Which DACs bypass digital filtering?

[Miska]

3 hours ago, ekpln said:

Hello everyone, what is your comments about Cen.Grand DSDAC 1.0 ?

 

I found this part in the manual:

 

So it should be capable of DSD Direct. Would be nice to test one out!

 

[ekpln]

1 minute ago, Miska said:

 

I found this part in the manual:

 

So it should be capable of DSD Direct. Would be nice to test one out!

 

Yeah i saw that, thank you! Btw, The name, "dsdac" seems very familiar to me 😀 i really wonder that how much their design is close to your open-sourced dsd dac design. 

[Miska]

41 minutes ago, ekpln said:

Yeah i saw that, thank you! Btw, The name, "dsdac" seems very familiar to me 😀 i really wonder that how much their design is close to your open-sourced dsd dac design. 

 

At least they talk about shift registers etc, so likely it is pretty close to similar.

[skipspence]

54 minutes ago, Miska said:

Hello everyone, what is your comments about Cen.Grand DSDAC 1.0 ?

I had this DAC on test for a week, I even wrote a small review, not in english, so in short, it's main goal as advertised is internal conversion up to DSD1024 through it's onboard FPGA, definitely not HQP philosophy, and it can do direct DSD as well.. 

Generally I wasn't very impressed with its sound, especially DSD direct, regarding it's price, would recommend Holo May instead, which is about same price but much better in SQ, and also technically. 

[copy_of_a]

On 3/5/2024 at 12:20 PM, woshifeng3627 said:

I consulted Singxer customer service for the latest version of the SDA-6 PRO2 DAC, which also uses the AKM 4191+4499EX chip. Similar to the GUSTARD A26, the difference is that Singxer customer service told me that their SDA-6 PRO2 DAC can use the DSD bypass channel on DSD512, which means that DSD64~512 can use the DSD bypass channel and support 48K x512. The reason is that they have more comprehensive development of AKM chips. Isn't it true that the AKM4191+4499EX combination chip DAC can only go through the DSD bypass channel under DSD128-256?

maybe this is the reason why the Singxer SDA-6 pro 2 doesn‘t have said limitations?

(don‘t know… just came across this article)

 

https://www.stereocheck.com/news/hifi/singxer-sda-6-pro2-balanced-dac-based-on-ak4499ex/?amp=1
its D/A conversion stage is based on two AKM AK4499EX DAC chips combined with an AK4191 digital filter. They support high-resolution streams up to PCM 32Bit 768 kHz and native DSD up to DSD512. The DSD stream operates in NOS (Non Over Sampling) mode without passing through the AK4191 filter to preserve the original signal as much as possible.

[woshifeng3627]

38 minutes ago, copy_of_a said:

也许这就是 Singxer SDA-6 pro 2 没有上述限制的原因？

（不知道...刚看到这篇文章）

 

https://www.stereocheck.com/news/hifi/singxer-sda-6-pro2-balanced-dac-based-on-ak4499ex/?amp=1 其 D/A 转换级基于两个 AKM AK4499EX DAC 芯片和一个 AK4191 数字滤波器。它们支持高达 PCM、32 位、768 kHz 的高分辨率流和高达 DSD512 的原生 DSD。DSD 流在 NOS（非过采样）模式下运行，无需通过 AK4191 滤波器，以尽可能保留原始信号。

 

I don't know how they implement native DSD512 for DA conversion, and wait for the big guys to get the machine test to know the truth, if not through AKM4191, then through what path to send DSD512 to 4499EX for DA conversion? Is the problem with the chip pin connection between AKM4191 and 4499EX? On the A26, I prefer to upsample the sound to the DSD128 output, while the DSD256 is much duller and not as open as the DSD128 sound

[Miska]

6 hours ago, copy_of_a said:

maybe this is the reason why the Singxer SDA-6 pro 2 doesn‘t have said limitations?

(don‘t know… just came across this article)

 

https://www.stereocheck.com/news/hifi/singxer-sda-6-pro2-balanced-dac-based-on-ak4499ex/?amp=1
its D/A conversion stage is based on two AKM AK4499EX DAC chips combined with an AK4191 digital filter. They support high-resolution streams up to PCM 32Bit 768 kHz and native DSD up to DSD512. The DSD stream operates in NOS (Non Over Sampling) mode without passing through the AK4191 filter to preserve the original signal as much as possible.

 

I'm happy to test one if someone sends one to me for testing. But I'm reluctant to spend yet another significant amount of money on yet another significant disappointment too...

 

And one can safely ignore what AKM says on their datasheets - it doesn't work as documented there...

 

[woshifeng3627]

6 hours ago, Miska said:

 

如果有人给我发一个进行测试，我很乐意测试一个。但我也不愿意在另一个重大失望上再花一大笔钱......

 

人们可以放心地忽略 AKM 在他们的数据表上所说的内容 - 它并不像那里记录的那样工作......

 

 

I've been waiting for your test results, and if the Singxer SDA-6 pro 2 does solve the AKM4191+4499EX native DSD64-512 D/A conversion issue, I'll know whether to switch to Cyan2 or Singxer SDA-6 pro 2! Because I don't want to repeat the comparison of spending money to buy back, haha!

Or is it possible that the 4499EX chip in later production solved the previous problem? This hypothetical estimate is unrealistic! haha

[Miska]

18 hours ago, woshifeng3627 said:

I've been waiting for your test results, and if the Singxer SDA-6 pro 2 does solve the AKM4191+4499EX native DSD64-512 D/A conversion issue, I'll know whether to switch to Cyan2 or Singxer SDA-6 pro 2! Because I don't want to repeat the comparison of spending money to buy back, haha!

 

If you want peace of mind, just go for Cyan 2 or Spring 3 L2.

 

18 hours ago, woshifeng3627 said:

Or is it possible that the 4499EX chip in later production solved the previous problem? This hypothetical estimate is unrealistic! haha

 

No, that is very unlikely to happen. Given that making any change to the silicon masks costs at least 100k EUR or more... That is joy of chip manufacturing.

 

[woshifeng3627]

3 hours ago, Miska said:

 

如果您想高枕无忧，只需选择 Cyan 2 或 Spring 3 L2。

 

 

不，这不太可能发生。鉴于对硅掩模进行任何更改至少需要花费 100k 欧元或更多......这就是芯片制造的乐趣。

 

 

If you still have A26 in your hand, you can really test whether there is a significant difference between DSD128 and DSD256, I listen to DSD128 really much better than DSD256!

[Miska]

8 hours ago, woshifeng3627 said:

If you still have A26 in your hand, you can really test whether there is a significant difference between DSD128 and DSD256, I listen to DSD128 really much better than DSD256!

 

It depends what you consider a significant difference...

 

[copy_of_a]

On 4/7/2024 at 12:04 AM, Miska said:

I'm reluctant to spend yet another significant amount of money on yet another significant disappointment too...

My post wasn't meant as a call for you to buy and measure an SDA-6 pro 2 😀
Me personally I also have no desire at all to replace my SDA-6 pro (original AK4499).
I just came across that article and thought using 2 AK4499EX chips and bypassing the AK4191 part for the DSD pipeline might be a clever idea to avoid the issues you've discovered with the AK4499EX.

 

[Miska]

9 minutes ago, copy_of_a said:

I just came across that article and thought using 2 AK4499EX chips and bypassing the AK4191 part for the DSD pipeline might be a clever idea to avoid the issues you've discovered with the AK4499EX.

 

Could be yes... But I wonder how they pass the data to the 4499EX if they bypass 4191... If it is well done, it may work very nicely. But we would only know by measuring it.

 

[woshifeng3627]

3 hours ago, Miska said:

 

可能是的......但是我想知道如果他们绕过 4499，他们如何将数据传递给 4191EX......如果做得好，它可能会很好地工作。但我们只能通过测量它来知道。

 

 

I looked back at the chat record with the customer service of Singxer, and he said that AKM4191 has two channels, one is SDM and the other is DSD directly, both of which have been tested to support DSD512 without any problem, and also supports 48KDSD512. I also tested that forced 48KDSD512 can make sound normally on A26. It's just that you said that only DSD128 and 256 are in DSD direct mode. Now they're saying they bypassed AKM4191. I wonder if his NOS mode uses the official matching circuit directly, or uses Singxer's own unique NOS mode! Create another DSD-NOS path!

 

But now I prefer the DSD128 sound, this is something I am sure of, 4499EX in processing DSD128 data for D/A conversion sound is much better, rather than the higher the rate of sound the better!

 

[bogi]

21 minutes ago, woshifeng3627 said:

Now they're saying they bypassed AKM4191.

Could you provide an exact link or citation? They could mean that they are bypassing volume control and modulator of 4191EQ in direct DSD mode. That does not mean bypassing of 4191EQ.

[woshifeng3627]

1 hour ago, bogi said:

你能提供确切的链接或引文吗？它们可能意味着它们在直接 DSD 模式下绕过了 4191EQ 的音量控制和模量 r。这并不意味着绕过 4191EQ。

 

I am through the mobile phone Taobao APP and customer service dialogue, not convenient screenshot sharing! If you have access to the Chinese mobile Taobao APP, you can also search their store directly and ask Singxer customer service! It's not that hard!

[bogi]

2 hours ago, woshifeng3627 said:

you can also search their store directly and ask Singxer customer service!

It's about your standpoint, not mine, so I asked you to paste here the exact formulation they used. It's very easy to misunderstand  formulations written in technically unclear marketing language. Wishes create myths ...

 

The device description on Magna Hifi states: "The DSD stream operates in NOS (Non Over Sampling) mode, bypassing the AK4191 filter to preserve the original signal integrity." In my understanding that sentence is not about bypassing AK4191EQ as a whole, but about bypassing a part of it, which they for simplicity named 'the AK4191 filter'.

 

I provided a citation with link.

You provided only unclear standing without exact citation and source.

 

 

[Miska]

This again sounds like they are using the DSD Direct mode of 4191... But in that case, the question remains if they did actually figure out workarounds for all the bugs in 4191 and AKM's documentation so that it actually works correctly. Gustard figured this out, while SMSL failed to figure it out (they gave up)...

 

Setting the bit in chip registers for DSD Direct and checking that audio comes out of the DAC doesn't mean that the actual functionality has been tested to work correctly... (it doesn't actually work correctly that way)

 

[leter15]

Hi I am Leter from Singxer.

 

Our DACs can all work in DSD Direct mode. Including SDA-1, SDA-2, SDA-6, and the latest SDA-6 PRO2. Among them, SDA-1, SDA-2, and SDA-6 can only work in DSD Direct mode, and they both use AKM's DAC chip. The default setting is to work in DSD Direct mode, and the volume cannot be controlled. That is to say, the DSDD BIT of the AKM chip is set to 1.

SDA-1 uses the AK4490 chip, but it still supports DSD512 playback, including 48K DSD. This may be the only AK4490 DAC that supports DSD 512. Although the DATESHEET for AK4490 does not support DSD512.

[leter15]

About SDA-6 PRO2
USB can only support up to DSD512, and currently the firmware of USB can only support DSD 44.1K. We will soon update the XMOS firmware to support DSD 48K. AKM also has two DSD channels internally, one of which actually enters the SDM modulator and works like PCM, and the other is in DSD Direct mode.
We have opened two types of DSD channels, one is normal mode (SDM) and the other is NATIVE mode, which is DSD Direct mode. They can play DSD64-DSD512 well in both modes

You can connect to SDA-6PRO2 through the I2S of SU-2, which supports DSD1024. Both support DSD 48K.

[leter15]

We played the files of DSD1024 using HQplayer in both SDA-6PRO2 DSD modes. Playing music files is normal without any lag or noise.
We use AP's FFT to confirm that AK4499EX is operating in DSD Direct mode. FFT display shows that when playing the 1K sine wave of DSD 64 files, high-frequency noise can be seen when the frequency is greater than 20K.
Under the same settings, the DSD64-DSD 1024 file can be played smoothly.

[woshifeng3627]

15 hours ago, Miska said:

这听起来像是他们正在使用 4191 的 DSD Direct 模式......但在这种情况下，问题仍然存在，他们是否真的找到了 4191 和 AKM 文档中所有错误的解决方法，以便它真正正常工作。古斯塔德想通了，而 SMSL 没能弄清楚（他们放弃了）......

 

在DSD Direct的芯片寄存器中设置位并检查音频是否从DAC中传出并不意味着实际功能已经过测试，可以正常工作...（它实际上并不能以这种方式正常工作）

 

 

So far, I only believe Miska when he says that AKM4499EX can only correctly process DSD pass-through data from AKM4191 under DSD128-256. I don't really believe the others. Science is based on practice, and only after testing can we know its actual performance! I decided to sell my A26 and related supporting equipment, replace Cyan2 to listen to music, especially now the HQ Client is more and more perfect! R2R decoding also has a real NOS, is indeed the best HQ partner.

[leter15]

16 minutes ago, leter15 said:

s that when playing the

Regarding MISKA's statement, setting registers can play normally, but will AKM's internal registers automatically change?
I haven't verified this situation yet, but I think it's impossible.
Actually, I tested it and found that AK4191 works in DSD 1024 Direct mode, and its BCLK becomes 11.2M. Usually, its BCLK is 5.6M.

[Miska]

1 hour ago, leter15 said:

We use AP's FFT to confirm that AK4499EX is operating in DSD Direct mode. FFT display shows that when playing the 1K sine wave of DSD 64 files, high-frequency noise can be seen when the frequency is greater than 20K.

 

My standard test is to to play 0 - 22.05 kHz linear sweep, with spectrum analyzer in 5 MHz bandwidth, peak-hold mode, 1M point FFT.

 

[leter15]

Hi Miska

Thank you for your reply. Your testing method is very good.

 

Regarding the issue of AK4191+AK4499EX transmission bandwidth, I suspect that you are only based on DATESHEET's conjecture and have not conducted actual measurements.
In fact, during transmission, a clock can transmit 2 bits of data.

We know that when transmitting DSD1024, a bandwidth of approximately 100Mbit (90.3168M/98.304M) is required.
So when BCLK=11.2M, it has a total of 7 data channels, and at this time, its actual bandwidth is 156.8M bits.
This bandwidth can fully meet the transmission requirements of DSD1024.