XMOS-based Asynchronous USB to I2S interface - Page 32 - diyAudio
Go Back   Home > Forums > Source & Line > Digital Source

Digital Source Digital Players and Recorders: CD , SACD , Tape, Memory Card, etc.

Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 19th January 2012, 11:40 AM   #311
marce is offline marce  United Kingdom
diyAudio Member
 
Join Date: Jun 2007
Location: Blackburn, Lancs
Barrows
Quote:
as noted, I use u.fl. But there are coax alternatives which do not make the right angle turn. I am curious what experienced RF design engineers might have to say about this. I have heard that pcb board traces are often purposely designed with gentle curves rather than hard right angles for high speed signals. How much difference these choices might make are fun to speculate about, but without some serious measurements, it really is just all speculation.
90 degree angles are best avoided, 45 degree angles work perfectly, you dont need to do curved traces, it wont make any noticable difference and will make layout a complete pain.
RF PCB design is totaly different from high speed digital design.
We use a 13GHz scope to examine our high speed lines (DDR interfaces, ethernet etc).
The most important thing with high speed digital layout is keeping the impedances equal (around 50ohm single ended, 100 diff pair) and ensuring signal integrity, the best way to achieve that is to use SIV (signal integrity verification software) works with IBIS data available from most chip manufacturers. And lots of ground planes for return currents, and preferably all lines done as stripline routing (especially the higher speed ones) if possible.
  Reply With Quote
Old 19th January 2012, 01:58 PM   #312
diyAudio Member
 
Join Date: Oct 2011
Quote:
Originally Posted by rsdio View Post
Ok, fair points.

I just wanted to explain my point that it makes more sense to design a reasonably high-quality DIY interface rather than something cheap.

I have designed many custom boards, and they're all expensive to make in small quantities. That's true whether the designs are cut-throat cheap or super high-end. Thus, my philosophy is that you might as well make DIY boards as high-quality as humanly possible. They're going to cost about the same anyway. Besides, cheap designs are churned out all the time by mass-market electronics shops, so you might as well let someone else spend the time making cheap stuff and focus your efforts on something a little better than average.

Where DIY folks tend to draw the line is with the number of layers and the cost of the parts. A few DIY designers try to stick to 2-layer boards because they're so much cheaper. Personally, I've done so many 4-layer designs with solid power planes that I'm not turning back to 2-layer unless the design clearly would not benefit (*). On the subject of exotic parts, I tend to like designing for multi-site and optional SMD placements. That way, budget-conscious DIYers can leave off the costly parts and still have a working design............

Long story short: I fully support efforts by the contributors to this thread to put as much design excellence as possible into this project.

...

You know what....... excellent point! I guess I am just afraid to go all out on first try only to strike out with a nonworking design and hundreds of $ in wasted money.... but yeah PCB's aint cheap :/

So yeah.... what kind of clock does OP use for his design? I mean part number. Any PLL clock in the future?

Yeah it is a good idea to leave empty slots for higher end components, it will be like having two designs in one only that it will be much cheaper just to get it working with the basic parts and than add on

Quote:
Originally Posted by Wolfsin View Post
When you noted that SACD and HDCD did not sound much different it occurred to me that you might have some pretty good equipment. That is the case IMO. I think some of your units probably have pretty good clocks. On bad equipment the difference between SACD and CD is glaringly apparent. The other possibility is that one part of the chain is bad enough that you cannot appreciate what you have.

I cannot think of a $400 DAC that offers "everything and more". What I want to do is ripBitz to an HDD and then squirt them into a DAC with timing as good as SACD media bitz get squirted to the built-in DACs in a couple of your components. That is a lot harder than it might seem on the surface.

Internally your components communicate using i2s. Getting a computer to emit i2s as a well clocked parallel 'stream' is the key. That is what Lucian's board can do because of an improved USB service class.
yes i do have an HDCD that i bought by mistake lol and it is the best sounding out of all the CD's I have but when i put it in a non HDCD player i hear the difference but the SACD and the HDCD...not so much and some CDs I have sound awesome...


Oh yeah it's not that simple at all but I'm getting the block diagrams organized in my head first and see how everything flows.

Lorien's WaveIO removes the clock inteference from the USB and converts to I2S, right? He achieves that via the XMOS asynchronous mode and running the XMOS as a master with it's own stable clock....

I read somewhere that they buffer the USB input and further remove any jitter or signal interference.

What the hell are GMRs?!? anyone have a part number for me or a datasheet? I cant find them on digikey... something to do with magnetic isoltion?

Ground isolation, any done on WaveIO?
__________________
This is not reality...
  Reply With Quote
Old 19th January 2012, 02:36 PM   #313
diyAudio Member
 
Join Date: Nov 2007
Huh a lot to catch up...
@ zxgravediggerxz: There are a lot of manufacturers that are using XMOS chips in their designs. I just wanted for WaveIO to be a good alternative to
stock board provided by XMOS at the same price. I do confess that at the beginning I didn't took into account the fees paid for money transfer, shipping. so I am where I am with the price. And, like rsdio I don't see me going back to 2-layer boards even if I wanted to...
My experience with XMOS firmware is.. rather limited even if I have more than 10 years experience in uC programming. The firmware is.. how do I call it... fragile. Any important changes into it has caused BSoDs to my laptop so I tried to keep it as much as possible close to the original form.
On the other hand, there were no one for me to give any advices on how do I have to do with XMOS to reach my goal other than XMOS itself. All your questions will find the answers there, just make an account and ask.
Your goal in essence is noble but going cheaper will not bring you performance unless you are really know what you're doing and this task is rather a time consuming one (at least in my perspective). I also have this in mind... but in my case it will show-up in a different form. I'm in process of designing it, hoping it will work at the end

@ audiodesign: I have the new batch of boards at disposal, ready for assembling. I need some time though for few old orders to catch up first.

@ marce: nice 'things' you're playing with

Cheers,
L

Last edited by Lorien; 19th January 2012 at 02:38 PM. Reason: typo
  Reply With Quote
Old 19th January 2012, 03:12 PM   #314
marce is offline marce  United Kingdom
diyAudio Member
 
Join Date: Jun 2007
Location: Blackburn, Lancs
Lorien, I do the PCB's and basic simulation, we found we had to start using simulation, some clocks and data rates are just getting silly, and using rules of thumb just dosn't work. Then we found that our normal scopes and more imortatly scope probes were not up to it, so we hire the 13GHz scope when we do a high speed interface to check the real world waveforms with the simulations. One caveat, you have to model the loading of the scope probe when doing simulations, as at the higher frequencys they load the line and effect the results. This also means that what you see on a scope is not always reality, due to the probe loading. There are numerous data sheets on Tektonix and Agilent sites, that cover this problem.
  Reply With Quote
Old 19th January 2012, 06:03 PM   #315
diyAudio Member
 
Join Date: Nov 2007
@ marce: I have one question for you that squeezes my brain for some time now: it's about switching master clock signals for WaveIO accordingly to the incoming sample rate: I have some ideas to use RF relays instead of any mux chip alternative. Doing this, there will be any "visible" degradation in signal's integrity using RF relays? Signals will have frequencies between 45 and 100 MHz.
Thank you,
L
  Reply With Quote
Old 19th January 2012, 07:15 PM   #316
marce is offline marce  United Kingdom
diyAudio Member
 
Join Date: Jun 2007
Location: Blackburn, Lancs
That is one question I'd have to pass on, I do the PCB's, I probably spend hours with data sheets and in meetings with the whole design time discussing in detail every part of the circuit, making numerous notes and references to look up. As I dont know the ineterface I cant coment. Some more specific links would be interesting.
As to clocks, what I can say is there are numerous clock drivers apearing, designed with the right drive for todays industry standard high speed boards, where 50 ohms impedance is quite easilywith 12+ layers quite often being the norm. Boards are becoming that complex and packaging that tight, that designs are done in a virtual world, with 3D modeling of the boards and components (IDF interface to mechanical CAD), so the board fits in the enclosure..and the electronic design being simulated and tested, its not uncommon to take two weeks routing a complex clock round a board.
Sorry, can only comment on stuff I've realy worked with or general high speed and complex layout problems.
Cheers Chaps.
Marc

Last edited by marce; 19th January 2012 at 07:29 PM.
  Reply With Quote
Old 21st January 2012, 05:34 PM   #317
diyAudio Member
 
Join Date: Sep 2005
Send a message via MSN to audiodesign Send a message via Skype™ to audiodesign
Lorien,
please assemble only board for me and tell me the cost.

Today I have compared the M2Tech with OpenSource Qnktc and there are many differences changing only the USB->I2S module so I need to test soon also the XMOS to decide which is better.
  Reply With Quote
Old 22nd January 2012, 10:56 AM   #318
diyAudio Member
 
Join Date: Nov 2007
@ audiodesign: you have PM. I'll do as you ask, starting from tomorrow.
Kind regards,
L
  Reply With Quote
Old 22nd January 2012, 12:31 PM   #319
regal is offline regal  United States
diyAudio Member
 
Join Date: Jan 2004
Location: MD
Quote:
Originally Posted by zxgravediggerxz View Post
You know what....... excellent point! I guess I am just afraid to go all out on first try only to strike out with a nonworking design and hundreds of $ in wasted money.... but yeah PCB's aint cheap :/

So yeah.... what kind of clock does OP use for his design? I mean part number. Any PLL clock in the future?

Yeah it is a good idea to leave empty slots for higher end components, it will be like having two designs in one only that it will be much cheaper just to get it working with the basic parts and than add on



yes i do have an HDCD that i bought by mistake lol and it is the best sounding out of all the CD's I have but when i put it in a non HDCD player i hear the difference but the SACD and the HDCD...not so much and some CDs I have sound awesome...


Oh yeah it's not that simple at all but I'm getting the block diagrams organized in my head first and see how everything flows.

Lorien's WaveIO removes the clock inteference from the USB and converts to I2S, right? He achieves that via the XMOS asynchronous mode and running the XMOS as a master with it's own stable clock....

I read somewhere that they buffer the USB input and further remove any jitter or signal interference.

What the hell are GMRs?!? anyone have a part number for me or a datasheet? I cant find them on digikey... something to do with magnetic isoltion?

Ground isolation, any done on WaveIO?
I think the GMRs are the GMR's most are using. Typically they add debatably as much jitter as a really good spdif. Just see Art's detailed measurements on audio circle if you question that, I don't mean to debate it in Loriens thread, his product looks to be the best design for the diyer at this time and a 3ps rms jitter spdif from a computer is expensive and arguable not woth the price..

However there is a means proposed to get jitter down to the single digit level with this unit. You use the GMR's on the i2s but put the master clock in the external DAC with a final synch alighnment right at the dac entry. The clock then has its gmr feeding the xmos in the opposite direction. This approach is about as good as it is going get as the remaining jitter is going to be from your clock quality, power, and DAC chip. The issue is you need to be stuck with rbcd and its mutliples or 48k and its mutliples unless you have some means to switch clocks based on sample rates which a tough task for the hobbiest. Personally I find hirez is so esoteric and often fraudulent, I will be satisfied with a 44.1 based clock at the dac and simply resample the hirez material to 88.2 or 176.4. But you could do it the other way. Whichever you chose includingjust using the gmrs in original original config will put you miles ahead of anything out there for the price.

I am also tempted to try forgoing the GMRs, use quality power and share the ground up to the dac output with galvanic isolation to the amp provided by classic line livel transformers.

Final option is a good sych alighnemt of the i2s and send it to the dac via a
dit4192 and pulse transformer as the Legato does, Still miles ahead of most transports and sounds amazing.


Any way you look we are lucky to have a card like this Wave I/o for our hobby/projects. My only concern is M$'s refusal to give a native usb2.0 Ausio driver, so we are at the mercy of a third party driver. Linux and Mac users don't have this risk.


BTW your HDCD sounds good because it was transfered to digital with an R2R ADC

As far as SACD, does this really require a DAC? I mean what is the dac doing to a 1 bit signal? One guy on here has DSD playing right to his speakers with no DAC involved, we are probably a long ways from really unlocking the right way to handle DSD without the corporate BS. You should be aware there usb device's that spit out a dsd stream but for I2s they aren't in the same league as the WaveIO which I believe is the only computer transport device which allows the Sabre PLL to tun at its tightest level (indicating lowest incoming jitter.)
  Reply With Quote
Old 22nd January 2012, 05:49 PM   #320
adelias is offline adelias  Greece
diyAudio Member
 
Join Date: Feb 2005
Location: No Pasaran
Have any changes been made to the new pcb boards or are they identical to those on the original WaveIO?
  Reply With Quote

Reply


Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
exaU2I - Multi-Channel Asynchronous USB to I2S Interface exa065 exaDevices 1357 3rd March 2014 08:51 PM
Introducing miniStreamer: Native 24/96 USB to I2S / SPDIF interface minidsp miniDSP 39 5th January 2014 11:00 PM
Ultimate USB to I2S interface sampler Digital Source 206 30th January 2012 03:45 PM
Is it possible to develop a ASIO driver for PCM2900 based USB Audio interface? cxhawk Digital Source 7 3rd December 2010 02:30 PM
interface I2S with USB mermoz Digital Source 0 21st February 2003 10:34 AM


New To Site? Need Help?

All times are GMT. The time now is 01:20 PM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright ©1999-2014 diyAudio

Content Relevant URLs by vBSEO 3.3.2