DIYINHK Isolated XMOS 768kHz DXD DSD512 review anywhere?

[barrows]

10 hours ago, Mihaylov said:

You are wrong. Masterclock come from oscillators X45 or X49 (pins 3) to the output of the DIYINHK board through chip U5 (pin 1) or in "extreme mode" (see description in DIYINHK site) - directly by cutting the trace under /R2 and short /R1. I have tried it practically.

That is good news as it means one could use the masterclock feed to directly re-clock the I2S lines and get pretty good performance.  I still do not understand why DIYINHK did not do it right, as they have a pretty clean master clock feed on the clean side of the board all they would have to do is add a Potato flip flop to their I2S output and they would have a really nice, re-clocking, isolated USB interface.

[Mihaylov]

2 hours ago, barrows said:

I still do not understand why DIYINHK did not do it right, as they have a pretty clean master clock feed on the clean side of the board all they would have to do is add a Potato flip flop to their I2S output and they would have a really nice, re-clocking, isolated USB interface.

Maybe because it is not necessary when all the elements of the digital path are synchronized from one masterclock?

[barrows]

2 hours ago, Mihaylov said:

Maybe because it is not necessary when all the elements of the digital path are synchronized from one masterclock?

Nope, here is why:

 

The isolators add ~200 pS of jitter, and the output of the XMOS also has significant jitter.  That is a lot these days.  What needs to be done here is to re-clock the I2S lines right before output via the clean masterclock feed (direct from the clocks).  This could be accomplished with the addition of a single Potato chip right at the output.

[Quadman]

I extensively used the non-isolated version of this board a little over a year ago http://www.diyinhk.com/shop/audio-kits/107-xmos-dsd-dxd-768khz-high-quality-usb-to-i2sdsd-pcb.html  I purchased it solely for it's ability to handle DSD512 via it's I2S output, which at the time was the only DDC I was aware of that could do 512.  I owned a heavily modified Gustard X20 and pulled it's USB card so I could tap to the I2S lines thru the PCI header.  I powered the DIY 3.3V from a DIY belleson based PSU.  I also used the dual DIY PSU 3/5/7 V  to power the 5V handshake voltage, and 3.3V tap to power a OXCO clock I had installed closer to the dac chips. I started using the 3.34 unsigned driver DIY had up on their site for a while and then moved to the signed 3.38 driver specifically for this card.  After a few failed attempts i finally got this card to work at 512 with the Gustard and it's dual ESS 9018 chips.  To me the SQ jump from 256 to 512 was night and day, everything just became alive, more human as I started calling it. The stage width and depth both increased the image solidity and presence also increased.  Sound was more there then listening to a recording.  Dynamics were excellent.  Those who say DSD512 is soft and not as dynamic as PCM just don't have their systems set right.

 

A few months after getting 512 to successfully play I built a server for a friend and he elected to purchase a T+A Dac8 DSD.  I had both this new server and T+A in my system and compared this to my server and Gustard X20 at 512.  It was not a night and day difference the gustard acquitted it self quit nicely.  Tho the T+A was slightly better when I packed everything up to send to my friend I was not left crying with the sound of my Gustard at 512.  Around late January the gustard developed short dropouts, less than 1 second and only occasionally.  I trouble shot everything in front of the DIY USB card.  The dropouts occurred at various rates, not just 512.  Everything in front of the card checked out just fine, so one day I pulled the card and re-inserted the stock USB card, now limited to only DSD256 but the dropout disappeared.  So something in the DIY USB card caused the dropouts.  it now sits in a box in my parts bin.  My buddy gave up Computer audio, I bought the T+A from him and now run DSD512 thru that.

 

While the DIY USB card worked it made wonderful sound in my system, 6 months worth anyway before the dropouts started.  I wasn't going to pay $98 for another board to have that happen again.  The T+A just works and no DDC needed.

[Whitigir]

Good info .  I couldn’t make it work with r2r7 and hesitated to try again anytime soon.  Amanero just works

[mivarpet]

On 8/17/2018 at 1:38 AM, Whitigir said:

Good info .  I couldn’t make it work with r2r7 and hesitated to try again anytime soon.  Amanero just works

 

@Whitigir I've used both the standard and isolated versions of this board on my Master 7 with the digital board upgrade and it works well.  The standard version I felt was better than the Amanero and the isolated version even better still.

 

Did your USB cable have straight through power?  I believe the USB input side of the module needs USB power as the 3V3 power to the module is only for the isolated side.

 

I can send you pin outs tonight when I get home.

[Whitigir]

2 hours ago, mivarpet said:

 

@Whitigir I've used both the standard and isolated versions of this board on my Master 7 with the digital board upgrade and it works well.  The standard version I felt was better than the Amanero and the isolated version even better still.

 

Did your USB cable have straight through power?  I believe the USB input side of the module needs USB power as the 3V3 power to the module is only for the isolated side.

 

I can send you pin outs tonight when I get home.

 

Yes! Please do! And thank you

[mivarpet]

@Whitigir These are the pinouts I used are -

 

Audio GD > XMOS
DA > Data
BCK > BCK
LRCK > LRCK
MLCK > MCK
GND > GND
GND > GND
VCC > 3.3V

 

I've also attached some pictures from the board and Audio GD digital board to show you the pins.

 

I'm using a Linux computer to drive the USB, so I don't need drivers.

 

Hope this helps.

 

[Whitigir]

Thanks! I did this but was unable to get intelligent sound.  It was more like a lot of static and some sound with r2r7, perhap it was a bit different 

[mivarpet]

21 hours ago, Whitigir said:

Thanks! I did this but was unable to get intelligent sound.  It was more like a lot of static and some sound with r2r7, perhap it was a bit different 

 

Thats a shame.  What are you driving it with?  I have the same digital board as the R2R7 & am running the latest Smooth Accurate firmware

 

The last two nights I’ve had some issues with mine too - sound distorted on some tracks and also playing slow.  I had to reset the DAC & source and it came good.  I certainly never had that problem with the non-isolated board.  I’ll remove the FIFO board and see how that helps.  The DIYINHK site says the FIFO board is only good for about 5% improvement on PCM1704 based DACs in any case.

[barrows]

If not using a re-clocker with this interface, it would be better to just use the non-isolated version, as the isolation chips typically add >100 pS of jitter.

[Whitigir]

I see, I am using the isolated version and no wonder why it is so bad...I am using my built PC as a source.  Anyways, it look like I will need to wait until newer version is coming out and look further.  For now, I will be using i2s Pinkfaun as what my pc was built for.  Thank you for the confirmations on the Pins out and the performances.  I am using Smooth accurate FW hybrid 

[mivarpet]

@Whitigir  I just tried my board without the re-clocker and got the same distorted sound as you.  I put the S2 jumper on (to disable the "hi-exact" mode on the input) and the music is playing, but not sounding great.  I guess this proves what @barrows mentioned in the previous post.

 

When I had the FIFO module on it was glitching a lot and struggling with 192K audio.

 

Looks like I may go back to the non-isolated version.  Sigh.

[barrows]

1 hour ago, mivarpet said:

@Whitigir  I just tried my board without the re-clocker and got the same distorted sound as you.  I put the S2 jumper on (to disable the "hi-exact" mode on the input) and the music is playing, but not sounding great.  I guess this proves what @barrows mentioned in the previous post.

 

When I had the FIFO module on it was glitching a lot and struggling with 192K audio.

 

Looks like I may go back to the non-isolated version.  Sigh.

Well, this is not what I would expect to hear with a properly working interface.  When I suggested that the isolated version should be used with the re-clocker (as otherwise the added jitter from the isolator will not be removed) I did not mean to say you might hear actual distortion.  With just the isolated interface alone it still should sound decent (not audibly distorted), but there would just be a slight loss of resolution due to the increased jitter.

The problems you are having mean something is wrong, very wrong, and you should figure out what the problem is.  Note that these DIYINHK interfaces use the higher power XMOS chip and it does require quite significant current, are you sure your power  supply is capable of enough current for the interface?

[Whitigir]

59 minutes ago, barrows said:

Well, this is not what I would expect to hear with a properly working interface.  When I suggested that the isolated version should be used with the re-clocker (as otherwise the added jitter from the isolator will not be removed) I did not mean to say you might hear actual distortion.  With just the isolated interface alone it still should sound decent (not audibly distorted), but there would just be a slight loss of resolution due to the increased jitter.

The problems you are having mean something is wrong, very wrong, and you should figure out what the problem is.  Note that these DIYINHK interfaces use the higher power XMOS chip and it does require quite significant current, are you sure your power  supply is capable of enough current for the interface?

So, without enough current, it would still be running but distorted ? I thought it wouldn’t be running at all.  I think it is a matter of communication errors 

[barrows]

@Whitigir,  So, to be clear,  how long is your I2S wiring and what does it go through?

[Whitigir]

9 minutes ago, barrows said:

@Whitigir,  So, to be clear,  how long is your I2S wiring and what does it go through?

The bridge from diyinhk to my board I2s wirings were about 3 inches.  It was never any problems with amanero384 though 

[mivarpet]

@barrows The Audio GD R2R-7 DACs have what is called "High Exact" mode on I2S inputs that offer a very precise lock on the input that can only be used with the best sources.  Please see this link from Audio GD (about half way down in "2018 version firmware setting guide" - 

 

For the HDMI or IIS input, if the sources or transmit cause the signal had not enough good, the sound had noise or distortion, push into the jumper for fixed the issue.

 

I ran the non-isolated version fine for many months with no problems, so I'm not sure if it's a power thing.  Would I be correct in saying the isolated version may draw less power as it's partly USB powered?  The telling change though was when I put a jumper on the Audio GD, the distortion went so it would appear it's a jitter issue.

 

After I posted, I put the FIFO board back on and the sound quality was restored without any need for the jumper.

[barrows]

54 minutes ago, mivarpet said:

I ran the non-isolated version fine for many months with no problems, so I'm not sure if it's a power thing.  Would I be correct in saying the isolated version may draw less power as it's partly USB powered?  The telling change though was when I put a jumper on the Audio GD, the distortion went so it would appear it's a jitter issue.

 

After I posted, I put the FIFO board back on and the sound quality was restored without any need for the jumper.

OK, so it sounds like the DAC is using an adjustable bandwidth for the PLL.  This would them to open the PLL bandwidth a bit to accommodate higher jitter sources.

Yes, the isolated version would draw way less from the local supply because the input portion is powered by the USB supply.  But this might put a big demand  on some USB sources, so with the isolated version, I would check to be sure that the USB source is capable of supplying sufficient current (some computers do not meet the 500 mA USB spec for their 5V outputs).

 

Good news that the re-clocker eliminated the need for opening the PLL bandwidth, if I understand you correctly, this indicates the re-clocker is functioning well.

 

The general (very general, not sure what this is based on?) is to keep I2S lines at <10 cm, which is about 4", but shorter is better.  Also, quality of wires and type matters, are these twisted pairs with a ground paired with each wire (good) or mini coax cables (u.fl, better)?  I2S wiring is a bit fussy, so I would always do everything possible to get it short and use quality cabling.  I prefer u.fl, if I use plain wires, I usually keep them less than 2".