I use a Chord Mojo 2 paired with a Poly as my daily driver for listening at home. It’s connected to my Roon Core (a late 2014 MacBook Air) and I use the Sennheiser HD 660 as headphones attached to the Mojo 2.
I don’t use the Roon DSP and I don’t upsample anything. What Qobuz plays is what I hear with lossless Signal Path.
While browsing to the web I’ve read the following statement:
Rob watts recommended not to upsample.
All Rob Watts dacs , upsample . this is in fact where their pride and achievement is, as they (Mojo included) upsample in a unique way, with finece , in an artful way .
so any software upsampling cuts into that.
you paid extra for Mojo to do this, using a a separate software to upsample, I’d cheating yourself.
When I play a 16/44.1 qobuz flac using the Mojo 2 in the setup detailed on the top of this post the Sample Rate button on my Mojo will be red, indicating the rate of the file played.
But in regards of the statement on Watt’s
Dacs, does it mean that the Mojo 2 is anyway internally upsampling the file in order to deliver the best quality available, even if the Sample Rate button on the Mojo is red or anyway indicating the exact sample rate of the Qobuz file played in Room (in this particular example is red, but it could be blue for 192 khz files, it doesn’t matter)?
With my Chord products the colour indicates the input/output resolution, so if it’s red it is receiving and outputting 44/16 files and the same for each colour/input & output file up to DSD.
I think that your information is flawed, to my knowledge they don’t upsample they use the FPGA system to process digital signals in a way that not many other products and companies choose to; this can lead internet reviews to be a little confused.
Chord products upsample internally. Their contention is what they do is better than other software.
As far as I understand the color is for the incoming signal, not what is being processed. For example, there is a color option for DSD, but, aside from Dave, all Chord products convert the incoming DSD to PCM for processing.
Thank you all for your answers!
But then I wonder is there a way to know exactly what’s the output?
And again, I never upsample, especially with the Mojo 2, but do you guys would say that the best way to go is to play the file whatever the sample rate is and lets the mojo do its magic?
Mojo actually upsamples and filters everything to 2048 FS (eventually getting to 104 MHz) and it does this in two stages. Firstly upsamples to 16 FS via the WTA filter, then there is a second filter that then takes you to 2048 FS.
Its the upsampling to 16FS where the clever stuff happens - that’s where the 44 dsp cores are used - and this process is orders of magnitude more advanced than what a PC can do. So what we are talking here is not the need to upsample, but how well that upsampling is done. So if you want the best sound quality, let Mojo have the original file and let it do the upsampling.
But what does it actually mean?
The output is 104Mhz?
I’m not sure what FS stands for (imagine you’re talking with the average guy who doesn’t know anything about tecnical details).
Thank you so much, that was exactly what I was curious to know.
Primarily because I don’t see the need to do so and second reason is that I just bought a Nucleus (not a Nucleus +) mainly because I don’t do dsp and I have a very beginner-like library.
I’m curious to hear what more knowledgeable people have to say about this. My understanding is that before the digital signal exists, there is sound. A continuous waveform. In the real world this has basically infinite resolution. Converting this analog waveform to digital requires sampling the analog signal at regular time intervals (sample rate) with some resolution (bits per sample). The job of the DAC is to convert this digital signal back into (hopefully) the same analog wave form captured by the microphone. This cannot be done perfectly as computers don’t deal with infinity very well. So the best discrete electronics can do is reconstruct the analog waveform from the digital samples with higher resolution than what we can hear.
The best we can hear is better than individual samples at 44.1khs, so the DAC must upsample and produce a closer to infinity waveform. Some smart guy named Nyquist claimed that 16 bits-per-sample at 44.1khs sample rate provides a DAC with sufficient digital resolution to extrapolate between samples and produce an analog signal that is indistinguishable by humans from the real wave form.
Every DAC upsamples to produce this analog waveform. And the DAC upsamples more than whatever you can upsample from your PC. The question is: does the DAC upsample better than whatever external upsampler you have? And since it will upsample more, does the head start help.
Rob Watts says it does not for his products. I don’t know if that’s true or not. On my Benchmark and Topping DACs I have not heard a difference between red book and upsampled input. But my ears are old.
No quite. We can’t hear “individual samples”, they don’t even exist as sound. And the DAC does not have to upsample. Sony’s first DACs didn’t (and gave digital a bad rep), so of course now NOS DACs are very popular again.
It is and does, assuming the DAC is close enough to ideal performance.
Not every, but oversampling is a decent way to make a well-performing DAC cheaply.
Nope, another smart guy named Turing proved that anything computable can be computed on any, well, Turing machine, with enough storage. Just won’t necessarily be in real time, and DAC, with dedicated hardware can do it cheaper and “fast enough”. THeoretically,m though, you could upsample anything to anything on a Commodore 64.
I’m lost here
First of all thank you both for your posts, this is turning out in a very educative conversation (at least to me, with my immense lack of knowledge).
What I don’t get is what are you saying, @Kelly_Burkhart, when you say “it does not for his products”. What’s that is not done for Watt’s products?
And @Boris_Molodyi what are you implying with your latest comment? You mean that Watt says is disputable (of course everything could be, we all agree on that but you know what I mean )? How is it in the context of this topic?
He claims his DACs upsample better than any external process and consequently, it is unnecessary to upsample a digital signal prior to sending it to his DAC.
Something from my post got removed, but basically, Mr. Watt is saying that something at the -300dB level is audible. Which is significantly below the noise you get from brownian motion of air molecules, so I find it rather… unlikely.
Any processing that Chord’s FPGA does can be done on any general purpose computer. That’s what Turing had proven long time ago. Now, how quickly it can be done is another question, dedicated hardware, or even a custom-programmed FPGA can do some impressive things, and he’s likely using some custom algorithms that you need to know in order to implement. But theoretically any upsampling Chord does you could do with e.g. HQPlayer and a fast enough CPU and graphics card.
How much of that is really audible, let alone “better” is some meaning of the word is another can of worms. Any time some solid proof has been asked (i.e. showing better measurements or results of proper blind tests) there’s a lot of hand-waving but no actual data presented.
If you like the sound of the Mojo, that’s all that should really count. In any event when the manufacturer says that e.g. not doing upsampling is the best for their product, it’s because they, of course, think they are doing it the best way possible. You could run signal through some different processing and find results more pleasing…
Quoting just the last part but you’ve been very clear on your post, now I see what you mean and have a better picture on this matter.
I love how my Mojo 2 sounds, I tried different DACs during the past couple of months (such as the Dragonfly Red or the Hifiman EF400) but always ended up going back to the Mojo.
It sounds more natural and musical to me, meaning that I’ve the impression that’s much more true to the original source than the other two DACs. Both sound amazing but with an heavy signature applied to all music I play. I don’t feel that way with the Mojo.
And that’s without touching anything a part from playing from time to time with the crossfeed mode of the Mojo.
Mojo does have its own DSP engine and can apply quite a bit of processing to the sound, which definitely can make itr more (or less) pleasing, but probably less accurate in sensu stricto, and with what passes for a user interface with Chord I have no idea how one can know what on Earth that thing is actually doing…
Some people seem to forget that the point is to enjoy the music, not chase some (possibly ephemeral) changes in sound quality…
I’d also note that there’s a distinction between “accurate” sound and “sound one likes”. While we can derive statistical data that in blind tests people tend to prefer more accurate sound, this is just general statistical preference; it does not mean that you will prefer one to the other. And while we can measure how accurate a device is rather well (and Mojo 2 is very good at it; not that you necessarily would need all the black magic Rob Watts put in it to achieve similar accuracy, but it is very good), we can’t really measure what you will like.
And once again thank you for making this distinction more clear.
Of course at the beginning of this journey I was very intrigued by statistics and objective data but the more I listen to music, the more I realize that they do not matter that much to me.
One thing is to choose your gear being aware of what you’re getting, another one is to chase the unrealistic perfection that you will never end up catching.
One thing to note about using roon or HQPlayer’s upsampling is you have to set the output to the maximum level your dac can take in order to bypass the internal upsampling. With Chord products it varies but generally 768k. This will bypass Chord’s first stage. This is what the Chord mscaler does as well. If you set the output lower than 768k, let’s say 96k, Chord will then upsample that again. So you never want it to upsample twice. When I had the Dave, I compared the mscaler vs HQPlayer and mscaler won for sound quality and ease of use, but together with the Dave, it’s a very expensive setup.
Some dacs don’t even allow you to bypass their internal upsampler even feeding it the highest level they can take. And if they do, it’s probably best to invest in HQPlayer for much better results than roon’s upsampler.
I’m current using a Naim Uniti Atom HE and they use an off the shelf dac chip but then do their own internal custom upsampling and filtering. Many dacs just use off the shelf dac chips that perform all the steps. And that’s where Chord claims what they do is much better. And same for HQPlayer’s claims. I prefer to find a setup that just sounds great without having to mess around with external upsampling. Actually Chord claims what they do is better than anything else out there.
