XMOS DSD 384 kHz / 32bit USB

LT1818 datasheet has not specified THD @1kHz, only at 5MHz, being 85dBc, and LME49710 has -130dB @ 1kHz.

I'm wondering how much impact could be 8 octaves above 20 kHz in hearing experience. 😱

A circuit that doing well 5MHz can only work so much better at 20kHz..

For LME we still do not know two basic things, schematic and gain-phase vs frequency diagram..
 
Yoda-The-Empire-Strikes-Back.jpg
 
Im asking because I feel Lucian board looks more "professional" to me. I had it some time ago and it was really good. Also support for Xmos drivers is very important thing.
Anyways this board looks also very nice and its cheaper I assume? Also there is some advantage of pairing it with dedicated DAC board with nice and close connection. I got Subbu DAC and dont know how it sounds to JLSounds solutions.
But what about drivers? Is it supported?
I contacted with JLSounds already but its always nice to hear someone who compared both converters, paired it with Subbu for example, heard dedicated JLSound dacs etc.
Regarding measurements I believe both converters are very good.
Is that converter also power hungry like WaveIO?
 
Last edited:
Spoken like a true diyAudiophile. Even though Joro's board has better specs in every category, Lucian's board has gold plated traces. Bling aways wins.

Actually, both boards have their pluses and minuses. I have both, and there are things that could be improved with both. But the JL Sounds board is quite good. I do wish that when feeding an external clock that it would go through the isolator. I think he ran out of isolator channels. Also, am not terribly fond of the SiliconLabs RF transmitter/receiver style isolators he uses. The have about 3.5 times more jitter than the NVE GMRs.
But at least Joro put them on the right side of the clocks! (That is a big fail on the WaveIO, and may be the reason I prefer the un-isolated output of Lucian's board.)
 

Attachments

  • P1080626.JPG
    P1080626.JPG
    131.8 KB · Views: 756
the output pins arrengement seems better on the Wave board : GSGSGSGS nothe SSSSG of the Joro's (would try if only !). Alqo better ndk clocks on the wave I/O!

What a pity Joro does not fix this output with a more serious layout here with ufls or a plug with a gnd return path for each I2S signal ! I asked last summer... but had just an answer about the clock board 🙁. So not easy to use with two multibit dac chips because the interlink missing quality (imho)
 
Hi,

Also support for Xmos drivers is very important thing.

I have a licensed XMOS driver - it's free for downloading. Of course the author rights for this driver belongs to XMOS, I just represent the driver according to the conditions from XMOS.

I do wish that when feeding an external clock that it would go through the isolator. I think he ran out of isolator channels.
The external MCLK 45/49MHz is also through the isolator.

Also, am not terribly fond of the SiliconLabs RF transmitter/receiver style isolators he uses. The have about 3.5 times more jitter than the NVE GMRs.
You are right. NVE GMRs isolator has less jitter than Si8661, which I use. The NVE has according to it's datasheet 100ps jitter, which is a lot for audio according to my opinion.
No matter which isolator is used, there is still need for reclock in order to minimize the jitter, that's why I use Si8661 and after that reclock.

Alqo better ndk clocks on the wave I/O!
On the last batch of boards there are NDK NZ2520SD 45.158MHz and 49.152MHz oscillators. I use a transition board, because the NDK oscillators need a capacitor near them. The measurements show worse work, when this capacitor is at a distance.

What a pity Joro does not fix this output with a more serious layout
There is an OEM board with GSGSGSGS pinout. This board is for people like you, who wants to build all around the I2Soutput according to your own visions 🙂

Regards,
Joro
 

Attachments

  • NDK_boards.jpg
    NDK_boards.jpg
    479.3 KB · Views: 721
Hello Joro,

Yes I remember you answered the same as above to me the last summer about the OEM board + the standalone crystals board as a more efficent answer to my needs.

But it was for me just the half of a solution as an external crystal board does introduce some other problems.

Btw your cute OEM board was advised by me for the Soekris Dam1021 R2R DAC as a good Qaulity/Price for an USB to I2S device instead an Amareno as the timing was not important because the DAC reclocks with a FGPA and buffers with a FIFO the input I2S signal !

Yes the USB OEM board has more gnd vias which permitt an easier way to have a standalone wire gnd for each acive I2S wire. I'm not a technician, but some readings at Diyaudio showed It's more than very important when we play with several boards and RF not to introduce worst things than the ones we try to solve !

But understand, some needs also more usefull choices ! I asked you (it was for the TDA1541 your development can not apply but also for two AD1862 I have if you remembered it). In those cases I needed a all in one board with both good crystals and good output layout as I described in the post above ! Plus, the vias without plugs and uf-l pads is not very usefull and consproductiv for quality as well

I say again than a better philosophy, both technical (for what it worth due to my poor level) and commercial (it's certainly a good humble advise as I didn't buy any of your boards last summer despite the need), would be to swap the crystals of your actual boards by NDKs as Lucian does with the Wave I/O and to solve the layout problem instead to push the diyer towards some other quality problems that the OEM board + the crystal board could introduce !

Is it so hard to solve those little problems with a new board ? You made already the hardiest part !

Your main board is certainly good as SuperDad testimonied, but again both the lake of better crystals and a more usefull layout for output (GSGSGSGS with more universal possibilities like uf-l plugs - already soldered please ! -) was a no go for me ! And as said two boards to solve that is not a real solution for the customer and the audiophile involved by technic !

I think with all those last stories of drivers, parts sourcing with crystals and layout design drawging put the balls into the side of developpers... not the customers ! All of you are more clever than us poor skilled enthusiasts ! Could be fantastic also than all those different plugs to link all the stuffs sold here at diyaudio(Fifo, usb, etc) would be more universal ! I only see the ud-l plugs to solve this problem as a spare solution on the boards as IanCanada is using himself with his devices (plastic plugs +uf-l plugs as well).

At least my humble opinion as a lurker... hopping to be customer later !

regards
 
Last edited:
Hi,

..it was for the TDA1541 your development can not apply but also for two AD1862 I have if you remembered it

There is no problem the TDA1541 to work on I2S mode, but it can’t be driven on simultaneous mode. This protocol is a little bit different. The XMOS board can drive two AD1862 (one for each channel), as described.

...
I say again than a better philosophy, both technical ...

I’ve designed the XMOS board for my needs, because I haven’t found suitable for me 🙂 In this design I wasn’t looking for a commercial, after that many people like the board which is another thing. It’s obvious that the board won’t be liked from all in this stage, but one can find critics for every device 🙂

Regarding the oscillator board, the goal was to give a possibility for using different kind of oscillators, not to stick with one kind. For example on many listening session comparing Crystek CCHD 957 with the NDKs over 70% of the listeners like more the Crystek. I must say that the oscillators were compared in the same conditions, i.e. only the oscillators were changed. It is obvious that the Crystek cost a lot, but it’s good one to have a choice.

...
Is it so hard to solve those little problems with a new board ? You made already the hardiest part ! ...

I don’t think that I’ll be able to satisfy all demands/requests of every diyer, but there are many different kinds of transport in which the designers rely on a different states, so everyone could find a suitable device for his needs 🙂

... both the lake of better crystals and a more usefull layout for output (GSGSGSGS with more universal possibilities like uf-l plugs - already soldered ...

As I mentioned the boards are equipped with NDK oscillators (the waiting was long). Regarding GSGSGSGS you are correct, in terms of EMS, this kind of interface is better, but in terms of jitter according to me – no. The GSGSGSGS increases the capacity of the signal wire (almost double), this leads to slant fronts of the signal, which is precondition for bigger jitter in consequence of the non avoidable noises in the system. In many of the top players made in the beginning of 90s the GSGSGSGS is not applied, but this doesn’t stop them to play gorgeous. For example you can check the service manual of AKAI CD93, which was referent for a quite long time. Why is that? – the audio transfer on I2S and the transfer to the DACs work synchronous, as on falling edge the info is switched, i.e. the noise common form every bus is on the same moment and does not interfere the fronts of the signals next to it, that’s why the above player and many other plays good.

Simple experiment: Harman Kardon HD970 – on the board there are two impulse PSU, but they are synchronized on Fs, if we take the synchronization the level of the noise increases with 15 to 20 dB, I know that the example is far away, but we can’t take the synchronization of the I2S protocol 🙂

uf-l plugs – they usually are used on GHz systems, there is no issue to use them in audio. Who wants to use this kind of connection could use an adapter – I think that on the forum there are some of them for selling.

... And as said two boards to solve that is not a real solution for the customer and the audiophile involved by technic !...

Ohhh …. As much audiophiles as much and the requirements 🙂 one wants all to be on one board and to be finished device, the other doesn’t like “all in one PCB board”, because they can’t combine/experiment with different nodes.

...
Can you post the instruction of how to accomodate NDKs to older boards...

Yes, of course. I’ll need some time to make a tutorial.

...
Also, can we purchase these tiny pcbs?...

The boards will be free of charge for people already has the XMOS board, there is an option the oscillators to be soldered on the boards (this also will be free of charge).

Regards,
Joro