• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). There are daily reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

CHORD M-Scaler Review (Upsampler)

Rate this product:

  • 1. Poor (headless panther)

    Votes: 333 89.3%
  • 2. Not terrible (postman panther)

    Votes: 11 2.9%
  • 3. Fine (happy panther

    Votes: 7 1.9%
  • 4. Great (golfing panther)

    Votes: 22 5.9%

  • Total voters
    373

Jimbob54

Master Contributor
Forum Donor
Joined
Oct 25, 2019
Messages
9,593
Likes
12,466
@amirm I've signed up here at the encouragement of some of your subscribers to raise a few questions about the measurements of the M-Scaler. For those who haven't seen my video where I interviewed Rob Watts to ask him about some of the results in the measurements you produced, my intentions are as follows:
  • I am not looking to make anyone right or wrong - I am seeking open discussion for clarity and exploration of what I believe to still be some ambiguous areas of measurements and their relation to human auditory perception
  • I am asking these questions to explore, not to criticise and I openly acknowledge that I do not have an engineering background or qualification. For the record, my primary qualifications and studies are in hearing and speech (speech pathology/therapy or pre-audiology) and organisational psychology
  • I am writing a full explanation of the concepts I am discussing for clarity of my perspective, not to suggest that anyone reading it doesn't understand
I should add before I continue that I may use the wrong terminology or (more likely) avoid it altogether. My aim is to discuss the concepts, not get caught up on whether I used the right specific word.

With that said, let's start with one specific measurement (or conclusion) that I felt was misleading in relation to an upsampler like the M-Scaler - specifically the performance with the 1kHz test tone. My belief is that an upsampler's purpose is to convert the incoming signal (let's say 48kHz) to a higher sample rate (ideally 768kHz in the case of the M-Scaler). In doing so, every frequency being received in the digital input signal should be faithfully recreated with no alterations to the frequencies in terms of frequency response.

As I understand it, there are two potential benefits of upsampling:
  1. That it improves the available range in which to apply noise shaping to push processing noise outside the audible band and allow for steeper attenuation filters
  2. In the specific case of the M-Scaler, the purpose stated by Chord / Rob Watts is to improve the timing accuracy of transients in a musical signal (i.e. where the DAC is reconstructing the rapid fluctuations of the output waveform of complex musical signals)
So, with all that in mind, I thought it was misleading to criticise the M-Scaler for providing no benefit when upsampling at 2x to 4x with a 1kHz tone. That is 100% accurate performance as far as I understand it. What benefits would you expect to see from an upsampler with a 1kHz test tone?

Welcome. If I were you I would be more interested in the $14k Chord device that doesnt seem to live up to expectations https://www.audiosciencereview.com/forum/index.php?threads/chord-dave-review-dac-hp-amp.35974/ than the $4k device that no one seems to know how to prove the worth of.
 

PassionforSound

Member
Reviewer
Joined
Jul 24, 2022
Messages
45
Likes
12
The 1kHz test is a standard test to measure distortion. In this case we notice "damage being done", as you called it. The "no benefit of upsampling" statement is still done in the context of evaluating distortion, meaning the increased distortion is still there. It's not meant to say anything about other performance criteria.

An evaluation of functional benefits of the M-Scaler is done with "Let's go back to basics and see if the M-scaler is doing what it is supposed to be doing by running our filter test". This test confirms the upsampling provides a sharper filter response.

There was no damage done to the 1kHz test tone as I understand it (referring to the testing with the Hugo 2). I'm still unclear on why someone would say the upsampling provided no benefit when it's not expected to and when the test is for distortion which is not what the M-Scaler is intended to influence. I understand running the test to check for distortion. My ongoing concern is the implications of the 'no benefit' statement.

I think it's best if we leave this for @amirm to address as they are his words and anyone else (including me) is just speculating as to why he chose those words.
 

PassionforSound

Member
Reviewer
Joined
Jul 24, 2022
Messages
45
Likes
12

mansr

Major Contributor
Joined
Oct 5, 2018
Messages
4,692
Likes
10,592
Location
Hampshire
I don't understand. If, on one hand, you're saying that an upsampler can't improve a 1kHz tone, then why say 'no benefit'? From what you're saying, the test should be considering if damage is being done and calling out 'no reduction in signal quality' or a reduction in signal quality.
Those are just different ways of saying "does nothing." Whichever phrasing you prefer, I fail to see how spending $5k to achieve it could ever be considered sensible.
 

Geert

Major Contributor
Joined
Mar 20, 2020
Messages
1,367
Likes
2,203
There was no damage done to the 1kHz test tone as I understand it (referring to the testing with the Hugo 2).

You specifically referred to the 1kHz graph where Amir made the 'no benefit' statement. That's the test with the Topping D70, and it shows extra distortion. If this is not what you referred to then better provide a link or quote, otherwise it's going to be difficult for Amir to address your concerns.
 

Jomungur

Member
Joined
Jan 3, 2022
Messages
91
Likes
574
There was no damage done to the 1kHz test tone as I understand it (referring to the testing with the Hugo 2). I'm still unclear on why someone would say the upsampling provided no benefit when it's not expected to and when the test is for distortion which is not what the M-Scaler is intended to influence. I understand running the test to check for distortion. My ongoing concern is the implications of the 'no benefit' statement.

I think it's best if we leave this for @amirm to address as they are his words and anyone else (including me) is just speculating as to why he chose those words.
I understand what you are getting at. By using the words 'no benefit', Amir is implying (intentionally or not) that the "upscaling" provided by the M Scaler could have benefited the 1kHz test tone when it's impossible to do so. So those words insinuate the M Scaler should have been doing something to the 1kHz test tone that it could never have done, which would be an unfair thing to say. I don't think it's a large point in the grand scheme of things. Could be a poor choice of words, or maybe there's a deeper explanation. My own thought when I read it was that Amir is overstating his point a little but for the purpose of demystifying the M Scaler, which is often hyped as a revolutionary black box that improves sound quality like no other product can. It fits the theme of his review that there's no magic in the M Scaler, it cannot do what is physically impossible.

Your post, while very polite (thank you), also raises the question of whether Amir is biased against Chord and takes unfair potshots that he wouldn't against another brand. Putting aside his positive reviews of other Chord products, it's possible, sure; there's a lot of vitriol in both directions on different sites and I myself would find hard not to be affected by that. Still, I don't read any review these days assuming zero bias by the reviewer. He has his own cutting editorial style which people may or may not like, but I've always thought the real point is to look at the measurements provided and then interpret them yourself. Just like you did here with the 1khz tone point.
 
Last edited:

PassionforSound

Member
Reviewer
Joined
Jul 24, 2022
Messages
45
Likes
12
I understand what you are getting at. By using the words 'no benefit', Amir is implying (intentionally or not) that the "upscaling" provided by the M Scaler could have benefited the 1kHz test tone when it's impossible to do so. So those words insinuate the M Scaler should have been doing something to the 1kHz test tone that it could never have done, which would be an unfair thing to say. I don't think it's a large point in the grand scheme of things. Could be a poor choice of words, or maybe there's a deeper explanation. My own thought when I read it was that Amir is overstating his point a little but for the purpose of demystifying the M Scaler, which is often hyped as a revolutionary black box that improves sound quality like no other product can. It fits the theme of his review that there's no magic in the M Scaler, it cannot do what is physically impossible.

Your post, while very polite (thank you), also raises the question of whether Amir is biased against Chord and takes unfair potshots that he wouldn't against another brand. Putting aside his positive reviews of other Chord products, it's possible, sure; there's a lot of vitriol in both directions on different sites and I myself would find hard not to be affected by that. Still, I don't read any review these days assuming zero bias by the reviewer. He has his own cutting editorial style which people may or may not like, but I've always thought the real point is to look at the measurements provided and then interpret them yourself. Just like you did here with the 1khz tone point.
Thanks for taking my point as intended. I have no problems with editorial commentary when it's in a subjective context - my whole channel is based on my personal views and subjective experiences.

I think you've actually better clarified my concerns with the statement in that it either deliberately or accidentally editorialised an otherwise objective measurement.

I also agree that it's a very small part of the bigger picture of the review. It was just the place I chose to begin because it is a little clearer cut as a measurement of the M-Scaler than some of the others.
 

Shadders

Member
Joined
Jan 5, 2019
Messages
20
Likes
18
Location
Londinium
There was no damage done to the 1kHz test tone as I understand it (referring to the testing with the Hugo 2). I'm still unclear on why someone would say the upsampling provided no benefit when it's not expected to and when the test is for distortion which is not what the M-Scaler is intended to influence. I understand running the test to check for distortion. My ongoing concern is the implications of the 'no benefit' statement.
Hi,
My interpretation is that the m-scaler does what it is supposed to do, and for a 1kHz tone, does not degrade the signal, yet, what else can it do to make it better ?

Nothing is the answer. So what benefit is it as a device ?

If there are benefits of the device, where are they from an objective analysis ?

Is it possible that the benefits are there, but there is no known test to prove that they exist and we therefore have to accept subjective reviews and opinion ?

Which then leads to the technical design question, if there is a benefit to the m-scaler which the designer has implemented, what tests did they use to prove the benefit is apparent, and will they release the test criteria and results of the tests to prove said benefit(s) ?

Regards,
Shadders.
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
12,993
Likes
29,293
Location
The Neitherlands
I feel like this test is both largely irrelevant (which you have sort of confirmed) and also creating a source of negative bias towards the meaning of the measurement by saying 'no benefit' instead of something that confirms that it is working as it should.

Amir stated:
Then again there was no benefit to upscaling either (related to Hugo2).
In the plots it states: no difference.

In the digital domain (so AP-> M-scaler -> AP) the signal became worse (jitter and noise).
Of course RW says, jitter is removed by the DAC (which DACs do) and the noise is intentional (24 bit dither).
 

PassionforSound

Member
Reviewer
Joined
Jul 24, 2022
Messages
45
Likes
12
Ok, so while we wait for @amirm to reply to my original question (not rushing you, Amir) let me pose another question and I'd ask you all to consider this objectively before responding.

The jitter measurements from the M-Scaler were quite shocking to me when I first saw them and I wondered how on earth Rob Watts could release a product with that type of jitter output. And then I read this from Amir in the review:

Fortunately any half-respecting DAC will filter this and not let it in the output but still, what is the claim to fame of M-scaler if it is not precision of its digital transformation? Let's save our frustration for other test results.

If this is the case then why are we judging a product for a measurement that is objectively irrelevant to its intended use with DACs that are unaffected by this measurement?
 

Geert

Major Contributor
Joined
Mar 20, 2020
Messages
1,367
Likes
2,203
Fortunately any half-respecting DAC will filter this and not let it in the output but still, what is the claim to fame of M-scaler if it is not precision of its digital transformation? Let's save our frustration for other test results.

If this is the case then why are we judging a product for a measurement that is objectively irrelevant to its intended use with DACs that are unaffected by this measurement?

Amir measures different parameters that give an indication of good design and engineering. The results of these measurements are what they are. If some of them are bad than don't blame the messenger.

And why not measure jitter? A lot of audiophiles believe jitter performance has a big impact on digital sound quality, and spend money on dedicated devices to improve on it. Now it suddenly does not matter anymore? You can't have it both ways.
 

PassionforSound

Member
Reviewer
Joined
Jul 24, 2022
Messages
45
Likes
12
Amir measures different parameters that give an indication of good design and engineering. The results of these measurements are what they are. If some of them are bad than don't blame the messenger.

And why not measure jitter? A lot of audiophiles believe jitter performance has a big impact on digital sound quality, and spend money on dedicated devices to improve on it. Now it suddenly does not matter anymore? You can't have it both ways.
I didn't question taking the measurement. I agree wholeheartedly with that part. My question was, on a forum that seems to pride itself on objectivity, why is that measurement considered negative when it actually doesn't influence the objective performance of any decent downstream DAC?

As you said, you can't have it both ways. Is this about objective measurements and conclusions based on the objective meaning of those conclusions or just subjective judgements? If it's the latter, that's fine, but let's call it what it is.
 

voodooless

Master Contributor
Forum Donor
Joined
Jun 16, 2020
Messages
6,070
Likes
10,052
Location
Netherlands
why is that measurement considered negative when it actually doesn't influence the objective performance of any decent downstream DAC?
Because it’s shit engineering! It’s out of spec both timing and voltage wise. This is objectively a bad SPDIF implementation, no way around that.
 

Geert

Major Contributor
Joined
Mar 20, 2020
Messages
1,367
Likes
2,203
why is that measurement considered negative when it actually doesn't influence the objective performance of any decent downstream DAC?

Like I said, on this forum devices are also judged on technical qualities, audible or not. That goes for everything that's tested over here. In this case these qualities are more than relevant, given the cost of the device and the designer making claims of other artifacts that, in general, are considered to be inaudible. The review doesn't consider this issue absolutely bad: "Well, if you can hear that well, then maybe you can hear this jitter too".
 

the_brunx

Active Member
Joined
Apr 7, 2020
Messages
273
Likes
709
I didn't question taking the measurement. I agree wholeheartedly with that part. My question was, on a forum that seems to pride itself on objectivity, why is that measurement considered negative when it actually doesn't influence the objective performance of any decent downstream DAC?

As you said, you can't have it both ways. Is this about objective measurements and conclusions based on the objective meaning of those conclusions or just subjective judgements? If it's the latter, that's fine, but let's call it what it is.
From a man who has been constantly called out on many of his videos on his channel for running flawed "Objective" tests to try to convince his audience about the difference USB cables make. You must think everyone else is playing the same game.
 

Ornette

Member
Joined
Jun 19, 2021
Messages
35
Likes
111
You specifically referred to the 1kHz graph where Amir made the 'no benefit' statement. That's the test with the Topping D70, and it shows extra distortion. If this is not what you referred to then better provide a link or quote, otherwise it's going to be difficult for Amir to address your concerns.
Actually, it wasn't. The sequence was:

1) M Scaler into Topping produced unexpected spikes in the FFT plot. Amir commented: "What the heck happened here?"
2) Amir stated: "Fortunately there was no problem using M-scaler with Hugo 2", followed by graphs of the M Scaler into Hugo 2 response.
3) Amir then stated: "Then again there was no benefit to upscaling either" followed by comparative FFT plots of 2X and 4X upsampling with M Scaler driving Hugo 2.

All that aside, I don't see any real benefit to anyone in trying to interpret the connotations of Amir's use of "no benefit" (particularly as a rejoinder to his own "no problem"). The crux of the complaint seems to be that the audible (according to Watts) benefits of the M Scaler are chiefly related to its improvement of transient timing accuracy, and Amir didn't measure that specifically. Perhaps a valid argument. I would suggest that it is Chord's job (rather than Amir's) to show us graphs or other technical data to support their case. As it is, all we have is Watts claiming he heard differences during the design and testing phase of the product.
 
Last edited:

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
12,993
Likes
29,293
Location
The Neitherlands
My question was, on a forum that seems to pride itself on objectivity, why is that measurement considered negative when it actually doesn't influence the objective performance of any decent downstream DAC?
This has nothing to do with pride and objectivity.
Jitter is something that can be measured.
There are tons of subjective audiophools claiming they can hear the tiniest amount of jitter.
Of course, in the case of DACs and SPDIF it is totally up to the DAC in question HOW well jitter is removed.

The measurements are there and after the the M-scaler it is worse. What's objectively wrong with stating that and quantifying it ?
What's even worse is the short peaks that are weird and may have something to do with the non standard spec output format but is unclear (not further investigated).
Why not mention that.

With the Chord DAC it worked fine. That is mentioned too. But obviously the Chord DAC works fine even with out of spec (but within their own range) input signals.
No reason not to mention that.

The fact that most decent DACs remove jitter is mentioned so there's that too. An objective statement.

As you said, you can't have it both ways. Is this about objective measurements and conclusions based on the objective meaning of those conclusions or just subjective judgements? If it's the latter, that's fine, but let's call it what it is.

It is about objective measurements and findings and what can be concluded about that. Nothing to do with subjective judgements. Most other sites and owners do that.
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
39,921
Likes
184,767
Location
Seattle Area
With that said, let's start with one specific measurement (or conclusion) that I felt was misleading in relation to an upsampler like the M-Scaler - specifically the performance with the 1kHz test tone. My belief is that an upsampler's purpose is to convert the incoming signal (let's say 48kHz) to a higher sample rate (ideally 768kHz in the case of the M-Scaler). In doing so, every frequency being received in the digital input signal should be faithfully recreated with no alterations to the frequencies in terms of frequency response.

As I understand it, there are two potential benefits of upsampling:
  1. That it improves the available range in which to apply noise shaping to push processing noise outside the audible band and allow for steeper attenuation filters
  2. In the specific case of the M-Scaler, the purpose stated by Chord / Rob Watts is to improve the timing accuracy of transients in a musical signal (i.e. where the DAC is reconstructing the rapid fluctuations of the output waveform of complex musical signals)
Hi there. Welcome to the forum. I have seen your video.

I think the concept of a digital processor is not understood. The M-scaler has no intelligence. It does the same thing on every input signal. As such, that 1 kHz set of digital samples were modified due to interpolation, dither applied (again), and output over a S/PDIF link which itself can cause timing variations (jitter). So it is very important to run the standardized 1 kHz test. And good thing too because I saw that it substantially degraded the performance of Topping D70s -- something the company does not warn about.

On your #1, DACs internally already upsample to very high frequencies. And provide noise shaping if they want. By your argument I should stop running 1 kHz test for all DACs then. Let's remember that by forcing the DAC to run at higher sample rate, the M-scaler is causing the DAC to operate differently. So it is important to measure its analog performance for that tone as well.

Also, when people make claims of improvements for M-scaler, they don't at all limit it to transients. They may do so after reading Rob Watt's argument but they are not in absence of that.

That aside, Rob provided no evidence that said transient response is a) improved and b) it is audibly beneficial. The word transient is thrown around in audio as standard marketing buzzword. As in that context, it has no meaning.

Let's remember that a true transient has infinite bandwidth and energy! No way then that it can be contained when sampled with 44.1 kHz which is 99% of what people listen to. Harmonics of any signal over 11 kHz and change is chopped off and gone forever from the format. And even if it weren't, we would not hear it anyway as it is outside of audible range.

It is trivial to improve transient response by increasing bandwidth. But per above, you are reproducing things that you don't hear. And your speakers likely won't reproduce (or distort if they did).

All this said, I did run upsampler specific test in my filter measurements:

index.php


There, we again discovered deficiencies in the M-scaler in the way it outputs over its single link. Its attenuation is worse than the naked DAC, something that Rob accepted in your interview but was absence in all prior talk about this device.

All in all, I found his answers in your interview very disappointing. He starts by saying he has not read my review. Really? He came for an interview on the topic with no knowledge of what I tested? Maybe if he had read the review he would have realized the questions you raised were not valid. Second, he didn't bring any objective measurements, or blind controlled tests to back his claims. So it was just a lot of talk with no reliable evidence.

As a reviewer, you need to be on guard to not be manipulated by a manufacturer with fancy buzzwords and unsupported claims. You started to ask hard questions but then accepted everything he said. That is not right. When he says he has conducted hundreds of even thousands of listening tests, you need to ask him if these were documented in any form so that we could see how valid they are (and try to replicate). At end of one of his talks at an audio show where he repeated his "structured/controlled" listening tests many times, I asked him if his tests were blind. He said no and that blind tests are stressful! Well, they are stressful if there is no difference to be found!

If his answers were too technical, you could have sent me a link to the video in advance of publishing to get my response. But you didn't and published his one-sided response. Before you say my review was also one sided, I provided tons of evidence including listening test samples. He didn't do any of this.

I am happy to see you here so that we can properly answer back. But wish you would put your consumer hat on and defend their interest first and foremost.
 
Top Bottom