OK, at long last, here is my review of the Berkeley Audio Design Alpha USB adapter. For people not familiar with this class product, this is a box that acts an external sound card. Its input is USB which means it nicely works with desktops and notebooks. The output is digital audio. In this case, the box supports both S/PDIF in the form of Coax BNC connector and AES/EBU which is a balanced audio output using XLR connections. You would connect this device to your favorite DAC, bringing it to the modern age of PC based music servers.
The other key differentiator here is the asynchronous interface to the PC. Instead of extracting its output clock from USB bus with its inherent instability, the machine turns the PC/Mac into a data pump, reading pure data with no sense of timing. The audio samples are stored in memory and then “clocked” out using a precision, custom oscillator (and other goodies they don’t let me talk about ). In addition, there is electrical isolation between the PC and digital audio outputs eliminating ground loops and noise from being transmitted through to the DAC.
The reason we do all of this is that when it is all said and done, we need a clean clock signal and power supply to convert our digital audio samples to analog. Better hygiene upstream goes a long way toward improving the performance of DACs. Paradoxically, even high-end DACs benefit from better interfaces. The only time I have seen no benefit was in a low end processor.
The Berkeley is a rather small box but it is big enough to have a power supply built in which is nice. And just heavy enough to support the cables connecting to it. In some way I wish it was the size of a normal Audio equipment (e.g. the Alpha DAC).
With that introduction, let’s get into what it sounds like. But first, a couple of important notes:
Note 1: This is a purely subjective review. It is entirely possible that my conclusions are wrong. Certainly the power of placebo is far higher than the small differences I hear. In my short testing, and the way these devices work, it is hard to set up the type of objective tests I like to have to reduce the effects of bias.
Note 2: My company carries the Berkeley audio products and in previous life, I acquired Pacific Microsonics/HDCD which was founded by the same people behind this product.
My test set up was the following:
1. Desktop PC with an SSD boot drive. Music on hard disk. Played using Windows Media Player. Control panel set to 24-bit, 44.1Khz for all devices for CD material. I find that 24-bit setting eliminates a lot if issues with Windows audio pipeline as the dither added is below the 16-bit audio samples.
For 24-bit material, I had 96 KHz samples I had saved from old days of Audio DVDs (not to be mistaken with DVD-A) which lacked encryption and hence were easy to rip. I set the control panel for all devices to 24-bit/96Khz to assure no resampling.
2. I had three connections on hand: Toslink out of the PC motherboard. Audiophilleo2 Asynch USB adapter (the one without the display). And Berkeley Alpha USB adapter. The latter connected using generic/random USB cables.
3. The DAC was my old friend, the Mark Levinson No360S. This is nice unit in that it has every kind of input. This allowed me to hook up the Audiophilleo using BNC connection, the Alpha with AES/EBU and the optical using that input.
4. The output of the ML DAC went into another old standby, a Rotel pre-amp. This gave me analog volume control so that I could set the volume control to max for all devices and player to 100%.
5. The output of the Rotel drove my NHT monitor speakers and sub. This is my home workstation setup which is where I was testing tonight. At some point I will also test in the office with the ML amplifiers, Revel speakers and Berkeley DAC.
Switching between connections involved having the sound control panel open and making the device I wanted to hear the default while WMP was playing and then switching the input on the DAC. This meant a 3-4 second gap. Pretty annoying and likely reduced my ability to here differences by a good bit. It also made the testing long and tedious.
This presented a good scenario. If all works as in the theory, the toslink should be the lowest performing, followed by Audiphilleo and then the Berkeley. Well, that is not how it went. At least not initially.
I played track after track. And while the toslink would often end last, in most tracks the Audiophilleo beat the Berkeley. I went through a ton of material and I would either hear little to no difference between devices or the Audiophilleo would win. I even ripped out the audio system and put in my Stax headphone system and still no cigar.
I was almost going to come and report everything as such and then I realized something. I had used an audio cable for AES/EBU connection. I swapped that out with an AES/EBU cable and I was taken back by the difference it made. It completely reversed the roles of the Audiophilleo and Berkeley with the latter always sounding better when there was a difference.
Reason for the difference in performance is that the asynch interface cleans up the PC interface but you are still stuck with this ill designed interface on the output, even in its pro version. Waveform distortion caused by cable mismatch impedance results in increased jitter feeding the DAC.
What did I hear? First, on many tracks nothing. As I mentioned, on a lot of material the difference is either not there or if it is, it is too small to rely on it existing in such an ad-hoc test. Where it did surface, it was predictable. The high frequencies were slightly cleaner. And ambiance and decay more pronounced which impacts the perceived soundstage of music/instruments. I hate to use this word but I can’t think of any other to way that the overall experience was more “analog like” in the sense of it sounding more pleasant.
I found the difference easiest to hear on my audiophile A-DVD 24-bit, 96 Khz tracks. Perhaps due to higher sampling rate, jitter reduction matters more. Or else, these are just better recordings so the difference is easier to hear.
For the objectivists in the crowd, the difference was never night and day. All three interfaces were dynamic, and were enjoyable to listen. The Berkeley and Audiophilleo simply added a bit more fidelity on top. It is like putting some pepper on a good stake .
My general opinion is that these devices improve the performance of any DAC you already have, even if it is a low end one in an AVR. They perform a nice function of providing a digital audio interface for your PC. They cost a few dollars, but in the grand business of high-end audio, they are a reasonable way to tweak your system.
So if you have the cash, by all means go with the Berkeley. It retails for $1,895. The Audiophilleo2 I tested goes for $579. The Audiophilleo1 which has a nice little jitter simulator and such, costs $949.
I realize that this is a 6-year old reply, but I'd like to reassure you (if all remains the same -lol) that no one will be the wiser when it is disclosed: " It is entirely possible that my conclusions are wrong." (Were you doing a dissertation at the time -or now, by chance ?)
It kind of reminds me of The "Fonz", being unable to mutter the word 'wrong' !
Right then, let's carry on ...