Hi, thought I'd share some experiments I did with my NFB 11.28 (Amanero interface with TXCO upgrade) in optimizing for my subjective audio preference.
I used a pair of RBH HP-2 with very fast beryllium dynamic drivers for monitoring.
When I first started using the NFB (ESS SABRE 9028), I was very impressed with the transparency it was able to reproduce. Switching back between the NFB and my Dragonfly Red (ESS SABRE 9018), which I've been evaluating against a Stoner Acoustics UD125 (AKM "Velvet Sound" AK4490) for office listening, I noticed I enjoyed music from the DFR more. Careful A/B listening made me realize that while both were more than adequate in detail retrieval, clarity, etc., the DFR produced what felt like warmer, more lush, more liquid sound. This was particularly evident when listening to the track
Luminous by
Alice and the Glass Lake. Sourced from the DFR, vocals were smooth and often times euphoric and the piano was liquid smooth, while the NFB produced what felt like relentless analytical superiority--there was a slight overall level of graininess and I could hear the hissy breath intakes and leaks even during the airy vocal parts heavy with reverb, which distracted from any feeling of euphoria.
Hypothesis 1: Warmness
My first hypothesis was that DFR produced a warmer sound. Fortunately, the NFB is
configurable! I set the jumper in place for the warmer sound signature.... and it was disappointing. A/B testing it several times, I sounded like that it merely recessed the treble range rather than change or improve the timbre of vocals and instruments. While I could see how that would help with brighter headphones (as the feature is stated to be designed for), the HP-2 is definitely
not a bright headphone by any measure, and the result was softer and quieter details in the music, which resulted in an overall feeling of relative muddiness compared to stock. I took the jumpers off.
Hypothesis 2: Digital Filter
(The above displays the PCB of the NFB-1, which is similar to but not exactly the same as the pin layout in NFB-11.28. Specifically, FIR_0 and FIR_1 are switched on NFB-11.28. You can tell by reading the pin labels.)
Digital filters can have subtle and meaningful impacts to the sound, especially around frequency transitions. Looking inside my NFB, I realized that the NFB 11.28 stock filter setting is for
"soft roll-off, minimum" with a jumper already on FIR_0 from the factory. According to this
incredibly helpful article, minimum phase soft roll-off filters are one of the newer filtering methodologies that adds a bit of (not very audible) phase distortion for better time response. Better time response (and drivers fast enough to take advantage of it) makes sense for a more analytical sound, but I really wanted that liquid smooth sound in my music. Keeping the stock jumper in place, I pushed another jumper into FIR_1 for a soft roll-off, linear phase filter setting. (See Appendum section below for an in-depth explanation between minimum phase vs. linear phase filtering.)
And
wow. The sound my NFB produces now is very close to the DFR. A/B listening, I don't think they're
exactly the same, but I really can't pinpoint where they differ. Both sound lush, liquid, and smooth and when listening to
Luminous both send shivers down my spine. The graininess and articulation I complained about previously have softened up. I'm loving it!
APPENDUM:
Here is another great article on minimum phase filtering vs. linear/zero-phase filtering:
https://www.dsprelated.com/freebooks/filters/Minimum_Phase_Filters.html. Specifically the section titled "Is Linear Phase Really Ideal for Audio?"
TL;DR: More or less aligned to the article in the OP: Minimum phase removes pre-ringing by reproducing an impulse with minimum delay, whereas linear/zero-phase filters equally space out every impulse and evenly divides the ringing before and after the impulse. The result is the former results in no audible pre-impulse ringing while the later reproduces the wave form to the maximal extent.