[Newk Yuler]

 
Quote:

 
Now a real SPDIF upgrade would improve this further but I'm keeping that one for a possible SPDIF replacement  board that you guys might be interested in.
 
Since receiving Woods board & analysing it, I would implement the SPDIF much more optimally than the current arrangement & with some tricks added too.

What?  What???  I almost snapped my neck on that double take!
 
This new javascript-laden message composition is driving me bat-crap crazy.  It's a pain to manipulate and seems buggy as hell.  I use Firefox with NoScript and new HeadFi does not play well.  Firefox's excellent spelling feature is ruined by the composition window's simple right click menu. It's too easy to accidentally make message compositions irretrievably vanish.  This is progress?
 
Guys might want to take a good look at the way the new setup is handling avatar images.  I need to take a look at the FAQs but it seems to be resizing avatar images to 100 pixels and they can look bad if you don't already have a 100 pixel image in your profile.

 

[Bill Allen]

The Black Art is getting it to sound awesome. Proper or not.
 
You got to love us DIY guys ... always doing things for the wrong reasons.
 
Yes I did indeed cut the main ground for the Spdif input board by cutting the trace at the number 5. Woops again! When I float ... I float big! This new mod now falls into "try your own risk" or perhaps "don't try this at all". I can't say there is any logic to why it sounds better in my system but it clearly does.  Isolated Floating Uber Mod ... IFU Mod

(You can fill in your own words there) Looks like rather than floating just the Spdif input I inadvertently moved the ground for the WM8805 receiver chip, its 3.3V regulator, 12mhz clock and primary on the Pulse over to my source. In my case the Pace Car.
 
The secondary of the Pulse is still grounded directly to the DAC via a jumper trace on the board as is the 3.3V regulator for the Tenor USB chip and associated bits though a different board trace. Maybe spitting up the grounds for these two major chips is actually beneficial, strong reasoning for dedicated digital input board. (Yea you are floating something now Bill.)

 
I for one definitely vote for a simpler dedicated implementation of Spdif & USB for the Chameleon.
Bring it on Jkeny.

 

[mmerrill99]

 
Quote:

 

What?  What???  I almost snapped my neck on that double take!
 
............

Don't injure yourself - I'm just gauging what level of interest there might be first before any further work?

 
Quote:

...............
 
I for one definitely vote for a simpler dedicated implementation of Spdif & USB for the Chameleon.
Bring it on Jkeny.

What about a separate USB board & a separate SPDIF board - both can be optimised to their own operation rather than a compromised implementation trying to deal with both. Also gives the option to just choose the functionality needed! How many people are using the ASRC?

 

[skibum]

Just ordered a Ultravox.  Going to use it with my HiFace and will report how the 2 work together

 

[Newk Yuler]

Quote:

Don't injure yourself - I'm just gauging what level of interest there might be first before any further work?

What about a separate USB board & a separate SPDIF board - both can be optimized to their own operation rather than a compromised implementation trying to deal with both. Also gives the option to just choose the functionality needed! How many people are using the ASRC?

Briefly, there was a push to get Michael at Teradak to make a best possible quality but simple SPDIF input board (with no TCXO) for users with high quality sources.  Possibly made with the proven DIR9001.  He's considered it and may include it with a new model using a new jack plate but there's no schedule for it and there's been no serious development.  I personally would prefer just the optimized SPDIF board for use with a Hiface or similar.
 

 

[mmerrill99]

Quote:

Quote:
Briefly, there was a push to get Michael at Teradak to make a best possible quality but simple SPDIF input board (with no TCXO) for users with high quality sources.  Possibly made with the proven DIR9001.  He's considered it and may include it with a new model using a new jack plate but there's no schedule for it and there's been no serious development.  I personally would prefer just the optimized SPDIF board for use with a Hiface or similar.
 

I think the WM8805 is probably a better, lower jitter solution than the DIR9001. Why no TCXO? A good local XO will be far superior than one recovered from the SPDIF stream!
 
I don't know what you mean by a SPDIF for use with a Hiface - they would be completely separate boards & have no relationship to one another except how to switch between the two? 
 

 

[Newk Yuler]

Quote:

I think the WM8805 is probably a better, lower jitter solution than the DIR9001. Why no TCXO? A good local XO will be far superior than one recovered from the SPDIF stream!  
I don't know what you mean by a SPDIF for use with a Hiface - they would be completely separate boards & have no relationship to one another except how to switch between the two? 
 

You undoubtedly have a much better understanding of component relationships.  There have been brief discussions with Steve Nugent (Empirical Audio) about TCXOs and how they can jade or degrade an established very low jitter source and have been considered an issue since the Valab had one dropped into the signal path before the DAC board in April of last year.  Bill Allen could chime in on this issue.
 
If interest in Chameleon can be guaged by the number of participants in this thread in relation to the participation in the older Valab modders threads, there may presently be far fewer Chameleon modders.  Many of the core guys here are actively striving to feed the DACs by I2S (Wood with your I2S Hiface).  In fairness to you, anything you develop strictly for Chameleon may get very little attention until Chameleon becomes more popular.
 
I personally jumped ship on Chameleon and bought an Audio-gd DAC19DSP, but that's because Michael chose to not pursue the new SPDIF option.  If I knew the option existed I would take another hard look at Chameleon and I expect it would be a angle for selling them to people with a Hiface or similar.  Michael knows that too.
 

 

[teraguy]

The basic assumption is the very low jitter source input to the DAC.  In this case, the TCXO may have higher jitter than the input, but who can measure the jitter degrade because of it.  The stock TCXO is not expensive but is good enought for most consumers.  Many users use it to upgrade their consumer CD player or DAC.  However, it may not good enought for the specific users  From the wm8805 and Dir9001 datasheet, they all offer the 50 ps jitter accuracy.   Which implementation is better in general conditions?   Maybe smeone can show it.   For the specific users, they may just want the I2S input for the DAC if they have very low jitter I2S signal. They hope no other components degrade the input I2S signal.  In mrketing, it should be a different segment.
 
For the DAC design, it has 2 different scenario.  One is to reduce the jitter for the input. The other is not to introduce jitter for the cleam input.  For different scenario, it will have different implemenation. Their purposes are to prepare low jitter input for the DAC chips as well as possible.
 
Technically,any implementation for the Teradak is not an issue.  The Teradak is to make a successful product, not the best product.  Undoubtedly  the Chameleon DAC shipping volume is much lower than that of the V2.5 version (The Valab DAC).  It may be because the V2.5 DAC has much lower pirce and long history.  It can also be believed that more expensive DAC will be shipped much less.
 
Michael is still thinking how to make a successful and better product.  Cuurently, the following items are in consideration for the versatile input. 
 
   1) The asynchronous 192k USB input or normal 96k adaptive USB input
   2) Dir9001 input without TCXO 
   3. Wm8805 + TCXO (or OCXO)
   4) Direct I2S input
 
The Teradak has all of the solutions.  What they concern is the price and the shipping volume.  Maybe you guys have some idea for it.
 
 
 
 
 
 
 
Quote:

Quote:
You undoubtedly have a much better understanding of component relationships.  There have been brief discussions with Steve Nugent (Empirical Audio) about TCXOs and how they can jade or degrade an established very low jitter source and have been considered an issue since the Valab had one dropped into the signal path before the DAC board in April of last year.  Bill Allen could chime in on this issue.
 
If interest in Chameleon can be guaged by the number of participants in this thread in relation to the participation in the older Valab modders threads, there may presently be far fewer Chameleon modders.  Many of the core guys here are actively striving to feed the DACs by I2S (Wood with your I2S Hiface).  In fairness to you, anything you develop strictly for Chameleon may get very little attention until Chameleon becomes more popular.
 
I personally jumped ship on Chameleon and bought an Audio-gd DAC19DSP, but that's because Michael chose to not pursue the new SPDIF option.  If I knew the option existed I would take another hard look at Chameleon and I expect it would be a angle for selling them to people with a Hiface or similar.  Michael knows that too.
 

 

[Wood]

Yes I agree, the Valab at $200, with a few cap swapping mods that a hobbyist like me can handle, is a true bargin, and so its no wonder it sells like hot cakes. The Chameleon is a better dac still, and invites simple mods. But at twice the price of the Valab, launches itself against  a very competitive, saturated, market.. I suspect its a lot better then a lot of the competition at its price point, but its has to be more then that to enjoy the Valabs success. Redesigning digital boards is beyond me, but this is what is requried to make this dac a true super bargin. The higher resolution of the dac board demands a higher quality input.  Hopefully we can persuade jkeny to stick around, and get the input board sorted out. Then I can be left in peace to do the real work of swapping coupling caps

.

 

[teraguy]

The Hiface uses local crystals to asynchronously ask the audio data from the USB and then buffer them.  The audio data are then packed to I2S format.  The I2S signal and the local crystals are used for the DIT4192.  The DIT4192 then converts the input I2S to SPDIF based on the local crystals (MCK input). The output SPDIF quality is highly dependent on the local crystals.  The asynchronous mode is possible to make low jitter output.  But it should be conditioned on the accuracy of local crystals.  Is the derivation correct?

Quote:

I think the WM8805 is probably a better, lower jitter solution than the DIR9001. Why no TCXO? A good local XO will be far superior than one recovered from the SPDIF stream!
 
I don't know what you mean by a SPDIF for use with a Hiface - they would be completely separate boards & have no relationship to one another except how to switch between the two? 
 

 

[teraguy]

Maybe in the future, there will be OPT, tube outputs or active ouputs.   However, they are not new idea for the DAC.

 
Quote:

Yes I agree, the Valab at $200, with a few cap swapping mods that a hobbyist like me can handle, is a true bargin, and so its no wonder it sells like hot cakes. The Chameleon is a better dac still, and invites simple mods. But at twice the price of the Valab, launches itself against  a very competitive, saturated, market.. I suspect its a lot better then a lot of the competition at its price point, but its has to be more then that to enjoy the Valabs success. Redesigning digital boards is beyond me, but this is what is requried to make this dac a true super bargin. The higher resolution of the dac board demands a higher quality input.  Hopefully we can persuade jkeny to stick around, and get the input board sorted out. Then I can be left in peace to do the real work of swapping coupling caps

.

 

[Bill Allen]

All interesting and valid points. However the pre-production Chameleon DAC currently sits between a rock and a hard place. In stock form it is an upgraded 16 chips Valab with 3 inputs, outboard power supply, 96/24 input capability, and a simple reclocker module that actually works. The Chameleon DAC at $400 is an fantastic value, swap in some high quality coupling caps and leave it alone. It sounds very good.
 
The current issue is most people on this forum are former Valab modders and we understand that "true" NOS DAC's can sound really outstanding, easily outperforming high dollar upsampling DAC's. However the limiting point with NOS DAC's has always been how to feed them ultra low jitter, be it source, interface, or transmission induced jitter. There is a lot of theory out there on how to best deal with jitter but very few solutions that actually work. Most cost a lot of money. However I feel the Chameleon is already utilizing the easiest and lowest cost jitter reduction solution, reclock just before the DAC chips. The best overall bang for the buck will come from upgrading this 24.576 clock. I think this is where TeraDak should focus their current resources.
 
What is the future:  The Chameleon DAC is not a Chameleon just because we mod it,  there should be different input and re-clocking boards available to customize it. This is how it will stand out from the competition and how it can continue to evolve. The customer can choose when he orders or easily upgrade his Chameleon when he is ready.
 
USB/Tolink/Coax input module (standard)
Dedicated Spdif input module.
Dedicated USB input module.
Dedicated I2S input module.
Upgraded Reclocking module.
 
On another front the Chameleon DAC can also help do marketing and research for future TeraDak products. So from my vantage point the Chameleon DAC is already a great product, with upgraded dedicated modules it can provide both outstanding sound and value. (Not to mention make it easier for us DIY types to modify.)

 

[mmerrill99]

Quote:

Quote:
You undoubtedly have a much better understanding of component relationships.  There have been brief discussions with Steve Nugent (Empirical Audio) about TCXOs and how they can jade or degrade an established very low jitter source and have been considered an issue since the Valab had one dropped into the signal path before the DAC board in April of last year.  Bill Allen could chime in on this issue.
 
If interest in Chameleon can be guaged by the number of participants in this thread in relation to the participation in the older Valab modders threads, there may presently be far fewer Chameleon modders.  Many of the core guys here are actively striving to feed the DACs by I2S (Wood with your I2S Hiface).  In fairness to you, anything you develop strictly for Chameleon may get very little attention until Chameleon becomes more popular.
 
I personally jumped ship on Chameleon and bought an Audio-gd DAC19DSP, but that's because Michael chose to not pursue the new SPDIF option.  If I knew the option existed I would take another hard look at Chameleon and I expect it would be a angle for selling them to people with a Hiface or similar.  Michael knows that too.
 

There is no doubt that a good, low jitter, well implemented local clock is a better option than a recovered clock from SPDI, really. This is not just my opinion - ask experts, they will say the same. I'm not sure what the issue with the "jaded" clock was but I bet it was implementation. One of the problems is the PLL needed to recover the clock - it is to be avoided if possible - even good chips like WM8805 & DIR9001 - it introduces jitter. Second problem is you are relying on excellent 75ohm cable & connectors along with proper termination to get the low jitter from the source to the SPDIF receiver for recovery - I think you have an idea of how hard this is. A well implemented local clock has fewer issues to deal with & they are pretty much under the control of the designer.
 
Thanks for the heads up on the interest & numbers here!
 
You wont go far wrong with a Audio-gd DAC BTW!

 

[skibum]

M2Tech USB to I2S Module
 
Found this over on diyAudiuo forums. 
 
http://www.diyaudio.com/forums/group-buys/166354-hiface-usb-i2s-24bit-192khz.html
 
I am not sure if it will give a 5 volt I2S output that we need.  I have asked audiodesign.
 
Here is the data sheet:
 
http://www.audiodesignguide.com/M2TOEM-01USB-to-I2SPrBNov-08-2009.pdf
 
looks pretty good for $50!
 
Update - the datasheet is no longer up there.  But it appears you can use a 5 volt buffer (like a SN74HCT125)
to level shift the 3.3 volt I2S to 5 volts...

 

[Newk Yuler]

Quote:

There is no doubt that a good, low jitter, well implemented local clock is a better option than a recovered clock from SPDI, really. This is not just my opinion - ask experts, they will say the same. I'm not sure what the issue with the "jaded" clock was but I bet it was implementation. One of the problems is the PLL needed to recover the clock - it is to be avoided if possible - even good chips like WM8805 & DIR9001 - it introduces jitter. Second problem is you are relying on excellent 75ohm cable & connectors along with proper termination to get the low jitter from the source to the SPDIF receiver for recovery - I think you have an idea of how hard this is. A well implemented local clock has fewer issues to deal with & they are pretty much under the control of the designer.  
You wont go far wrong with a Audio-gd DAC BTW!

I was just surfing information and ended up looking at the Audio-gd FUN.  It's made to use a DIR9001 plug-in module as an input option.  I noticed the 24 MHz clock on the board, then curious I looked for an image of the DAC19 without the DSP board where the DIR9001 is normally hidden.  It has a similar clock similarly laid out to the module.  So Kingwa, this brilliant Chinese engineer apparently agrees that a local clock is the way to go.  And I went that way in spite of Chameleon without considering it.  So I'm having a good laugh at myself right now.

 
How about this Audio-gd DIR9001 module?  Could it be a Chameleon modder's answer to the SPDIF simple input board?  He's selling them as an option to the FUN for $22.50 USD.  Notice the pins 13 and 14 option appears not used.  Does that board look well designed to you?
 
If it can be related to Chameleon, note Kingwa doesn't use a pulse transformer on SPDIF, explaining his perspective specifically for the DAC19:
 
The DAC19 input without transformers to isolated. Because most standard source coaxial output and AES output have transformers for isolated and setup the output impedance. So don't need the DAC has the input transformers. And if the source has transformers , the DAC has transformer is unuseful, will degrade the sound unless the design or owner want more musical but slight less detail. Some few source without transformers, in this station, add a transformer at DAC19 maybe can slight improve. But if add the transformer at the source will be better, it can setup the output impedance at 75 ohm.
 
If I understand that translation correctly, Kingwa believes most SPDIF sources are already isolated and doubling it at the DAC will degrade the signal, so he goes with what he believes will be a better common match.  Sounds familiar.  Quote credit to slim.a in his DAC19DSP/C2 review thread.
 
I was using an Off Ramp that provides galvanic isolation from the source into my Valab with a pulse transformer.  There was a noticeable improvement after the pulse transformer was installed.  I bought another from Digikey when I ordered the DAC19.  After burn-in I'm going to put it behind the DAC19's coax input and see how it goes.  It's easy to undo.
 
Didn't someone say the Hiface provides galvanic isolation?