Jump to content
IGNORED

Chords New M -Scaler


Recommended Posts

Interesting unit,  which has over 1 million TAPS.  

https://chordelectronics.co.uk/product/hugo-mscaler/

 

https://darko.audio/2018/07/chord-electronics-announce-hugo-m-scaler/

 

So why does it matter how many TAPS ( FIR's) you have and why is it mo better.  https://dspguru.com/dsp/faqs/fir/basics/

I mean when I listen to a Halo LV 3 in NOS or the PS Audio Direct Stream DAC they sound pretty good.  Not sure how many of these TAPS ( FIRS) they have.

The Truth Is Out There

Link to comment

Rob Watts is technically correct in that a longer filter gives a more accurate reconstruction. However, I'm certain that nobody can hear the difference between 100k and 1M taps. Conferring such great importance on the filter length falls, in my opinion, in the same category as talking about skin effect at audio frequencies: real phenomena of limited or no relevance to audio applications.

Link to comment
2 hours ago, mansr said:

Rob Watts is technically correct in that a longer filter gives a more accurate reconstruction. However, I'm certain that nobody can hear the difference between 100k and 1M taps. Conferring such great importance on the filter length falls, in my opinion, in the same category as talking about skin effect at audio frequencies: real phenomena of limited or no relevance to audio applications.

 

If taps are so "important to chord, I wonder why the other DAC manufacturers don't list their TAP quantity. OR is this Marketing at it best.

The Truth Is Out There

Link to comment

Proof is in the pudding.  Tested by others and found to make a big difference, number of taps.  So big, that it has been recommended by those using the Blu mscaler and now the Hugo mscaler, to be a far more important SQ upgrade over any server tweaks upstream.    Basically a must for those with the coin and in the Chord system build.  

Anyone else that says otherwise is obviously talking from back seat science and not from actual listening.  Thus disregard them.

(JRiver) Jetway barebones NUC (mod 3 sCLK-EX, Cybershaft OP 14)  (PH SR7) => mini pcie adapter to PCIe 1X => tXUSBexp PCIe card (mod sCLK-EX) (PH SR7) => (USPCB) Chord DAVE => Omega Super 8XRS/REL t5i  (All powered thru Topaz Isolation Transformer)

Link to comment
2 hours ago, mav52 said:

 

If taps are so "important to chord, I wonder why the other DAC manufacturers don't list their TAP quantity. OR is this Marketing at it best.

 

Different designers/listeners have different areas of focus.

 

What is true is that for the most accurate reconstruction from a digital sampling perspective, the closer to the infinite sinc, the better.

 

However, some folks prefer the sound of certain other filter approaches, some people prefer to emphasize the noise shapers, R2R, voltage regulation, etc.

 

Chord's direction has been towards infinite sinc and with pulse arrays -- part of the synergy of the latter is they believe they can build medium power (like in the tens of watts per channel) "digital amplifiers" that can be fed directly from a 16FS digital signal (or perhaps better stated -- high power DACs minus the primary reconstruction filter).

 

Folks who do software upsampling to output to a DSD DAC are similar but, of course, DSD is 1 bit whereas the Chord's pulse arrays are 4 or 5 bits (I forget) to speed up the error factor correction delay.... 

Link to comment
59 minutes ago, Miska said:

 

Funny thing is that computers doing stated upsampling to 16x with 1M taps doesn't even break a sweat. It is so easy to pull it to completely different figures with computer, if one wants....

 

Now I would like to see Rob Watts (Chord) and Bob Stuart (Meridian/MQA) arguing about this filter length stuff, they are in completely opposite camps... :D  Both of them fixing timing problems...

 

I personally prefer to sit somewhere between and find other aspects of filter design more interesting than number of taps, which is as meaningless figure talking about filters as MIPS is when talking about computer performance.

 

 

Sometimes we pay for the packaging...  E.g., Build me an upsampler with a USB in/out so I can play Tidal through it, and I am will gladly try it.

 

Though I could write equivalent code (since he disclosed that he ended up at equivalent of sinc, so the calculations are public domain) to upsample my stored .wav/.flac files, that is < 5% of my listening, so I need a solution that works with streaming.... so, an early adopter I became...

Link to comment
2 hours ago, Miska said:

 

Which doesn't really make practical difference compared to the modulator design. Chord uses so simplistic modulator that the noise floor looks more like something between DSD128 and DSD256 even though it runs at 100 MHz.

 

That’s an excellent point and thanks for sharing your Mojo measurements in the past. That said, I think in terms of noise shapers: 

Mojo/Hugo: 5th order noise shaper
Hugo 2: 11th order noise shaper
Hugo TT 2: 12th order noise shaper
DAVE: 17th order noise shaper

so maybe it’s better now?

Link to comment

Hi Guys - Thanks so much for the discussion thus far. It's really cool to see people offer opinions about sound quality after listening and for others to offer their expertise based on what they know about the technology.

 

All information is welcomed.

 

Since there's no free lunch, I'd love to know about the Pros and Cons to the Chord approach. 

Founder of Audiophile Style | My Audio Systems AudiophileStyleStickerWhite2.0.png AudiophileStyleStickerWhite7.1.4.png

Link to comment

I will say, after experiencing the Chord stack at RMAF 2017, my immediate reaction was to set about selling my DAC stack and acquiring a Chord Blu2/DAVE one and am happy despite paying the early adopter tax. To me, even under show conditions, it was convincing enough to set a plan into motion.

 

For me, the big con, besides $ (much reduced now with standalone M Scaler), was conceding that I am OK with focusing energy on a PCM-centric system w/o MQA (and w/DSD as a second class citizen).

 

I have not personally experienced MSB -- that's the big comparison unknown for me. (Not that I would spring for a Select 2 given the price, but a Reference would be within the ballpark.)

Link to comment
4 hours ago, ecwl said:

That’s an excellent point and thanks for sharing your Mojo measurements in the past. That said, I think in terms of noise shapers: 

Mojo/Hugo: 5th order noise shaper
Hugo 2: 11th order noise shaper
Hugo TT 2: 12th order noise shaper
DAVE: 17th order noise shaper

so maybe it’s better now?

 

Order is just as useless number to describe a modulator as is number of taps to describe a filter.

 

Higher order modulator makes the increasing noise slope steeper, which in turn puts more demand on the analog filter.

 

So just blindly increasing number of taps or modulator order is not necessarily best solution...

Signalyst - Developer of HQPlayer

Pulse & Fidelity - Software Defined Amplifiers

Link to comment
6 hours ago, rayl1234 said:

Sometimes we pay for the packaging...  E.g., Build me an upsampler with a USB in/out so I can play Tidal through it, and I am will gladly try it.

 

That's what I've been doing with software upsampling to DSD512 (of course one can do any PCM rate too). Playing Tidal doesn't require USB though.

 

Signalyst - Developer of HQPlayer

Pulse & Fidelity - Software Defined Amplifiers

Link to comment
16 minutes ago, ecwl said:

This is the reason why Rob Watts prefers the thermometer code approach which is basically having multiple elements of PWM/DSD DACs running parallel to solve the DSD timing issue. He prefers using discrete elements, like dCS, because he thinks there is simply too much noise that would be present on DAC chips (which he used to design). This would technically solve the small amplitude transient problem of DSD and the level matching problems of R2R DACs. But more importantly, running multiple elements (5-bit 2.8MHz or 5.6MHz like dCS), still would not solve the noise floor modulation and jitter problem completely. Because the amount of switching activity would still be signal dependent. As a result, Rob Watts developed Pulse Array DAC using thermometer code with constant switching activity. Because the elements are constantly switching, the switching no longer is signal-dependent so it significantly reduces signal-dependent noise, noise floor modulation and makes the DAC design much more jitter-immune.

 

And the DSC1 DAC design I published for playing up to DSD512 (24.576 MHz) uses 32 discrete elements with constant switching...

 

dCS has 24 and ESS Sabre has 64.

 

Anyway, there's really no timing problem and reason for using multiple elements is completely different from any timing stuff.

 

Signalyst - Developer of HQPlayer

Pulse & Fidelity - Software Defined Amplifiers

Link to comment
11 minutes ago, Miska said:

 

And the DSC1 DAC design I published for playing up to DSD512 (24.576 MHz) uses 32 discrete elements with constant switching...

 

dCS has 24 and ESS Sabre has 64.

 

Anyway, there's really no timing problem and reason for using multiple elements is completely different from any timing stuff.

 

I think you’re the expert on the matter. You’re right, my understanding for multiple elements is to avoid slight mismatch between each element. 

But I am not sure if Rob Watts description of constant switching is the same as yours. I was under the impression that in dCS, ESS, and presumably DSC1, you’re using thermometer code or something similar to do DEM (dynamically select elements) so that you’re not switching the same elements each time to avoid element mismatch. But you would still get signal dependent switching. Whereas Rob Watts describes matching elements that switch so that the same number of elements would always switch at the same time all the time so that you always get the same amount of noise from switching.

But I might have understood it wrong.

Link to comment
45 minutes ago, ecwl said:

I think you’re the expert on the matter. You’re right, my understanding for multiple elements is to avoid slight mismatch between each element. 

But I am not sure if Rob Watts description of constant switching is the same as yours. I was under the impression that in dCS, ESS, and presumably DSC1, you’re using thermometer code or something similar to do DEM (dynamically select elements) so that you’re not switching the same elements each time to avoid element mismatch. But you would still get signal dependent switching. Whereas Rob Watts describes matching elements that switch so that the same number of elements would always switch at the same time all the time so that you always get the same amount of noise from switching.

But I might have understood it wrong.

 

Multiple elements are used for other reasons, and DEM is used to decorrelate mismatch between the elements.

 

Some DEMs may not always switch all elements, but DSC1 does switch all elements every time (although it is easy to modify not to).

 

Signalyst - Developer of HQPlayer

Pulse & Fidelity - Software Defined Amplifiers

Link to comment
3 minutes ago, Miska said:

Some DEMs may not always switch all elements, but DSC1 does switch all elements every time (although it is easy to modify not to).

Very neat. I’ve always wanted to listen to HQPlayer playing through the DSC1.

 

And you’ve finally inspired me to understand how this constant switching Rob Watts talks about works. And it made me realize even though Mojo/Hugo are 4-element DACs, they are really functioning like a 1-bit/element DSD DAC at 104MHz. Although I’m guessing this is how Rob Watts does it as he was never explicit about it.

Basically, I think he uses the elements to represent either 1 or 3 and nothing else

That means when turned on, lets say we start with

0001 or 0010 or 0100 or 1000

And if the next signal is also 1 then he switches 2 elements to still stay at a sum of 1

e.g. 0001 to 1000 or 0100 or 0010

And if the next signal is 3 then he switches 2 elements to get to 3

e.g. 0001 to 1101 or 1011 or 0111

And if the next signal is 3 then he switches 2 elements to stay at 3

e.g. 0111 to 1011 or 1101 or 1110

 

So the DAC modulates between 1’s and 3’s (equivalent to 0’s and 1’s) to generate the signal. But then the DAC is constantly only switching 2 elements all the time.

 

I have to admit, I don’t fully understand how DSC1 works as I don’t fully understand circuit diagrams since I never took electrical engineering. But it’s impressive how you manage to do something similar with 32-elements and switching all of them at the same time.

Link to comment

Regardless how many values you want to represent, this is generally nice property of thermometer code (scrambled unary coding) used with SDM; you can represent the same value in many ways. Because all bit positions have same weight. This is something that is not possible with PCM/R2R. In addition, with PCM/R2R weight of the element error increases exponentially towards MSB.

 

Modern DAC technology allows a lot of flexibility on how one does the conversion, from DSP to the actual conversion stage.

Signalyst - Developer of HQPlayer

Pulse & Fidelity - Software Defined Amplifiers

Link to comment
1 hour ago, Miska said:

 

Order is just as useless number to describe a modulator as is number of taps to describe a filter.

 

Higher order modulator makes the increasing noise slope steeper, which in turn puts more demand on the analog filter.

 

So just blindly increasing number of taps or modulator order is not necessarily best solution...

 

Is there any significance to even vs. odd order modulators? 

One never knows, do one? - Fats Waller

The fairest thing we can experience is the mysterious. It is the fundamental emotion which stands at the cradle of true art and true science. - Einstein

Computer, Audirvana -> optical Ethernet to Fitlet3 -> Fibbr Alpha Optical USB -> iFi NEO iDSD DAC -> Apollon Audio 1ET400A Mini (Purifi based) -> Vandersteen 3A Signature.

Link to comment

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now



×
×
  • Create New...