Grimm Audio MU1 and MU2 Music Players

[PYP]

3 minutes ago, The Computer Audiophile said:

 

I'v been thinking about that paragraph above and don't think it's totally correct. In fact, I'm not sure there is a current and correct definition that delineates between up and over sampling. The MU1 calls its capability Oversampling (2fs and 4fs). But, this is nowhere near what's "needed" to run the D/A conversion (if sticking with the paragraph above). Almost all DACs will go much higher whether the users likes it or not. 

 

Perhaps a digital expert like @Miska could lend us a hand?

Not being a technologist, it is all confusing to me.  That paragraph is followed by a section on Upsampling and the relationship between the two.  

[The Computer Audiophile]

8 minutes ago, Mike123 said:

as a DAVE DAC owner, Ive found 4X to be a bit too overwhelming in midrange. 2X seems to be more immersive. Anyone else with DAVE thats experienced same?

That's an interesting observation. I'm using different DACs, but I can see/hear what you're taking about. However, I'm also hearing differences based on the music / instrument. 

 

 

[PYP]

7 minutes ago, Nicholas Dong said:

What power cables do you use with Mu1?  I am using Vovox excelsus power cable for now. I have not tried other cables. What do you think is the most suitable power cable for MU1?  Eelco claims not to use HIFi cables, its 1 is that will damage the power jack, its 2 I think most power cables add clarity, and MU1 upsampling is doing the same thing, so is there a conflict between them  , what do you think?

The MU1 doesn't seem as vulnerable to noise via the power cord as other digital devices I've used, so I can understand Eelco's preference to keep this part simple.  I use a Kubala-Sosna cable that has worked well with other equipment too.   But there are lots of options for power cables.  

 

If you believe the theory of power cables, then the cable not only filters out noise coming into the component, but noise leaving the component back into the power conditioner/electrical line.  If the MU1 is mostly immune to incoming noise and doesn't inject noise back into the electrical circuit, then that is the result of a well-built power supply (Grimm Audio apparently spent a lot of time and ingenuity on getting this part right).  

[Miska]

2 hours ago, The Computer Audiophile said:

I'v been thinking about that paragraph above and don't think it's totally correct. In fact, I'm not sure there is a current and correct definition that delineates between up and over sampling. The MU1 calls its capability Oversampling (2fs and 4fs). But, this is nowhere near what's "needed" to run the D/A conversion (if sticking with the paragraph above). Almost all DACs will go much higher whether the users likes it or not. 

 

Perhaps a digital expert like @Miska could lend us a hand?

 

I think the article you posted is fairly accurate in terms of definition. HQPlayer supports both, but the rate conversion for DSD outputs is specifically categorized as oversampling in HQPlayer, since the digital filter output rate is at minimum 2 MHz. But the two terms can be many times used interchangeably.

 

I would call 2x and 4x upsampling, it is not as such sufficient for running D/A conversion without further processing. 2x would be like very first CD players from early 80's. And 4x would be like CD players from late 80's and very early 90's.

 

Unless you are running a NOS DAC, it will do further processing from 2x/4x inputs before conversion.

 

[Evo1668]

Eelco  replied to one of the first UK reviews of the MU1 by The Ear two years ago in it he states

 

https://the-ear.net/review-hardware/grimm-audio-mu1-digital-hub-network-streamer-server/

 

“it is interesting to read that Jason prefers to disable the MU1 upsampling, and also the internal fpga upsampling of the iFi DAC (“bit perfect’ mode). The iFi DAC has a custom ‘interleaved’ construction of 4 Burr-Brown DAC chips. It makes me wonder if this DAC topology for some reason performs better when fed with a lower sample rate. It is hard to tell. In any case just like iFi we offered the option to turn the fpga upsampling off to check which setting offers the best sound quality. Usually upsampling wins, but you never know. For example, with some pop albums, the ‘edge’ that lower quality upsamplers in DAC chips offer was part of the decision making during mastering and could be missed by some.”

Eelco Grimm, Grimm Audio

[pmorali]

5 hours ago, The Computer Audiophile said:

I've got a good handle on this now. 

 

Question 1: Do you guys change the MU1 upsampling setting (No, 2fs, 4fs) for different music, or leave it alone once you set it?

 

Question 2: Is anyone using convolution with the MU1?

1. I set it at 4FS and I left it alone since.

2. I am not using any convolution with the MU1.

[FredM]

4 hours ago, The Computer Audiophile said:

Question 1: Do you guys change the MU1 upsampling setting (No, 2fs, 4fs) for different music, or leave it alone once you set it?

 

Question 2: Is anyone using convolution with the MU1?

 

I’m using the DAC’s in the Grimm LS1’s (your earlier question)
1: 4fs, all the time

2: not that I know 

 

4 hours ago, The Computer Audiophile said:

 

In fact, I'm not sure there is a current and correct definition that delineates between up and over sampling. The MU1 calls its capability Oversampling (2fs and 4fs). But, this is nowhere near what's "needed" to run the D/A conversion (if sticking with the paragraph above). Almost all DACs will go much higher whether the users likes it or not. 

This might be interesting: 

Your questions remind me of this interview. It’s in Dutch, but has very decent English subtitles. Eelco explains the importance of a clean clock, jitter treatment, Grimms preferred transport to external DAC’s (AES, which carries the clock) and which effect has been noticed when a DAC ‘chip/calculator’ is feeded with a 176.4 or 192 kHz signal. The video link has a timestamp to this specific subject. Of course the other parts of the interview might also be interesting and relevant.
 

Trying to summarise that part of the interview: The 176.4 and 192 kHz signal are closer to the ‘native’ format of the DAC ‘chips’. Then less digital computation from the ‘DAC chip’ is needed, the DAC’s calculator/internal upsampler is less stressed, less rounding errors, which lead to beter sonic performance.
 

 

@The Computer AudiophileSome posts ago you also mentioned dCS upsamplers, I believe that dCS are using their own DAC design with patented and home build DAC ‘chips’ /internal upsamplers!? (Don’t know for sure). Perhaps the effect of a 176.4 or 192 kHz feed works different with dCS DAC’s (when they used a native format which is not close to 176.4 or 192 kHz). Would be interesting how the MU2 design will look like (where the MU2 is equipped with a DAC), if I understand Eelco in the interview above correctly the output of the FPGA board is directly connected with the DAC (where only the FPGA board processes the calculation, which would remove the intermediate step / calculation which usually happens in the DAC chip). 

 

[The Computer Audiophile]

51 minutes ago, FredM said:

 

I’m using the DAC’s in the Grimm LS1’s (your earlier question)
1: 4fs, all the time

2: not that I know 

 

This might be interesting: 

Your questions remind me of this interview. It’s in Dutch, but has very decent English subtitles. Eelco explains the importance of a clean clock, jitter treatment, Grimms preferred transport to external DAC’s (AES, which carries the clock) and which effect has been noticed when a DAC ‘chip/calculator’ is feeded with a 176.4 or 192 kHz signal. The video link has a timestamp to this specific subject. Of course the other parts of the interview might also be interesting and relevant.
 

Trying to summarise that part of the interview: The 176.4 and 192 kHz signal are closer to the ‘native’ format of the DAC ‘chips’. Then less digital computation from the ‘DAC chip’ is needed, the DAC’s calculator/internal upsampler is less stressed, less rounding errors, which lead to beter sonic performance.
 

 

@The Computer AudiophileSome posts ago you also mentioned dCS upsamplers, I believe that dCS are using their own DAC design with patented and home build DAC ‘chips’ /internal upsamplers!? (Don’t know for sure). Perhaps the effect of a 176.4 or 192 kHz feed works different with dCS DAC’s (when they used a native format which is not close to 176.4 or 192 kHz). Would be interesting how the MU2 design will look like (where the MU2 is equipped with a DAC), if I understand Eelco in the interview above correctly the output of the FPGA board is directly connected with the DAC (where only the FPGA board processes the calculation, which would remove the intermediate step / calculation which usually happens in the DAC chip). 

 

Yes, this is the reasoning for upsampling to 2 or 4fs.
 

If you look at Chord DACs and upsample to the max input rate of it’s usb input, the DAC works half as hard to upsample to its final rate. I believe this can be seen in power supply draw (I could be mistaken on the objective measure). 
 

The same goes for dCS. The upsampler is used to feed the DAC to make it easier for the DAC to get to its native rate. 
 

I was just trying to find out if that’s what Grimm is going for with the MU1 because almost no DACs top out at 4fs internally. It would make sense the MU1 is acting similarly to the dCS upsampler. 
 

I also think Grimm should change the terminology to upsampling. It says oversampling in the web interface, but 4fs is a long way from an oversample these days. 

[The Computer Audiophile]

Hi Guys the review is done. I just sent it for fact checking and will publish as soon as I get it back.

[FredM]

6 hours ago, The Computer Audiophile said:

because almost no DACs top out at 4fs internally.

I thought it’s not about what is maximum spec is, but what format the DAC ‘chip’ is less stressed. Another reason seems to be that at 4fs the MU1 is able to transfer the jitter treaded clock signal (via AES) to the external DAC.

 

Looking forward reading your review 👍

[aangen]

I have my MU1 —>Stealth Octavia-T AES Cable —>Gryphon Ethos —>Gryphon Commander —>Gryphon Apex powering Vivid G1 Spirits. All audio cables are V17 Sakras. I just leave it at 4FS and occasionally I try the +3dB function. If I did a blind test I am not certain I could reliably pick 2X or 4x more than 50% of the time.

 

What kind of upgrade do I need to finally hear the difference? The MU1 just works for me, I have zero interest in trying anything else. I made a Pink Faun server be my Roon Server and used the MU1 as an endpoint. That was amusing but unnecessary. The Pink Faun didn’t make a difference I could hear.
 

I have a pretty quickly growing library of DSD256 files. It moves through the MU1 smoothly and sweetly.

 

I believe everything the MU1 sends out that AES port is blessed by Eelco and that will be enough for me.

[bluebeat]

 

@The Computer Audiophile

I'm using the MU1 with an Ayre QX-5 Twenty. It pushes the DAC to unsuspected heights, most notably with 16/44 files. Always 4fs, no convolution.

[Miska]

6 hours ago, FredM said:

I thought it’s not about what is maximum spec is, but what format the DAC ‘chip’ is less stressed.

 

Reason to upsample/oversample is not about how stressed DAC chip or DAC is. But to use technically best possible computation algorithms and filter design for doing the filter that is needed to reject out of band images so that the DAC's analog reconstruction filter can the remove the rest.

 

What is best possible algorithm depends on first on objective features of the filter, such as amount of pass-band ripple, roll-off characteristics and amount of stop-band attenuation, in addition to highest possible rate it can output. And secondly it depends on listener's personal sensitivity to different aspects of the the filter and the music genre being listened. As different music genres have different signal properties.

 

[The Computer Audiophile]

My review is up.

 

 

[Nicholas Dong]

What does MU1's convolution mean?

[The Computer Audiophile]

2 minutes ago, Nicholas Dong said:

What does MU1's convolution mean?

MU1 doesn’t have a convolution engine, so it doesn’t support convolution filters. 

[Nicholas Dong]

1小时前，计算机音频爱好者说：

MU1没有卷积引擎，因此它不支持卷积滤波器。

Is the convolution engine similar to the filter of the DAC? I don't quite understand the meaning of the convolution engine

[The Computer Audiophile]

2 minutes ago, Nicholas Dong said:

Is the convolution engine similar to the filter of the DAC? I don't quite understand the meaning of the convolution engine

Very different. 

 

Here is an interview I did with @mitchco, in which he explains convolution. 

 

A convolution engine is software the processes a convolution filter. A convolution filter is used for digital room correction. 

 

 

 

 

 

[Nicholas Dong]

7 hours ago, The Computer Audiophile said:

非常不同。

 

这是我做的一次采访@米奇科，其中他解释了卷积。

 

卷积引擎是处理卷积滤波器的软件。卷积滤波器用于数字房间校正。

 

 

 

 

 

I finally understand, it turned out to be similar to the Roon DSP room calibration function, thanks for the answer

[Nicholas Dong]

Regarding the upsampling of MU1, do you prefer Sharp or slow for the DAC filter? I personally like MU1 upsampling 4S with DAC Slow filter

[The Computer Audiophile]

16 minutes ago, Nicholas Dong said:

Regarding the upsampling of MU1, do you prefer Sharp or slow for the DAC filter? I personally like MU1 upsampling 4S with DAC Slow filter

When using the T+A DAC I like Bez 2. 
 

“Pure Bezier interpolator. This process delivers a perfect reconstruction of the original music signal. It exhibits no pre- or post-echoes of any kind, and does not add coloration or timing errors to the original signal. In sonic terms this method offers an impressiveblend of naturalness, good dynamics and accuracy.

[audiobomber]

On 10/25/2022 at 2:14 PM, The Computer Audiophile said:

I've been thinking about that paragraph above and don't think it's totally correct. In fact, I'm not sure there is a current and correct definition that delineates between up and over sampling.

I researched this some time ago and came up with a definition from somewhere that made sense to me. I don't think it was an audio site (sorry, no reference).

 

For audio:

- Oversampling is sampling at a higher frequency than needed, e.g. 16/44.1 to 16/192

- Upsampling is adding bits that were not present in the original signal, e.g. 16/44.1 to 24/44.1

 

According to this definition, going from 16/44.1 to DSD or 24/176.4 is upsampling and oversampling, but I am satisfied to call it upsampling.

 

Dan (not an expert)

[Caso]

2 hours ago, audiobomber said:

I researched this some time ago and came up with a definition from somewhere that made sense to me. I don't think it was an audio site (sorry, no reference).

 

For audio:

- Oversampling is sampling at a higher frequency than needed, e.g. 16/44.1 to 16/192

- Upsampling is adding bits that were not present in the original signal, e.g. 16/44.1 to 24/44.1

 

According to this definition, going from 16/44.1 to DSD or 24/176.4 is upsampling and oversampling, but I am satisfied to call it upsampling.

 

Dan (not an expert)

Is there a 192 kHz frequency at 16 bit?

[The Computer Audiophile]

Just now, Caso said:

Is there a 192 kHz frequency at 16 bit?

Technically there’s nothing stopping someone from creating a 16 bit file at any sampling rate. 

[audiobomber]

16 minutes ago, Caso said:

Is there a 192 kHz frequency at 16 bit?

Sure. As an example, Windows 10 oversamples to 16/192 in the Sound Settings menu, as well as 48, 88.2, 96, 176.4 @ 16 bits.