Benchmark DAC1 USB D/A processor & headphone amplifier Manufacturer Comment
Editor: Thank you for reviewing the DAC1 USB. Your keen observations and measurements of the USB's performance were a bit of a surprise to us, given the intense testing that we had performed during development. For a variety of reasons, I believe that we have developed the best USB audio technology currently available. The problem is that operating systems and media players can and do change frequently. We have re-created your tests, created some additional tests, and have performed a rather exhaustive (and exhausting) test of the available players and operating systems.
Based on our latest tests, we believe that you were hearing the effects of 16-bit truncation and/or operating-system–based sample-rate conversion. We have discovered a very serious bug in iTunes 7.5. We are in the process of updating our audio wiki pages with the latest information regarding iTunes. Fortunately, the truncation and sample-rate-conversion problems can be avoided in both Apple and Windows systems. Unfortunately, the iTunes 7.5 bug makes iTunes unsuitable for playback of 44.1kHz CD material. We hope that this bug will be fixed in the near future (details are listed below). There are several players that work flawlessly. Unfortunately, iTunes is no longer one of them! We will be reporting the media-player bugs and workarounds on our computer-audio wiki pages (www.benchmarkmedia.com/wiki).
The fade-to-noise tests reveal an entirely different issue that will not affect the audio in most circumstances (details below). These test signals activate our USB auto-mute circuit, and have several characteristics that make them different from music signals: they have 0DC offset, no noise other than TPDF dither, no high-frequency content, and a single low-level test tone. In trials with actual CDs we have not been able to find any tracks that activate the mute circuit. The only exception to this is that some 16-bit end-of-track fades may mute when they fade below –70 to –80dBFS. Please note that this can occur only when the track was mastered without the use of shaped dither. Under these limited circumstances, the mute circuit can cause a loss of the last fraction of a second at the end of the track.
We were able to correct this problem in firmware and will be shipping you a new unit. Again, we do not believe that this is ever an issue with CD playback. We have a test setup that monitors the mute signal while executing a playlist of 16-bit CD tracks. Nevertheless, we have decided to remove this mute function to ensure that it will never be an issue. You will find that the new unit will operate as expected with any fade-to-noise signal you send to it.
Here are more details:
1) The current version of iTunes (7.5) for OSX is very broken. Some older versions work properly, but 7.5 introduces truncation and DSP artifacts that can measure as high as –80dBFS. These levels are easily audible. There is no system configuration that will permit transparent playback from 44.1kHz sources with iTunes 7.5. Curiously, the only format that works properly in 7.5 is 24-bit/96kHz. This iTunes problem has nothing to do with USB communications, as the results are the same when using the built-in optical port on our PowerBooks. As a workaround, we suggest using the free VLC for Mac player. VLC works perfectly with all formats (provided OSX is set to match the sample rate of the files being played). We highly recommend the VLC player for OSX users, and strongly advise against using iTunes. We will attempt to identify which versions of iTunes work and which ones don't. But beware, iTunes updates itself without user intervention. One of our test computers "upgraded" itself to the dysfunctional iTunes 7.5.
2) OSX still does not have the ability to follow sample-rate changes. We consider this a nuisance for most users and a show-stopper for users who want to play a mixed list of 44.1kHz and 96kHz material. An OSX sample-rate mismatch will invoke the very-poor-quality sample-rate conversion that is built into OSX. The iTunes 7.5 bug seems to be locking on this sample-rate conversion at all sample rates other than 96kHz. Until this is fixed, OSX is not the operating system of choice for audio playback.
3) The Windows Media Player volume control truncates to 16 bits when playing 16-bit rips, but operates at 24 bits when playing direct from CD. Workaround: Do not use the WMP or Windows System volume controls when playing from a 16-bit rip (use the analog volume control on the DAC1). The output word length of the volume control should match the word length of the output device, not the input file. Essentially, this is a Windows Media Player bug, and we will report it to Microsoft. If WMP is fixed to provide 24-bit output, the Windows System volume controls will function in 24 bits. No operating-system changes should be required. I expect that this problem will be resolved in the near future.
4) Windows Media Player performs HDCD decoding when playing directly from a CD, but will not perform HDCD decoding when playing from a rip. Certain HDCD features can make very noticeable changes in the audio. I have one CD in my collection that plays about 12dB quieter when HDCD decoding is active. I suspect that the CD was mastered improperly, as this HDCD gain-reduction feature should not have been turned on. In this mode of operation, HDCD will actually reduce resolution by 1 to 2 bits! CDR burns of lossless rips behave exactly the same as the original tracks when played directly by WMP. In other words, WMP will decode HDCD tracks after a rip-and-burn cycle, provided the audio is played directly from the CDR. Rip the CDR and HDCD turns off when playing the rip. Burn the rip back to a CDR, play the CDR, and HDCD is active again. Bottom line: rips can sound different from the original because of a lack of HDCD decoding when playing rips. We believe that this HDCD-decoding issue is a WMP bug. We see no reason to turn off the HDCD decoder when playing lossless rips. We suspect that this is an error in Windows Media Player. Again, we will submit a bug report. WMP will have a unique advantage over all other players if they fix the HDCD playback from lossless rips.
5) The latest version of Foobar is nearly flawless and foolproof. We highly recommend it for Windows platforms. The output has a 24-bit path and the volume control operates with full 24-bit resolution. It supplies a 24-bit signal to the Windows master volume control, and this forces the Windows volume control into a 24-bit mode of operation. With Foobar, it is perfectly okay to use the Windows volume control.
6) Lossless rips made under WMP seem to be truly lossless. Multi-generation lossless rips and burns will still activate the HDCD decoder in WMP when played directly from the CDR. Our tests have not revealed any loss in quality.
7) Stereophile's fade-to-noise test (from the CBS test CD) revealed the presence of our USB auto-mute circuit. This circuit is activated by a string of successive samples having a value of 0 in all bits. We tested this circuit with many CDs when we developed the product, and tested it again after Stereophile's review. In no case could we find a CD that would inadvertently trigger the mute circuit. However, we did find rare circumstances in which it could mute a fraction of a second of a track fadeout if the fade was created in mastering rather than in an earlier mixdown. The test track from the CBS disc has some unique qualities that differ significantly from normal music tracks. The test track has absolutely no DC offset, no high-frequency content (other than dither noise), and has absolutely no noise other than mid-tread digitally generated TPDF dither. We modified the USB mute function to ensure that it will never mute when the USB interface is receiving audio data. All DAC1 USB units are now shipping with the modified firmware.
8) Windows Media Player still does not support playback of 24-bit files without the installation of third-party codec plugins. We recommend the Combined Community Codec Pack. Once this codec is installed, WMP plays 24-bit files properly. We expect that Microsoft will add a 24-bit codec soon.
9) Installation of a DAC1 USB is fully automatic under Windows Vista, XP, and 2000. Under these operating systems, the DAC1 USB can be plugged in or removed at any time without changing any audio settings.
10) Installation of a DAC1 USB under OSX requires manual selection of the audio output device. Under OSX, the DAC1 USB can be plugged in or removed at any time, but the audio output device must be manually selected. We will encourage Apple to fix this problem, which applies to all audio devices and is not unique to the DAC1.—John Siau, Director of Engineering, Benchmark Media Systems
Manufacturer's Comment #2
Editor: After extensive testing and communicating directly with the engineering team at Apple, some of these initial observations have been explained. We now know the reason for the poor performance observed in the initial tests, and we have conclusive information about the operation of iTunes 7.x on Mac OS X.
iTunes 7.x can work very well on Mac OS X with the proper configuration, but the wrong configuration can cause serious distortion. It is important to understand the fundamental modes of operation of both iTunes and CoreAudio, specifically regarding sample rates.
CoreAudio and iTunes can simultaneously operate at independent sample rates. At all times, the sample rate set in AudioMIDI Setup dictates the sample rate at which CoreAudio is operating. When iTunes is launched, iTunes locks to the sample rate at which CoreAudio is currently operating (which is the sample rate that is set in AudioMIDI) and does not change until it is closed and re-launched. However, after iTunes launches and locks its sample rate, its sample rate will not change thereafter, even if CoreAudio's sample rate setting in AudioMIDI Setup is changed. To change the sample-rate of iTunes, iTunes must be shut down, and then restarted after the desired sample rate is set in AudioMIDI Setup.
The audio being played in iTunes will always stream from iTunes at the locked-in sample rate. In other words, if iTunes is locked to 44.1 kHz, all audio with other sample rates will be converted to 44.1 kHz by iTunes. iTunes then streams this audio to CoreAudio to be mixed and/or streamed to the hardware.
What does all of this mean for the end user? If the user changes CoreAudio's sample-rate in AudioMIDI Setup to something different than what iTunes is locked to, CoreAudio will convert the sample rate of the audio that it is receiving from iTunes. In this case, the audio may be undergoing two levels of sample-rate conversion (once by iTunes and once by CoreAudio). (The SRC in iTunes is of very high quality (virtually inaudible), but the SRC in CoreAudio is horrible and will cause significant distortion.) If the user wants to change the sample rate of CoreAudio, iTunes should be restarted so that it can lock to the correct sample rate.
We are suggesting two different recommended solutions to our customers:
1) The "Set It And Forget It" solution for iTunes 7.x: Before opening iTunes, set the sample rate of CoreAudio (in AudioMIDI Setup) to 96kHz. Do not change the sample rate of CoreAudio unless iTunes is restarted after the change is made. This solution will prevent CoreAudio from applying SRC, as the quality of CoreAudio's SRC is horrible. Also, by having iTunes locked at 96kHz, all audio with sample rates below 96kHz will be up-sampled to 96kHz. This will cause virtually no loss in sonic quality, as the quality of iTunes' SRC is very good—virtually inaudible. Also, by avoiding down-sampling by iTunes, this setting will never result in a loss of bandwidth.
2) The 'Bit-Transparency For Each Sample Rate' solution: (Note: This solution is rather cumbersome, offers virtually no quality improvement over the first solution, and can easily be mis-configured, which will cause severe distortion.) Before opening iTunes, set the sample rate of CoreAudio (in AudioMIDI Setup) to that of the audio you will be playing. Do not change the sample rate of CoreAudio unless iTunes is restarted after the change is made. This solution will prevent CoreAudio from applying SRC, and avoid SRC by iTunes for all audio with the same sample rate that iTunes is locked to.
Also, the end user should not hesitate to use the volume control in iTunes 7.x, as it is very well designed and operates at 24-bits for audio devices that support 24-bit operation.
For iTunes 6 and earlier, simply set the sample-rate in AudioMIDI to match that of the audio being played, and keep the volume control at 100%.—Elias Gwinn, Benchmark Audio