[rb2013]

I am in the process of returning my new ifi ipower 5v and have a lps similar to the one in the SU-1 coming from china.

Would love to see someone build one of these with the LT3042 LDO's.
 
So far the best of the Chinese LPS I have tried is the R-Core TeraDak DC-30W.   I modded with Nichicon HWs.
 

 

[rwmj5]

I'm sorry if I'm completely clueless on this topic, but are the XMOS U8 coverters that are on ebay for less than $100 any good, or are they just junk?  Also, are there any issues with drivers running these on Windows 10?  The whole driver situation with this chip is very confusing with usb audio 2.0 only being supported for pros at some high price etc.  Not sure if it yould make any difference to someone just listening to playback.  I would like to get high quality output from Spotify...

 

[drez]

  Would love to see someone build one of these with the LT3042 LDO's.
 
So far the best of the Chinese LPS I have tried is the R-Core TeraDak DC-30W.   I modded with Nichicon HWs.
 

 
I'm glad you posted this.  I was looking for replacement for the Panasonic FC on the DIYNHK kit and those might do the trick.  Thanks.

 

[rb2013]

  I won't put it that way, it's more of a market decision. Remember I read an interview about F-1 earlier this year, the interviewer specifically point out B0505S and
ask why they put such a sub-optimal component there, reply from Singxer is ".....considering the cost..., to lower the sale price...". Also from their advertisement,
one of its target market is replacing the Amanero module built into a lot of DACs --- compatible mounting hole, compatible pin out, etc. And the +5V, Gnd power through
holes right beside B0505S give strong hint to owners - you could remove B0505S and use DAC's internal power or a better source to power the clean section.
 
Right. For I2S users, this is a non-issue; for SPDIF users, I won't worry too much, the noisy XMOS is already blocked out, so the clean section is pretty quiet.
 
 
No. It has built-in LPS using LM2940 to power the clean section; As analyzed in previous post, the power driving the XOs is a solid 9uV.
 
 
Yes, I agree with you, see post #3058
 
Other factors need to be considered too, ex. does 1uV LDO power both XOs and XMOS, or there are separate LDOs for each. Noisy XMOS will feed its generated noise back to its power line, usually bypass capacitor won't be able to constrain all the noise locally, so it will pass to the XOs if a single LDO is shared by both. Layout and board size also will determine if 1uV noise level can really be achieved throughout the board.
Personally I will wait till end of this year. I believe a lot more XU216 based DDC will come out pretty soon. Also I will wait until comparison against SU-1, not F-1 which definitely need modification to really show its potential.
 
I thought you already decided to upgrade your office setup to AOIP too?
 
 
Just the metal cover, due to the cramped space in my DAC (F-1 is installed inside my DAC), and constant removing and installing for modification, taking photo, etc.
the cover just come off, soldered it back many times, finally decide to give up. Worry about EMI shielding initially, luckily no negative influence can be detected so
far.

Thanks for the information.  Esp on the DC-DC stuff.
 
I'm still working out the mods on the BURL - so running USB in my office until then- and until I sell the rest of my USB chain stuff.
 
I'm assuming the F-1 and SU-1 are designed to shut down which ever clock is not running  to min noise. 
 
Looking at the Digikey info on the CCHD-575 - it does not appear to have the Tri-State Standby (enable/disable) function.  The CCHD-975 does.
 
This was from the Hydra Z website:

TECHNICAL IMPROVEMENTS

Hydra Z benefits from all the qualities of the

X+

, and even more. New clocking scheme allows shutting down one of the two main oscillators when not in use. By reducing the mutual interference between clocks, better phase noise performance can be obtained. The power supply section was moved to a different chassis for best results. 

 
The Hydra Z like the Berkeley use the 975.

 

[rb2013]

   
I'm glad you posted this.  I was looking for replacement for the Panasonic FC on the DIYNHK kit and those might do the trick.  Thanks.

De Nada
 
PS
Just a comment on the Talema - not really anything that special - just better then the BingZi that Gustard used in the U12 and the std version of the DU-U8.
 
Once I find some time and funds will go with a Simon Tuned or Lite R-core and build a PS.
 
Interesting debate on the best design of PS transformer: EI, Toroidal, R-Core.
 
I have always understood that the R-core have the best noise filtering ability:
https://www.audialonline.com/topics/psu-transformers-ei-vs-toroidal-and-more/
http://www.stereo.net.au/forums/topic/55835-why-a-torroidal-instead-of-an-r-core-transformer-in-digital/
 

What you have stated is mostly true. However, the reason why R-core transformers are the best choice for ALL low level audio applications, is due to their far lower coupling. This means that any mains-bourn interference will be greatly attenuated by an R-core transformer, compared to a toroidal. E-I type transformers lie somewhere in the middle. Better than toroidals, but not as good as R-core types. And yes, radiation is higher with both R-core and E-I types, but that can effectively dealt with by either good shielding, or placing the transformer in a separate enclosure. 

 
For the R-Core and EI - best to have it well away from the DDC electronics:

 

[rb2013]

  I'm sorry if I'm completely clueless on this topic, but are the XMOS U8 coverters that are on ebay for less than $100 any good, or are they just junk?  Also, are there any issues with drivers running these on Windows 10?  The whole driver situation with this chip is very confusing with usb audio 2.0 only being supported for pros at some high price etc.  Not sure if it yould make any difference to someone just listening to playback.  I would like to get high quality output from Spotify...

No worries on WIN10 and the drivers - on my system they worked fine.
 
Under $100?  Maybe a bit more for the Breeze DU-U8 - with the Talema is still a darn good DDC.

 

[bButcher]

  I'm assuming the F-1 and SU-1 are designed to shut down which ever clock is not running  to min noise. 
 
Looking at the Digikey info on the CCHD-575 - it does not appear to have the Tri-State Standby (enable/disable) function.  The CCHD-975 does.

Yes, from CCHD-575's spec, it doesn't have enable pin and the list typical consumption is 15mA. Since on my unit the measured current draw is 14mA for the XOs so only one is running while working and on Singxer's website, it mention "Customized CRYSTEK femto clocks are used". So maybe the customization part is the added enable pin.
 
O.K. confirmed. Just measure it, the two enable pins are connected to CPLD's pin 18 & 19.

 

[Superdad]

  Looking at the Digikey info on the CCHD-575 - it does not appear to have the Tri-State Standby (enable/disable) function.  The CCHD-975 does.
 

 
Hi Rob:
 
I promise you you the 575 is the better clock.  Crystek will even say so themselves.  They have improved their process a great deal.  The 3 samples they sent us last year each came with their own phase noise plots.  At 10Hz offset the samples were -108, -110, and -112dB  Much better than their own published graphs and much better than the 957.  Pretty fantastic for a $9.40 clock (that's 1K price direct).  Plus it is physically much smaller which also has its advantages.
 
--Alex C.

 

[Superdad]

  Once I find some time and funds will go with a Simon Tuned or Lite R-core and build a PS.
 

 
Go for the R-core Rob!  When we were designing the choke-filtered JS-2 in early 2014, we compared a really good toroid to an R-core and there was absolutely no contest.  The R-core smoked the toroid in every way.  The ones we use are copper shielded, but here is the pic from that fateful day (toroid version up on wood blocks):
 

 

[rb2013]

   
Hi Rob:
 
I promise you you the 575 is the better clock.  Crystek will even say so themselves.  They have improved their process a great deal.  The 3 samples they sent us last year each came with their own phase noise plots.  At 10Hz offset the samples were -108, -110, and -112dB  Much better than their own published graphs and much better than the 957.  Pretty fantastic for a $9.40 clock (that's 1K price direct).  Plus it is physically much smaller which also has its advantages.
 
--Alex C.

Hi Alex - Nice to see you back here!  We may have had our battles, but I always respect your knowledge and input.  Good info on the 575 - strange Digikey did not show a version with the Tri-state feature, or maybe I missed something.  Or maybe it's not as important as Audiobyte says?  That is the ability to disable the unused audio clock.
 
As you can see a few talented folks here are working some very interesting mods to the F-1.  Maybe Uptone could do a Uber-F-1?  That would be very interesting.
 

   
Go for the R-core Rob!  When we were designing the choke-filtered JS-2 in early 2014, we compared a really good toroid to an R-core and there was absolutely no contest.  The R-core smoked the toroid in every way.  The ones we use are copper shielded, but here is the pic from that fateful day (toroid version up on wood blocks):
 

Nice!  Your concurrence is appreciated.  I can see the use of a toroidal in a small one unit case design - due to the EMI issues.  But that can easily be a non-factor with an ext LPS box.
 
Researching yesterday I came across this comment:

Thanks Zaphod, but isn't mains bourne interference dealt with by chokes and capacitors?

It should be, but every bit helps. And R-core transformers are around 100 times better than toroidals. They're not just a little bit better. They are spectacularly, brilliantly better. The reality is, that for anyone purporting to build the best product possible, then the best components should be used (within cost constraints, of course). I will remind you once more: Toroidals are the very WORST choice of power transformer for low level (audio) components. ANY transformer will make a better choice.

http://www.stereo.net.au/forums/topic/55835-why-a-torroidal-instead-of-an-r-core-transformer-in-digital/
 
Also looked at shielding materials - came on this - looks like copper is pretty good.
http://www.ets-lindgren.com/pdf/emctd_1293_weibler.pdf

 

[rb2013]

  Yes, from CCHD-575's spec, it doesn't have enable pin and the list typical consumption is 15mA. Since on my unit the measured current draw is 14mA for the XOs so only one is running while working and on Singxer's website, it mention "Customized CRYSTEK femto clocks are used". So maybe the customization part is the added enable pin.
 
O.K. confirmed. Just measure it, the two enable pins are connected to CPLD's pin 18 & 19.

Good stuff - those Singxer folks are smart...so I guess a special spec'd version vs the OTC.

 

[MaXaZoR]

I just picked up a SU-1 to use with my HTPC -->Intona-->Schiit Gumby chain.
 
I'm curious what cable and input should I use on the Gumby, AES?   Anyone have any decent relatively cheap cables they would recommend for this chain?

 

[gldgate]

While there will never be a consensus, I know of many folks with Yggy/Gumby's who seem to prefer AES - me included. However, don't expect the difference to be "night and day".  
 
I actually like these for well made inexpensive AES cables.
 
https://benchmarkmedia.com/collections/cables/products/benchmark-digital-xlr-cable

 

[mordicai]

  I just picked up a SU-1 to use with my HTPC -->Intona-->Schiit Gumby chain.
 
I'm curious what cable and input should I use on the Gumby, AES?   Anyone have any decent relatively cheap cables they would recommend for this chain?

bluejeans cable

 

[Martinrey]

Sorry to come into this thread so late.  Does the Singxer F-1 XMOS XU208 USB Digital Interface 384K Coaxial I2S DSD256 With case work well to upsample to DSD 256 using Foobar into the Gustard X20 U.  Right now I am upsampling to DSD 128 into the Gustard.  Does the Singxer support I2S?  If so what type of cable shoud I use to interface between the Singxer and the Gustard?  
 
Thanks for you indulgence...(I know these are newbee questions)...
 
- Martin