differences. But a $10 length of 12 gauge heavy "zip cord" style speaker wire will sound and measure very similarly at audio frequencies to an exotic braided, oxygen free, high purity unobtanium, $1000 length of 12 gauge speaker cable.

    And for those wondering if there's something unmeasurable that matters, there are several blind tests that demonstrate nearly all cables sound the same. Some tests were even done in audiophile's own homes with their own systems. See the Wired Wisdom link in Subjective vs Objective for one example.

  20. I'm interested to see your audio interface reviews because I just got the M-Audio DSM1, which are usually used in the recording world. However, I'm only using them for play back and am considering a S/PDIF connection instead of analog.

  21. I've seen reports of E10 users noticing that there's a delayed cut-in from silence that cuts the beginning of the first spoken word on some VoIP connections. They say different headphones bring out the problem more. I'd be interested to see if that's captured in your testing.

  22. @akgk171, many DACs mute in various ways when there's no bit stream to decode, or even when the bitstream is there but is all zeros (i.e. "digital silence"). Some take longer than others to unmute. VoIP has so many other quality issues I would think the DAC would be the least of them. But you never know I guess.

    I'll think about a way to easily test for the problem. One that comes it mind is playing a very short test tone that might last only 0.5 seconds and measuring the length of tone coming out of the DAC. If it clipped off the beginning, the resulting tone would be less than 0.5 seconds. I suspect many DACs will shorten it to some degree.

  23. If cables make no difference, why does Canare turn an HD-650 into a muffled mess and roll off the highs of so many headphones? Why does a specific headphone cable make some headphones unlistenable? Have you actually tested multiple headphone cables (DIY cheap stuff) with your OWN ears and ignored this "science". How is sound clarity and detail measured? If all wire sounded the same, it'd make my life sooo much easier. I don't believe in expensive rip-off cables, but cheap quality wire such as Mogami or Belden. You should do some testing and report back. Maybe recable about 20 KSC75s for your next experiment. Thanks.

  24. @Anon, I can offer the following:

    If you compare 3 wire headphone cables to 4 wire cables you might possibly hear some subtle differences. But that's due entirely to both channels not sharing a common piece of wire.

    Speakers are generally a much more challenging "load" than headphone drivers. They're lower impedance making them, proportionately, much more sensitive to any impedance effects in the cable. They're also typically a much more reactive load due to multiple drivers, the crossover network, etc. So it's reasonable to assume speakers are more "cable sensitive" than headphone drivers.

    The above is important because while people think they hear all sorts of differences between speaker cables--much as you describe with headphone cables--when you simply remove the knowledge of which cables they're listening to those difference disappear. All I have to do is lay both sets of cables along the floor to the speakers, stand behind the speakers, and swap cables without the listener knowing which pair is connected. Just by doing that, all the differences they were hearing just minutes earlier are gone. They can't tell which is which.

    The above is exactly what's documented in the Wired Wisdom article I mentioned above. And it was done in the listener's own home, with their own systems, speakers, music, and high-end cables. They simply swapped some cheapo speaker wire for the expensive wire without being able to tell which was connected. All the differences the owners of these systems attributed to their expensive cables magically disappeared.

    Headphone cables are difficult, for obvious reasons, to compare blind. Even if you blindfold someone the cable stiffness, microphonics, and other physical characteristics usually give away which cable you're listening to without even playing any music.

    Find the link in the Subjective vs Objective article to the BBC McGurk Effect video (or just search for it on YouTube). There you will find out just how readily our ears can deceive us when there's other knowledge about what you're listening to. It's a very well proven fact. It's just how our brains are hardwired.

    There's a well run study documented in an AES paper where the exact same musical tracks were played repeatedly for many different listeners. The listeners were asked if they preferred A, B, or they couldn't tell any difference. 3 out of 4 listeners expressed a clear preference when it was really just the same track being replayed. They heard a clear difference where none existed.

    You don't have to believe me that speakers are far more demanding of cables than headphones, but they are. So if speaker cables don't make any audible difference, neither do headphone cables. There also won't be any measurable differences at the headphone drivers themselves between different cables of any rational design regardless of price.

  25. will the ODAC's usb input sound just as good as the spdif input? is it async?

  26. thanks for the response nwavguy!

  27. The ODAC (in this iteration at least) is unlikely to have an S/PDIF input. But I can say most well designed USB DACs will perform better fed from a computer via USB than via the computer's S/PDIF output or by using a USB -> S/PDIF converter. The reason is the built-in S/PDIF in many computers is not very good, and adding an external converter just forces an extra clock recovery which is likely to add jitter.

    The ODAC is not async. But it has very respectable jitter performance. And reduced jitter is the only benefit of async USB. Some async solutions, most notably those that use USB Audio Class 2, have some disadvantages--not the least of which being no native Windows drivers.

    I'm not aware of any inexpensive way to do true async well. I'm not sure how HRT is making any money at $149 with their DAC but that's the least expensive true async DAC I know of that will do 24 bit over USB. The TI chip it's based on is classified as obsolete and scheduled to be discontinued so it seemed like a poor choice for the ODAC.

  28. NwAvGuy for those of us who have noise from our power supplies in our computers going through the USB port as a whine, or worse, will there be some sort of USB isolation available for that DAC, would there be a way to reject this noise instead?

    "Read somewhere that newer high efficiency PSU dump a lot of noise back into the earth."
    Would this be accurate or myth? We were discussing our Seasonic PSUs introducing noise into the system.

  29. @Michael, if you're referring to the ODAC, it does have USB power filtering. Some higher-end PC power supplies (including some from Seasonic) have PFC--Power Factor Correction. That can be implemented in different ways, but in general I would expect PFC to make the supply less likely to cause problems. PFC should cause the power supply to draw current from the AC line in something closer to a sine wave lowering the radiated and conducted noise. But there are lots of variables involved with power supply noise so it's difficult to make general assumptions. If a particular PC has really noisy USB ports an inexpensive powered USB hub is generally an easy cure.

    With the next prototype of the ODAC I'll be testing it from USB power using several different PCs to see how much the noise and jitter performance differs.

  30. Headphone cables, due to how reasonably thin / light / flexible they might need to be, tend to become quite a compromise in design.

    I think high capacitance and crosstalk associated with miniaturization tend to make headphone cables audible to an extent, especially if the driving amp is marginally stable or if high output impedance forms a high pass filter that cuts into the audio band.

    If space is of no concern at the amp side, I tend to treat headphones just like speakers:
    reverse biased diodes to the rails, RC zobel, and most importantly a small air core inductor // a small resistor before the headphone jack.

    Without these measures I don't feel guaranteed safety when plugging in just any new pair of headphones. There are consequences playing with high bandwidth discrete / composite designs, and I don't see why headphone don't deserve these "royal" treatments like speakers.

  31. Thanks for your reply, I had received a response from Seasonic saying they couldn't tell if the Power supply or Motherboard was to blame, my experience was swapping only the power supply from a 330W to 520W, with the 330W it's[noise] barely audible while the 520W is very apparent, at least with my 110 dB/mW IEMs.

    My three(RMA'd the 1st 520W, 2nd does the same thing) of my units have "Active Power Factor Correction [99% PF Typical]" as well as the person I talked with.

  32. @nereis, there isn't any "compromise" in headphone cable design. 1 meter of 20 AWG cable (which is on the small/light/flexible side) isn't enough to change anything audible. Your "high capacitance" theory is not justified. The same is true of crosstalk for 4 wire headphone cables. If you believe otherwise, share the references that support your view.

  33. @NwAvGuy: just a minor nitpick: PFC isn't only found on high end PSUs, since it's a legal requirements in some countries (including all of the European Union). Low end units often only have a passive PFC though, and their power factors aren't quite as good as those with active PFC (which should all have a PF above 0.99).

    Anyway, keep up the good work - your blog is a fascinating read!

  34. @NwAvGuy: I'm huge fan of your no bullsh*t approach and lenghty articles. :)

    Thank you very, very much for your tytanic (I hope it's the right word, I'm no native speaker) work, and looking forward to read more from you!

    In the matter of audio I have no technical knowledge, but i'm keen on doing DIY projects.
    I've built complicated DAC based on WM8804 and WM8742 with Ne5532 and AD797 analog stage (it still needs proper power though) and I'm very eager to compare it to ODAC, when it'll be ready!

    In the matter of E10: I've borrowed it together with uDAC2 from befriended local audio shop for Chirstmas to compare it. (I've previously listened to uDAC2 wich TF10's and D2000's)
    for me, my brother (who owns SRH440's) and my GF (using K81DJ's) E10 is clearly better soundwise than uDAC2. It also lacks noise and channel imbalance, and doesn't warm-up so much.
    Last but not least it's much cheaper here in Poland (88USD vs. 147USD)
    I think it's a keeper!
    Hopefully, your measurements will say the same :)

    I also wan' to ask you for a test of two other cheap but capable USB DAC's: Audiotrak's Maya U5 and Cube.

    Thanks again and keep up the good work!

  35. @ov, Thanks for the comparison of the E10 vs NuForce uDAC-2. I'm not surprised you like the E10 better. Audiotrak products are difficult to buy in the USA. Most of them are not sold by the usual vendors. They seem to mostly be popular outside the USA.

  36. Just wanted to wish you Happy Holidays too. You've been a huge inspiration and have really made a good show for objective audio practices. I hope you continue benefiting the hobby with your knowledge and patience for years to come, it is much appreciated as always.

  37. Nitpicking the nitpick, but it seems like most computer SMPS with active PFC don't actually reach 0.99 power factor at lower loads, like below roughly a third of the rated capacity. And that's where most computers end up most of the time, since most sit around with mostly idling CPUs and idling GPUs all day long, when they're on.

    Anyway, this side track about noise was regarding USB DACs. The concern is then AC crud on the USB power, and not stray EMI, right? I'd think that an external DAC would be isolated reasonably far away from the major offenders. That's assuming some USB DAC apparently does not do a good enough job filtering the rails and rejecting the noise there.

    At least from the ODAC preview, noise levels are looking great, so whatever it is you're doing seems to be just fine. If just as an academic exercise, I'd be curious at whatever difference in DAC noise levels you can find plugged into different computers (or the powered hub). If there's a nontrivial difference, maybe you can try one of the sub-$40 DACs as well as a comparison, to see its relative susceptibility to different power supply noise.

    Anyhow, I'm sure this is priority #284991 for you right now, so get back to doing whatever it is you want to do.

    Merry Christmas, Happy Holidays, and all the rest. Regardless of whatever it is you decide to do (or not do) for next year, thanks for all the contributions! =)

  38. I need to buy either a Clip+ or a Clip Zip, but I'm holding off until you publish your thoughts. When you do, please talk about usability differences with the new screen and the new big button.

    I also have an O2 being built for me. Thanks for bringing such quality to the masses!

  39. Whether it's of any interest to you or not I don't know, but Head-Fi have just felt it necessary to close down their most recent O2 thread. A number of contributors (some whose names I see among the commentators here) had for some time been expressing rational views pretty much in unison, which was heartening to see.

    On the one hand it's disappointing that the thread was shut down, but on the other it's of some satisfaction that the administration there are unable to contain or tolerate the opinions being expressed other than by blanket censorship. I can't help but feel that they're storing up trouble for themselves in the future. I certainly hope so.

  40. Looking at your comment that the TI chip is obsolete, would you think the HRT Streamer 2 is a good buy for a DAC on an extreme budget for 24/96?

    I would also like your opinion on the Dacport LX, the cheaper version of Centrance's Dacport, without the headphone amp, but otherwise identical. The Dacport measures darn well in the Stereophile review, especially in terms of jitter and noise, which I think is emulated in the LX, and you seem to be quite impressed with the work Centrance does. I have a good deal for an unopened LX with local warranty.


  41. Looking forward to it!

  42. Nwavguy,

    I find the low end of the volume just a tad too loud when I'm listening at night.. any chance that we could get a pot that would let me listen on lower volumes without worry about channel imbalance?

    Don't get me wrong, the o2 is worlds better than my matrix was about channel imbalance,I hated listening to that amp at night, it was just too loud. I just figured I'd let you know and maybe see if you can find a remedy :)

  43. Apparently the E10 can output 100mW into 32 ohms, and I calculate a peak to peak voltage of 2.53V. Does this mean it is boosting the power supply a bit, or what? (and I know that FiiO's power specs are honest RMS specs, judging by the existing reviews here) \Greg.

  44. @Anon re Dacport LX, I think it's a solid product.

    @Chris, your O2 is configured for too much gain if you're using it with the volume that low. See the O2 Details article gain sections. You can probably just clip two of the gain resistors and be much happier. To answer your question, there are no better pots I know of that will fit. The O2 uses the exact same pot as the $1600 Benchmark DAC1 Pre.

    @SullivanG, just like the E11 I reviewed, the E10 has a DC-DC converter. Li-Ion batteries are only about 3.3 volts near the end of their run time. And even rail-to-rail op amps don't really swing all the way to rails when driving a real load (i.e. headphones).

  45. From the official site, E10 outputs 150mW into 32ohm (but I know somehow the package says 100mW). Voltage swing figures can be read from this PDF file http://www.fiio.com.cn/upfile/File/2011/20111207110053.pdf

    16ohm:voltage swing 5.1Vpp (or 1.8034Vrms), max power 200mW
    300ohm:voltage swing 8.2Vpp (or 2.8996Vrms), max power 28mW

    That's pretty good for a USB-power device.

  46. @NwAvGuy
    That's some expensive equipment for audio analysis in that 'Do Specifications Lie?' article, is that THD minus Noise a feature on only the best equipment?
    19,000~30,000USD used/referb, hate to see its Retail tag.

    would you consider measuring that way in the future [if you could] or would it be impractical? I doubt you've amassed a treasure trove of headphones to test with.

  47. Thanks for all you have done/will do: I give you permission to take a short year-end break :-)

    I liked Benchmark's comparison with 2 other high-end amp/DACs: it seemed objectively written and researched, and no subjective voodoo-speak. A quick Google of the two price points quoted in the review ($1148 and $1695) shows they were likely the Lavry DA10 and the Metric Halo ULN-2; these were reviewed together on Stereomojo.com, where the reviewer, one Dr. John Richardson, thought they were good values at those price points (??!!). Lessee... that would make a $230 or so ODAC a ginormous value!!!

  48. Hey NwAvGuy

    Commenting all the way from Singapore. Love your blog. Keep up the awesome work.

  49. @Michael, I'm not aware of any low priced options for "THD minus noise". The dScope I use does it and, while dScope's are a bit more reasonably priced than similar products from Audio Precision, the Series III with both analog and digital I/O is still five figures.

    While THD-only is useful for some things, a decent argument can be made THD+N is generally the more useful measurement. What most care about is everything the gear under test is adding to the music. THD+N comes closer to including "everything" than THD only.

  50. Hi NwAvGuy,

    Thanks for your great job thus far. I'm looking very impatiently forward to your promised review of E10, as I'm about to get one.

    Cheers mate.

  51. are you going to review the dacport? it's less than $400 and it's capable of 24-bit resolution..

  52. @Anon, Stereophile already tested the DACport and it performed fairly well. I try to focus on testing products that have never been independently tested rather than duplicate other test results.

  53. Do harmonic distortion and noise decrease when driving a higher impedance load?

  54. @Akg, Distortion usually decreases with rising impedance. And often, below some impedance, distortion rises dramatically as the impedance drops (see the AMB Mini3 review in May for an example). Noise usually doesn't change much.

  55. Hey NwAvGuy

    Keep up the good work!!!!!!!!!!

  56. FiiO just keeps churning out new product. I just saw that the E17 is the new one on the way. http://www.fiio.com.cn/news/index.aspx?ID=154

    You know people are going to hit you up for this one in the coming months. The amp stage looks pretty powerful compared to the E7.

  57. Following your blog for a period, you gave the most information about the devices which should be known.

    Waiting for the article of balanced audio to be launched. I am so curious what is the difference between balanced & unbalanced, and is it worth upgrading to be balanced system?

  58. @Akg, FiiO just needs to do the E-O2 (correctly) and be done with it... ;)

    @Nick, generally no, it's not worth upgrading to balanced for home headphone use. Balanced audio is best used for professional audio (i.e. recording, live sound systems, etc.) or where there are otherwise noise problems with unbalanced connections. Balanced audio doesn't magically somehow sound better unless it's getting rid of audible noise.

  59. Looking forward to seeing what you think of the E10. I bought one a few weeks back, and for £60 think it's a good little device, it certainly lifts the performance of the average PC or laptop audio output.

    Would have been interested in your own views on the DACport as this is likely to be my next purchase, but appreciate that you'd prefer to cover products which haven't had a decent qualitative review and measurements elsewhere (and I think John Atkinson does a good job over at Stereophile). Still it's got Benchmark Media blood in it's veins, so I'm pretty confident it'll sound good.

    All the best



  60. First of all happy new year!

    I have a few doubts:
    1) Playing music at a low volume set by software would reduce the sound quality? If I understand it right, it would reduce the bit depth, and it would be better to play it at max software volume and to control volume analogically with a pot.

    2) Can you just use an equalizer to counter the alinearity of the freq response of headphones? What I mean is amplify the frequency ranges that are low in the freq response and reduce the ranges that are too high, to make it as flat as possible. Would that cause lots of armonic distortion?

    3) Is there much difference in using a software equalizer or a hardware one?

  61. Happy New Year to all....

  62. NwAvGuy, I hope you and yours had a great Christmas and saw the new year in in safety. ( I still can't play the "Holiday" name game).

    I want to let you know how much I am looking forward to the progression of the ODAC design and build project. I am a little torn between buying a DAC (Burson being a proud Aussie and based on reviews) or await the ODAC. Price mit be the decider.

    Regardless I wonder if you looked at or would look at their discrete componet (op-amp) ? The idea that one can produce an op-amp from discrete componets is not strange, that is what was done before high-density IC based devices. What I am interested in is the asserted performance improvement and I would like to see the objective measurement that might provide evidence ne way or another.

    Op-amp rolling with oneof the Burson discrete op-amps could be interesting but I dont have a good enough reference amp to undertake this. Regardless, I am curious enough to buy one of their dual op-amp boards but would really value your objective measured assesment if you can ever find the headroom to undertake it.

  63. @Anon, With a good 24 bit DAC the PC volume bit resolution is a non-issue. Note, however, a lot of USB DACs that run 24 bit over S/PDIF and/or have a 24 bit DAC chip, degrade to 16 bits over USB.

    You can only partly correct headphone problems with EQ. It's a complex topic but EQ is not without audible side effects. So you're essentially trading one set of problems for another. But, ultimately, what matters most is what sounds best to a given person. If you prefer the sound with EQ, then by all means, use EQ. :)

    @Rouston, I'm not sure what to make of Burson. I've never tested any of their gear, but the discrete op amp thing is 99% snake oil in their applications. See my Op Amp Myths & Facts article for why. Much like Schiit Audio, Burson is a big Head Fi sponsor. And, like Schiit, they've been caught with their pants down at least once that I know of.

    The article linked above has information about the much-loved-at-Head-FI AudioGD discrete op amps. Despite all the praise, they measure horribly. Subjective bias can be a very powerful thing.

    I seriously doubt Burson has the engineering talent, multi-million dollar labs, and other resources, to out-design the op amp guys at National, Analog Devices, etc. It's just a marketing gimmick.

  64. when will your ODA and ODAC be available to build? i know you don't know the exact date, but in about a month? 2 months? 6 months?
    i want them so bad i can't wait!! :)
    thank you so much for releasing these great designs to the public, you're awesome!

  65. @Anon, the ODA/ODAC schedule really depends on what comes up, etc. But we're aiming for "1st quarter"--i.e. before April--for the ODA construction documents to be fully published and the ODAC to be available as an assembled board.

  66. Hope you and yours had a happy holiday and happy new years from me too. Thanks very much for all the good stuff in 2011, just keep it coming. Every article has had something useful even if it wasn't directly in my 'field of interest' and that in itself is really remarkable.
    Also with the intent of helping to 'keep it coming' if you wanted to put up a paypal donate button I would be more than happy to chip in something to help defray the cost of the testing bench. I support the shareware I use and I've learned so much here I feel the need to give something back.
    In any event, thanks very, very much.

  67. I clipped R19 and R23 to bring my gain levels down to 1x and 2.5x, it worked very well. thanks for the suggestion nwavguy! I can listen at night with no worries of channel imbalance now, and make better use of the volume knob.

  68. Thank you for posting the article by John Siau. It is one of the best articles about distortion that I have come across,

    After building an amplifier, and playing around with the feedback, I became convinced that distortions below the noise level are significant. Our ears can pick this up. With a analog meter this effect is completely hidden. Even with a scope it would be impossible to pick it out of the complex trace.

    I was amazed to read that the damping effect with headphones or earbuds is so important. You have pointed the way for further investigation.

    As an analogy, I have stated that when an orchestra plays at full volume, a single instrument out of tune will be heard by our ears. I cannot imagine how I could pick this up on a measuring instrument.

  69. @Cornelis, you're correct about analog meters and oscilloscopes. But the dScope (or an Audio Precision) audio analyzer can pick distortion out down to better than 108 dB below the signal. A compact disc only has 96 dB of dynamic range so that's way below the best noise floor possible with 99% of digital music. To put -108 dB in perspective, for a 1 volt audio signal, that's 0.000004 volts worth of distortion. Or, put another way, it's 1/250,000 of the signal.

    The dScope measures below 0.0004%. Blind listening tests have shown that distortion while playing music tends not to be audible until levels of 0.05% under the most ideal conditions. Under more typical conditions it's often not audible until at least 0.1% or higher. See: Music vs Sine Waves for more.

    So a strong case can be made the dScope can measure distortion at least 100 times lower in level than a human can hear (0.0004% vs 0.04%).

  70. ^ I love how you have an absolutely rock solid answer for every freakin nitpick, or claim, or question , or whatever crosses your path. Its no wonder the O2 sounds so damn good! You clearly know what you are doing.

    I too clipped the gain down to 1 and 2.5 - I'm now getting a plenty loud 75db average spl at around %40 volume on my LCD-2's. perfect!

    now if NwAvGuy could just take some of the noise floor out of my office we'd be all set!

  71. Greetings!

    Fiio announced around 2 weeks ago that there will be some modifications to their current lineup.
    The E10 is included there as well, and I'd like to know if the version you're reviewing is the old one or the new one, as I have a feeling that maybe it's more than just a cosmetic change (cutting corners).

  72. The dScope might do this for a pure sine wave. Can it do it for a complex signal?

  73. @Anon, the E10 I have is brand new but apparently the older version. There have been complaints about the headphone jack. From the pictures FiiO has posted, they're now using a different gold jack which hopefully is an improvement. There's nothing to otherwise indicate a change in audio performance (i.e. it's the same circuitry, etc.).

    @Cornelis, please see the Music vs Sine Waves article. The article addresses exactly the "complex signal" vs sine wave question. Techniques such as audio differencing allow using complex signals for amplifier distortion analysis and the results are consistent with sine wave testing.

  74. Hi Nwavguy!

    First of all, happy new year to you and your loved ones!

    I'm looking forward to the measurements of the E10. I'm refreshing your blog in every 2 hours since dec. 28.
    Please save me some time and tell when you will publish the measurements. :D

    I haven't got any reference dac/amp to compare, but i think the E10 sounds good (and the HeadStreamer as well).

    We did some controlled blind testing (3 person - 30 tries - matched volume)- HRT Headstreamer vs. E10, but none of us were able to spot any differences (the result was nearly 50%-50%).
    I repeated the blind test with several headphones and earphones, but... failed every time.

    So i'm really curious, that both of them have the same flaws, or they equally good, without major problems.

    Keep up the good work! We need you! ;D

  75. Ah I see. Thank you.
    I had thought maybe there was more to the change, because the position of the headphone out and bass boost switch have moved, but I now have the peace of mind knowing that the modification is for the better. I was going to jump on the last of the older batch if something indicated otherwise.
    Anyways, eagerly waiting for the review now.

  76. @slfln, interesting comparison test. I'm sure in sighted listening some will claim all sorts of "obvious" differences between them. ;) I'm late on the E10 review, I was hoping to have it up yesterday but things have been kind of crazy catching up after the holidays.

    @Anon, I really can't say what all has changed with the E10 inside. It's sort of odd FiiO would make a bunch of changes to several products at once. I wonder if they switched factories? They might be using a new "ODM" (Original Design Manufacture) that wanted to tweak the designs--likely to lower costs as you suggested. I doubt FiiO owns their own factory and things are really changing fast in China with labor costs rising sharply.

  77. Hello NWAVGUY,

    I'm a fan of your reviews. I just got an E10 and and also eager to read about it. So far it is excellent, I have a question though about sample rate and bit setting for the dac :

    do you recommend setting 24/96 in the sound properties in windows for the E10, or just leave it to 16/44?

    I have tried A/Bing both settings and im not sure if the switch to 24/96 made it refined or my mind is fooled by the lost details when it upscaled to 24/96. I have no 24/96 music files and am concerned which setting is best for 16/44 music.

    Thanks in advance, and happy new year to you!
    Michael DG

  78. Hey NwAvGuy, Just posted to say I've read most of your site and read a lot of your pre-ban HF posts also, it's good stuff. Thanks for what you do, really looking forward to your E10 review, keep it up.

  79. @Anon, you want to always use the native sampling rate of your music whenever possible to avoid re-sampling (which usually causes more harm than good). But there can be some benefits to using 24 bits. So I would suggest, when possible, configuring Windows for 24/44 assuming you mostly listen to typical music. Only DVD audio, SACD, and high resolution downloads are natively sampled at higher bit rates.

  80. Hello NwAvGuy. I am eagerly awaiting the arrival of your O2 amp from JDS labs who offered a complete unit for a relatively good price for us unable to build our own. I am fairly new to this hobby and we newbies owe you many thanks for your informative posts. Personally you have saved me a great deal of money as I move to higher end equipment. Many thanks.

  81. Sorry forgot to ask if it is true that the Darth Vader of the audiophile forums is now offering your 02 for sale. Talk about irony after all you went through if it is true

  82. I've read that its best to set windows to the highest rate supported by the dac (not including oversampling). I have a crappy udac2, so I have set windows to 24/96 and use foobar to dither/upsample everything to 24/96. (Software volume controls everywhere set to 0dB). I think I understand why 24 bit would be useful, but does upsampling really mess things up so bad? And whats the deal with software equalization? (Maybe it would be worth doing an article on these topics:))
    An example: the built-in equalizer in foobar had problems with "complex" music (mostly electronic), I could really easily hear distortion/cracks/clipping, whatever that was(no,not the usual udac2 clipping), but with the better 'Graphic Equalizer' plugin these problems are gone.

  83. @Anon, if by "Darth Vader" you mean Jude at Head-Fi, he's not (to my knowledge) selling the O2 but despite banning all discussion of the O2 when it was first announced, he seems to have eagerly accepted advertising revenue for the O2.

    @Anon, you need to read some factual info on sampling rates, how DACs work, etc. You've apparently been reading myth instead. Any sampling rate higher than the music is upsampling. And 99% of all music is CD audio which is 44 Khz.

    Furthermore, your uDAC2 can't run at 24/96 over USB. So you're probably forcing your 44 Khz music to be upsampled to 24/96, and then it's downsampled back to 16/44 or 16/48 by the operating system/driver before it's sent to the DAC.

    Yes, sample rate conversion really is generally a step backwards. And why do it if there are no benefits? Even 44 Khz has been proven in extensive listening tests to be entirely transparent to even skilled listeners (see the Meyer & Moran as just one example).

    Software EQ works great when it's done right. The problem is some software doesn't automatically adjust the level to prevent clipping.The loudest something can be is 0 dBFS and most recordings hit 0 dBFS without any EQ. If you boost some frequencies with EQ, the music will now exceed 0 dBFS and distort badly. That's what you're hearing. It has nothing to do with the complexity of the music. There's an overall level slider in the built-in Foobar EQ. Use it to reduce the level to keep the signal below clipping.

  84. @Anon, I highly recommend xnor's EQ for foobar2000 instead of the default EQ: http://www.hydrogenaudio.org/forums/index.php?showtopic=88505

  85. It's a small company called Epiphany Acoustics that's selling the O2, and they're a new sponsor. Surprised that Jude allowed it in the first place...

  86. Your second link "Sonic Advantages of Low Impedance Headphone Amps" points to an article referring to the damping factor many times. I think anytime the damping factor is used to make a point without also giving consideration to the DC resistance of the voice coils, something is wrong. I am all for it when the output impedance of an amplifier is reasonably low, but as soon as the damping factor is larger than, say 20 to maybe 100, it becomes a non issue.

  87. Thank you for the response. I used this guide (for my win7 machine):
    this is the EQ I mentioned:
    I think I'll wait for the ODA, then throw that crappy udac out the window (literally).

  88. @eE10000, the DC resistance of headphone drivers is usually fairly close to their rated AC impedance. But headphones don't operate at DC, so their AC impedance is the more valid measure for damping factor. So there's nothing wrong with Benchmark's paper as he discusses the AC impedance.

    I agree with you on damping factors, and my worst case rule of thumb is only 8 (see: Output Impedance).

    Using your value of 20 causes a huge number of headphone sources to fail with most headphones. For example, the iPod Touch is around 7 ohms so that means it shouldn't be used with any headphone under 140 ohms. But it doesn't have enough output to even drive most headphones over 140 ohms.

    Or the FiiO E9 desktop amp has a 10 ohm output impedance. So it would fail with any headphone under 200 ohms. Etc.

  89. Thanks for the link to the "Do Specifications Lie?" article. It's long been my belief that measuring audio amplifiers into purely resistive loads is doing a disservice to consumers and to those companies that produce top-quality audio amps. That article gave me some strong evidence that my lazy-engineer conjecture might have more than a little validity.

    The industry should be specifying performance into reactive, real-world loads. If the automotive industry measured car performance analogous to the way that the audio industry measures amplifier performance, all test tracks would be completely flat, level, and straight, without so much as a pebble on them.

  90. @Fred, as the output impedance of an amp approaches zero its performance into real-world vs resistive loads becomes essentially the same. That's what Benchmark documented in those articles and is consistent with my own measurements. So it's acceptable to measure using resistive loads with such an amplifier. The main issue is that higher output impedance hurts performance with real world loads in multiple ways.

  91. @NwAvGuy, John Siau wrote "All three products have a high-quality headphone output with very low output impedance..." He even refers to them at "'0-ohm' headphone amplifiers" in his other article. Yet when presented with a real-world, reactive load, two of the three units under test diverged significantly from their performance into a purely resistive load. So there's more at work here than just the output impedance, though I grant that high output impedance is almost assuredly going to compromise performance.

    He concluded with "Our measurements, taken under typical listening conditions, confirmed that audible differences should exist. The published specifications did not lie, the manufacturers did not lie, and our ears did not lie. The truth lies here: Specifications must be measured under typical operating conditions if they are to be useful in predicting audible differences."

    I agree with John Siau: We need performance measurement standards that reveal and predict, rather than obscure, real-world performance differences.

    (Clutch in, shift gears, clutch out)

    I received my O2 assembled board and front panel from JDS Labs yesterday and put it into the specified case, wired up a power supply, and started listening with my Sennheiser HD580s. My hat is off to you. Spectacular, particularly the dead-silence with no input signal. My immediate reaction to the silence was to conclude that it might not be working! I'll be taking it into work today to share with one of the other electrical engineers, who is also an audiophile.

    I have to give JDS Labs kudos for the workmanship on the board and front panel as well as their top-notch customer service. I have no commercial affiliation with them other than being a satisfied customer.

    One tip for others: I used a couple of 1/4" squares of soft foam weather stripping on the inside of the back panel (next to the screw holes) to prevent the board from moving front-to-back. Rock-solid.

    Now that I've got one, I'll probably get a bare PCB and build up a second, probably with power jack and signal inputs (RCA) on the back panel for non-portable use.

    Thank you again for your generous contributions to the audiophile community.

  92. I love your blog, especially the "geeky" bits. However I am unsure how to correlate the measurements to the perceived quality of the sounds I hear. What is your opinion of/reaction to the GedLee metric http://www.gedlee.com/distortion_perception.htm
    and the underlying criticism of THD and IMD as predictors of perceived sound quality?

  93. @Fred, I'm glad you like your O2! I see your point about real loads. I believe the differences in measurements between resistive and real loads fit into one of two categories:

    - Significant and potentially audible differences due to the output impedance being too high and/or stability issues (the amp has troubles with reactive loads).

    - Inaudible, but perhaps still measurable, differences when the output impedance is sufficiently low and the amp is sufficiently stable into reactive loads.

    I agree it's worthwhile to make more measurements with real loads--particularly for amps that are borderline "sufficient" in the above categories. In my experience, measuring the output impedance and the stability into a reactive load will largely predict an amp's behavior into real vs resistive loads.

    It's also worth pointing out some tests are difficult or impossible with real loads. And Benchmark's own published measurements for their headphone outputs (i.e. the Audio Precision results in the user manuals) mostly follow the industry standard practice of using a resistive load.

    As I've said, it's something I plan to explore further. I have a DAC1 Pre, and the HD650 headphones, so I can re-create several of John's tests and see how the O2 measures up. As for the O2 being quiet, you probably noticed in the first O2 article, but it's substantially more quiet than even the DAC1.

  94. @Anon, I had read the GedLee AES papers but not the multimediamfg article. Thanks for the link.

    In a few places I've discussed how a single THD and/or IMD number often doesn't tell the whole story. It's important to also inspect the spectrum and evaluate the nature of the individual harmonics. In effect, GedLee's method does that in a more objective way.

    I comment on specific harmonics and "harmonic signatures" in several of my reviews. Some forms of distortion are much more likely to be audible than others. That's what GedLee, and others, are trying to quantify with improved distortion analysis.

    It's especially critical to evaluate the nature of distortion (i.e. the harmonic signature) during product development. The spectrum provides many clues to what's likely causing the distortion so the designer knows where to look for improvements.

    There's a lot of conflicting information on distortion perception--even in AES papers. I read an article that tried to summarize much of the research and the author expressed frustration with how different some of the findings were.

    But, despite their differences, most of the distortion authors will agree if the total distortion plus noise (i.e. everything the piece of equipment is adding to the audio signal) is -80 dB or lower it's very likely inaudible when listening to music. That's 0.01% THD+N. And the few who may not agree, should believe if THD+N is below -96 dB (the total dynamic range of CD audio) it's inaudible. That's 0.002% THD+N.

    Based on the above, one can strongly argue: Below a certain level, the distortion signature no longer matters. Above that level, however, it can matter.

    The GedLee method is certainly an improvement over traditional measurements for when distortion is above the threshold of audibility. It would be great if their method (or something similar) is widely adopted. Audio Precision and Prism Sound could update their audio analysis software to provide a standardized way of implementing GedLee-like measurements.

    But that hasn't happened--probably because most manufactures prefer to downplay specs. They want to compete in much more subjective ways. Land Rover doesn't want you to know their $80,000 SUV is slower than a $25,000 V6 Honda Accord in a 0-60 test. They prefer to emphasize the looks, leather interior, status, pedigree, sense of adventure, image, etc. It's the same with a lot of audio gear.

    In effect, the GedLee method subjectively weights distortion measurement to hopefully be more consistent with perceived distortion. It does that by objectively quantifying the harmonic signatures. Using their method could explain why two amps that both measure 0.2% THD+N can sound notably different.

    An interesting analogy is there are now labs that can analyze wine down to even trace amounts of virtually everything in it. They have developed statistical data and modeling that lets them predict how the wine will score in subjective testing (i.e. "90 points", etc.) without ever tasting it. If I remember correctly, it's reasonably accurate something like 80% of the time. Wine makers like it because it removes the subjective bias, and day-to-day variability, of human tasting.

    Another analogy would be the music genome project that lets services like Pandora predict what music you like by digital analysis of tracks you enjoy. There's even such a project now for visual art. Having more computing power, and massive databases, has enabled software to better predict what humans perceive. It's a fascinating topic.

    But, unlike wine, art, cars, etc., a lot of audio gear only has to clear certain hurdles to be entirely transparent. Improving it further doesn't make it sound any better. The GedLee method is most valuable for lesser gear that trips over the traditional hurdles.

  95. @NwAvGuy, I believe you are correct that most of the significant, audible differences are due to marginal designs of output stages.

    I was once able to hear the difference using two different interconnect cables in blind testing -- with 100% accuracy. One was an "audiophile" cable and the other was generic, as supplied with most consumer audio gear. I always attributed my success to a marginal output stage in the high-end CD player, possibly a reactive and/or too-low-impedance in the audiophile preamp, and a reactive boutique audio cable. The end result was a really expensive, and unpredictable, version of tone controls.

    Perform that same test with a CD player incorporating a better output stage and I'd probably fail to have any success at discerning which cable was in use. While I know that most subjectivist claims are based on wishful thinking, imagination, and preconceptions, I bet that there are some that are based on hearing actual, but unmeasured, performance differences.

    I bet that the O2 performs very well into reactive loads.

    No criticism was intended of any manufacturer that publishes specifications measured into resistive loads; that's the established industry standard. But a set of standards that fails to uncover significant, and audible, faults, needs revising. When a manufacturer designs an audibly superior product, that should be reflected in superior measured performance specifications. How will we beat the subjectivists into submission when they can hear differences when the standard specifications predict none?

  96. @Fred, I agree. And I've written a lot about how many audiophile myths often have some grains of truth behind them or there was some applicable truth at the myth's origin. But many audiophiles eagerly apply those grains of truths in ways that just don't make any sense. That's true for cables, vibration damping, power conditioning and other similar high profit items that generate massive amounts of revenue. Almost none of it can survive a blind listening test.

    When it comes to trying to persuade subjectivists into being more rational and objective, a new distortion measurement isn't likely to help. They very often insist they hear differences where literally none exist. One AES documented study played a track as "A" and "B" asking audiophiles which they preferred. Over 70% had a strong preference when, in fact, both A and B were the same track being played over the exact same hardware. Most were expecting to hear a difference so that's what their brains served up. It's like watching the McGurk Effect video and having your eyes trump your hearing.

    So until more people participate in some proper blind listening tests, I really don't see those spending hundreds or thousands on cables, power conditioning, etc. letting go of the fantasy. They want to believe in Santa Claus for a variety of reasons--not the least of which is admitting otherwise would mean they've wasted a lot of money on needless audiophile "upgrades".

  97. If possible, I'd really like to see a review of the Focusrite Scarlett 2i2 next to the Roland UA-55 Quad-Capture. They have a similar size and price bracket, but the Roland has much more diverse connections and lacks a separate volume control for monitor output, which I find distracting. If possible it'd be an interesting comparison.

  98. @Anon, I have a Quad Capture. It's one of the most full featured USB powered hi-resolution interfaces I could find. But I'm not that impressed with it so far.

    While I have not yet fully tested the Quad, I can say the digital I/O is NOT bit accurate which is disappointing. And the D/A and analog audio performance is decent but not great. I've also found some bugs in the driver/control panel software. And, like pretty much every pro audio interface, the headphone output has too high of an impedance to be useful for high quality listening.

    I don't plan to buy a Focusrite 2i2 but I did briefly test a couple of the similarly priced current Tascam interfaces and their bloated USB 2.0 drivers are a mess. On two different Windows PCs I couldn't get the Tascam interfaces stable enough to even test. They did, however, work as native UAC2 devices on my Mac but with limited functionality (at least using my usual audio tools).

    I'm increasingly convinced most of the companies turning out reasonably priced pro audio interfaces are not interested in spending the time and money to properly develop solid drivers--especially for Windows. They seem to get them sort of working for 11 seconds on a single in-house PC, they call it good, release the software, and move onto cramming more features only the marketing department wants into the next release/product. It's sad.

    I have encountered various driver problems with:

    Pre Sonus
    M Audio
    Native Instruments

    The only drivers I've found to be relatively solid are:

    Avid (Mac only)
    Apogee (Mac only)

    The above is a long way of saying I'm a big fan of interfaces that will work, even with reduced functionality, with native UAC1 or UAC2 drivers included with the PC's operating system. The proprietary drivers are often a mess. And, with some, you can't install just the drivers as they're embedded in a big install package. So you're also forced to also install lots of bloatware most don't want or need.

    If you read the interface reviews on Amazon, Sweetwater, Musician's Friend, etc. you'll find people complaining their entire PC was rendered unstable or even unusable just by installing a pro audio interface. Most of the companies don't bother to get their drivers certified by Microsoft (WHQL) and there are lots of unhappy users with driver-induced blue screens, etc.

    The Roland drivers don't seem to be that bad on my PC, but they have some rough edges. It's a decent interface for the price and for being USB powered. It has better input metering than most in its price range. But it's not audio nirvana.

    1. Thank you for the swift and in-depth reply! You are a real hero of the modern electronics and audio world, I'm constantly linking this blog to everyone interested.
      My headphones aren't the most difficult load I suspect, with their 55 ohm nominal impedance; the AKG K 272 HD.
      I think I can't go wrong with either one, Focusrite 2i2 and Roland Quad Capture both have a plethora of good features and plenty of quality to spare for my modest requirements (balanced main outs, headphone out, low latency, durable build, small form factor, decent mic pre-amps with phantom power, less than $250).

      Time will tell which one ends up on my table, I'll definitely keep following your blog whatever the case. Cheers!

  99. NwAvGuy!

    A few months ago while looking up some audio information, I stumbled across your blog on accident. Your posts were eye-opening, and I was soon hooked into browsing your archives every spare minute I had. I decided to give your O2 amp a try.

    I built my O2 the other day, and WOW. What a difference! I'm using the X-Fi Go Pro DAC (which I see you have reviewed) so I have quite a decent economy headphone setup, thanks to you!

    I'm currently studying to become an electrical engineer, and seeing your great work and generosity is inspiring. I've been through all of my math, physics, and basic circuits classes and I am currently enrolled in signals and systems. Not until reading your material have I actually had an in-depth understanding of many aspects of audio circuitry. Your article "All About Gain" was especially captivating. In my current level, nobody has ever properly explained gain to me! I've spent lots of time in labs on frequency selective devices but no time on any kind of amplification! (Also turned out that my instructor SKIPPED the two chapters on op amps!) And you have exposed me to this extremely rewarding hobby which may possibly become my profession.

    Thank you NwAvGuy.
    I am truly grateful.

  100. re: Windows driver quality - if you're suspicious of a driver's stability, you can use the Windows "driver verifier" functionality to put it under tighter verification. Type "verifier" at a CMD window to open the configuration. Essentially will find certain classes of driver defects much faster. Note that usually this means it'll bluescreen your system faster - it's for test/development - but the problems found are real defects.

  101. Hi Nwavguy!

    You've talked about the "point of diminishing returns" before, and how the Benchmark PRE is well past it. Well, i wanted to get your thoughts on the current points of diminishing returns in the amp/DAC products today. Right now, I have roughly a $500 budget or so, looking to replace my uDAC1 with something that does my Sennheiser HD600s some justice.

    Browsing through old posts, I saw that you liked these items, but its still hard to get a sense of where the point of diminishing returns is, aka the point at which the majority of people can't tell the difference between higher priced products through double-blind testing.

    The points where they could possibly be:

    Fiio E7 or E10 and O2 amplifier = $80 + $150 (preassembled) = $230ish
    HRT Streamer II and O2 amplifier = $150+ $150 (preassembled) = $300ish
    Centrace DACport = $400ish
    HRT Streamer II+ and O2 amp = $350+$150 (preassembled) = $500ish

    (prices may be off, going by memory)

    These are pretty much the options i'm considering right now. Do any of these come near what you feel is that special point? If not, do the improvements sort of scale up relative to money spent, in these instances? Am I missing a notable product? Or is that point near the DACmini price range right now?

    Even if you could just give a general sense of where you think those points are right now, i'd think it'd help me tremendously. I realize you're trying to do this very exact thing with the ODA/ODAC, but where do you think the market shakes out to right now?

    Thanks in advance! Love the blog!

  102. @Luke, if you don't mind a two box solution, the E10 or HRT plus the O2 is a reasonable solution. If you want a single device, the DACport might be worth the extra money. If it's only for desktop use, you might want to wait for the ODA/ODAC.

    The weak area in many products is the headphone output more so than the DAC. For example, the 10 ohm standard output impedance of the DACport (you have to pay extra to get a 1 ohm output) can cause audible differences with many headphones. With the HD600, however, even the 10 ohm DACport will be OK. The E10 doesn't have enough output for many headphones, including your HD600s. The O2 (or ODA) solves the headphone problem for nearly any headphone you might want to use now or in the future.

    As for DACs, I'm not done measuring the E10, but what I've seen so far is at least respectable. I would rule out the E7 as it can only do 16 bit audio. Based on other reviews, the HRT it's also respectable. The E10 and HRT are right around the point of diminishing returns and the least expensive DACs I know of offering 24 bit over USB.

    Without running blind tests I can't say for certain the E10, HRT and DAC1 would sound the same in blind testing but I suspect any differences would be much less significant than many would expect. The plan is for the ODAC to perform at least as well overall as the E10 and HRT and it will be blind tested against the Benchmark DAC1 and likely other more expensive DACs. And, best of all, it can be packaged with a very solid headphone amp--something that still eludes nearly all the similarly priced competition.

  103. thanks for the response!

    if the E10 and HRT-basic are around the point of diminishing returns, I think I may opt to be a little safe and go ahead with the DACport. I don't want to be continually buying a bunch of DACs because im still not satisfied, so I'm willing to pay a little more now. If the DACport is better than the E10/HRT (it should be, right?), i think that i'll have more insurance that I'll really be satisfied.

    As for the output impedance, the only instances I can imagine this being a problem is for Grados and IEMs, correct? You're usually not looking to use a DAC or amp on a lower impedance headphone, because they're usually not good enough to deserve that treatment anyway. And since I'm not a Grado or IEM guy anyways, I should be fine.... even if i go that route in the future, this "problem" is an O2 amp away from being solved, no? :P

  104. There are lots of headphones that will suffer at least a bit with the 10 ohm output impedance of the standard DACport. They include many of the 60 ohm and under AKGs, 32 ohm Grados, 32 ohm Ultrasones, 20-something ohm Denons (like my AH-D2000), the 32 and 80 ohm Beyers, and virtually all balanced armature IEMs. My 80 ohm Beyer DT770 Pro 80s, for example, sound best from a zero ohm source.

    You're correct that many low impedance heapdhones are less likely to need an amp. But if your PC is the source, what is your option? The built in headphone jack? It may have an even higher impedance than the 10 ohm DACport.

    As for not deserving a good source, many of the headphones above are $200+ and easily benefit from a high quality source. But if you shop carefully, you don't have to spend a lot to get such a source.

    I just finished testing the E10 and it's fairly respectable. The max output is around 2.5 Vrms which is borderline for your HD600's depending on how loud you like to listen and what sort of music you listen to (see my More Power article). It also hates 24/88 content. But, if you can live with those limitations, it's a relative bargain.

  105. The 24/44 recommendation for Windows is especially important for Vista and Windows 7, when listening to YouTube, or using applications that use the "WaveOut" API. The reason is that if Windows has to resample, the fidelity is not as good as it should be. (XP is fine). Microsoft have confirmed this bug, but I'm not aware of a fix yet. \Greg.

  106. Umm, I am pretty sure that it is Windows XP's Kmixer that has the problems resampling and would not resampling 16bit to 24Bit also have bad effects?

  107. Its simply ridiculous how bad the USB/sound driver/API implementations are. Like we are still in the 90s...Wake up microsoft!

  108. The XP audio "Kmixer" indeed is the one with well documented problems. See the Computer Audio link in the right hand sidebar of this blog.

    As for 16 bit to 24 bit, or even vice-versa, that's not re-sampling at all. In the first case you just pad out the 16 bit value with a bunch of zeros in the least significant 8 bits. In the second case you just truncate the 24 bit value to a 16 bit value. Both are trivial operations and neither changes the timing of the sampling which is where all the problems come from.

  109. br777 here - hi, i know you've probably covered this but i still dont get it.
    why is does the volume into my headphones vary depending on my source.
    example - using a dac via my O2, is louder than using an ipod with LOD via my O2.
    i have to turn the volume up more to achieve the same level when using the ipod.
    I realize this is not something that only happens with the O2.

  110. @Br777, back in the early days of digital audio the Redbook standard specified digital playback devices should have an output of 2 Vrms for a 0 dBFS signal (the loudest signal possible on a CD or digital recording). If all manufactures would have stuck to that standard, you wouldn't notice a level difference between sources.

    But, sadly, many manufactures have deviated from the accepted standard. Portable devices, due to their low voltage batteries, often have much less output. LODs are typically about 4 times lower at 0.5 Vrms. And some "audiophile" vendors have decided more is better and have gone well above the 2 Vrms spec with their products--likely to make them seem louder in comparison to other products or as a cheap way to buy a few more dB dynamic range. In A/B comparisons the louder device will almost always be preferred.

    So, in reality, There's a range of digital source levels from about 0.5 Vrms to 3+ Vrms or 6X. That's nearly a 16 dB difference which is relatively huge.

    The above is why it's best for headphone amps to have selectable gain like the O2 does. You often need more gain with a portable source--like an iPod LOD, and less gain with a home source--like your DAC.

    1. @NwAvGuy, you wrote: "And some 'audiophile' vendors have decided more is better and have gone well above the 2 Vrms spec with their products--likely to make them seem louder in comparison to other products or as a cheap way to buy a few more dB dynamic range. In A/B comparisons the louder device will almost always be preferred."

      From the film This is Spinal Tap,

      Marty DiBergi: Why don't you just make ten louder and make ten be the top number and make that a little louder?
      Nigel Tufnel: [pause] These go to eleven.

      To address the problem of wildly varying line output levels, I've been toying with the idea of making a preamp in which each input has its own dedicated op amp (NJM2068 now ranks high on the list). Component switching would place after the input op amps and before the volume control voltage divider and output stage. That would let me trim the volume individually for each component, probably using resistors plugged into a DIP socket.

      A very good example of where this would help: I use a Yamaha AV receiver for HDMI switching and Dolby surround decoding. The preamp outputs from this go to my preamp input (I built it years ago around a now-obsolete Precision Monolithics BUF03 -- is 220V/uSec faster slew than needed for audio? LOL!), which in turn drives a Hafler PRO2400 amp. So when I switch the preamp input to the Yamaha, I have to turn the volume knob way up as the Yamaha output is variable (by remote control), rather than fixed.

  111. I repeat - the problem with Windows Vista and Windows 7 has been confirmed by Microsoft. I noticed the problem the very first time I played a YouTube clip on my netbook, and the default sample rate on the netbook was 48kHz. Changing the sample rate to 44.1kHz fixed the problem.

    Yes, Kmixer has issues as well, but XP does not suffer from the above problem. (at least, not to nearly the same extent) \Greg.

  112. @SullivanG, On my list of future articles is one on USB audio that will cover sampling rate conversions, bit accuracy, bit depth, and more. I'll be running some tests of operating system re-sampling. If you have any references you can link as to the "confirmed problem', please post them here or send them to me privately. Thanks.

  113. @NwAvGuy: Please read this thread on the MSDN Developers Forum:

    It goes down many tangents, but Microsoft eventually acknowledged the resampling problem in the message timestamped Tuesday, March 29, 2011 3:56 PM (I can't link to a specific reply in the thread) \Greg.

  114. Thanks SullivanG! It looks like it's specific to WaveOut. That's important as the dScope and some of my other software can bypass WaveOut. Based on what's in that thread, those who absolutely have to let the OS perform sample rate conversion might want to look into using something like Foobar and using DirectSound or other output methods.

    It would also be interesting to know if Microsoft has fixed the bug yet--at least for Win7 (I suspect they'll never fix Vista). One could hope it would at least be in SP2 for Win7.

    Finally, I want to stress everyone should be avoiding sample rate conversion if at all possible in any operating system including OS X and Linux. It's always an imperfect process--especially to do it on the fly in real time with minimum CPU loading on slow PCs (i.e. the criteria for OS-based SRC).

  115. Yes, WaveOut and WaveIn. Another issue mentioned in that thread is that Windows 7 (and Vista, presumably) no longer changes the sample rate of the hardware when WaveOut applications request it. This is very important sometimes, for example when testing soundcards! We now have to make sure that whatever hardware settings we want to use, are manually set in the playback/recording device properties, because otherwise the soundcard tester will ALSO be testing the quality of Windows sample rate conversion, and of course the results will be terrible. Also note that someone else in that thread seems to have found a similar resampling problem with DirectSound, but only on Vista. They could not reproduce this problem on Windows 7. It is good that DScope can bypass WaveOut. \Greg.

  116. Some time ago, somebody pointed me to your blog, and to an article that said that sine waves were useful because you can create any signal (really any finite energy or finite slope signal, which includes all real electrical signals) from a sum of sine waves.

    This is true. However, it's also true that you can create an infinite number of such "basis vectors".

    Sine waves are, however, important, because your ear does a mechanical frequency analysis on the cochlea (literally) and thus the sinusoidal domain relates very nicely to human hearing.

    If you run into deep foolishness, please feel free to drop me a line, I'd be glad to help stomp that out.

  117. If CPU load is any indicator, then Windows Media Player on Windows 7 doesn't perform SRC, so there's a mechanism in Windows 7 for calling native DAC rates. It's the only media player I've used that maintains ~3% CPU usage with lossless playback. All the others use far more.

  118. @Anon, thanks for the input on sine waves, etc. If you want to contribute an article on the topic I'll gladly publish it.

    @akg & Sullivan, I can attest that Win7 can use native device rates. I can play a special dScope generated test file and the dScope will determine if it's bit accurate. Several USB audio devices I've tested maintain bit accuracy in Win7 automatically at different file sample rates even played back in Windows Media Player or Foobar. So it's clear Win7 is not re-sampling under those conditions. When you play a different file the output automatically switches to the new native format--i.e. from 16/44 to 24/96.

    I think the problem comes in when you play a file in a format not natively supported by the device AND use WaveOut. I have not verified it, but the thread referenced above implies a bug degrades the quality of the SRC under those conditions.

    The comments to this article are probably not the best place to discuss the intricacies of PC audio. Please stay tuned for a more closely related article where the comments are more likely be found in the future by those curious about such things.

  119. @NwAvGuy - yes - it's WaveOut, but the file format does not matter I don't think. Unfortunatey, a LOT of web audio uses WaveOut - YouTube & FLASH content in general. For example, the SoundCloud audio hosting site suffers from the problem. In my testing, Microsoft Silverlight does NOT suffer from the problem. I think some sound card tester utilities use WaveOut as well. As I say, for the time being set your soundcard to 24bit/44kHz for best web audio quality. \Greg.

  120. I was surprised to read the following in an earlier post of yours: "Furthermore, your uDAC2 can't run at 24/96 over USB. So you're probably forcing your 44 Khz music to be upsampled to 24/96, and then it's downsampled back to 16/44 or 16/48 by the operating system/driver before it's sent to the DAC."

    On NuForce's website, they clearly state that 24/96 is one of the "native" modes: "USB DAC: USB 1.1, 2.0 compatible. USB native bit rate: 32, 44.1, 48, and 96 kHz, 24-bit"

    Are they just flat-out lying there? Or are they cleverly stretching the truth by stating the *internal* DAC is 24/96 capable while never actually sending it 24/96 over USB?

  121. Hi there NwAvGuy,

    I have been eagerly waiting your FiiO E10 review, as I'm sure have many others. And as we have been looking forward to your more trustworthy and detailed review, FiiO released yet another Headamp/DAC, the E17.

    Are you also planning to review the E17 in the near future? I ask you, because although it is said to sell for some 150 USD -almost the double of the E10-, it certainly poses a dilemma for those of us who have been waiting to read your more substantial and trustworthy review on the E10, in order to make a safer bet in getting one.

    I don't want to use this thread for what it isn't meant for, but since you mentioned to have concluded the tests on the E10, I thought it might not be totally out of the question. Another question I would ask is that, since the E17 will sell for a similar price to the Audinst HUD-mx1, would it make sense to compare them before spending the money?

    I also wanted to thank you for your great blog and for your sincere efforts to set the record straight when it comes to what one can really expect from headphone amplifiers, DACs and the listening experience in general, as well as sharing the great O2 and the ODAC designs.


  122. @Paul, I was confusing the uDAC-2 with another product. You are correct it supposedly does support up to 24/96 over USB. Sorry about the confusion. It still, however, is a significantly flawed product unless they have changed the design from last year when I tested it.

    @Anon, the FiiO E17 isn't yet generally available in the USA. If it's anything like the introduction of the E10, it may be six months or longer before it is. If there's enough interest I may test it someday. it seems to be the replacement for the E7.

    I would generally trust FiiO over Audinst. Both have made significant mistakes with products, but FiiO is clearly a huge cut above Audinst with more dealers and much more advanced products in terms of packaging, features, etc. Audinst is basically just another "me too" eBay vendor and the one Audinst DAC I tested had serious problems.

    1. NwAvGuy,

      I've come across a Korean review website that tries to objectively measure the gears, and the website seemed to think Audinst performed pretty well based at least on the criteria they've set. How does your result compare with that of this website's?: http://goldenears.net/board/1577408

      I don't have the audinst and am not trying to defend the product. It was one of those products that I was thinking about getting in the future, so I just want to confirm whether it really is just a mediocre product or if your audinst was a lemon.

      I'm looking very forward to your ODA and ODAC btw!

  123. "Thank you for your mail and support to FiiO!

 Yes, you are right , our E10 will cut off the music about 0.7mS to solve the phase shift distortion of the WM8740 with the TE7022 USB receiver.

 and it can not solve by any firmware or fixed it by hardware. but our E17 have not such problem because the structure is different, the digital signal

 will go through wm8804 so we don't need to reset the WM8740 each time when there are new signal to make sure it work perfect. 

If you have any other question, please feel free to contact us!


    FiiO's reply to a head-fier's inquiry about the initial delay in playback. Just thought I'd let you know about this issue. I assume all E10s have that issue (including yours) And does you E10 have the loose jack issue like quite a few people?

    I'm still waiting for my E10 to arrive, as the seller forgot to pack it with the headphones I bought.

  124. @Anon, it's possible Audinst makes some decent products but I have not seen credible measurements of their gear and the one product I tested was disappointing. Like so many similar manufactures, they seem more interested in using popular FOTM chips and pretty PC board layouts than sweating the design details that result in the best performance.

    @kingpage, I'm not sure what FiiO is trying to say using Chinese English in that quote. I suspect they're talking about configuring the WM8740's digital filtering for the correct sampling rate but I'm not sure how they would do that without a microcontroller. I would have to spend some time with the WM8740's datasheet, etc.

    The headphone jack on the E10 I tested likes to eject the plug with even a slight tug which happens often with headphone listening. It's not so much that the jack is loose, but the spring loaded contact is not aligned correctly inside. Instead of firmly seating into the notch at the tip of the plug, the spring tension is always on the verge of pushing the plug out of the jack. Oddly the line out jack on my sample was fine.

    A side effect of the above is the AD8397 op amp is briefly short circuited every time the plug pops out with music playing. The AD8397 is not short circuit protected and, due to its high current capability, is well known for blowing up when shorted. I assume FiiO has added some protection of some sort against shorts but it's still not an ideal situation to have the jack pop out while listening.

    If I keep this up I won't need to publish my long overdue E10 review ;)

    1. they have since changed the jack. http://www.fiio.com.cn/news/index.aspx?ID=146&page=1

  125. I have the same issue with the E10. The jack always feels really springy and likes to pop out the plug

  126. Hi NwAvGuy,

    I have recently purchased Denon AH-D2000 headphones and a FiiO E10.
    So far I am very pleased with the pairing, although I find I am still adjusting to the bass, coming from Sennheiser HD555s previously. I am planning to build the ODA / ODAC once its design is complete.

    I am however having a small problem with my E10 and am wondering if you might have any idea what could be causing it.

    Whenever I set the E10 to use any form of 24bit sampling within Windows 7, I find that on inception of playback from no signal the volume drops heavily and the sound distorts for several seconds before returning to normal. This is particularly noticeable with push to talk applications in which the microphone transmitting 'beep' is constantly dipping in volume and crackling.

    Apologies for picking your brain, I realize this is not a help desk but can not find much information on the subject and was wondering if you have ever encountered this problem with your E10 and 24bit sampling?

    Best Regards

  127. I've already returned the E10 I tested to its owner but I didn't notice that problem. As mentioned above it does have a slight delay for a fraction of a second before it starts playing. I also know it hates 88 Khz sampling rates but that seems unlike in your case. If you're using it for chat/voip/skype/etc I would suggest just running in 16 bit mode if you can't figure anything else out.

    Also, to be entirely honest, for the D2000 the ODA/ODAC may not be a very audibly significant upgrade over the E10. The D2000s are relatively easy to drive for full size cans and just need a low impedance source. The E10 fits that bill. So unless you get some more hungry cans, you might want to save your money and just keep the E10.

  128. Hello NwAvGuy, when would you be able to post your E10 review? Thanks, looking forward to it!

  129. Would like to ask for your opinion if in a DAC, assuming somewhat proper implementation, would the same DAC chip and USB receiver mean 2 different DACs are indistinguishable, even if their basic circuit differs.

    I ask this as I am interested in Audioengine's new D1 DAC. It is based on the TAS1020B and the AKM4396, similar to the DACport, but at a cheaper.price point.

  130. @mykeldg, the FiiO E10 Review is PUBLISHED (countless typos and all)!

    @anon, the short answer is "no". As I'm fond of saying: "Implementation is everything." A perfect example is the FiiO E11 is a virtual clone of the AMB Mini3. They use the exact same op amps and basic toplogy. But the E11 far outperforms the Mini3 because FiiO did a better job with the PCB layout, power supply, grounding, etc. I've seen several other examples with various eBay clones of respected audio products. They typically use all the same chips, and look great, but perform poorly because of the PC board design, layout, substitution of some cheap Chinese parts, power supply problems, etc. The differences can be huge and audible.

    As yet another example, my own O2 headphone amp design uses really inexpensive basic components that many audiophiles would never consider suitable for a high performance amp. Yet the O2 outperforms anything even close to its price and many headphone amps costing many times more that use far more exotic parts. The reason is I took the time to address all the implementation details. In contrast, a lot of audio designers and manufactures just slap parts on a board and ship the product hoping nobody will ever fully measure its performance.

    So, long story short, I have no idea how the Audioengine D1 performs. My experience with Chinese/Asian designed audio products has been hit or miss. Many such products seem to emphasize the looks and chip brands/part numbers over proper design and genuine performance. They often provide minimal and/or vague specs and are never independently tested. Despite a strong cult-like following on certain forums, many such products proved to have serious flaws. Look up my NuForce uDAC-2 review and the Banned at Head-Fi articles for just two examples.

  131. Hi NwAvGuy, in reference to your above comment, my take on the difference between the Mini3 and Fiio E11 was the power supply to the opamp chip/s, I thought fancy PCB layout are only there to look pretty. :) It seems some people believe the deciding factors of sound quality in a DAC are the power supply, DAC chip, capacitors and opamp, is this incorrect?

    Looking forward to those coming attractions! Perhaps you can add audiophile software to the list? Such as Jplay.

  132. @kiteki, You're partly right about the Mini3 vs E11. In this case, the "third channel" is part of the Mini3's power supply (it's a virtual ground or rail splitter) and AMB did a poor job implementing it. The rest of the power supply in the Mini3 is actually better than the E11.

    Look at my MacBook Air review for just how important PCB layout is. One channel of the Air performs far worse than the other. Yet they both use the same chips and the same power supply. PCB layout can be more important than even the power supply. It can easily be the "weakest link" in some designs.

    The capacitors and op amps are usually among the least critical items in a lot of audio gear. See my Op Amp Myths & Facts article for more about the op amps. And if you want to know more about capacitor "sound" see my O2 Design Details article.

    Most modern DAC chips are capable of transparent performance when they're properly implemented. The problem is they're often not properly implemented. The implementation usually matters more than the chip itself. And a big part of that is the PCB layout.

    When op amps are directly driving headphones, as they are in most Cmoy-like designs including the Mini3, the choice is important because most op amps are not well suited to driving such a low impedance load. Many assume, for example, a OPA2134 Cmoy will sound better than a 4556 Cmoy because the 2134 is more expensive and has a better reputation, but the opposite is true.

  133. i plan on using an old laptop (trusty thinkpad t42) as source for your ODAC, do you have any recommendation on OS, player, plugins or any triks to optimize it just for "UBS streaming"? ...one more question, any benefit on using a usb isolator as http://www.circuitsathome.com/measurements/usb-isolator

  134. Hey NwAvGuy,

    How is the ODA/ODAC coming along? I know you said you'd have the update on them soon in January, but I was hoping to get a simple idea on your progress!

  135. There hasn't been much ODA/ODAC progress to report and it's purely because I haven't had as much time as I hoped to devote to the project. I also haven't been keeping up with the diyAudio and other threads. But hopefully I'll have more free time soon.

  136. For ODAC sources, see the Computer Audio article perma-linked in the right hand column of this blog. USB tricks are only necessary if you end up having an audible noise problem. No noise, no worries. Headphone gear often isn't grounded in anyway so you don't have to worry about the sorts of ground loops that can happen when connecting computer audio to home systems with speakers, A/V components, etc.

    There hasn't been much ODA/ODAC progress to report and it's purely because I haven't had as much time as I hoped to devote to the project. I also haven't been keeping up with the diyAudio and other threads. But hopefully I'll have more free time soon.

  137. Hi Mr. N,

    I have a noob question. I remember reading you saying that the driver of Apogee is decent and that you recommend Duet for Mac. But since Apogee Duet is quite expensive, relative to Apogee One (and of course to ODAC too), I am wondering if One serves the same function for audio playback (I read people saying that One only has mono recording whereas Duet has stereo...).


    1. I haven't tested one, but I've heard only good things about the Apogee One as long as you're a Mac user. If you want it for recording, rather than playback, you have decide what you're really going to do with it.


PLEASE NOTE! If you see "LOAD MORE..." OR "LOADING..." just above this paragraph, you need to click that link to load the newest comments. It may take several seconds. When all the comments are displayed the Load More/Loading link disappears.

COMMENT APPROVAL: Comments may take anywhere from a few minutes to a few days to be approved. You might want to save a copy (highlight Ctrl-C) of your comments before submitting in case something goes wrong. The approval process is necessary to prevent obscenity, etc. Please keep comments on topic. Comments promoting commercial interests will not be approved and may be reported as SPAM.