New Audio-gd R-7, R-7HE R-8, R-27, R-27HE, R-28 Flagship Resistor Ladder DACs and DAC/amps

[DACLadder]

I wrote a sound report to Kingwa about the TDA Asy firmware and sound differences with HDMI I2S source input using specifically the Gustard U16 and SIngxer SU-1. His thinking, if interpreted correctly, is the TDA Asy firmware is more transparent and differences in sources easier to recognize. Kingwa suspects the U16 has lower phase noise than the SU-1 which he believes affects sound more than jitter. Clock phase noise as a whole is also affected by circuit design, power supply, and PCB layout. I would go on to say FPGA firmware implementation as well.

In Kingwa's Email reply I corrected some grammatical phrasing, spelling, and added notes () to make it more readable. But Kingwa's English has improved dramatically!

"I got a lot of good feedback on the tda version firmwares.
I think it (TDA asynchronous f/w) is more transparent causing the input sources (U16 vs. SU-1) to show more difference in sound quality.
If the high level (U16) and low level (SU-1) source sound just the same, I think the firmware fails (not transparent).
The source clock phase noise has more effect on sound quality than jitter.
Besides the oscillators, clock phase noise is also dependent on the power supply and the layout design, but it is more difficult to reduce in design alone (starts with good oscillators).

I think the U16 must have lower clock phase noise than SU-1.

Kingwa"

 

[borrego]

I am confused by Kingwa's comments.

Isn't the purpose of the Asyn firmware MCLK reconstruction feature is to make the DAC more immune to less than ideal input signal ? (i.e. the SU-1's higher phase noise MCLK signal)

By right, the Asyn firmware will be a success provided it can bring the SU-1 input closer to the quality of the U16 input quality, and maintain the same (if not improve) the quality of the U16 input comparing to the V3A firmware.

If the Asyn firmware makes the relatively bad (SU-1 input) sounds worse, and the good (U16 input) sounds better, it should be considered as half success.

Or is it the Asyn firmware makes the relatively bad (SU-1 input) sounds no worse, and the good (U16 input) sounds better?

 

[DACLadder]

@borrego Has me scratching my head... Doesn't make much sense either comparing to the info in the TDA firmware/ DAC project write-up. But as @FredA reported a few days ago Kingwa has been doing something similar since the M7 days and presume all other inputs besides I2S are already asynchronous.

I think on this trial TDA firmware Kingwa just discovered the I2S PLL does more harm than good. Just getting rid of that may make I2S better sound capable. Why the big sound difference I don't understand at this point? Used the SU-1 yesterday and U16 today. Night and day difference with TDA Asy firmware. Even more so versus V2 and V3. The PLL on I2S started with the Singularity board design (Hi Exact mode, etc).

@JaMo What input/ source are you using with TDA Asy firmware? Thank you!

 

[JaMo]

I am confused by Kingwa's comments.

-Don't be. We are talking about taking things further.... Nothing is less good than earlier. This is good. With an excellent oscillation (XO like the Accusilicon) You can reclock and cut loose from MLCK and get a much better result. (MLCK can pick up noise/interference) It has to be done with a thought through software. Rockna is doing this and now Kingwa have found the track..

@DACLadder: I use the U16 with the 1.4 PSA I2S fw. I only use HDMI/I2S. Excellent sound from linux based PiCorePlayer (Pi3+) I use LMS Nightly Build onmy NAS'es that has Native DSD support and much, much more.

I think we all can help Kingwa with feedback now. I want to advice us all not to tweak things right now. Stay where You are now (sound systemvise) and try to describe to Kingwa what You hear and what You think can be better and how... I think we all are providing to make history here in some sense...

/Jan

 

[FredA]

-Don't be. We are talking about taking thing further.... Nothing is less good than earlier. This is good. With an excellent oscillation (XO like the Accusilicon) You can reclock and cut loose from MLCK and get a much better result. (MLCK can pick up noise/interference) It has to be done with a thought through software. Rockna is doing this and now Kingwa have found the track..

@DACLadder: I use the U16 with the 1.4 PAS I2S fw. I only use HDMI/I2S. Excellent sound from linux based PiCorePlayer (Pi3+) I use LMS Nightly Build onmy NAS'es that has Native DSD support and much, much more.

I think we all can help Kingwa with feedback now. I want to advice us all not to tweak things right now. Stay where You are now (sound systemvise) and try to describe to Kingwa what You hear and what You think can be better and how... I think we all are providing to make history here in some sense...

/Jan

I think you are right, Jan. As Kingwa is perfecting his knowledge of English, he will become more knowledgeable and pick up design elements he did not think of here and there. 3 month ago he would not want to try the Accusilicons, now he has realized they are clearly better and improve the sound. So my feeling is that the r7 will improve much within the next months.

 

[Toni-Mang]

The clock bundle is on the way, and i got both tda firmwares. The i2s board for input 5 was delivered with a 7 pin connector, but on the R8 MB it is 6 pin connection, for input 4 and 5.
I compared the i2s on input 4 (gustard u16) with the isolated usb amareno on input 5...i seems to sound better, but not clearly imho....yes, more bass, but better in kind of three dimensionality?...i am still unsure.And i still dont understand, how the i2s (mckl) input will sound better with better clocks on the board of the r7/8...so pll should be set to on? What is the benefit of the u16 than....i'm lost...

 

[DACLadder]

@Toni-Mang In the two TDA firmwares Kingwa may have borrowed DSP 'pll' jumper (S3) for his 16-bit TDA DAC chip simulation mode. I don't see any write-up regarding DSP pll mode in the TDA description. Also the HDMI and USB synchronizer functions (new jumpers S1 and S2) only work with the Syn version. The I2S synchronizers have no function with Asy version firmware.

 

[FredA]

The clock bundle is on the way, and i got both tda firmwares. The i2s board for input 5 was delivered with a 7 pin connector, but on the R8 MB it is 6 pin connection, for input 4 and 5.
I compared the i2s on input 4 (gustard u16) with the isolated usb amareno on input 5...i seems to sound better, but not clearly imho....yes, more bass, but better in kind of three dimensionality?...i am still unsure.And i still dont understand, how the i2s (mckl) input will sound better with better clocks on the board of the r7/8...so pll should be set to on? What is the benefit of the u16 than....i'm lost...

Have your burned the u16 for 150 hours? I should assume you did. It also depends on the hdmi cable you use. The f-1 was slighlty better the amanero in my setup. The u16 is clearly better.

It depends also on how reveiling you equipement is.

By the way, with a usb cable swap, I got rid of the playback issues with 188 and 192k files. I still have an issue with having to reboot my computer (usbridge running volumio) upon skipping once in a while. The playback will stop and not resume even if i skip.

Using the new tda asynch fw, things could get more stable as the world clock is no used.

 

[Toni-Mang]

Not really burned in yet (aprx. 100h), In and out is aq coffee. Do you think this will increase the soundstage significantly if the burn in process will go on?
I wanted to get a second I2S Input on Input 5 (swap the USB board out and the second I2S Card in) for the U16.
But isnt the MCKL over I2S critical? Is there a benefit for the I2S ports with the new Clocks? Talking about the R8...in my understanding, sync fw, I2S and the source Clock is critical…
so no Benefit from the new Accusilicon Clocks. And only if it is async, the new Clocks are critical...
Or do i missunderstand the functionality?

 

[DACLadder]

Kingwa and I exchanged Emails yesterday. He is concerned forum users are confused on Asy firmware and the I2S MCLK. So I will try to expand on my comments (and avoid further confusion). Source generated MCLK is traditionally used by DACs as it makes DAC designs much easier (and cheaper). MCLK is synchronous with the other I2S clocks (BCLK, WCLK) so really no synchronizing of the received data is necessary. But, as Kingwa pointed out, using the source generated MCLK has its perils as it is not as pristine or pure as locally generated clocks. Makes sense as source MCLK is sent over a cable and through data transmitters/ receivers, separate power supplies, etc. Jitter and phase noise add up as it is transferred from the source to the DAC.

Asy firmware eliminates the need for the source MCLK. Source I2S BCLK and WCLK are only used to move the source data into the FPGA and not used again. The received I2S source data is then reclocked using locally derived clocks of better quality (from the two onboard DAC oscillators). OK, that makes sense. The mechanics of Kingwa's resynchronizing remains proprietary though.

At this time Kingwa doesn't really fully understand why using the Gustard U16 I (we) get better perceived better sound quality versus the SIngxer SU-1. The resynchronization of data should minimize this. So he wants me to back up and test the SU-1 against V3 and Asy firmware. Does Asy improve sound against V3 when using the SU-1? How much, etc? Will be working on this today..

 

[FredA]

Not really burned in yet (aprx. 100h), In and out is aq coffee. Do you think this will increase the soundstage significantly if the burn in process will go on?
I wanted to get a second I2S Input on Input 5 (swap the USB board out and the second I2S Card in) for the U16.
But isnt the MCKL over I2S critical? Is there a benefit for the I2S ports with the new Clocks? Talking about the R8...in my understanding, sync fw, I2S and the source Clock is critical…
so no Benefit from the new Accusilicon Clocks. And only if it is async, the new Clocks are critical...
Or do i missunderstand the functionality?

The burn in improves pretty much everything. See my older posts on the u16 thread.

MCKL is needed indeed for the synch fw.

The accusilicons should show benefits with both fw. The asynch should be the best one in the end. Depending on the setup and preferences. Improving XOs is alway beneficial to sound in my experience.

 

[lvc10000]

On a different note, I've been trying out the new TDA Asyn software on my M7S, coming from V3 NOS3 and it is quite something I didn't expect.
I've been trying the various configurations but the 4X Oversampling just hits the spot in my case. It may sound silly but I can't exactly put my finger on it because I feel its just "right". Difficult to describe. There is a righteousness to it in comparison to all others. I'm not sure if it may also be a question of speed/establishment/immediacy. Anywho, I went back to check the others again but that "righteousness" calls me back to 4X oversampling (S7 jumper only). I thought it'd be worth sharing in case some others can get the same enjoyment I am.

I'm running my system differently from others it seems, upstream I'm coming in via a MicroRendu (with LPS) via USB into my M7S and its Amanero with isolation input. Not via an I2S interface. And with stock oscillators at this time.

 

[borrego]

By design I2S was not really meant for external connection between devices. It was designed to be the link between the "Transport" circuit and the "DAC" circuit of a player device. For example: (i) between the "Transport" circuit board and "DAC" circuit board inside a CD player, (ii) between the Amanero circuit board and the R7 digital board. In these "by designed" applications, where the cables are no more than a few inches long and shielded by the player/DAC outer case, the MCLK signal won't get interfered much to the point of requiring reclocking/regenerating to gain performance. In this application, it makes more sense to improve performance by making the "Transport" circuit outputting a higher quality MCLK signal..

It is possible for one finding the "Syn" firmware sounding better than the "Asyn" firmware if the Transport output MCLK signal is of higher quality than the regenerated MCLK.

I actually think Kingwa should work on a new ES8620 + Accusilicon AS318-B internal USB transport board (like an internal version of U16), replacing the existing Amanero board, to realize maximum improvement.

 

[DACLadder]

@borrego You are correct as I2S is just a data transfer method. Just looked at the official I2S spec and MCLK is not even referenced in the document. MCLK was just added on at some point to make DACs cheaper to manufacture.

 

[DACLadder]

Let us (and Kingwa) know what you think about Asy and Syn firmware. I have too many DACs, sources, and firmware to sort out for my opinion. The Accusilicon oscillators shipped today so next week have those to evaluate as well. Sheesh! Sounds like fun but will all take some time.

But a big question about the Asy firmware.... For a given source input (S/PDIF, USB, I2S) does Asy sound different than V3? I2S may obviously. Kingwa implies S/PDIF should be better in his TDA DAC project write-up. @lvc10000 likes Asy on the M7S using Amanero USB (isolated version).