Did you happen to compare the sound of the offramp direct from your computer into the I2S, with the PWD MKII's USB input without, your Trinnov in the mix?
Audio Voodoo - A tale of Offramp 5 and PWD MKII
I own a perfectwave dac MKII, and recently inserted a Trinnov all digital room correction / cross over system into the digital chain before the DAC. The upshot is I can no longer use the PWD bridge or I2S input, because the Trinnov only has AES/EBU digital in and out.
Thanksfully, the MKII upgrade came along just in time, because the difference between AES/EBU and I2S are now very minimal.
Since in NativeX mode the PWD MKII is asynchronous, you would expect the PWD MKII has become largely immmune to jitter and noise in the upstream digital signal path, and any money spend on very good clocking and jitter and noise reduction upstream is a waste of money at worse, and a very low return proposition at best.
While a plausible argument, I wanted to test the theory and got myself Empircal Audio's new Offramp 5 USB converter with turbo-clock upgrade (30 day trial). This replaced a hiface EVO with bolder power supply. My source is a CAPS 2.0 server, running JRiver with Jplay plugin. Boy was I in for a surprise. The offramp 5 was a big step up.
Next, a used monolith battery power supply for the offramp shows up on audiogon, and naturally I scoop it up. Now things get really interesting - this was an ever bigger step up that than the Offramp 5 itself.
Bottom line: even though the science of digital audio signal processing has improved vastly, apparently there is still a voodoo element to it. Theory can only go so far explaining what will sound better. No substitute for trying things out yourself. I could not be happier with the Offramp 5 and monolith.
Thanksfully, the MKII upgrade came along just in time, because the difference between AES/EBU and I2S are now very minimal.
Since in NativeX mode the PWD MKII is asynchronous, you would expect the PWD MKII has become largely immmune to jitter and noise in the upstream digital signal path, and any money spend on very good clocking and jitter and noise reduction upstream is a waste of money at worse, and a very low return proposition at best.
While a plausible argument, I wanted to test the theory and got myself Empircal Audio's new Offramp 5 USB converter with turbo-clock upgrade (30 day trial). This replaced a hiface EVO with bolder power supply. My source is a CAPS 2.0 server, running JRiver with Jplay plugin. Boy was I in for a surprise. The offramp 5 was a big step up.
Next, a used monolith battery power supply for the offramp shows up on audiogon, and naturally I scoop it up. Now things get really interesting - this was an ever bigger step up that than the Offramp 5 itself.
Bottom line: even though the science of digital audio signal processing has improved vastly, apparently there is still a voodoo element to it. Theory can only go so far explaining what will sound better. No substitute for trying things out yourself. I could not be happier with the Offramp 5 and monolith.
6 responses Add your response
Glad it worked out well for you. I recall the related discussion a while back in this thread. I wouldn't call it voodoo, though. It just means that like pretty much every other component design, the PerfectWave DAC is not, er, perfect :-) I don't know specifically how the NativeX function works, but I would expect that if detailed design information were available it would not be hard to hypothesize ways in which your findings could be explained. For example, extraneous high frequency spectral components on the incoming signal, associated either with jitter or simply with digital noise, could to some degree find their way through stray capacitances, grounds, power connections, etc. such that they bypass (couple around) the jitter reduction circuitry, and thereby affect the clocking of the dac chip to some degree. As you said, though, there is "no substitute for trying things out yourself." Those kinds of effects will generally have little if any predictability. Enjoy! Regards, -- Al |
