Which DACs bypass digital filtering?

[Miska]

1 hour ago, woshifeng3627 said:

Now my question is, since there is a problem with the 4499EX chip, why is there not uniform performance in DSD direct mode? At the same time, why is it possible to play 48x DSD and DSD512 when turning off DSD direct mode? This design is a bit contradictory. AKM engineers probably have a short circuit in their minds. Providing such an immature DA converter is completely incompatible with AKM4191. At least in terms of specifications, AKM4191 is 64 bit and supports PCM1.5M and DSD1024, but its performance is halved in DSD direct mode! I originally thought that AKM's new combination chip could support DSD1024 like HOLO's R2R, but the result was a bit funny!

 

In non-direct mode AK4191 converts DSD to it's internal presentation and converts the sample rate to the one AK4499EX is set to run (either 5.6M or 11.3M). This way it can do DSD1024, etc.

 

In direct mode this cannot happen, thus only DSD128 or DSD256 can be played in direct mode.

 

[woshifeng3627]

19 minutes ago, bogi said:

如果基于 48k 的 DSD 速率可用，则直接 DSD 对于基于 44.1k 和 48k 的 DSD 速率应具有相同的行为。
关于直接 DSD 路径功能，基于 48k 的 DSD 速率仅带来 44.1k 基于速率。

自适应速率复选框旨在通过避免跨系列转换（44.1->48 或 48->44.1）来降低 CPU/GPU 负载。

 

I just retested the DSD direct mode of A26. DSD256 and below can only play 44.1xDSD, DSD512 can only play 48xDSD, and 48x256 cannot play!

[woshifeng3627]

2 minutes ago, Miska said:

 

在非直接模式下，AK4191 将 DSD 转换为其内部演示，并将采样率转换为AK4499EX设置为运行的采样率（5.6M 或 11.3M）。这样它就可以做DSD1024等。

 

在直接模式下，这不会发生，因此在直接模式下只能播放 DSD128 或 DSD256。

 

 

I just retested the DSD direct mode of A26. DSD256 and below can only play 44.1xDSD, DSD512 can only play 48xDSD, and 48x256 cannot play!

[Miska]

9 minutes ago, woshifeng3627 said:

I just retested the DSD direct mode of A26. DSD256 and below can only play 44.1xDSD, DSD512 can only play 48xDSD, and 48x256 cannot play!

 

DSD512 inputs switch to some kind of non-direct mode. 48k-base DSD is not supported there.

 

[bogi]

9 minutes ago, woshifeng3627 said:

DSD512 can only play 48xDSD

Your observation does not look as correct. Set lighter filter/modulator, set adaptive rate, play 44.1k based source content and you will see that 44.1k based DSD512 will most probably play without issues, although not in direct DSD mode.

According to Miska no direct DSD is possible at DSD512.

[woshifeng3627]

2 minutes ago, Miska said:

 

DSD512 输入切换到某种非直接模式。那里不支持基于 48k 的 DSD。

 

 

As you said, I don't know how SDA-6 PRO2 implements DSD512 and 48KDSD to run in DSD direct mode, because its DSD only has one mode, NOS, and I don't have a machine, so I don't understand.

[woshifeng3627]

4 minutes ago, bogi said:

Your observation does not look as correct. Set lighter filter/modulator, set adaptive rate, play 44.1k based source content and you will see that 44.1k based DSD512 will most probably play without issues, although not in direct DSD mode.

According to Miska no direct DSD is possible at DSD512.

 

After enabling DSD direct mode in A26, for DSD256 below, 48xDSD256 cannot be selected, and can only be played normally under 44.1xDSD256. For DSD512 audio above DSD256, only 48XDSD512 can be selected, and 44.1xDSD512 cannot be selected. HQ cannot choose whether to enable 44.1 or 48 through adaptation!

[woshifeng3627]

7 minutes ago, Miska said:

 

DSD512 inputs switch to some kind of non-direct mode. 48k-base DSD is not supported there.

 

Isn't DSD only available in two modes, either direct or non direct? After enabling DSD direct, other paths can still be taken? It seems that there is only one direct path for AKM4191.

[Miska]

13 minutes ago, woshifeng3627 said:

Isn't DSD only available in two modes, either direct or non direct? After enabling DSD direct, other paths can still be taken? It seems that there is only one direct path for AKM4191.

 

AK4191 supports direct from DSD64 to DSD512. But AK4499EX only at DSD128 or DSD256. With DSD512 input the result is kind of non-direct path.

 

If you want to use A26 correctly in direct mode, set filter to "wide" and always use 44.1k x256 (or 44.1k x128) fixed output rate. Anything else will give you something strange.

 

[woshifeng3627]

47 minutes ago, Miska said:

 

AK4191 支持直接从 DSD64 到 DSD512。但AK4499EX仅在 DSD128 或 DSD256 上。使用 DSD512 输入时，结果是非直接路径。

 

如果要在直接模式下正确使用 A26，请将过滤器设置为“宽”，并始终使用 44.1k x256（或 44.1k x128）固定输出速率。其他任何事情都会给你带来一些奇怪的东西。

 

 

Okay, I don't plan on playing the AKM4191+4499EX combination DAC anymore. I plan to switch to the holo Cyan2 decoder to see if I can get rid of this A26 device this year! I hope Cyan2 doesn't disappoint me. So if Cyan2 upsamples to DSD256, DSD512, and DSD1024, from the measurement results, which DSD rate has better sound quality?

[copy_of_a]

5 hours ago, bogi said:

No, usual terminology 5-10 years ago was native/DoP about DAC USB input and "direct DSD path" as a term used in many DAC datasheets for a path bypassing oversampling, volume control and modulator.

You are right!

I forgot the term "direct DSD" is used to describe chip/DAC behavior (volume bypass etc.) as well as it is colloquially used to describe the connection path from computer to DAC.

I use the iFi audio app quite a lot and my connection path is called "direct DSD", so this is what I am stupidly used to in the meantime.

Apologies!

[Miska]

3 hours ago, woshifeng3627 said:

from the measurement results, which DSD rate has better sound quality?

 

DSD256 or DSD512, depending on how you weight things.

 

[cpcat]

 @Miska have you ever measured a Lampizator DAC?

 

They are pretty secretive about what chips they use, only saying “ modern sigma delta “ or something to that effect.  Mine is designated “digital engine 53” and their most current is now “digital engine 11”.   Cryptic really, almost like there is a railroad inside there. 🤪

[Miska]

3 hours ago, cpcat said:

 @Miska have you ever measured a Lampizator DAC?

 

No... A bit too expensive to buy just for measuring. 😅

 

If someone ever happens to be around capital area in FInland, I can arrange measuring it. Or alternatively when I'm traveling somewhere around Europe, I can take my portable measurement kit with me.

 

[woshifeng3627]

2 hours ago, Miska said:

 

不。。。有点太贵了，不能只买来测量。😅

 

如果有人碰巧在芬兰的首都地区附近，我可以安排测量它。或者，当我在欧洲的某个地方旅行时，我可以随身携带便携式测量套件。

 

 

If you travel to China, I believe you can measure many devices! Haha

[dericchan1]

Hi folks, the Holo Cyan 2 should make it on both the pcm and dsd list.

 

A question @Miska, is the Cayin RU7 not a discrete 1bit dsd dac that bypass digital filtering?

[sunjam]

Sorry for my ignorance, is DAC based on CS43131 (that supports NOS for PCM) be considered as "DACs bypass digitial filtering"?  If yes, it could be add to the PCM list too.

[bogi]

Cyan 2: +1

 

RU7: According to Miska there is some processing done on input data making it not bit perfect.

https://audiophilestyle.com/forums/topic/31941-which-dacs-bypass-digital-filtering/?do=findComment&comment=1265284

 

[cpcat]

On 12/27/2023 at 10:47 AM, Miska said:

 

But seems to be never bit-perfect... So I don't think it should be listed here

Is HQP conversion to DSD bit perfect ?

Sorry if it seems like a stupid question ., just curious from an academic standpoint .  I would assume HQP could convert to DSD then back to the original file if needed- not that you would necessarily want to. 😀

[bogi]

1 hour ago, cpcat said:

Is HQP conversion to DSD bit perfect ?

This topic is about DACs doing direct D/A conversion without digital resampling or additional DSP which could change audio content in digital domain against what was input into DAC.

Processing in HQPlayer is of course not bit perfect, because bit perfect would mean the opposite - no digital processing. Roles of HQPlayer and direct DSD capable DAC are intentionally split: HQPlayer doing only digital signal processing and DAC doing the conversion. The base idea is that computers provide much more suitable environment (resolution, instruction set efficiency, processing speed) for mathematical operations what DSP is.

 

I suggest not to continue here about software DSP. Better to stay on topic. There are many threads from the previous years on this forum which discussed the bit perfect topic in relation to software. Just use one of them or create new one which would fit your intention what to discuss.

To your bare question: with PCM -> DSD -> PCM you never get the original PCM file, so it is never bit perfect. That was discussed in other threads already.

[cpcat]

Well, sorry for that!

[Miska]

11 hours ago, cpcat said:

Sorry if it seems like a stupid question ., just curious from an academic standpoint .  I would assume HQP could convert to DSD then back to the original file if needed- not that you would necessarily want to. 😀

 

Yes, as academic experiment, it is possible to start for example with 192/24 source file, convert it to DSD512 and back to the original 192/24 data.

 

[Miska]

20 hours ago, dericchan1 said:

A question @Miska, is the Cayin RU7 not a discrete 1bit dsd dac that bypass digital filtering?

 

It always runs everything through some PCM-DSD or DSD-DSD conversion. So it is never bit-perfect. (and RU6 seems to always do some processing on PCM, so that's not bit-perfect either)

 

[ekpln]

Hello everyone, what is your comments about Cen.Grand DSDAC 1.0 ? As I see it is not a NOS dac, but I wonder about the comments especially with the HQPlayer DSD upsampled audio. 

 

Sir @Miska , do you recommend this DAC?

 

Thanks!

[Miska]

37 minutes ago, ekpln said:

Hello everyone, what is your comments about Cen.Grand DSDAC 1.0 ? As I see it is not a NOS dac, but I wonder about the comments especially with the HQPlayer DSD upsampled audio. 

 

Sir @Miska , do you recommend this DAC?

 

I have not tested it myself, so I'm not able give any particular recommendations about it. I've seen a video review where someone mentioned that it has optional DSD Direct support. But I don't have any firsthand information about this.

 

One could at least ask the manufacturer.