[Sonic Atrocity]

   
 
Well its good to know that info. thanx

Does this have an affect on playback?

 

[saratoga]

  So, in short, does rockbox on the dx50 deliver a neutral response or...? 
 
I'm sorry. I'm still learning about how all this stuff works.

 
Aside from additional features, its no different than the stock firmware.  

 

[ratinox]

Second, the post you're quoting is misunderstanding the spec sheet.  When one uses an oversampling DAC, the analog low pass filter is set at maybe 1/4th the oversampled frequency, not the base frequency.  So at 44.1k and 128x oversampling, the analog filter cuts on at about 1-2MHz.  The digital filter handles lower frequencies, and that is adjusted to match the sampling rate.  In rockbox we actually do support higher sampling rates when possible, we just restrict the playback engine to 44.1k or 48k for efficiency reasons and because higher sampling rates are generally useless given how modern oversampling devices work.

First, I was misunderstanding nothing. Simplifying, perhaps, but not misunderstanding.

Second, I was describing the DAC in DX50 as an adjunct to how the player operates because regardless of what Rockbox and Mango do the Wolfson DAC will cut anything above 20 kHz assuming iBasso configured it to spec.

Third, nobody oversamples 128 times base sample rate. Two times 48 kHz and four times 48 kHz get you -- do the math while I type -- 96 kHz and 192 kHz. Why multiples of 48 kHz? Because that's the base sample rate that professional gear in the early days of digital recording (DAT for example) and multiples of that are really easy to work with.

Fourth, 1 MHz? That's 1000 Hz. That's AM radio, not audio.

Anyway, to answer the questions:

No, oversampling does not exaggerate anything. It's math. If you double the number of samples per second (2x oversampling) and then duplicate each sample then you get the same curve and thus the same analog signal out of the converter. The benefit to oversampling is that it increases the analog frequencies where aliasing artifacts appear during digital to analog conversion. If you oversample high enough then you can push those artifacts above the threshold of human hearing and cut them off with a low-pass filter so that they don't introduce unwanted distortion or damage the equipment. It's much easier and less costly in terms of power and components doing it this way than using analog anti-aliasing filters.

The output frequency range of the DX50 hardware is 20 Hz to 20 kHz regardless of the player software (with the same caveat above, that iBasso implemented the DAC chip to Wolfson's spec).

Rockbox is, as I've repeated written here, transparent. It does nothing to alter or modify what you hear unless you configure it to do so.

DX50 may be transparent. It probably is but I can't be certain until someone runs it through a thorough test and demonstrates that it is and I see the results.

 

[jj69]

  See, this makes me suspicious of rockbox, and makes me wonder if you're getting more detail, or just a boosted response in the higher frequencies. 

 
It is definitely NOT a result of boosted high frequencies.  If you had to live with the DX50 from its initial release, having its sound change completely with each new firmware update, with none of them ever sounding quite right, then you would understand.  The stock firmware (all of them) was a mess sonically, and never sounded much better than a Clip+
 
With Rockbox, all the hidden potential of the DX50 is finally unleashed.  As I said previously, if you haven't heard your DX50 with Rockbox, then you've never really heard the DX50. 

 

[saratoga]

Second, I was describing the DAC in DX50 as an adjunct to how the player operates because regardless of what Rockbox and Mango do the Wolfson DAC will cut anything above 20 kHz assuming iBasso configured it to spec.

 
 
You're misunderstanding the spec sheet.  The digital cutoff is 0.45*fs, while the analog has a 3dB cutoff at 195 kHz.  So at 44.1k, the cutoff is min(0.45*fs,195khz) = 19.85khz.  At 192khz it is 86.4kHz.
 
This is how oversampling DACs work.  The cutoff is defined in terms of the non-oversampled rate.  It cannot be fixed like you were thinking.
 

Third, nobody oversamples 128 times base sample rate. 

 

Maybe you are not aware with the term oversampling DAC.  Basically it is a DAC that oversamples the incoming sampling rate a few hundred times then uses a relatively low bit DAC running in the MHz.  Noise shaping is then used to shift the quantization error out of band.  In this case, the DAC actually runs at 128*fs, so 5.6 MHz for a 44.1k input.

FWIW, wikipedia has a great article explaining the math.
 
 

Fourth, 1 MHz? That's 1000 Hz. That's AM radio, not audio.

 


MHz = 1,000,000 Hz.  KHz would be 1,000 Hz.
 
 
 

No, oversampling does not exaggerate anything. It's math. If you double the number of samples per second (2x oversampling) and then duplicate each sample then you get the same curve and thus the same analog signal out of the converter. 

 



 
I guess it doesn't really matter, but this is not how oversampling works.  Samples are not duplicated, rather they are zero stuffed.  So to oversample 2x you would do s1, 0, s2, 0, ...  4x would be s1,0,0,0, s2, ....
 
You can do it the way you're thinking, but if you look at the math it rolls off higher frequencies pretty badly, so in practice you would not want to do that.  But yes, you are right, besides greatly reduces quantization error, oversampling also protects against aliasing, which is a very nice advantage.  
 
 
 

Rockbox is, as I've repeated written here, transparent. It does nothing to alter or modify what you hear unless you configure it to do so. 

 



Absolutely agree 100%.  This is one of our design goals.  Everything that can alter the sound can be disabled, and is generally off by default except for replaygain.

 

[cooperpwc]

  It is definitely NOT a result of boosted high frequencies.  If you had to live with the DX50 from its initial release, having its sound change completely with each new firmware update, with none of them ever sounding quite right, then you would understand.  The stock firmware (all of them) was a mess sonically, and never sounded much better than a Clip+
 
With Rockbox, all the hidden potential of the DX50 is finally unleashed.  As I said previously, if you haven't heard your DX50 with Rockbox, then you've never really heard the DX50.

 
I agree.

 

[ratinox]

Maybe you are not aware with the term oversampling DAC.  Basically it is a DAC that oversamples the incoming sampling rate a few hundred times then uses a relatively low bit DAC running in the MHz.  Noise shaping is then used to shift the quantization error out of band.  In this case, the DAC actually runs at 128*fs, so 5.6 MHz for a 44.1k input.

Okay. I see where you're coming from. You're working from 1-bit (delta-sigma?) digital to analog converters where the operating frequency of the DAC chip is relevant. Which covers everything (or nearly so) that Wolfson makes, and Cirrus Logic, and ESS, and everyone else.

Thanks for catching me on the kHz vs. MHz screwup. That was utterly bone-headed of me.

 

[saratoga]

Okay. I see where you're coming from. You're working from 1-bit (delta-sigma?) digital to analog converters where the operating frequency of the DAC chip is relevant. 

 
Sort of.  They're not one bit anymore, although back in the 1990s those did exist.  Now a days I think 4-5 bits is most common.  So the core DAC has maybe 24-30 dB SNR.  Then throw in 128x oversampling and 2nd order noise shaping, which is 15 dB per octave, and you get 24-30+log2(128)*15 = 129 to 135 dB SNR, which is below the thermal noise limit.  
 
In practice, its very hard to make a portable player with more than about 100 dB SNR, just because everything is crammed so close together and there is very little room for components like capacitors which are needed for good filtering.  Hence most only do about 16-17 bits before their theoretical resolution is swamped by some other factor.  

 

[headwhacker]

 
Aside from additional features, its no different than the stock firmware.  

 


I don't if RB use parts of the stock FW or completely takes over digital playback and feed the DAC. But, from my subjective impression the stock firmware sounds slightly different than RB at default settings (all at 0dB). The obvious being the bass. The stock FW seems to have a bit of elevated low end compared to RB.

 

[GasMaskMan]

Does rockboxing the dx50 break warranty? 

 

[Sonic Atrocity]

 
 
It is definitely NOT a result of boosted high frequencies.  If you had to live with the DX50 from its initial release, having its sound change completely with each new firmware update, with none of them ever sounding quite right, then you would understand.  The stock firmware (all of them) was a mess sonically, and never sounded much better than a Clip+
 
With Rockbox, all the hidden potential of the DX50 is finally unleashed.  As I said previously, if you haven't heard your DX50 with Rockbox, then you've never really heard the DX50. 

Ooops, sorry. I just realized my post came off as bit terse. I stand corrected.  
 
I'm really interested in this player. Really hoping to see someone do a full review of it soon.
 
edit: by "full review" I mean with technical tests and stuff

 

[kingdixon]

Does this have an affect on playback?

No i dont think so, this is only related to the clock.

 

[mrbunch]

I've mod the WPS of AluminArt and CenterArt,
see if they work for you
(I don't have the skill to mod linux kernel, so I mod the theme to show 99% as 90% etc.)
AluminArt: https://dl.dropboxusercontent.com/u/17013940/AluminArt_Edit.7z
CenterArt: https://dl.dropboxusercontent.com/u/17013940/CenterArt_Edit.7z

 

[modulor]

  I've mod the WPS of AluminArt and CenterArt,
see if they work for you
(I don't have the skill to mod linux kernel, so I mod the theme to show 99% as 90% etc.)
AluminArt: https://dl.dropboxusercontent.com/u/17013940/AluminArt_Edit.7z
CenterArt: https://dl.dropboxusercontent.com/u/17013940/CenterArt_Edit.7z

Nice! I'll test them out when I have some time later today and report back 

 

[mrbunch]

Nice! I'll test them out when I have some time later today and report back 

If you had fine battery reading, you'd better stay with original themes.
This mod is simply matching
99 with 90
98 with 80
...
91 with 10
90 with 0
As I've exactly the same problem as kingdixon.
Or you can check the battery level for full drain, then use notepad++ to open wps and change the battery part