Why negative reviews on USB with newer USB DACs?


I switched over from a CD transport as primary music source to a MAC Mini music server about 1 1/2 years ago using the USB output to my DAC via a HAGUsb converter. I found the results far superior than any cd transport I used with S/PDIF coax connections to the same DAC. Over the past six months I keep reading professional reviews of USB DACs like the Stereophile review of the Bel Canto DAC 3 and the Six Moons review of the Cullen Modified PS Digital Link III where the reviewer was unimpressed with the USB input vs. the S/PDIF input of the same DAC. I find this a little bewildering with all the reported successes on this and other forums with Lossless based hard drive music servers being vastly superior to tradional CD drives via the USB output. The professional reviews lack specifics so it's hard to know if the computers used were optimal in setup, e.g., loss-less files, cables, etc.. I'd be interested in hearing from other forum members who are dissappointed that professional reviews are sending a message that USB is not the way to go when if fact it is!

A few other details on my systems: HagUSB is AES/EBU version. Connected to the XLR input of my PT 1A/PT 3A/ Monolithic P3 digital front end. The USB cable is a Belkin Pure AV. The power cord for the MAC Mini is BPT L10 with Oyaide 079 ends connected to Oyaide AC outlet. BTW, the PC upgrade from stock PC brought a remarkable improvement to the sound from MAC Mini. And all files are Apple-Lossless.
ed_f
I agree with gmood1. Also not all usb implementations are created equal and not all dacs are created equal.

I used to use an Emmlabs CDSD transport and DCC2 dac. My PC->USB->Empirical Audio Turbo-2->DCC2 produced the same result. The sound was indistinguishable. So I sold the CDSD because it was less convenient. Part of this was using software that did bit-perfect rips and bit-perfect playback, and bypassing the kmixer, and using a battery power supply, and no ASIO driver, and no resampling. As long as the bit stream from the PC is bit-perfect and the USB interface produces the same or less jitter than your transport, then the PC should sound the same or better than the transport. But it takes all of these things to be in place. You can't play a song through Windows Media Player and a poorly-implemented USB interface and expect great results.

I also owned a Benchmark USB dac at the same time as my CDSD/DCC2. It didn't come close to Turbo-2 + DCC2. I don't know if the problem was the Benchmark USB interface or just the Benchmark dac, but it was nowhere near the same league as Turbo-2 + DCC2. I ended up returning the Benchmark within the 30 day return period. Granted, it was a much cheaper product, but frankly I don't see where all the great reviews are coming from. Very bland sound.

So, you still need a great sounding dac for great sound, and you also need a well-implemented USB interface to let the dac shine, and you need bit-perfect output from the PC.
Most of the "other" USB implementations you mention use one of the family of TI PCM270X chips.

These are plug-and-play but not very good jitter-wise compared to the TI TAS1020 or the TUSB3200. The 270X chips also dont support 24/96 operation. The TAS1020 and TUSB3200 do. The latter require USB firmware to be written, so they are more costly to implement. Most manufacturers went the cheap easy way and now most of them regret it. Some are buying USB modules from me now that use the TAS1020.

Steve N.
Empirical Audio
Steve,

My experience with USB is limited to the AES/EBU version of the HagUSB which I believe uses the 2704 chip. I haven't experienced any downside. In fact, I am quite pleased with the HagUSB both from a performance and operational standpoint. It may perform well in my system because the PT P1a reduces jitter down to a very low 2 picoseconds and I have my Mac Mini on a dedicated line with Oyaide outlet and L10 power cord.

Which leads me back to my original question about the Bel Canto and modified PS Audio DAC reviews. Is it your feeling that since they strickly use 270x chips they can't provide USB performance on par with S/PDIF input? Or it also a matter of poor jitter reduction after the 270X that's not as good as the jitter reductions for their respective S/PDIF inputs?
Ed - it is all relative. I have lots of customers with HagUSB and other converters. Once they get one based on the TAS1020 they never look back. Big difference. This is what the reviewers hear in the BelCanto and others that use the 270X chips. The AirPort Express uses the 2705 and even with a ton of mods and a Superclock4, it still does not hold a candle to the TAS1020. As a result, I discontinued it as a mod.

Steve N.