I'm going to try to respond to a few things you've said here... (see my little (#) flags...
(1)
I agree with the principle of "the cleanest signal path"...
However there are two catches to implementing that goal.
The first is that "the signal path with the least parts" may not be "the signal path that adds the least coloration" (which is the
real goal).
This fact is often used to mislead by inference.
For example, some boutique DACs have "only a a passive transformer between the DAC chip and the output".
And, while this is a
simple signal path, it has
FAR higher levels of several types of distortion than even a medium-grade op-amp.
So you will get
FAR better performance by using active op-amps instead of that passive transformer.
In many situations adding parts actually improves performance.
The second catch is that, unless you have a schematic, and know how to read it, you really don't know what's in the signal path.
For example, a "simple DAC without a volume control", does not necessarily have "a simpler signal path" than one with a volume control.
There could be a lot of other components in the signal path - or relatively few.
(2)
That can be true... but you do not automatically get that advantage simply by using a DAC chip that supports a higher bit depth.
The digital volume control in the Sabre DAC chips, and many other modern DAC chips, does take advantage of that ability...
However, the DAC chip must "know enough to shift the incoming audio up and pad it with zeros", and not all of them do.
That also may or may not apply to a digital volume control in a digital audio player... depending on how
that is programmed.
You are also confusing ANALOG POTENTIOMETERS with ANALOG LADDER VOLUME CONTROLS.
Almost all analog volume controls are voltage dividers...
An analog potentiometer is literally a long thin resistor with a contact that slides along it (usually curled into a horseshoe shape).
The position of the slider determines "how much of the resistor is below the contact and how much is above it" which sets the divider ratio.
The reason they tend to have tracking issues, especially at very low settings, is essentially mechanical.
(It's difficult to make a single resistor where the resistance value varies precisely enough along its entire length.)
A "ladder network volume control" solves this problem by using a whole series of separate resistors.
Instead of the position of a slider the ratio of the voltage divider is chosen by an actual selector switch.
(Each step delivers a slightly different output level; the selector switch lets you pick "which step on the ladder" to use.)
Because separate resistors can easily be made very precise ladder network voltage dividers are also usually very precise.
But ladder network volume controls are usually limited by the number of steps that are practical to include.
(If you want a control with 50 steps, you need at least 51 separate resistors, and a selector switch with 50 positions.)
The "analog ladder network volume controls" we use in many of our products are of this type.
The incoming analog signal for each channel is fed into an analog voltage divider...
The voltage divider for each channel is made up of a
LOT of very precise discrete resistors...
(Most ladder networks actually use a pair of resistors
for each step - because that makes the choice of individual resistor values simpler.)
Then a selector switch chooses which point on the "ladder" to connect to the output for whatever level you choose.
There are actually a few variations on this theme... but they all do effectively the same thing.
Because each step is actually created using fixed resistors the actual level for each, and the matching between channels, is precise to a small fraction of a dB.
This used to be done using separate resistors and selector switches... with the results being very large, very expensive, and still having a limited number of steps.
However, with modern versions like we use, all of the resistors, and the analog selector switches, for multiple channels, are all included in a single IC chip.
(That chip also includes the digital control circuitry that enables us to tell the switches what level to select.)
So, for any specific setting on that volume control, you have an analog signal, whose level is being set by a pair of resistors, chosen by an analog switch.
(You just have a whole lot of them in a single chip.)
Therefore you get all of the benefits of a potentiometer - with none of the drawbacks.
(3)
Different DAC chips most certainly do sound a tiny bit different... although the differences can be somewhat difficult to describe.
They can also vary rather widely depending on the other circuitry, like the power supply, and any audio buffers in the signal path.
(And these absolutely can overshadow the inherent differences between the chips themselves.)
However I do find how you grouped the various ones interesting.
Until recently, most people I know would have said that the AKM and Burr-Brown DACs were quite similar, and both quite neutral.
But that many devices that use the ESS Sabre DACs often tend to be "bright" and "excessively detailed".
(That is usually the complaint from people who are critical of them.)
More modern designs with Sabre DAC chips, like our XDA-3, manage to avoid this "distinction" and deliver a very detailed but neutral sound.
(But most people still consider the Burr-Brown and AKM DACs to be most similar... and slightly different than the Sabre DACs.)
(4)
And, of course, many people like B&W speakers... but some also consider them "a tiny bit too detailed".
(I personally prefer as much detail as possible - as long as it really belongs there - but not everyone agrees.)
Thanks Keith, all very good thoughts/information.
But the reality is that those analog ladder network volume controls are VERY transparent...
So even going through two of them isn't going to significantly affect the audio quality.
(1) I'm sure that's true, but the purist in me wants the cleanest signal path possible
(2) With a 32-bit DAC you should have a lot of extra zeros to adjust volume without losing any quality. Given that analog volume pots are quite difficult to get right, I was a bit surprised to see the XDA-3 use an analog ladder. Not a complaint, but certainly a curiosity. I'm mostly playing 24-bits via Apple Music these days.
(3) I know this
shouldn't be audible, but I've listened to a of DACs and the ones that I always want to come back to are the ESS Sabre and old Burr-Brown DACs. I've come to enjoy the AKM DACs in the RMC-1L for longterm listening, but historically they've never been my cup of tea in Rotel equipment or Topping DACs. I think the RMC has a favorable implementation to me.
I gotta tell you, the RMC-1L with AKM DACs, Purifi class-D amp, and Bowers 800 series is hands down the
cleanest setup I've ever had.
And I think here is the reality -- I'm probably going to order an XDA-3 anyways just to hear the difference
(4) Thank you, that was a lot of technical reasons to consider the XDA-3, but I think for me it's more simple. I care about 1) music quality at all cost 2) a thin simple system that fits into my life. My main concerns are 1) music quality above all else 2) center channel for clear dialog 3) sub and surrounds for improved home theater experience.
The RMC-1L is a big chunky system with a billion channels and features I don't need and I worry about taking quality away from music. The XDA-3 seems to hit all my marks for a compact 2.0 (or 2.1 since my sub has it's own crossover) system.
