Review and Measurements of Schiit Yggdrasil V2 DAC

[gvl]

No, I want to see if your 0.1dB marker slides when using dithered signal for the linearity test.

 

[DonH56]

Since dither is generated within the DAC, not provided by the source, I am having a hard time seeing the relevance of this test?

I'd rather Amir spent his time testing something, anything, else...

 

[gvl]

I want to better understand the test data AP sends to the DAC during the linearity test, does it always set a single bit to 1 and others to 0, or it uses values, perhaps random, that can result in multiple bits set to 1 in individual samples?

 

[Wombat]

I want to better understand the test data AP sends to the DAC during the linearity test, does it always set a single bit to 1 and others to 0, or it uses values, perhaps random, that can result in multiple bits set to 1 in individual samples?

Ask AP https://www.ap.com/contact-us/technical-support/ or browse their on-site info.

https://www.ap.com/technical-librar...ibrary_category=application-notes&load_page=1

 

[amirm]

I want to better understand the test data AP sends to the DAC during the linearity test, does it always set a single bit to 1 and others to 0, or it uses values, perhaps random, that can result in multiple bits set to 1 in individual samples?

The "data" sent to the DAC is a simple sine wave. Linearity test keeps changing its level and measures the output of the DAC. At all times the DAC sees a sine wave.

 

[amirm]

Since dither is generated within the DAC, not provided by the source, I am having a hard time seeing the relevance of this test?

I'd rather Amir spent his time testing something, anything, else...

I have recently learned that purgatory for audio is having to measure Schiit audio gear over, over and over again...

 

[Blumlein 88]

I have recently learned that purgatory for audio is having to measure Schiit audio gear over, over and over again...

Purgatory or maybe a Sisyphean task?

 

[derp1n]

Pushing Schiit up hill?

 

[Blumlein 88]

Pushing Schiit up hill?

I think with a little photo-shopping we could turn that into a new Avatar.

 

[bennetng]

Dither will only improve performance if input bit-depth is higher than output bit depth. That means a 20-bit device can playback any bit-depth equals to or less than 20 bits without causing any truncation errors.

As long as there is no additional DSP (e.g. volume control, EQ etc) between a 16-bit file source and the input of a DAC, dithering will not make any improvement other than adding a layer of useless noise.

 

[Thomas savage]

I have recently learned that purgatory for audio is having to measure Schiit audio gear over, over and over again...

 

[Thomas savage]

Pushing Schiit up hill?

There’s something magic, heroic even about the dung beatle , I can’t help but think nature played a cruel joke on the little blighter though, poor bastard ...

Dung is fairly amazing, it can be used for all sorts of things and is vastly superior to schiit.

 

[The Dragon]

???

Here is the comparison of 16 bits versus 24 bits:

View attachment 13705

The noise floor jumps up by 30 dB when using 16 bit samples. Is this what you were asking about?

I would expect the same sort of thing when playing 24 bit files through the Modi Multibit.

 

[Deleted member 2348]

Dither will only improve performance if input bit-depth is higher than output bit depth. That means a 20-bit device can playback any bit-depth equals to or less than 20 bits without causing any truncation errors.

As long as there is no additional DSP (e.g. volume control, EQ etc) between a 16-bit file source and the input of a DAC, dithering will not make any improvement other than adding a layer of useless noise.

I don't think that this is 100% correct. Any upsampling and filtering ("oversampling") before the DAC will increase the number of bits because it involves (digital) multiplications. For example, if you multiply 16 bit data with 10 bit filter coefficients you will get 26 bit data. Of course this data will not contain more than 16 bit "real" information, but if you truncate without dither you will get low-level artefacts from the digital filtering (e.g. aliasing components). Any truncation will create distortion which is correlated with the sample frequency.

Edit: Schiit claims to leave the original sample values untouched with their super-burrito-whatever-filter. I wouldn't be surprised if this explained all these low level problems of their multi-bit DACs.

 

[Wombat]

There’s something magic, heroic even about the dung beatle , I can’t help but think nature played a cruel joke on the little blighter though, poor bastard ...

Dung is fairly amazing, it can be used for all sorts of things and is vastly superior to schiit.

It reminds me of the perpetual audio upgrade process - never ending. Like in audio magazines they are selective re the shit they push.

https://repository.up.ac.za/bitstream/handle/2263/30394/07chapter7.pdf?sequence=8

 

[mindbomb]

I would expect the same sort of thing when playing 24 bit files through the Modi Multibit.

The bifrost multibit and modi multibit are 16 bit dacs, so the noise wouldn't lower with 24 bit files.

 

[bennetng]

As long as there is no additional DSP

Any upsampling and filtering

DSP includes upsampling and filtering.

 

[gvl]

The "data" sent to the DAC is a simple sine wave. Linearity test keeps changing its level and measures the output of the DAC. At all times the DAC sees a sine wave.

Don't you think that measuring low level linearity of a technically 20-bit DAC with 24-bit input data, especially undithered, is fundamentally wrong?

 

[Blumlein 88]

Don't you think that measuring low level linearity of a technically 20-bit DAC with 24-bit input data, especially undithered, is fundamentally wrong?

Of course not. The DAC could tell owners it only accepts 20 bit or less inputs or quality degrades. It could do the proper thing and dither 24 bit input to 20 bit internally.

20 bit inputs aren't common on any formats. So Schiit does say this is optimized for 16 bit CD sources. You could do a series of 20 bit tests, but why? You don't have any such sources unless you use some playback software that can do that for you. I suppose you could dither in that software as well if it does that. Or you could convert all your digital files to 20 bit with dither and hold them separately on your hard drive.

 

[amirm]

Don't you think that measuring low level linearity of a technically 20-bit DAC with 24-bit input data, especially undithered, is fundamentally wrong?

What do you mean undithered? All signals I send to it is dithered.

That aside, my linearity measurements *stop* at 20 bits/-120 dBFS. If it is true 20 bits, it needs to nail that. It is Jude/Atomicbob who measure to -140 dB/24 bits.

I am not going to send it 20 bits as that is not a music format.