CHORD M-Scaler Review (Upsampler)

[MacCali]

It’s less of an issue at high sampling rates, the nasty stuff gets pushed out higher in the frequency range where we don’t hear it.

Yes that was pointed out, but the one specifically with the ps audio studio recording which was being upsampled or recorded in the improper DSD format by them was showing a significant rise and no smoothing cause the band is much more narrow.

Again this is not my main question, my question is does using OS on a dac create the same artifacts as a song that is upsampled incorrectly

 

[gvl]

Yes that was pointed out, but the one specifically with the ps audio studio recording which was being upsampled or recorded in the improper DSD format by them was showing a significant rise and no smoothing cause the band is much more narrow.

Again this is not my main question, my question is does using OS on a dac create the same artifacts as a song that is upsampled incorrectly

A well behaved upsampling digital filter shouldn’t produce any artifacts, but if it comes from ps audio then who knows

 

[solderdude]

my question is does using OS on a dac create the same artifacts as a song that is upsampled incorrectly

Not all software re-samplers are created equal so it makes sense those from DAC(chip) manufacturers also are not equal.

SRC Comparisons

src.infinitewave.ca

So how any DAC that uses a re-sampler does this exactly and what filter/up-sampler they use may differ.
One would have to look at the waveform outputs.
How audible this is (errors that should not be there that reach audible thresholds under specific circumstances/songs) will depend.

 

[DonDish]

Are they really “nasty”? Do they have any audible impact?

NOS filters sound like shit!

 

[Sir Sanders Zingmore]

NOS filters sound like shit!

Perhaps, but that wasn’t my question

 

[Galliardist]

NOS filters sound like shit!

And so we write off the first generation of CD players: which generally show good numbers, and which won over tens of thousands of people, and were highly reviewed subjectively before the "digital bad" idea worked its way through the press. In practice the newer NOS devices are too often made to sound "different" according to the ideas of their designers: we pay our money and we take our choice. It's the implementation that counts.

 

[AudioSceptic]

Can we stope calling filterless DACs, NOS? Please.

Plenty of natively running 16/44 DACs can reproduce 10kHz perfectly well, because they have a well designed LPF.

I've been wondering about that. Surely there have been non-oversampling DACs *with* reconstruction filters, and over-sampling DACs *without* filters?

 

[charleski]

I've been wondering about that. Surely there have been non-oversampling DACs *with* reconstruction filters,

Absolutely. All DACs need some form of analog filter on the output stage, but a NOS DAC needs a higher-order filter.

https://www.ti.com/lit/an/sbaa001/sbaa001.pdf

and over-sampling DACs *without* filters?

The whole point of oversampling is to allow most of the stopband rejection to be performed in the digital domain with only a gentle analog filter on the output. So this wouldn't make any sense at all.

 

[AudioSceptic]

You keep saying these things, and they keep being untrue. The incorrectness of your statements is exceeded only by the certainty and frequency with which you are making them.

"Delta-sigma dacs/oversampling will only ever be useful as long as high res music is not played into it" - this is nonsense. As in, it literally makes no sense. DS DACs process higher sample rates and higher bit-depths just fine, and conversely a non-DS DAC (or an outboard upsampler like the Chord) does absolutely nothing to make "high-res" sources any better. The limitations of high-res digital sources in improving upon redbook are dictated by digital sampling theory and the range of human hearing, not by the type of DAC you run them through.

You say that you are using "(stairstep)" because "other people call them that." But they are not that. If you don't know that, then you misunderstand the basics of digital sampling theory and therefore of what allows this stuff to work at all.

If you do know that - and your last comment implies that you do - then you are not comprehending how this fact undercuts your argument here, as you have used the stairstep fallacy as the foundation (whether you know it or not, and whether you are willing to admit it or not) for your claims about "time resolution" being a problem that needs to be solved in DACs, and a problem that can somehow be solved in incremental, gradual ways.

RE a NOS DAC, you keep banging on about how such a device is not "broken" because an oversampler can be "added" to it. This is semantics - it's restricting the definition of "broken" until it's meaningless. If you have to augment a DAC's functionality by grabbing a digital signal out of your source in order to remedy a built-in deficiency in the DAC's design and performance, it is reasonable to call that design broken - "built-in deficiency" = "broken by design." And even more importantly, it is manifestly unreasonable to try to argue that it isn't. The only way out of that is to claim that NOS produces superior analogue output - which it most certainly does not. It's analogous to how any DAC that benefits from a "galvanic isolator" (and to my knowledge Amir has found only one, the Modi 2) is not "okay but can be taken to the next level" - rather, it is a broken design. It simply fails to do something basic that electrical, electronic, and/or digital data knowledge tell us can be done, and done fairly simply.

Now, when it comes to R2R vs Delta-Sigma, sure, R2R does not equal NOS - there are indeed oversampling R2R DACs. But that's not the point here at all.

Oh - and we still can't hear, detect, or sense 100kHz. That hasn't changed since your last incorrect comment. We'll keep monitoring the situation, though.

Re "stairsteps", can there be anyone who hasn't seen the excellent video by Monty of xiph.org? <

>

 

[AudioSceptic]

Absolutely. All DACs need some form of analog filter on the output stage, but a NOS DAC needs a higher-order filter.

https://www.ti.com/lit/an/sbaa001/sbaa001.pdf

The whole point of oversampling is to allow most of the stopband rejection to be performed in the digital domain with only a gentle analog filter on the output. So this wouldn't make any sense at all.

It would make no sense but isn't it possible? Didn't Pioneer make CD players like that at some point (late 80s-early 90s) or did I misunderstand what they were doing? They had a name for it that I just can't remember right now.

Edit: Legato Link!

 

[charleski]

Oversampling inevitably involves filtering the signal. You're adding extra data points, what values should they have?
From Ken Pohlmann's book:

 

[Galliardist]

Absolutely. All DACs need some form of analog filter on the output stage, but a NOS DAC needs a higher-order filter.

https://www.ti.com/lit/an/sbaa001/sbaa001.pdf

The whole point of oversampling is to allow most of the stopband rejection to be performed in the digital domain with only a gentle analog filter on the output. So this wouldn't make any sense at all.

Hmm... I owned the Consonance 120 Linear CD player, no digital filter, supposedly NOS but it actually had a 2x option. So of course I blind tested (didn't level check accurately, which may be an issue) and could reliably pick and preferred the 2x option. Supposedly it had a steep analogue filter.

Later I got an Oppo BDP-105 and it was indistinguishable playing CD - or indeed hybrid SACD where layers were the same - from the Consonance with the 2x option. So I guess I have to question your "no sense at all" comment.

Of course without measurements I can't be aware of what the 2x option did - it may well have inserted a digital output filter as well as oversampling. The player got fried a few years later, so too late to find out now, I guess.

 

[Mnyb]

Can we stope calling filterless DACs, NOS? Please.

Plenty of natively running 16/44 DACs can reproduce 10kHz perfectly well, because they have a well designed LPF.

It’s less of an issue at high sampling rates, the nasty stuff gets pushed out higher in the frequency range where we don’t hear it.

Yeah the point the whole time is to get oversampling going so that required and necessary filter can be made both simpler and better and partly as digital implementations .

The new brethren of Filterless NOS DAC’s does a double take here hoping two wrongs makes it rigth. It does not.

The first generations of audio DAC’s had their share of engineering challenges it took more engineering do do properly and results where good at the time ,It took some engineering to get the lpf rigth when it was as close to the audio band as it was in the early days.
It’s very typical of audiophile fads to bring back old solved problems to the scene again

A long the way their where good reasons to choose the technology we ended up with . Today a good engineer can design a DAC unit with off the shelf standard components with standard tolerances and the thing can be mass produced and hit its design specs.

No individualy laser trimmed resistor ladders or other kinds of tuning involved

 

[restorer-john]

The whole point of oversampling is to allow most of the stopband rejection to be performed in the digital domain with only a gentle analog filter on the output.

No. The side effect of oversampling was the simplification and reduction in costs of LPFs. It was driven by ENOB resolution, not 'gentle' filtering.

 

[charleski]

No. The side effect of oversampling was the simplification and reduction in costs of LPFs. It was driven by ENOB resolution, not 'gentle' filtering.

A low-order filter is, surprise surprise, cheaper. I’m not sure what your point is.

 

[restorer-john]

A low-order filter is, surprise surprise, cheaper. I’m not sure what your point is.

Necessity is the mother of invention. Look it up.

Philips had a 14 bit converter (the TDA-1540D), committed to silicon and they had to bring a 16 bit ENOB CD player to market. They almost made it with 4xOS.

All of a sudden, the expensive, carefully tuned analog/active LPFs were not needed. The rest is history. Why not take OS to the extreme? And they did, with PDM (the first bitstream), then PWM.

Sigma Delta was created for one reason only. Cost. Not performance. It was inferior and Philips knew it. But, it got better.

 

[charleski]

Cost. Not performance.

ENOB is a primary measure of a DAC’s performance level. 16bits is the target they had to hit. They wanted to get there as cheaply as possible, but that’s just stating the bleedin’ obvious.

Digital technology accelerated so rapidly that the margins involved in getting to 18, 20bits and higher through delta-sigma became minuscule, and designs that could boast higher bit depths had a clear selling-point that made them more marketable. Again, none of this is a revelation.

You’re just being argumentative.

 

[AudioSceptic]

ENOB is a primary measure of a DAC’s performance level. 16bits is the target they had to hit. They wanted to get there as cheaply as possible, but that’s just stating the bleedin’ obvious.

Digital technology accelerated so rapidly that the margins involved in getting to 18, 20bits and higher through delta-sigma became minuscule, and designs that could boast higher bit depths had a clear selling-point that made them more marketable. Again, none of this is a revelation.

You’re just being argumentative.

No he isn't!

 

[MaxBuck]

No he isn't!

 

[DonDish]

And so we write off the first generation of CD players: which generally show good numbers, and which won over tens of thousands of people, and were highly reviewed subjectively before the "digital bad" idea worked its way through the press. In practice the newer NOS devices are too often made to sound "different" according to the ideas of their designers: we pay our money and we take our choice. It's the implementation that counts.

Implementation? Something like a polished turd?