Thanks Kai for taking your time to explain this. I knew about the PCM conversion for metering, as stated in the manual. But never had a clue about how volume control was done. With my limited knowledge, I could only imagine PCM conversion for volume control in the outputs as well. But now I know, that it’s done inside the ESS chip - in what ever magical way.
I am sure, that the RMe engineers were perfectly up to the task to properly implement it.
Thanks for the link: it was an interesting read, including some additional links there!

This is not quite true.
The ESS DA-chips do handle DSD natively, do NOT convert to PCM.

They have the big advantage over AKM-DA chips that they offer a chip-internal volume control, that is seamlessly integrated into ADI-2.
I’d consider the ESS the more advanced solution.

It‘s not disclosed how exactly the ESS chips does handle volume control, but measurements hint to, that DSD‘s advantages of very high audio bandwidth and others are maintained, see here:

https://archimago.blogspot.com/2022/06/ … i-dsd.html
The ESS chip behaves like any other “DSD-direct”-chip when feed with DSD.

Look at the link to the measurements:
The whole ESS behavior is DSD-direct-like, e.g.:

• The jitter does not cause sidebands like unavoidable with PCM, but ends up in the broadband noise-floor, like is typical for DSD.

• Ultrasonic bandwidth isn’t PCM-like brick-wall limited, just rolls off like low-order analog filtered.

Practically, having a volume control when using DSD (without PCM conversion) is very handy and increases the usability of ADI-2 for most.

The other way round: ADI-2’s DSD-direct (on AKM) is useless if you want a headphone amp, as 'phones out is purposely switch off in this mode on AKM-equipped ADI-2’s (to protect 'phones and user’s ears).

You have to switch off DSD-direct then, which means all ADI-2 CAN play it on 'phones, but without the DSD-inherited advantages on the AKM-versions.

IMO ESS is the better choice, no matter what’s going on in the chip.

I don’t get the point in converting PCM to DSD by a player soft.

Why convert into a very feature limited audio format if the advantages of the format are already lost in the source?

Do you think the (lossy) conversion will bring back that advantages (rhetorical question)  – the answer is no, of course not, what’s lost is gone forever.

Inside the DAC chip such a handling is done anyway (*1), but with two advantages using PCM up to there:
• You still can use all the sound-improving features of ADI-2.
• You can use ADI-2’s volume control even on AKM-equipped models.

(*1)
That’s exactly what DSD is about, storing the working-format that oversampling AD- and DA-chips are using just next to their analog stage anyway, bypassing the noise shaping and other DAC-chip internal processings.

This makes sense only in a pure DSD signal path without any PCM inter-steps.
Native DSD recordings only very few exist, as it’s impossible to work with the DSD format in the typical way music is produced.

There are a few analog tape to DSD recordings.
E.g. MFSL, after some discussions, disclaimed their latest analog tape to analog vinyl re-issues incorporate a DSD inter-step.

I didn’t check if these transfers are published, maybe or maybe not, MFSL might not have the license to do so.

Personally I prefer either the original releases or very good remixes like the ones from Steven Wilson.

They are based on the original multitrack tapes, giving him the freedom to let everything shine in a new, clearer light while he’s still working faithfully to the original production.

Good example:
Jethro Tull, Songs From The Wood (The Steven Wilson Remix):
https://tidal.com/album/76176162

The original 1977 album (vinyl) sounds a bit muffled, which made me not picking it too often.
The remix is pure fun to listen, if you like the music.

Things like that make a real “night and day difference”, not mostly theoretical advantages of format switches.

Find good music, listen and enjoy!

In my studio I can clearly distinguish between native recording with sample rates of 44.1 kHz versus everything higher.

44.1 kHz can be improved by choosing a DA-filter that has a better compromise between frequency- and impulse-response and aliasing suppression.

The standard “Sharp” filter is frequency linear up to 20 KHz audio bandwidth, almost completely avoids aliasing, but has a tendency to smear the treble.

I prefer “Slow” with the AKM equipped ADI-2 Pro, and SD LD and Brickwall with the ESS ADI-2/4 Pro SE.

In exchange for a tiny treble rolloff and a little more aliasing (except Brickwall) in the top octave they give me better resolution of treble instrument’s transients, e.g. it’s easier to distinguish between hihats and shakers in a mix, and the overall sound is more transparent.

I clearly dislike the “NOS”-filter, it puts a layer of “fog” on the music that I attribute to aliasing noise which this filter doesn’t avoid at all.
It’s too far optimized towards impulse response for audio listening.

But, NOS is a good choice if one uses ADI-2 as measurement frontend to generate short impulses.

This all is relevant with sample rate of 44.1 kHz, lesser so with 48 kHz, but not at all with anything higher.

I agree and that is why I don't upsample nor use DSD conversion on my Hi-Fi system. I can't even distinguish between 44,1/48 kHz sample rates and higher as KaiS does. So, for me (and my ears) there is no point in upsampling.

Ironically, I discovered and bought an ADI-2 DAC (AKM based) just because I wanted to experiment with DSD direct. I have ended using the ADI-2 DAC without upsampling nor converting to DSD. IMO, there is no need.

Hi DVDR,

I came across your topic, which is already some months old, decided to write my opinion, because you might be still interested in the DSD conversion theme, a theme, which is full of incorrect, sometimes misleading information.
Above all, it is good to understand that there is not such thing as DSD-chip. The nature of this format DSD is such, that it can be converted to analog form with the help of discrete designs, using resistors, capacitors or transformers. The mass available dac chips on the market always manipulate the DSD signal in order to convert it (they manipulate as well the PCM signal) with the so called modulators.
There are DSD-Only-Dacs available, without chips, but they are usually rather expensive.
But you are right that it makes only sense to have the DSD file converted in its original 1-bit form, no question about it.
Every digital-to-digital conversion brings loss of quality, whatever you might here or read...