[Shahrose]

For the length, Steve recommends 1.25 - 1.5m, similar to para's range above.

 

[IPodPJ]

Also another new finding for me (maybe already well known for other experienced head-fiers): according to Kingwa, the PCM1704U-K dac chip support 192KHz sample rate at 8x over-sampling, which is higher than the official specification (official data sheet here: http://www.ti.com/product/pcm1704). And since the S/PDIF decoder in Master-7 (I don't know the model) is different from that of Reference 7.1 (DIR9001), this enables the Master-7 to accept 192KHz sample rate.

That seems very unlikely. TI has always stated it supports 96K at 8x over sampling or 768k at 1x. Maybe he has figured out some clever way to achieve that by stacking the chips, though at the same time wouldn't it also negate the benefits of stacking 4 chips per channel to achieve greater SNR and be more like stacking only 2 chips per channel?

Perhaps he means the digital filter can accept up to that file resolution but it would likely have to be down sampled, or be played at 4x over sampling. I'm not sure how he can bypass the inherent design of the chip and if he can is it a solution that won't cause any issues at some point in time, or prematurely shorten the life of the chip? There was nothing wrong with the way it sounded at 96K at 8x, as the Ref 7.1 is a fantastic sounding DAC.

I2S is a protocol which was never designed for external hookup use, or designed to be used at lengths greater than 6" internally on a PCB. To use it to connect components requires a dedicated transmitting circuit and a dedicated receiving circuit which is more parts in the signal path. I am unsure how such a signal could provide any benefits over a solid S/ PDIF signal with precision clocking. Then of course you have the incompatibility factor between devices, as not every component uses the same pin outs or connectors. This is why S/PDIF was designed in the first place. Keep that in mind.

 

[lmswjm]

Regarding the AES connection; Kingwa always regarded it as a colored, almost tube like path. That's why it wasn't on the Ref 1. I believe he designed his original DAC's with S/PDIF in mind and added AES later for demand. That said, I have a pro audio AD/DA Crane Song HEDD 192 that the design was based around the AES connection which I only use, and added S/PDIF for convenience.

 

[tme110]

There's another reclocker, I don't remember the make or the exact issue but it had several outputs including AES.  The manufacture themselves stated that they would rate AES the lowest quality signal. Something like it needed to be buffered then converted to an AES signal or something...  I'm not sure how or why their implamentation would be different than others it sounds like though AES may be the best for long paths, it's not the best to have a signal converted into.
 

 

[hmouse]

That seems very unlikely. TI has always stated it supports 96K at 8x over sampling or 768k at 1x. Maybe he has figured out some clever way to achieve that by stacking the chips, though at the same time wouldn't it also negate the benefits of stacking 4 chips per channel to achieve greater SNR and be more like stacking only 2 chips per channel?
Perhaps he means the digital filter can accept up to that file resolution but it would likely have to be down sampled, or be played at 4x over sampling. I'm not sure how he can bypass the inherent design of the chip and if he can is it a solution that won't cause any issues at some point in time, or prematurely shorten the life of the chip? There was nothing wrong with the way it sounded at 96K at 8x, as the Ref 7.1 is a fantastic sounding DAC.
I2S is a protocol which was never designed for external hookup use, or designed to be used at lengths greater than 6" internally on a PCB. To use it to connect components requires a dedicated transmitting circuit and a dedicated receiving circuit which is more parts in the signal path. I am unsure how such a signal could provide any benefits over a solid S/ PDIF signal with precision clocking. Then of course you have the incompatibility factor between devices, as not every component uses the same pin outs or connectors. This is why S/PDIF was designed in the first place. Keep that in mind.

This is what Kingwa replied when I asked him why Master-7 can support 192KHz at 8x.

"The PCM1704 confirm can support 1536KHz sampling input so can support up to 192K at 8X times oversampling .
In the Reference 7.1 , it is applied DIR9001 for SPDIF so limit at 96K SPDIF input.
In Master 7 have different SPDIF decoder for the SPDIF input so some inputs can support up to 192K."

 

[shogo33]

Does anyone know which SPDIF input receiver that is used in the Master7? 
 

 

[Kingwa]

The Master 7 have one DIR9001 and 3 pcs WM8805 for SPDIF input.
This design allow the receivers can setting on the best performance for different inputs .
Except the BNC input applied DIR9001, others are the WM8805.

 

[IPodPJ]

This is what Kingwa replied when I asked him why Master-7 can support 192KHz at 8x.
"The PCM1704 confirm can support 1536KHz sampling input so can support up to 192K at 8X times oversampling .
In the Reference 7.1 , it is applied DIR9001 for SPDIF so limit at 96K SPDIF input.
In Master 7 have different SPDIF decoder for the SPDIF input so some inputs can support up to 192K."

I guess I'd have to hear that from a TI engineer. And didn't Kingwa formerly state exactly what the TI engineers had said as well? I don't believe anywhere in the PCM1704 literature does it say you can use that chip for 192kHz with 8x over sampling. The chip was designed in 2000, or earlier even.

 

[paradoxper]

Quote:

There's another reclocker, I don't remember the make or the exact issue but it had several outputs including AES.  The manufacture themselves stated that they would rate AES the lowest quality signal. Something like it needed to be buffered then converted to an AES signal or something...  I'm not sure how or why their implamentation would be different than others it sounds like though AES may be the best for long paths, it's not the best to have a signal converted into.
 

I2S can be directly connected to a DAC chip, so it does not have digital signal conversion like AES. But it really comes down to implementation.

 

[Kingwa]

Quote:

I guess I'd have to hear that from a TI engineer. And didn't Kingwa formerly state exactly what the TI engineers had said as well? I don't believe anywhere in the PCM1704 literature does it say you can use that chip for 192kHz with 8x over sampling. The chip was designed in 2000, or earlier even.

The Wadia 27 and the later version which is had pcm1704 built in have upsampling to 2.82M feed to PCM1704 since 1997. This sampling frequency just  like 352K X 8 oversampling .
The verify is quite easy, who have an oscillograph can check the DSP input sampling if is 192K, the PCM1704 input if is 1536K .

 

[FauDrei]

So TI created DAC chip that has better specs in real life than on paper. Perhaps 192KHz was science fiction back when PCM1704 was designed so they did not even bother to test it higher than 96KHz even though chip is capable of doing 8x 192KHz.
 
As I see it - KingWa is just taking advantage of that PCM1704 "undocumented feature".
 
Now, if 24/96 through SPDIF sounds excellent on A-GD RE-x flagships - I can only imagine how does 8x oversampled 24bit 192KHz music sound through USB-32 on Master-7...

 

[Kingwa]

I was surmise that time have not any digital audio chips can arrive the 192K playback ablity, and TI was design the DF1704 for cooperate with PCM1704 which is support up to 96K, so they intended or involuntary declare the max input is 768K on the file.

 

[IPodPJ]

I was surmise that time have not any digital audio chips can arrive the 192K playback ablity, and TI was design the DF1704 for cooperate with PCM1704 which is support up to 96K, so they intended or involuntary declare the max input is 768K on the file.

That's because many engineers don't believe there is merit for anything higher than 96kHz on play back, and in fact higher sample rates can be detrimental to SQ.

 

[lmswjm]

+1

 

[RedBull]

Quote:

The Master 7 have one DIR9001 and 3 pcs WM8805 for SPDIF input.
This design allow the receivers can setting on the best performance for different inputs .
Except the BNC input applied DIR9001, others are the WM8805.

 
That means, BNC limited to 24/96? 
I am just curious, since the rest are using WM8805, why don't just use WM805 for BNC as well?