Gustard U12 USB Interface 8 Core XMOS chip

[FredA]

I see the phase noise plot - the numbers do look very good.  And SK sourced - SK made?

Nice find!

Not sure where it is sourced from.

I asked Kingwa about the di-u8 by the way, he does not seem to be convinced it is much better than the di2014 if any. I much prefer the breeze but i have the cheap transfo in the di2014.

I am not sure about this clock mod, i think i have to figure a way to reduce ground noise furthermore. I am thinking about the 6-moons quote above. I think the problem is not noise in itselt but the fact that the zero reference is not perfectly equal between the components. Something like being on shaking floors with not perfectly syncronized shaking if you see what i mean. The ground reference is coming through 2 different path in the dac, especially with i2s. With the isolation transfo of spdif, it should not be such a big problem but then jitter cannot be better than that of the spdif receiver, with is way higher than the breeze's level in my case. Maybe the solution is to use multiple transfos with common reference in the i2s connection, i don't know. I would have to reflect on this. Maybe none of this is relevant, i am not an expert. Just some thoughts.

 

[FredA]

I looked for some spec for tcxo clocks used by audio-gd. Look like it is quite low, less than 0.3 ppm. Maybe i would get better results with a balanced connection. I might try to order the best bulk aes cable available and mount it on a pair of neutrik to give it a try. Might get an easy upgrade for not much.

On the orher hand, people seem to all agree that i2s is the best input on the master-7. Depends on the spdif cable however, which people tend neglect.

 

[hgpsemaj]

I2S supports 32B 384kHz and native DSD (i.e. not DOP.).

 

[abartels]

Bob,
 
I didn't put 10 R-Cores in my dac for their weight or fun 

 
And, those comprehensive EMI filters with a lot of cabling, no, not for fun, and wooden - seperated  - subchassis for psu and dac and I2S, not for fun,
even the copper plating/shielding was not for fun 

 
I proved the opposite, it definitely is possible to build a dac with integrated I2S-USB module with VERY high quality. It can be compared with analog sound, but with resolution and dynamics which exceeds it by far,,,
 
It all starts with extensive filtering, if noise from usb wants to interfere with one of the 9 other power lines, it has to travel thru 2x differential mode filters (in 2x psu's) AND 3x Common mode AND differential mode filters!
 
 
 
I need some comparison with other (very expensive) dac's, I will try to arrange this in the coming month.
 
But, thanks for clarification 

 Just couldn't understand why bringing up those Hugo dac's while review says it sounds Digital, and there wasn't any moment he could relax from it, hmmm,,,
 
 
 
Regards,
 
Alex

 

[ccschua]

 
Yes there is a trim-pot - but the adjustment requires a voltage meter and opening the case.  Of course doing the adjustment while powered up and the screw is very small - so not for the inexperienced - large voltages in there.  So I guess that's more then some buyers want to handle - understandably.  So I guess the sellers do the adjust themselves and just advertise the unit for one voltage to eliminate confusion.

 
I have done this several times, toggle between 12V and 5V by adjusting the trimpot. in the end, got tired, potential for lost track of the voltage. so in a rush, one may connect a 12V supply to a 5V device. warning !

 

[ccschua]

 
Well if the DXIO really turns out to beat a Crystek or NDK clocked Breeze - I might think of getting a Hynes for it.  $300+ for a power supply is a bit steep, but heck look at what I've spent on Cerious power cables.

 
I would think try to give diyinhk LT 3042 (1uV) reg a try. its USD 45 and gives you a dual for which you can buy a dual 6V output toroid and small enough casing to house the DXIO, reg and toroid inside.

 

[rb2013]

Not sure where it is sourced from.

I asked Kingwa about the di-u8 by the way, he does not seem to be convinced it is much better than the di2014 if any. I much prefer the breeze but i have the cheap transfo in the di2014.

I am not sure about this clock mod, i think i have to figure a way to reduce ground noise furthermore. I am thinking about the 6-moons quote above. I think the problem is not noise in itselt but the fact that the zero reference is not perfectly equal between the components. Something like being on shaking floors with not perfectly syncronized shaking if you see what i mean. The ground reference is coming through 2 different path in the dac, especially with i2s. With the isolation transfo of spdif, it should not be such a big problem but then jitter cannot be better than that of the spdif receiver, with is way higher than the breeze's level in my case. Maybe the solution is to use multiple transfos with common reference in the i2s connection, i don't know. I would have to reflect on this. Maybe none of this is relevant, i am not an expert. Just some thoughts.

You might want to try a Wyred for sound Remedy and a short spdif rca to rca connector. The Remedy reclocks the spdif with a Crystek CCHD-957 and acts as a spdif buffer and upsampler.  It worked great on my DAC60 (where I had already upgraded the clock to a TXCO Vanguard) - and slightly improved the APL NWOjr (which already has great clocks).
 
Interesting the Brreze was better then the di2014.  Do you have picture of the di2014 board - I thought that used the OEM TXCO JYEC Chinese clocks like the Breeze, Gustard, Melodious.  I know that Audiogd was even selling these clocks for awhile.  But on the DU-I8 the clocks are different

 

[rb2013]

I looked for some spec for tcxo clocks used by audio-gd. Look like it is quite low, less than 0.3 ppm. Maybe i would get better results with a balanced connection. I might try to order the best bulk aes cable available and mount it on a pair of neutrik to give it a try. Might get an easy upgrade for not much.

On the orher hand, people seem to all agree that i2s is the best input on the master-7. Depends on the spdif cable however, which people tend neglect.

You want to find a phase noise plot or at least some various freq measurements.
 
The generic OEM TXCO clocks like in the Breeze, Gustard, and Melodious are just like the JYEC or Vanguard (it seems the manufacturer will custom stamp the top cover with your logo if you order enough).
The only measurement I have seen is at 1kHz -125db.  Not great.  But they are rated at 0.1ppm
 

 

[rb2013]

  Bob,
 
I didn't put 10 R-Cores in my dac for their weight or fun 

 
And, those comprehensive EMI filters with a lot of cabling, no, not for fun, and wooden - seperated  - subchassis for psu and dac and I2S, not for fun,
even the copper plating/shielding was not for fun 

 
I proved the opposite, it definitely is possible to build a dac with integrated I2S-USB module with VERY high quality. It can be compared with analog sound, but with resolution and dynamics which exceeds it by far,,,
 
It all starts with extensive filtering, if noise from usb wants to interfere with one of the 9 other power lines, it has to travel thru 2x differential mode filters (in 2x psu's) AND 3x Common mode AND differential mode filters!
 
 
 
I need some comparison with other (very expensive) dac's, I will try to arrange this in the coming month.
 
But, thanks for clarification 

 Just couldn't understand why bringing up those Hugo dac's while review says it sounds Digital, and there wasn't any moment he could relax from it, hmmm,,,
 
 
 
Regards,
 
Alex

Now that's the approach!  Not shared power supplies with USB boards like you see on many DACs.  Where have you seen that kind of attention to USB board isolation any where else?  Nice work Alex.  But for those who are not up to building a DAC with 9 separate R-core power supplies - the separate USB bridge with a good dedicated AC filter is the way to go.  Those Art Audio PB4X4Pro's are only $80.

   
I would think try to give diyinhk LT 3042 (1uV) reg a try. its USD 45 and gives you a dual for which you can buy a dual 6V output toroid and small enough casing to house the DXIO, reg and toroid inside.

Good suggestion - Paul does do some neat work.  I think a modded/upgraded TeraDak DC-30W with a Cerious Graphene power cord should do the trick.

 

[FredA]

 
You might want to try a Wyred for sound Remedy and a short spdif rca to rca connector. The Remedy reclocks the spdif with a Crystek CCHD-957 and acts as a spdif buffer and upsampler.  It worked great on my DAC60 (where I had already upgraded the clock to a TXCO Vanguard) - and slightly improved the APL NWOjr (which already has great clocks).
 
Interesting the Brreze was better then the di2014.  Do you have picture of the di2014 board - I thought that used the OEM TXCO JYEC Chinese clocks like the Breeze, Gustard, Melodious.  I know that Audiogd was even selling these clocks for awhile.  But on the DU-I8 the clocks are different

Or maybe this: could be the breakthough ingredient I am missing: galvanic isolation of my rj45 i2s connection (clean ground at last!).
 
http://www.ttl-network.de/news-details/new-rj45-network-galvanic-isolator.html

 

[rb2013]

  I2S supports 32B 384kHz and native DSD (i.e. not DOP.).

I'll just quote Mike Moffat from Schiit Audio about the decsion to omit DSD native and i2s on their newest sota flagship DAC the $2400 Yggdrasil:
 

  What about DSD?
Yggdrasil won’t do DSD, no way, no how.
 
But what if I want DSD?
If DSD becomes a significant part of the market, where “significant” is defined as “more than 1% of total sales,” then we will support it via a separate dedicated DSD decoding box. Today, when it is less than 0.01% of recorded music, well, hey, we're not so interested.
 
What? But I hear DSD is the future!
Yeah, and reel-to-reel was the future in the 1970s, and it’s dead now, and DAT was the future in the 1980s, and it’s dead now, and HDCD was the future in the 1990s, and it’s dead now, and SACD was the future in the 2000s, and it’s dead now. But, let's say Sony suddenly opens their vaults and offers 30,000 DSD albums with guaranteed direct-from-DSD provenance at $5.99 each, and Apple starts streaming DSD as its de facto format for iTunes, (yes, we know, stop laughing) then hey, Yggy is fully upgradable...and then we just change it up to be the best DSD DAC out there. 

 
Oh and his comments on 32bit audio:
 

Wait, what is this about 32-bit music?
How much 32 bit music do you have? (Not that it will ever exist—we can't get the noise floor that low. Period. Unless Dr. Who pays us a visit and drops some alien tech on us...) Also, how many giga-rate DSD recordings do you have? None, because they don't exist.
 

But look at the DAC and filter technology they put into that unit - amazing stuff - and they did include SPDIF Coax, BNC, & AES - just sayn'

 

[rb2013]

Here is another highly acclaimed $10,000 DAC The TotalDAC D1- no i2s or native DSD std.  SPDIF coax Yes.
 
Do you think all these great designers omit i2s because it costs to much to include?
 
http://www.totaldac.com/D1-tube-eng.htm
 

 
Now where you see DSD and i2s offered a lot is on S-D DACs in the $400 to $900 range.  Buzz words sell.

 

[FredA]

You might want to try a Wyred for sound Remedy and a short spdif rca to rca connector. The Remedy reclocks the spdif with a Crystek CCHD-957 and acts as a spdif buffer and upsampler.  It worked great on my DAC60 (where I had already upgraded the clock to a TXCO Vanguard) - and slightly improved the APL NWOjr (which already has great clocks).

Interesting the Brreze was better then the di2014.  Do you have picture of the di2014 board - I thought that used the OEM TXCO JYEC Chinese clocks like the Breeze, Gustard, Melodious.  I know that Audiogd was even selling these clocks for awhile.  But on the DU-I8 the clocks are different

The di2014 is a single-clock design based on the via usb32 chip. I have the standard transfo but upgraded clock which i believe is 0.3ppm (compared to .1 for the breeze).

http://www.audio-gd.com/Pro/dac/DI2014/DI2014EN.htm

 

[rb2013]

The di2014 is a single-clock design based on the via usb32 chip. I have the standard transfo but upgraded clock which i believe is 0.3ppm (compared to .1 for the breeze).

http://www.audio-gd.com/Pro/dac/DI2014/DI2014EN.htm

I see - so for the di2014 there where two upgrade options - one was the R-core transformer and the other the TXCO clock - or dual clocks?
 
It looks like it just comes std now with the R-core and has three clocks - two audio clocks (on the lower left) and a USB clock (lower right).
 
It would be nice to know which LDO's he is using.
 

 

[FredA]

 
I see - so for the di2014 there where two upgrade options - one was the R-core transformer and the other the TXCO clock - or dual clocks?
 
It looks like it just comes std now with the R-core and has three clocks - two audio clocks (on the lower left) and a USB clock (lower right).
 
It would be nice to know which LDO's he is using.
 

That's it. R-Core and TCXO (1). The via chip cannot work with 2 clocks as far as i know. At least, not with the implementation used by audio-gd. 
 
I am looking for galvanic isolation now. That should do it for me. There is also this product to considerer (besides the rj45 isolator I mentioned earlier) although it does not seem to work consistently with all usb receivers according to a Computer Audiophile thread.
 
http://www.ttl-network.de/news-details/new-rj45-network-galvanic-isolator.html
 
I would rather go the i2s way. Much more likely to work properly IMO and thus i would still be using the best input of the M7 according to experts. I have tried spdif over bnc with a dhlabs cable and preferred i2s).