This is a review, listening tests and measurements of the CHORD Hugo M-scaler upsampling digital transport. It is on kind loan from a member and costs US $5,295 (some show it as $5,650).
We have the usual lightshow of buttons as other CHORD digital products but here, due to less overloading of the buttons, it is not hard to navigate. Still, it is annoying to have to keep looking up the color coding in the manual until you memorize them. Here is the back panel:
The unit accepts digital audio over either USB or S/PDIF inputs (coax and optical). Very oddly for a device in this class, a large PC laptop type power supply is provided, needlessly causing anxiety for most of their customers buying this tweak. An internal power supply should have been provided. Output is in the form of S/PDIF again or dual BNC outputs which is proprietary to CHORD. Fortunately the member kindly provided a Hugo 2 and all the necessary (high-end Nordost) cabling to make the system work.
If you are unfamiliar with the marketing of this product, it takes an input and upsamples it to higher rate using incredibly high number of taps. High taps allows sharper transition from pass-band to ultrasonic. Company claims that one needs huge number of taps for better fidelity, literally going to millions of them. This has a side-effect of creating fair amount of latency so there is a "video" mode where if it detects 48 kHz sampling, it will shut off the upsampling (or lowers its taps?). While 48 kHz is a common sampling rate for video, many other rates are used so this is not a full fix. Fortunately a "bypass" mode is also provided which basically disables the filter. However, it still lowers the levels to make it match the level when upsampling is used (needed for headroom as to avoid overflow). Company rightly realizes that if the volume is higher in bypass mode, customer may think that sounds better and reject the whole notion of the unit. If so, I wonder then why company doesn't believe in going the next step and providing level matched, controlled blind test to show the efficacy of the unit.
Back to upsampling, the rates provided are 2X, 4X and 16X with the latter needing dual BNC connection at 768 kHz.
CHORD M-Scaler Measurements
My first test was to examine the quality of the Coax output with respect to jitter. Results were disappointing and depended in how you output bits to the unit. Let's start with USB connection:
The green line is the Audio Precision analyzer measuring jitter spectrum of its own Coax signal. The red, green and orange lines are what we get when we send the same digital signal through M-Scaler, i.e. USB In/Coax out to the analyzer. We see a large increase (in relative terms) of the baseline noise indicating random jitter. The correlated spikes are elevated much higher as well.
Turns out the above was the "good case." Here is the bad case when we use Audio Precision's coax output to feed the M-scaler:
Good grief. What is going on here? Not only do we have the same baseline (higher) noise floor but now have distinct deterministic spikes coming out of the unit. I captured the AP and M-Scaler's computed jitter level and it is embarrassingly bad:
It is so bad that you can see the problem just looking at the waveform of jitter:
On the left we see AP analyzer's benign, very low level noise. On the right we see clearly, somewhat sinusoidal of M-scaler. I have tested $100 digital bridges that far outperform the M-Scalar on this front! As a way of reference, 500 picosecond/0.5 nanosecond is enough jitter to equal one bit of 16 bit digital audio sample. M-Scalar is outputting 8X that amount!
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.
Next I connected a Topping D70S to the analyzer and captured its output by itself:
We see the great performance that we expect from this DAC. Now let's put the M-scaler in the loop:
What the heck happened here? Even in pass through mode performance of D70s is heavily degraded. We have spikes at oddball frequencies which I can't explain. I see reviewers with company blessing using M-scaler with third-party DACs. I would say be cautious, very cautious doing so.
Fortunately there was no problem using M-scaler with Hugo 2:
Then again there was no benefit to upscaling either:
And here is the response with dual Coax:
Still the same.
I should note that the Hugo 2 was running on battery for all of these tests per suggestion from the owner.
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 with the D70s:
It is indeed. Both 2X and 4X modes provide an ultra sharp response. However they do so at the cost of higher noise floor/lower attenuation.
I ran a very high resolution frequency response test focusing on the end of the spectrum and results are basically the same in all modes:
The bypass mode has the sharp filter and the others do not because the DAC is operating at higher sample rate. With respect to audible band up to nearly 22 kHz though, the frequency response is essentially the same. So anyone saying tonality has changed with upscaling modes needs to re-think that.
CHORD M-scaler and Hugo 2 Listening Tests
Conveniently, the switch between bypass and 2X upsampling was seamless. So I used that to perform listening tests. I say some because the button cycles to higher rates and there, it causes a glitch/pause which completely throws off your mind. Going by forward mode of bypass to 2X, I detected no difference at all. Nothing changes as far as tonality, soundstage, etc. I captured a couple of samples for you to listen as well:
www.dropbox.com
www.dropbox.com
You can listen right in the dropbox or download them. If you are sure there is a difference, go ahead and say which one is M-scaler bypass and which is 2X.
FYI I could not run a null test with DeltaWave as there is too much clock drift and clips are not lined up. The difference shown was the same as two consecutive captures without changing anything.
Conclusions
I was very disappointed to see high noise levels and jitter on Coax outputs of such an expensive digital products. And one where its designer claims you can hear problems at 300 dBFS. Well, if you can hear that well, then maybe you can hear this jitter too so I suggest not releasing a product that performs this bad. Second, there is a serious compatibility issue with third-part DAC that I tested. Maybe it is OK with other DACs, I don't know. But grabbing a sample DAC and having such serious degradation in its performance with M-scaler is very concerning.
Putting design issues aside, I found no audible difference in upsampling with company's own DAC. As it should be. The differences in the filter are above audible band so they should not be audible. Is it nice to have a sharper filter? Sure. But I sure as heck wouldn't pay nearly $6,000 to get that.
Bottom line, I don't see a reason to own M-Scaler. It can damage the audio signal in some cases and in others, provide no audible benefit. If people think otherwise, I highly suggest performing a blind test with enough repetition to provide statistically valid results. Otherwise, the M-scaler remains as a product with no purpose.
Needless to say, I can't recommend the CHORD M-Scaler.
-----------
As always, questions, comments, recommendations, etc. are welcome.
Any donations are much appreciated using: https://www.audiosciencereview.com/forum/index.php?threads/how-to-support-audio-science-review.8150/
We have the usual lightshow of buttons as other CHORD digital products but here, due to less overloading of the buttons, it is not hard to navigate. Still, it is annoying to have to keep looking up the color coding in the manual until you memorize them. Here is the back panel:
The unit accepts digital audio over either USB or S/PDIF inputs (coax and optical). Very oddly for a device in this class, a large PC laptop type power supply is provided, needlessly causing anxiety for most of their customers buying this tweak. An internal power supply should have been provided. Output is in the form of S/PDIF again or dual BNC outputs which is proprietary to CHORD. Fortunately the member kindly provided a Hugo 2 and all the necessary (high-end Nordost) cabling to make the system work.
If you are unfamiliar with the marketing of this product, it takes an input and upsamples it to higher rate using incredibly high number of taps. High taps allows sharper transition from pass-band to ultrasonic. Company claims that one needs huge number of taps for better fidelity, literally going to millions of them. This has a side-effect of creating fair amount of latency so there is a "video" mode where if it detects 48 kHz sampling, it will shut off the upsampling (or lowers its taps?). While 48 kHz is a common sampling rate for video, many other rates are used so this is not a full fix. Fortunately a "bypass" mode is also provided which basically disables the filter. However, it still lowers the levels to make it match the level when upsampling is used (needed for headroom as to avoid overflow). Company rightly realizes that if the volume is higher in bypass mode, customer may think that sounds better and reject the whole notion of the unit. If so, I wonder then why company doesn't believe in going the next step and providing level matched, controlled blind test to show the efficacy of the unit.
Back to upsampling, the rates provided are 2X, 4X and 16X with the latter needing dual BNC connection at 768 kHz.
CHORD M-Scaler Measurements
My first test was to examine the quality of the Coax output with respect to jitter. Results were disappointing and depended in how you output bits to the unit. Let's start with USB connection:
The green line is the Audio Precision analyzer measuring jitter spectrum of its own Coax signal. The red, green and orange lines are what we get when we send the same digital signal through M-Scaler, i.e. USB In/Coax out to the analyzer. We see a large increase (in relative terms) of the baseline noise indicating random jitter. The correlated spikes are elevated much higher as well.
Turns out the above was the "good case." Here is the bad case when we use Audio Precision's coax output to feed the M-scaler:
Good grief. What is going on here? Not only do we have the same baseline (higher) noise floor but now have distinct deterministic spikes coming out of the unit. I captured the AP and M-Scaler's computed jitter level and it is embarrassingly bad:
It is so bad that you can see the problem just looking at the waveform of jitter:
On the left we see AP analyzer's benign, very low level noise. On the right we see clearly, somewhat sinusoidal of M-scaler. I have tested $100 digital bridges that far outperform the M-Scalar on this front! As a way of reference, 500 picosecond/0.5 nanosecond is enough jitter to equal one bit of 16 bit digital audio sample. M-Scalar is outputting 8X that amount!
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.
Next I connected a Topping D70S to the analyzer and captured its output by itself:
We see the great performance that we expect from this DAC. Now let's put the M-scaler in the loop:
What the heck happened here? Even in pass through mode performance of D70s is heavily degraded. We have spikes at oddball frequencies which I can't explain. I see reviewers with company blessing using M-scaler with third-party DACs. I would say be cautious, very cautious doing so.
Fortunately there was no problem using M-scaler with Hugo 2:
Then again there was no benefit to upscaling either:
And here is the response with dual Coax:
Still the same.
I should note that the Hugo 2 was running on battery for all of these tests per suggestion from the owner.
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 with the D70s:
It is indeed. Both 2X and 4X modes provide an ultra sharp response. However they do so at the cost of higher noise floor/lower attenuation.
I ran a very high resolution frequency response test focusing on the end of the spectrum and results are basically the same in all modes:
The bypass mode has the sharp filter and the others do not because the DAC is operating at higher sample rate. With respect to audible band up to nearly 22 kHz though, the frequency response is essentially the same. So anyone saying tonality has changed with upscaling modes needs to re-think that.
CHORD M-scaler and Hugo 2 Listening Tests
Conveniently, the switch between bypass and 2X upsampling was seamless. So I used that to perform listening tests. I say some because the button cycles to higher rates and there, it causes a glitch/pause which completely throws off your mind. Going by forward mode of bypass to 2X, I detected no difference at all. Nothing changes as far as tonality, soundstage, etc. I captured a couple of samples for you to listen as well:
Dropbox
www.dropbox.com
Dropbox
www.dropbox.com
You can listen right in the dropbox or download them. If you are sure there is a difference, go ahead and say which one is M-scaler bypass and which is 2X.
FYI I could not run a null test with DeltaWave as there is too much clock drift and clips are not lined up. The difference shown was the same as two consecutive captures without changing anything.
Conclusions
I was very disappointed to see high noise levels and jitter on Coax outputs of such an expensive digital products. And one where its designer claims you can hear problems at 300 dBFS. Well, if you can hear that well, then maybe you can hear this jitter too so I suggest not releasing a product that performs this bad. Second, there is a serious compatibility issue with third-part DAC that I tested. Maybe it is OK with other DACs, I don't know. But grabbing a sample DAC and having such serious degradation in its performance with M-scaler is very concerning.
Putting design issues aside, I found no audible difference in upsampling with company's own DAC. As it should be. The differences in the filter are above audible band so they should not be audible. Is it nice to have a sharper filter? Sure. But I sure as heck wouldn't pay nearly $6,000 to get that.
Bottom line, I don't see a reason to own M-Scaler. It can damage the audio signal in some cases and in others, provide no audible benefit. If people think otherwise, I highly suggest performing a blind test with enough repetition to provide statistically valid results. Otherwise, the M-scaler remains as a product with no purpose.
Needless to say, I can't recommend the CHORD M-Scaler.
-----------
As always, questions, comments, recommendations, etc. are welcome.
Any donations are much appreciated using: https://www.audiosciencereview.com/forum/index.php?threads/how-to-support-audio-science-review.8150/
Last edited:
