Hugo M Scaler by Chord Electronics - The Official Thread
Dec 27, 2019 at 7:07 PM Post #9,811 of 18,478
I should also say that at 10’, I could very clearly hear the difference between Chromecast Audio toslink, and HDMI from my NUC to Kanex toslink, and XSPDIF2 toslink to HMS. Distance made no difference to the impact I heard, even through OPtoDX to DAVE.

In my opinion, there is Definitely something going on with clock quality on toslink, and something other than radiated RFI at play. However I am hyper eager to be proven wrong by the RF stop box!
 
Dec 27, 2019 at 7:26 PM Post #9,812 of 18,478
I should also say that at 10’, I could very clearly hear the difference between Chromecast Audio toslink, and HDMI from my NUC to Kanex toslink, and XSPDIF2 toslink to HMS. Distance made no difference to the impact I heard, even through OPtoDX to DAVE.

In my opinion, there is Definitely something going on with clock quality on toslink, and something other than radiated RFI at play. However I am hyper eager to be proven wrong by the RF stop box!
Soz Ray, what happened when the sources were 10 foot away that didn’t happen when they were less then 10 foot away :)
I’m considering buying a 20 metre toslink cable to see if having the source very far away makes a noticeable difference to me .. as it only costs £15 to try ...
 
Dec 27, 2019 at 7:36 PM Post #9,813 of 18,478
As you can imagine, not the easiest thing to A/B. I moved my NUC end point, ISO Regen, XSPDIF 2, HMS, and Opto DX TX unit 10’ away from the Optó DX RX unit and my DAVE. Even at that separation, I heard different things very clearly on the digital source side (same as when the stack was next to my DAVE). I can’t say (yet) whether 10’ away is better than right next to DAVE, but I can say that in my setup, that it all the digital tweaks and changes were still as audible in both positions.

In my case, I used 3 meter optical link with the ODX to get my separation, not a long Toslink cable (I wanted to get HMS away from DAVE as well)

the real test will be with Dans RF stop box. That will actually make A/B testing easier, so I’ve been waiting to do those tests until then.
 
Dec 28, 2019 at 7:50 AM Post #9,815 of 18,478
Last edited:
Dec 28, 2019 at 8:51 AM Post #9,816 of 18,478
If anyone is bored in the run up to the new year, and looking for some audiophile project, the principles of making a DIY faraday cage are explained in various pages/videos on the web, for example https://backyardbrains.com/experiments/faraday
It doesn't look difficult, once one has got hold of the screen metal mesh.
The most difficult aspect is probably making the faraday cage stylish enough to be a permanent fixture on someones hifi stand

Alternatively here is a ready built faraday cage, that one can buy from Amazon. https://www.amazon.com/WiFi-Router-Guard-Large-Blocks/dp/B01IJD22YG

For RF it is a bit more complicated than that DIY set up I'm afraid but there are firms who sell RF isolation enclosures and who will install interfaces in the enclosure walls so that devices within the enclosure can be connected to equipment outside the enclosure. See for instance this link.

https://jretest.com/product/jre-1812/

For me I have sort of got to the point of my system sounding 'pretty darned good' and I am mostly content to just sit back and listen to it rather than tinker at the moment. I am pretty sure Rob Watts is tinkering away on our behalf anyway and I am hoping he will have some new toys for us to buy in the coming year . . . . .
 
Dec 28, 2019 at 10:17 AM Post #9,817 of 18,478
Hi everyone,

In the last two months, I tested the M Scaler in a couple of different setups and I think finally I figured out what it does with the sound. After I received the TT2 I had the same impression as some of you guys, that it doesn’t do anything but it dose.

So first of all in the first month I had only the stock BNC cables and was running the M Scaler with Schiit Yggdrasil A2, Chord Mojo and Hugo TT2.

With Schiit Yggdrasil on 192 kHz max sample rate, everything was clearer, more focused and precise, and the soundstage was a little bigger.

Using the Mojo at 384 kHz I had the impression that with M Scaler sound was more rounded on the edges and more 3D layering. The tonality was softer, like using a tube amp.

(most of my listening was made using AudioFly AF180 IEMs, Hifiman HE1000se, and Meze Empyrean).

With TT2 I could do fast A/B comparison using M Scaler (double BNC) and simple optic input from MacBook Pro (I equalized the volume and the TT2 remote help me switch fast between imputes). My first impression was that at 192 kHz sample, there was no difference between M Scaler and upsample using only Audirvana from Macbook. The sound was just a little smoother on M Scaler but slower. At the max sample of 768 kHz, I didn’t like the sound. When fast comparing it seems the same in detail, imaging but when listening for a long time something was not right. I couldn’t find what. TT2 is a remarkable DAC full of recording room details retrieval and good tonality (a hint of V shape sound is present depending on the unit).

After I introduce the Wave Stone BNC, I start observing with ease what the M Scaler was doing.

The most obvious setup was using TT2 and IEMs. On live recordings like Ariana Grande Vivo live (Ariana Grande - Dangerous Woman (Vevo Presents)), using M Scaler on max 768 kHz, the sound was more around my head, coming from behind and from the side, and the stage was closer, bigger. Bypassing the M Scaler, all the stereo presentation moved in front and at a big distance. With M Scaler in the chain, it was like you are there at the event in the second row. It was a very very cool effect, I was amazed. Using just one BNC at 384 kHz had the same effect maybe a little smaller. The effect disappears when the sample rate is at 192 kHz or low.

So I listen to the same thing using He1000se and had a surprise. With TOTL full-size open headphones this effect of around your head and a close stage like you're there is rendered without M Scaler. Adding the M Scaler in the chain moved the whole stage even closer and this effect was strange. All the front vocals or instruments were in my face and a lot of them appeared to cut my head in two. It was like you are entering too far with your head in the recording. Also, there was a perception of empty space left and right from the central singer and all the orchestra was behind my ears. This effect was also the same with Meze Empyrean and Sennheiser HD800.

I tried the TT2 and M Scaler at some friends on a couple of expensive speaker systems and they were very impressed. At 384 kHz and 768 kHz, the effect of the whole stereo scene coming closer was there. The sound was decoupling from speakers with more ease and it appeared more live.

I don’t know if listening to songs using this effect is a good thing or not. It alters the originally intended presentation. Especially in very expensive headphones, it seems a bad thing from my perspective.

These are my personal findings and impressions so please see if you guys also are heading these things.

034BF3F1-72EC-4EC5-B442-525F632C64B8.jpeg
 
Last edited:
Dec 28, 2019 at 10:36 AM Post #9,819 of 18,478
I’m getting the upgrade itch :triportsad:

For those that have heard the TT2 with the M-Scaler and the DAVE with the M-Scaler, how much better is the DAVE with the M-Scaler?
 
Dec 28, 2019 at 11:14 AM Post #9,820 of 18,478
I’m getting the upgrade itch :triportsad:

For those that have heard the TT2 with the M-Scaler and the DAVE with the M-Scaler, how much better is the DAVE with the M-Scaler?

A LOT BETTER. It is not subtle.
 
Dec 28, 2019 at 11:16 AM Post #9,821 of 18,478
Has anyone been able to compare the M-Scaler to HQPlayer4? If so, what were your thoughts?
For a true apples-to-apples comparison, always run signals through Mscaler from your source via USB. So it's Source (HQPlayer OFF)->mscaler and then Source(HQPlayer ON)->mscaler. Where OFF means original (say 44.1/16) music is passed to Mscaler for WTA1 upsampling, and ON means set
  • PCM
  • Bits=24
  • Max=768khz
  • Filter=poly-sinc-long-lp (million taps sinc linear phase response)
  • Dither=ns9 (9th order noise shaping)
  • Adaptive rate (so 44.1>705.6 and 48>768)
This is, IMO, so very close - technically and soundwise. I've asked Jussi to code up an 11th order noise shaper to get it even closer.

Dan
 
Dec 28, 2019 at 11:19 AM Post #9,822 of 18,478
Has anyone been able to compare the M-Scaler to HQPlayer4? If so, what were your thoughts?

I compared the two a bit, and as much as I'd really really like to, I couldn't quite convince myself that HQPlayer's output was as good as the M Scaler. I've grown pretty attached to the M Scaler's subtle, un-harsh transients and warm presentation, which I don't get quite as much with HQPlayer. That's about as far as I'm able to describe the differences at the moment. I'd like to do more comparing though.

Also, a few words on setup for those who are interested. It's easy to do an apples-to-apples comparison by simply connecting the PC to the M Scaler via USB, and then setting HQPlayer's filter to either whatever filter you want or "none". If a filter is set (assuming the upsampling rate is set to 708/768), then HQPlayer will be doing the upsampling; if no filter is set, then the upsampling will be handled by the M Scaler. The resulting volume will be identical.

The HQPlayer settings I'm using are:
  • Device: "Hugo M Scaler" (be sure to set this explicitly)
  • Bit depth: Default
  • Sample rate: 768k, with "Adaptive output rate" set to "match rate family"
  • Dither: NS9 (Jussi insists that dither should not be set to "none", FYI)
  • Filter: I've tried "poly-sinc-long-lp" and a few others. I liked "closed-form-m" quite a bit.
 
Dec 28, 2019 at 11:21 AM Post #9,823 of 18,478
For a true apples-to-apples comparison, always run signals through Mscaler from your source via USB. So it's Source (HQPlayer OFF)->mscaler and then Source(HQPlayer ON)->mscaler. Where OFF means original (say 44.1/16) music is passed to Mscaler for WTA1 upsampling, and ON means set
  • PCM
  • Bits=24
  • Max=768khz
  • Filter=poly-sinc-long-lp (million taps sinc linear phase response)
  • Dither=ns9 (9th order noise shaping)
  • Adaptive rate (so 44.1>705.6 and 48>768)
This is, IMO, so very close - technically and soundwise. I've asked Jussi to code up an 11th order noise shaper to get it even closer.

Dan
LOL GMTA. Dan, should bit depth be set to 24, and not "default" or 32? Just wondering...
 
Dec 28, 2019 at 11:38 AM Post #9,824 of 18,478
LOL GMTA. Dan, should bit depth be set to 24, and not "default" or 32? Just wondering...
Don't confuse 32-bits with the resolution of the content/stream you can send to the HQPLayer/MScaler. 24-bits refers to the output from HQPlayer to the DAC. Internally both HQPlayer and WTA1 use ultra high precision filter computation (80-bit floating point in the case of HQPlayer). This is then converted to the final output resolution of 24-bits PCM as required by the DX (dual coax BNC) interface. MScaler set this 24-bits automatically but HQPlayer provides a general flexibility here. I hard-code to 24-bits just to be sure. You can leave dither=none, which is ever so slightly bright sounding as it is just a truncation/rounding to 24-bits ; or you can use the dither=XXX setting to touch the lower order bit to move the near-ultrasonic noise up in frequency. HQPlayer offers various settings here but as Rob has indicated elsewhere, the higher order noise shapers sound best for depth preservation. (which, BTW, I find incredible that we can hear this).
 
Last edited:
Dec 28, 2019 at 12:40 PM Post #9,825 of 18,478
Don't confuse 32-bits with the resolution of the content/stream you can send to the HQPLayer/MScaler. 24-bits refers to the output from HQPlayer to the DAC. Internally both HQPlayer and WTA1 use ultra high precision filter computation (80-bit floating point in the case of HQPlayer). This is then converted to the final output resolution of 24-bits PCM as required by the DX (dual coax BNC) interface. MScaler set this 24-bits automatically but HQPlayer provides a general flexibility here. I hard-code to 24-bits just to be sure. You can leave dither=none, which is ever so slightly bright sounding as it is just a truncation/rounding to 24-bits ; or you can use the dither=XXX setting to touch the lower order bit to move the near-ultrasonic noise up in frequency. HQPlayer offers various settings here but as Rob has indicated elsewhere, the higher order noise shapers sound best for depth preservation. (which, BTW, I find incredible that we can hear this).
Jussi definitely says dither is required to prevent distortion. I am going to try the NS9. In my opinion HQPlayer sounds imperceptible to my mscaler.
 

Users who are viewing this thread

Back
Top