Things I don‘t know

Let me see if I can answer some of your questions -

1. Polarity - It has to be in software, reversing it in hardware means all your ones become zeros and zeros become ones.  This isn't analog!   Conceivably, a bit stream could be designed for that but to what end?  If it is done in hardware, then it takes some silicon to covert the phase, so SW is usually the way to go.

2. Clocking - 10m?  I am not sure what you mean by 10m.  I believe the question you need to ask is about jitter.

3. Galvanic Isolation - You are correct, galvanic isolation is the way to go.  The problem is, it is relatively expensive.  If they can achieve similar results in SW, then that is the direction they are forced to go.   Note that the same thing applies to balanced inputs and outputs in the analog domain.  Specs love to quote CMRR but it is CMB on the output device that dictates how good the CM rejection is because excellent CMB is much harder to achieve.  Galvanic isolation on the output is really the way to go, yet it is a cost issue, and many, if not most, audiophiles would ignorantly question a transformer in their signal path.

4. Class D - It is still a young and evolving technology, I do believe it is much too soon to expect standardized clock frequencies or power supply topologies.  Designers need to have the flexibility to improve their design as they see fit. 

I’ll have to admit I can’t reliably tell if the polarity is reversed or not, at least not with music. My equipment does allow me to switch the polarity.

&spatialking,

my point on polarity was that it is determinant of sound quality but being ignored by top equipment designers

10mhz is the standard frequenzy for masterclocks controlling several components in the same system, typically by a 50 or 75 Ohm BNC connection. Most top class devices do not provide a connection for this.

Galvanic isolation cannot be replicated in software.

Agreed on class D, however there seems to be no discussion about different approaches so far.

You have to look at this from a market point of view. Do the engineers at Toyota not know how to design a vehicle that rides like a Rolls Royce or a performs like a  Ferrari SF90? Whether they do or not is irrelevant because the market cannot bear millions of those cars on the supply side, so we are forced to compromise to a price point to sell to a wider market. Making more Ferraris would put pressure to lower prices to compete, yet the lower prices would preclude production of those cars.

Same applies to audio. Only a few manufacturers can compete to a very select buyer who demand the best. But sometimes, a Schiit Audio comes along and offers digital players for less than three grand that competes with (or even beats) these kilobuck digital players, but the audiophile community judges them by the price tag. "It sounds good for the money" is usually the dismissive review.

(1) Many listeners can’t hear polarity reversal, even listeners with good systems and musical knowledge. As to why hardware doesn’t always offer it, it’s another feature that costs money to include. Why don’t (most) preamps have tone controls?

(2) The 10 MHz clock, last time I looked, was more designed to enhance long-term stability; at least, that was the view of the digital expert whose essay I read 1-2 yrs ago. The clocks in digital devices are getting better at the more critical issue, short-term stability.

(3) Galvanic isolation is a tool to achieve an end: low noise, distortion, etc. Perhaps manufacturers can achieve that in other ways. Measurements and listening of recent relatively inexpensive DACs (by SMSL and Topping, e.g.) indicate that to be the case. It’s the result that counts, not the means to get there.

(4) See (3). There is more than one way to get the job done. Maybe the ’optimal’ value of those things depends on other factors. Also, manufacturers are always looking for ways to distinguish their equipment, and it’s easy to say "our circuitry is special" and it’s hard to contradict. It makes great ad copy and can be used by reviewers who want to sound knowledgeable ("Audio Idiot explains that their special HAN transistors are faster and widely used in aerospace applications, where absolute accuracy is vital"). Finally, audiophiles show a disturbing tendency, encouraged by some reviewers, to think that anything more complicated, exotic, and expensive is better, so that always generates debate.

Another unfortunate characteristic of audio discussions is what was mentioned above: the tendency to look at the means, rather than the ends. Whether it's "Class A is better than ....", or "The ESS chips are not as good as ..." or "MOSFETS are ...", or "NOS DACs are the only ones that ....," in my view, there is far too much obsession over the mechanics, rather than whether the result sounds like music.

Regarding 10mhz clocks, do you want a square wave or a sine wave?

Everything I've read is that external clocks are worse than a really good internal clock.  The distance in the leads makes them more susceptible to jitter issues than a low phase-noise internal oscillator.

Class D amps are not (as Atmasphere likes to point out) digital.  D was just the next amp type in line.  They are switching amps with analog comparators, but it is often argued that the switching frequency being higher pushes the noise out further.

Actually there is no sampling frequency for class D because class D is an analog encoding function. It is unfortunate that it's called class D. It was only called class D because it was the next amplifier format and it came after class C(lousy for audio).

Power supply is critical for class D. It must be very fast because class D goes from zero to max theoretically instantly(impossible but the closer it comes the better). That implies lots of current and a very low output impedance for the power supply.

@erik-squires

that is one line of arguments, indeed. However Innuos spends enormous efforts on reclocking both USB and Ethernet connections and best I know DCM offers full external clock support for all its units. There is also a vibrant market for master clocks from Esoteric, Afterdark, Mutec, Cybershaft, Antelope et al. I‘d argue the jury on clocking is still out. In my case reclocking my dac and Etherregen yielded major benefits.

On Clocks

Very few DAC manufacturers e.g. Esoteric and DCS, make their own clocks and most manufacturers don’t want you going off and buying someone else’s product. That’s aside from the fact that digital designers do differ in their views of the relative merits of onboard and offboard clocking.

I used a Mutec reclocker for a while. It helped an older piece of gear appreciably, at least to my ears, but the reclocking degraded the sound of a newer, better (but less expensive) piece. Why, I don't know.

The main use of external clocking (other than profit) seems to be in sound studios with numerous pieces of gear that need synchronization. That's why consumer gear rarely has clock inputs.

@mike_in_nc 

Did you just use an M3 or a ref10? makes a big difference

I had an MC-3+ Smart Clock USB. The REF10 is more than I was willing to pay for something that I might or might not consider an improvement. My conclusion was that differences caused by reclocking will be system-specific (a pretty bland and safe conclusion).

Those interested in the subject might look at this 2010 comparison of clocks in the pro sound journal Sound On Sound. It said, among other things:

There's a widespread notion that adding a high‑quality master clock to a digital system will somehow magically improve its overall performance. While that might possibly have been the case in the very early days of semi‑pro digital converters where, frankly, some of the internal digital clock designs were pretty ropey, it certainly isn't the case today. . . . Today's converter designs generally work best on their own internal clocks, and most will deliver a slightly poorer performance when clocked externally. The very best devices will show no change in performance at all, because they have superb clock-extraction circuitry that can remove all traces of clock jitter and other external clocking artifacts, so they work just as well as when running on their internal clock.

As always, the questions are: Is there a difference? Is the difference an improvement? Is the improvement worth the price?