Having tried about all I/V means known to the planet and refusing to ever try OpAmps apart from discrete ones which also did not work out, I ended up with something carrying a slew rate of 670V/us. I know, not 1K, but better for my environment (which is about noise figures, voltages used and some more configurative stuff like differential setup etc.) I could not find. And you know what ? it sounded better. Way better.
My reasoning back at the time was that the high frequencies were eating the current, especially when seeing that not the bass improved but the highs. This is sort of related to the images behind the mirror which are there in my NOS/Arc Prediction Filtered case and although they are sufficiently down, they eat energy and by a higher frequency than normal audio band frequencies, obviously.
When you dig deeper in my explanation, you can see that it is all over unofficial "reasoning out" what is for the better and what won't work. I mean, let pass higher frequencies beyond the audio band less distorted (see slew rate) on purpose ? Well, if we don't do that then the in-band higher frequencies become muffled. A nice natural sort of filter ? no way. No way since those "un-muffled" frequencies come out the most undistorted. And so the fun : instead of seeing a roll off in the higher frequencies dedicated to NOS/Filterless, we perceive a high frequency output (for level) which is unsurpassed and of which at least I dare claim that it is unsurpassed indeed.
And let's say (and hope) that by now I should have sufficient experience in knowing the difference between "freshness" coming from unfiltered NOS hence a mountain of piled up harmonic distortion on one side, and super silk and undistorted highs with a perceived maybe over 20dB more output than any normally filtered DAC on the other. Ever heard sheer SPL from brushes on a snare ? 5 years back the whole d*mn brushes weren't even auadible in my system. And yes, brushes may be the toughest "instruments" to render well.
Peter
Thanks for sharing your experience, Peter. I've been very curious about the Phasure DAC for some time, and I appreciate your experience and insight. That quality of "freshness" is exactly what was missing from the expensive delta-sigma DACs I was auditioning ... and is also absent from many DSD converters, as well. It's hard to describe, but when the "freshness" is missing, the music has a sort of monotone quality, and the performers sound bored.
I'm guessing the noise-shaping algorithms are doing strange things to the dynamics ... it certainly sounds like the dynamics are altered. With one visiting DAC, the dynamics were so different that I could never get it subjectively level-matched with my Burr-Brown DAC. I'd match 'em up on a quiet passage, then they would sound totally different when it got loud.
Regarding slewing ... well, that's another down-the-rabbit-hole topic all by itself. Slewing is a function of
current distortion, not voltage distortion. When an internal amplifier drives a capacitive load, and starts to lose linearity, it doesn't happen all at once. There can be a large region between the onset of slewing (gentle distortion) and hard slewing. Depending on the topology and the devices used, this region can be fairly small (1:1.5) or pretty big, like (1:10). It can be the difference between hard, gritty, 2-dimensional sound when the HF content gets loud, or a soft, syrupy sound that's dull and velvety sounding. Slewing can sound like either, depending on how the device reacts to driving current into a capacitance.
You might have noticed that's what many people describe as the difference between PCM and DSD. Uh, yeah. Which is why I don't think PCM vs DSD comparisons have much validity until we know more about how the analog circuits are responding to 1~20 MHz content. Typical audio circuits (with feedback) that have dominant poles anywhere from 10 Hz to 1 kHz get into real trouble at RF frequencies; feedback is mostly gone, the output stage is not happy driving the load of the feedback circuit (as well as external loads), and opamps typically have Class AB output stages as well as problems with thermal tails. (Transient gain shifts due to temperature changes on the die which occur in the 10 mSec range.)
We have two problem areas: the hard-to-measure things noise-shaping does to the signal, and the way the analog circuits respond to significant amounts of RF content. DSD converters have very substantial amounts of noise-shaping: by my rough calculations, 60 dB or more in the 1~5 kHz region.
But the 1~5 MHz decorrelated noise spectrum of DSD may be kinder to hard-slewing analog circuits than the spiky images of PCM. So right there we have two major sources of sonic difference between the two systems - in fact, we have two flavors of PCM: delta-sigma with noise-shaping, and ladder converters without noise-shaping.
