Over the course of the past year I continued to pursue the holy grail of absolute transparency. For me this describes a system that has crossed the uncanny valley of sound reproduction; a system where all sources of noise have been attenuated to such a degree that your brain suspends disbelief and you never doubt that you have been transported to the time and place of the original recording. This is Qualia, and I believe, at the end of 2022, I am almost there.
Refer to the attached drawing and annotations below. High resolution is here.
Ⓐ Speakers & Headphones
I sold my Focal SM9 reference monitors in search of more resolution. I chose the PMC MB2S studio monitors. They are monsters ...in size and capability. For amplification, I selected the Accustic Arts MonoII's and companion hybrid tube preamp II. See my blog. For headphones, I had a long audition of the Audeze LCD-5 but preferred to stay with my HIFIMAN HE1000se for sound and comfort.
I made the jump to a Sean Jacobs external power supply (Chord warranty be damned) and accompanying Sean Jacobs Capacitor board. There is no magic to good power for a DAC: just get the lowest noise and the lowest impedance and the SJ setup comes very close. However, I improved it still with a LFP battery bank (200Ah) and toroidal inverter plus multiple RF-STOP filters in series to clean up BMS and rectification noise. I also added a double application of DAC-WRAP plus DC-BLOCK and LP-FILTER and a separate highly filtered battery for OPTO-DX receiver power.
My RF-STOP Faraday box is still relevant despite all the other isolation. I have now gone exclusively with battery power inside my box. A single 12V LFP on a DC switch powers a NUC, Ethernet Fibre media converter and OPTO-DX. I swap two batteries: one inside RFSTOP and one on a charger. I am working on a way to ease this manual process.
I also added other pieces of my kit inside the box (DC-BLOCK and USB-FILTER) as I experiment on what level of noise manifest on the optical signal is still audible.
A new version of PGGB file-based upsampling will be released in the new year. This version uses a windowless Sinc filter (revolutionary), infinite taps (as many as needed), 256-bit floating point precision for reconstruction and noise shaping plus truly lossless equalization. This level of perfection demands a good system with lots of memory (64GB ram minimum) - and even then processing takes several minutes per track. For me, the results are so compelling that I only listen to PGGB files. Keep on the lookout for product release here.
Ⓔ Network & Other
My network is the one aspect of my reference system that has changed the least. I am happy with my Keces linear supply and since substantially all my chain is off AC mains, any power conducted network digital noise is not audible. Still I am wary and I am looking for ways to reduce or eliminate any impact.
What I have changed, however, is the amount of STABILANT-22 in my system. As transparency increases, the importance of the smallest issues grow in significance. So I spent the time and applied this to all connections.