Author Topic: Computer transport interface  (Read 72719 times)

Offline Tuyen

  • Hero Member
  • *****
  • Posts: 571
  • Liked: 221
Re: Computer transport interface
« Reply #20 on: May 23, 2014, 06:50:21 PM »
Slightly off-topic, but has any KDAC users tried hooking up the KDAC's finely tuned output stage on a different DAC board (i.e. a chip that supports hi-res/dsd input)?     If so,  is the 'magic' still there?      Or is the TDA1541 chip the crucial heart of the KDAC's signature sound?   

Is the idea of trying these new hi-res USB-I2s converters, to convert hi-res/DSD files down to 16/44 so the TDA1541 based KDACs can play?  downsampling via software/hardware?      Or is the intention to work on a hi-res supported DAC chip with the KDAC's tube output stage bolted on?

Offline Rob181

  • Jr. Member
  • **
  • Posts: 67
  • Liked: 14
  • Trust Your Instincts
Re: Computer transport interface
« Reply #21 on: May 23, 2014, 07:42:03 PM »
PC audio will surpass the KillerDAC - no question.  It's just a question of who can do it.

My money is on DSD replay being the best solution. 

T

That being the case...is it possible to "graft" a KD output stage to something like PS Audio Direct Stream DSD Dac....

Offline zenelectro

  • Hero Member
  • *****
  • Posts: 825
  • Liked: 177
Re: Computer transport interface
« Reply #22 on: May 23, 2014, 07:54:35 PM »
Hi Terry,

I am not sure I follow 100%. The picture I linked to has a Raspberry Pi connected to the DDDAC directly via i2s, no isolation.

I thought I had seen pictures of a isolation board using i2s but can't find the picture (it may have been just a USB to is2 board?).

Are you referring specifically to the KillerDAC and using an isolation board?

My lack of knowledge tends to make me view things in a simplistic manner. :o

regards

The link you posted is using WaveIO USB->I2S board. Ref picture below. It has both GMR isolated and direct I2S
outputs. The isolated OP is not re-clocked.

cheers

T


Offline zenelectro

  • Hero Member
  • *****
  • Posts: 825
  • Liked: 177
Re: Computer transport interface
« Reply #23 on: May 23, 2014, 07:58:52 PM »
That being the case...is it possible to "graft" a KD output stage to something like PS Audio Direct Stream DSD Dac....

Of course, but a PS audio direct stream DAC is a very expensive piece of kit that can be improved upon WRT performance.

You would also have to throw away most of the OP level to make it suitable for the KD triode OP stage.

There are better options than the PS audio DAC and for cheaper for DSD OP.

Stay tuned. :)

 T


Offline zenelectro

  • Hero Member
  • *****
  • Posts: 825
  • Liked: 177
Re: Computer transport interface
« Reply #24 on: May 23, 2014, 08:01:21 PM »
Now your talking but yet to hear it......................

Hey Mario,

Ultimately it will be subjective - but if you want to get closer to the sound of analog master tape, I
think DSDx128 will do it.

T


 

Offline zenelectro

  • Hero Member
  • *****
  • Posts: 825
  • Liked: 177
Re: Computer transport interface
« Reply #25 on: May 23, 2014, 08:03:36 PM »
Ok, cool.     For some reason I thought there was a restriction to an 11M master clock speed,  but you're saying it's less than 11M so it's fine.

Oz,

You are confusing master clock (mck) with bit clock. Bit clock of I2S runs at 2.8MHz.

cheers

Terry

Offline zenelectro

  • Hero Member
  • *****
  • Posts: 825
  • Liked: 177
Re: Computer transport interface
« Reply #26 on: May 23, 2014, 08:09:01 PM »
Slightly off-topic, but has any KDAC users tried hooking up the KDAC's finely tuned output stage on a different DAC board (i.e. a chip that supports hi-res/dsd input)?     If so,  is the 'magic' still there?      Or is the TDA1541 chip the crucial heart of the KDAC's signature sound?   

Is the idea of trying these new hi-res USB-I2s converters, to convert hi-res/DSD files down to 16/44 so the TDA1541 based KDACs can play?  downsampling via software/hardware?      Or is the intention to work on a hi-res supported DAC chip with the KDAC's tube output stage bolted on?

I think ultimately they are different beasts. The KD dual triode OP stage has a gain of around 30 so the IP level is pretty low for a few volts OP.

A DSD converter will have more OP level so there are more possibilities WRT analog stages. Also DSD is an inherently different beast
sound wise and will most likely require some OP stage tweaking to suit it's characteristics. There's also the RF noise to consider.

I would say leave the KD for what it does and start fresh - just gut feeling.

T

Offline Tuyen

  • Hero Member
  • *****
  • Posts: 571
  • Liked: 221
Re: Computer transport interface
« Reply #27 on: May 23, 2014, 08:33:27 PM »
Hey Terry,

Would one of these usb to i2s modules provide better performance 'out of the box' over the waveio im using ?   Or does it just have 'potential' to be better if I add external clocking or something along those lines?


Whats your chip of choice for highres/dsd?  Sabre es9018?

Hrmm.. got me thinking..  KillerDDDAC!   Hehehe

Offline kajak12

  • Hero Member
  • *****
  • Posts: 2468
  • Liked: 78
    • http://killerdac.com/forum/index.php
Re: Computer transport interface
« Reply #28 on: May 23, 2014, 08:46:42 PM »
Hey Mario,

Ultimately it will be subjective - but if you want to get closer to the sound of analog master tape, I
think DSDx128 will do it.

T
I think dsd is far from it unless you can come up with something we are phukedddd
still discovering the link between electronics and audio reproduction.so much to learn and so little time

Offline ozmillsy

  • Hero Member
  • *****
  • Posts: 2249
  • Liked: 277
Re: Computer transport interface
« Reply #29 on: May 23, 2014, 08:53:22 PM »
I think dsd is far from it unless you can come up with something we are phukedddd
He will.   All good things take time.   :)

Having to 'live' with 16bit for the time being is such sweeet sorrow.     ;D
It's all about the music,, not the equipment.

Offline Tuyen

  • Hero Member
  • *****
  • Posts: 571
  • Liked: 221
Re: Computer transport interface
« Reply #30 on: May 23, 2014, 10:01:16 PM »
The whole idea of 'computer audio' will only work if you yourself actually want it to.

If you are a person who doesn't like the idea of using computers in general or the idea of storing/accessing your music library from say an ipad,   I reckon even if the resultant sound is as good as the 'analog tape',   it still wouldnt be 'good enough'.

Another example is like vinyl rips.  When I had my turntable setup running, I thought it sounded great.   I played a copy of the 24/96 vinyl rip I downloaded and played it through my digital system and felt it sounded just as good. Possibly even better.  Yet there are vinyl advocates out there that still believe the sound of vinyl is (and will always be) more 'real' than any digital source.    What we audiophiles 'think' sounds more 'real' is more than just what comes out of the speakers. 

That's my observations from most of the people I've met in this hobby so far.


Offline dannydigital

  • Newbie
  • *
  • Posts: 28
  • Liked: 8
Re: Computer transport interface
« Reply #31 on: May 23, 2014, 10:08:43 PM »
Ok, cool.     For some reason I thought there was a restriction to an 11M master clock speed,  but you're saying it's less than 11M so it's fine.


The TDA1541 is not timed by the master clock (no master clock pin), just L/R select timing (Word), Bit Clock (Clock) and the Audio Data (Data) which together = I2S.

The Master Clock (11.2896MHz) in the Philips CD system is required to clock the decoder chip (A) and the oversampling /interpolation chip (B) which most people prefer to bypass (i.e. no oversampling). The decoder chip (A) then produces the above I2S signaling.

Offline Greg Erskine

  • Newbie
  • *
  • Posts: 48
  • Liked: 2
Re: Computer transport interface
« Reply #32 on: May 23, 2014, 10:47:08 PM »
The link you posted is using WaveIO USB->I2S board. Ref picture below. It has both GMR isolated and direct I2S
outputs. The isolated OP is not re-clocked.

cheers

T

Hi Terry,

Oh I see the confusion now. Although there is a WaveIO USB-I2S board in the picture it is not connected to the DDDAC. The computer it is plugged into is not powered up either.

The Raspberry Pi, which is underneath the WaveIO board has its I2S (P5) connected directly to the DAC.

regards

Offline dannydigital

  • Newbie
  • *
  • Posts: 28
  • Liked: 8
Re: Computer transport interface
« Reply #33 on: May 23, 2014, 10:51:52 PM »
Danny,

I'm not sure what you mean by 'regeneration' -  it is just a synchronous re-clocking arrangement.

The big issue with these boards is they require galvanic isolation, but the isolators add jitter.
As such the clock needs to be DAC side of the isolator and a/ be sent back through the isolator
b/ re clock signals coming out of the isolator.

Also not sure which isolators they are using as there are quite a few different technologies
ATM all with various trade offs. Theoretically if the reclocking is done correctly this shouldn't
matter. We will see.

They appear to have done this correctly (I think). Then it's a case of types of clocks, logic
used for reclocking, bypassing, layout etc etc.

The price is pretty amazing for what they do - so I'm getting one.

cheers,

Terry

Terry,

The device seems to be re-clocking on the DAC side, not sure if it feeds the timing back over the isolation though.
I can see that it uses an LC574A (D flop-flops) that looks like it's clocked with the oscillators on the isolated side to output the signals.
Two oscillators I can see there (isolated side), I think one for PCM stuff and the other for DSD maybe.

http://www.datasheetdir.com/LC574A+download

It will be interesting to play with the L & R DSD streams, supposedly DSD is just decoded by some LPF setup. Maybe this can be somehow discretely implemented.

D

Offline ozmillsy

  • Hero Member
  • *****
  • Posts: 2249
  • Liked: 277
Re: Computer transport interface
« Reply #34 on: May 23, 2014, 11:30:31 PM »
The Master Clock (11.2896MHz) in the Philips CD system is required to clock the decoder chip (A) and the oversampling /interpolation chip (B) which most people prefer to bypass (i.e. no oversampling). The decoder chip (A) then produces the above I2S signaling.
That clears it up, thanks Danny.
It's all about the music,, not the equipment.

Offline zenelectro

  • Hero Member
  • *****
  • Posts: 825
  • Liked: 177
Re: Computer transport interface
« Reply #35 on: May 24, 2014, 07:35:04 AM »
Terry,

The device seems to be re-clocking on the DAC side, not sure if it feeds the timing back over the isolation though.


Danny,

As you are probably aware, sending  the clock back through isolators is required for the reclocking to
work properly and everything to be synchronous. That's one thing I always look for. Also the clock
selection (44.1 / 48 multiples) hi/lo sig needs to be sent through isolators. It's all pretty complex to
do correctly.

Quote


I can see that it uses an LC574A (D flop-flops) that looks like it's clocked with the oscillators on the isolated side to output the signals.
Two oscillators I can see there (isolated side), I think one for PCM stuff and the other for DSD maybe.

http://www.datasheetdir.com/LC574A+download


One oscillator for 44.1 / 88.2 / 176.4 (and DSD) etc one for 48 / 96 / 192 etc.

Quote

It will be interesting to play with the L & R DSD streams, supposedly DSD is just decoded by some LPF setup. Maybe this can be somehow discretely implemented.


I had a look at this board and some of the feedback from users at DIYA. It seems that even though he claims
to have implemented re clocking, it is still very susceptible to cables, source etc etc. I'm guessing the
implementation is just not up to scratch.

I think best way is to implement the whole reclocking isolation scenario as a separate board. More work.

For the money it's worth having a play with though.

T



Offline zenelectro

  • Hero Member
  • *****
  • Posts: 825
  • Liked: 177
Re: Computer transport interface
« Reply #36 on: May 24, 2014, 07:49:39 AM »
Terry,

The device seems to be re-clocking on the DAC side, not sure if it feeds the timing back over the isolation though.

Looking at close up they have used Si Labs 8661BC isolator. This has 5 x 'forward' channels and 1 x 'reverse' channel.
That reverse one will be for the clock going back.

T

Offline omodo

  • Sr. Member
  • ****
  • Posts: 492
  • Liked: 104
  • Green tinted glow
Re: Computer transport interface
« Reply #37 on: May 24, 2014, 09:27:14 AM »

I had a look at this board and some of the feedback from users at DIYA. It seems that even though he claims
to have implemented re clocking, it is still very susceptible to cables, source etc etc. I'm guessing the
implementation is just not up to scratch.


IME even with Ian's FIFO (+isolator + si570 reclocker) changes to source (usb-i2s module - amenero vs waveio), software & even usb cable ($5 special vs furutech) was audible... and it's meant to be the DIY reference implementation?..

Offline zenelectro

  • Hero Member
  • *****
  • Posts: 825
  • Liked: 177
Re: Computer transport interface
« Reply #38 on: May 24, 2014, 10:18:36 AM »
IME even with Ian's FIFO (+isolator + si570 reclocker) changes to source (usb-i2s module - amenero vs waveio), software & even usb cable ($5 special vs furutech) was audible... and it's meant to be the DIY reference implementation?..

Yep I hear you.

Look - to do one of these boards -really- well would most likely require a 4 layer board
and a -lot- of experience / time to get everything obsessively right.

It's all pretty high speed with a lot going on and everything (noise) gets everywhere
very easily.

Would it be worth he time  - buggered if I know?  :) 


T

Offline dannydigital

  • Newbie
  • *
  • Posts: 28
  • Liked: 8
Re: Computer transport interface
« Reply #39 on: May 24, 2014, 02:46:35 PM »
O.k. I've got it,

We can put a black hole next to the isolator, suck all the noise out....and theen re-clock it!

Just gotta find the right black hole, vgot some datasheets here somwhere.  :P

D