Author Topic: DDDAC 1794-NOS 24-192  (Read 166532 times)

Offline hedalfa

  • Hero Member
  • *****
  • Posts: 1540
  • Liked: 326
Re: DDDAC 1794-NOS 24-192
« Reply #80 on: November 28, 2012, 08:50:09 PM »
Andy, what happened with the BLH boxes?  Have you tried them yet?
Thanks for giving us the boxes.

I would have been happy for andy to have them, but yes I can imagine the reaction from her in doors.

I will get around to trying them out for fun. If they were to be made usable they would need a lot of work. I wonder how they would go with goodmans 18 inch drivers, could plenty of mojo, but the box is likely to resonate a lot without better bracing etc.

Cheers

tuyen

  • Guest
Re: DDDAC 1794-NOS 24-192
« Reply #81 on: November 29, 2012, 07:02:46 PM »
Link to discussion thread on Doede's new designed DAC on DIYAudio forums:

http://www.diyaudio.com/forums/digital-line-level/224108-nos-192-24-dac-pcm1794-waveio-usb-input.html

Offline kajak12

  • Hero Member
  • *****
  • Posts: 2444
  • Liked: 66
    • http://killerdac.com/forum/index.php
Re: DDDAC 1794-NOS 24-192
« Reply #82 on: November 29, 2012, 08:10:53 PM »
Link to discussion thread on Doede's new designed DAC on DIYAudio forums:

http://www.diyaudio.com/forums/digital-line-level/224108-nos-192-24-dac-pcm1794-waveio-usb-input.html
Great read tuyen and congrats to doede's work having a go is better then putting $$$ into commercial pockets,looking forward to hearing your dac at my place with computer audio you never know i might be in for a treat...........
still discovering the link between electronics and audio reproduction.so much to learn and so little time

Offline springcreek

  • Sr. Member
  • ****
  • Posts: 475
  • Liked: 74
Re: DDDAC 1794-NOS 24-192
« Reply #83 on: November 29, 2012, 09:09:31 PM »
Hi Andy,

Most welcome if you ever stop by WA.   Doede's designs are excellent.  Well designed, fun to build, not silly priced, reference-level performance in my opinion.  The 120 chip paralleled TDA1543 DDDAC1543MK2 with the USB module was my fav DAC for long time.    This new hi-res usb dac he has designed plays on same level for 16/44 files. Hi-res recordings add that extra layer of fidelity that is quite enjoyable.

From my own testing, I felt USB via pc transport has already surpassed the performance of disc spinners long time ago. Hence why I decided to move on all my disc spinners.  Isn't it funny how there seem to be 2 separate groups. One who has moved to computer based transport because they feel it it brings sonic improvement.  The other group finding the disc spinner still sounds best and computer transport still has long way to go...     both groups listening to same music.    which group is deaf? or should i say, which group is more deaf then the other? :)

Me personally, I think I'm lucky to be in the first group. As using computer transport also includes the much-important convenience factor, without having to feel that I am compromising on sound quality :)
 
Re the bass horns: Nearly done. Probably 1 weekend of work/beer to go before we be rocking.   Indeed,  nothing like bass notes between 50-300hz through a front loaded horn hitting ya ears and body. The speed and scale of transients are pretty crazy.   Reinhard was right all-long about bass horns being the final frontier with regards to 'doing' horns.

Thanks Tuyen

Great news that computer audio is delivering the goods...I was hoping that was the case. I may have more of a chat to you soon about progressing this.

Also looking forward to hearing how the 50hz horns turn out. I would think PM2A Lowther or Field Coil Lowther in one of Martins 204 hz front horns and a 50hz bass could be very special...hopefully I will get to hear it. I also see that Martin can configure the 204hz horn for an Axiom 80 which would be very nice as well. Seems like a really nice guy.

Good luck with the new DAC and I look forward to reading about its developement as well as the bass horns.

Cheers, Andy  ;D

Offline rhlauranna

  • Hero Member
  • *****
  • Posts: 1197
  • Liked: 153
Re: DDDAC 1794-NOS 24-192
« Reply #84 on: November 29, 2012, 10:07:26 PM »
aaaahhhh, yesterday evening we had one of those sound checks, and just right in time before the great  "japanese session", presumably everybody is dreaming of...  oh, what fun and excitement !!! I remember having had this last time when I was young - and when it was just before Christmas...

...late in the afternoon my very unpatiently awaited 2nd, 3rd and 4th deck of Doede's new DDDAC1794 NOS DAC arrived, and what did I do? No, I didn't check it with my "radio", I immediately sat in my car and drove to Klaus. Why? Because there for rather a short time Doede's personal DAC with the Sowters is running for final testing purposes before our session next week...

and so for the second time we had the possibility to listen to this wonder very comfortably not only for a single track, or two, or three, and not in combination with so many other necessities to try out, but singularly and really intensively, and with all sorts of music that we like  a n d  at all loudness volumes we personally enjoy with different tracks and sources (and not only dreaming of how it would sound if we would have the right level - we are on stage)... this was simply mindblowing realistic...

as I told already, for me, this combo leads to HiFi-heaven, and this time we did no longer oscillate on the border, we were clearly in...

it was no question:

1. the multiple boards very independently from other tweakings lead to the generally desired positive effects (reduced jitter...) that we already are aware of the DDDAC 1543 and

2. the different controlled power supplies from Doede and Bernd in direct comparison showed the necessity of implementing one at all... yes, there were slightly audible differences, but not of an excluding sort, one can very sublimely "live" with both, no matter, in the end the differences here are audible but rather "neglectable" as a whole...

3. the biggest surprise was not that what we already knew and what we had in our minds and what we have been listening to before and what now very impressively was confirmed, the real icing on the cake this time was something completely different, something that we discovered by accident, and for me this is some sort of voodoo because it is totally within the digital area where only 0's and 1's reign...

perhaps you might remember that a year ago or even before I implemented the silver USB Kimber cable in my system ? Well this cable was heavily recommended by Chris Connecker over at www.computeraudiophile.com ... normally I am very hesitating in following such recommendations, but as there were so many DIYers responding about dramatic positive effects in all their whatsoever different systems (remembering me of the saying when judging a Grateful Dead show: "100.000 dead heads can't be wrong", I finally gave it a try and - was not disappointed... there was contrarily so much more: once listened to it, there was no way back...

this cable being implemented within my system. at that time into the DDDAC1543 DAC, brought an amelioration in sound that was simply incredible and must be heard to be believed - but - and here comes the voodoo - and what was totally not understandable for us - brought zero effect within Klaus's system, we had no explanation for it, but it was exactly like that...

and just yesterday - out of nothing - while talking on the telephone, Klaus asked me to bring this cable to him, and I was thinking to myself, that this would be of no use, but as it was Klaus asking, I did it of course... and after Klaus having implemented the cable in his system all of a sudden the same positive effects were immediately so clearly listenable as in mine. Klaus, who only very heavily gets really excited, only looked at me with that very rare smile in his face I like the most from him: satisfied down to the bone!!!

to make it short: in combination to all those wonderful ameliorations that we are experiencing with the DDDAC1794 this tiny little cable is highly effective, working wonders on to the sound, it simply puts away two thick curtains and adds an extra very welcome dose of openness, pulling the whole sound one meter more in front of the system, like a precise but smooth fortissimo coming out of nothing, and all in the intended direction.... normally for such an amelioration one has to pay lots and lots of money, and this time, if I have it right in mind, only some 65 $ !!!

well. with all fun that we had, what did we learn: we always have to try out, even if we are absolutely convinced that we already know... it seems we do not...
« Last Edit: November 30, 2012, 04:05:02 AM by rhlauranna »

tuyen

  • Guest
Re: DDDAC 1794-NOS 24-192
« Reply #85 on: November 30, 2012, 09:05:18 AM »
All good guys  :D

Nice stuff Reinhard.  I using a 'special' pure silver USB cable too. I think I could hear a slight improvement over the standard USB cable, but it could of been because the silver cable length is only 0.5m while the standard cable was 1m.   

Did you/Klaus keep the length of the 2 cables consistent?   I wonder which factor has more influence on the effects you noticed. The length, the shielding between the wires inside(one is carrying 5v power and others data) or the material that the wire is made of.

We would assume, the best USB cable, is no USB cable. Right?  So I wonder if there will most improvement if we could solder from the motherboard of the computer a 2-4cm length of wire to the USB module. Bypassing the standard USB cable and sockets may yield good benefits?

I remember seeing in the past someone doing exactly this with the old dddac1543. They built an all in one music server solution.

Maybe something to think about...

Offline ozmillsy

  • Hero Member
  • *****
  • Posts: 2163
  • Liked: 245
Re: DDDAC 1794-NOS 24-192
« Reply #86 on: November 30, 2012, 09:43:46 PM »
JKeny recommends a usb2usb connector, instead of USB cables (for his USB converter).

Extremely short 2" hard connection.   Makes it all a bit inconvenient placement wise.   But swears it makes an audible improvement.
It's all about the music,, not the equipment.

Offline zenelectro

  • Hero Member
  • *****
  • Posts: 776
  • Liked: 144
Re: DDDAC 1794-NOS 24-192
« Reply #87 on: November 30, 2012, 10:34:34 PM »
I didn't know where to put this - so here it is.

If any of you guys can sit through it, it is very good, especially around the middle.
One of the designers at ESS (Sabre) talks about SD dacs versus multibit, noise shaping, etc
Most interestingly your ability to hear all these things and why.

They certainly put a lot of work into the Sabre.

I was put on to it by abraxalito of diyaudo.


<a href="http://www.youtube.com/watch?v=1CkyrDIGzOE" target="_blank">http://www.youtube.com/watch?v=1CkyrDIGzOE</a>








Offline ozmillsy

  • Hero Member
  • *****
  • Posts: 2163
  • Liked: 245
Re: DDDAC 1794-NOS 24-192
« Reply #88 on: December 01, 2012, 02:08:43 AM »
F.a.r.k. Me.    I was having a restless sleep, and instead of turning on the tv and falling asleep, I made the mistake of checking in, and clicking play on this clip.   :-\

Utterly fascinating, thanks for sharing that Zen. Probably the most thought provoking YouTube clip I've ever seen.  I immediately want to watch it again, but I can't, otherwise I'd be up all night.

Now I know why you were inspired to try the Sabre.  

I come back to the concept of bypassing the filtering just doesn't seem right, if using a sigma delta Dac.   Well, it's flat out wrong, according to this clip.    Is that what you're doing Doede?   Are you really bypassing the filtering?  Or have I got it wrong?    Are you bypassing oversampling only, but there is still noise shaping and dithering filters in another part of the circuit?     But hang on, it's a sigma delta chip,   So by definition,  it *has* to be resampling the 24bit input signal.     Hmmm.      ???

I think I'm starting to realise why a tda1541a in nos mode with tube outputs works well.   There are short comings that the (now dated) oversampling tries to overcome (likely to be primarily HF nasties),  but the oversampling introduces it's own kind of issues (veil) that wasn't there before.    Remove the oversampling step, removes the veil,  but what about the inherent nasties that are still there?    The tubes on the output stage filters them out.  [light bulb moment] Could be barking up the wrong tree, and joining the wrong dots,,,,, not sure.

Coming back to sigma delta.   If the Sabre really does have superior filters to all the other sigma delta chips on the market,  maybe tube outputs aren't really needed?   They might still add that euphonic romance,  but maybe the inherent HF filtering provided by the tubes isn't needed?

Where does that leave us?   Scratching our heads, trying to figure out the right settings to use, on the myriad of programmable settings on the Sabre.    It could be all too hard Zen,  but I get the feeling the guys at Sabre have figured this all out, and they don't publish it.    

Ps: loved the references to the various 'customers' having the 'audio dude' who does the listening test, and gives the final tick in the box.   Very good.   :D
« Last Edit: December 01, 2012, 11:06:09 AM by ozmillsy »
It's all about the music,, not the equipment.

Offline ozmillsy

  • Hero Member
  • *****
  • Posts: 2163
  • Liked: 245
Re: DDDAC 1794-NOS 24-192
« Reply #89 on: December 01, 2012, 02:49:28 AM »


If there is a way to bypass that oversampling module, and get straight to the dac modulator,  then it's undocumented.  

If that modulator handles the re-sampling from 16/24bit into whatever multibit wordlength the chip converts at (?), then maybe the modulator also has its own noise-shaping/dithering algorithms?  

Doede, chime in here, and tell us what you know the 1794 chip does in each of these modules,  so I can understand the approach better.
It's all about the music,, not the equipment.

Offline zenelectro

  • Hero Member
  • *****
  • Posts: 776
  • Liked: 144
Re: DDDAC 1794-NOS 24-192
« Reply #90 on: December 01, 2012, 11:43:25 AM »


If there is a way to bypass that oversampling module, and get straight to the dac modulator,  then it's undocumented.  

If that modulator handles the re-sampling from 16/24bit into whatever multibit wordlength the chip converts at (?), then maybe the modulator also has its own noise-shaping/dithering algorithms?  

Doede, chime in here, and tell us what you know the 1794 chip does in each of these modules,  so I can understand the approach better.

Oz,

The important thing to understand is the 8 x OS filter and the modulator are separate things.

The 8 x OS filter has it's own low pass brick wall filter, and the SD modulator has it's internal, in the Sabres case, 5th order
LPF. The modulators filter is working much higher and as you can see from the vid in a different way.

In both Sabre and 1794 you can disable the OS filter and just use PCM straight into the modulator - I think.

However I'm not sure of the real consequences of running a DS DAC at say 44.1 or even 96 / 192 without
upsampling to say 352kHz. There must be some impact WRT the OOB noise.

That's why I asked the question WRT wide band plots of 1794 running 'zero OS'.

I'll just have to do all this myself in the usual zen - no BS fashion, this is what it does, this what I
understand, this is what I don't  etc etc.

The video was most interesting WRT a) how modulators react to non steady signals, b) humans perception
of music through various DAC / electronics.

The other thing Sabre has is an inbuilt ASRC. This recalculates the samples based on a separate and independent
clock at the DAC. So you can feed anything into Sabre and the OP will always be locked to a fixed clock.
These (ASRC's) are notoriously bad for sound. The Sabres is reportedly very good but it is still a question.

The secret is you can also disable this feature and run the chip synchronously.

The last thing is that because these SD DACs run at very high speeds, I believe the clocking becomes
very important.

Once you consider all these things, the bog simple R2R looks like a complete doddle to get right in comparison. :)


Offline zenelectro

  • Hero Member
  • *****
  • Posts: 776
  • Liked: 144
Re: DDDAC 1794-NOS 24-192
« Reply #91 on: December 01, 2012, 01:32:06 PM »
OK Oz,

How's the headache going? :)

Here's some more. The first section is on digital vol control versus analog.

The second half is on jitter. This is worth persevering with as it explains various clocking systems,
PLL's, ASRC's etc.

<a href="http://www.youtube.com/watch?v=JYjHKv2_OqQ" target="_blank">http://www.youtube.com/watch?v=JYjHKv2_OqQ</a>




Offline ozmillsy

  • Hero Member
  • *****
  • Posts: 2163
  • Liked: 245
Re: DDDAC 1794-NOS 24-192
« Reply #92 on: December 01, 2012, 02:38:10 PM »
The important thing to understand is the 8 x OS filter and the modulator are separate things.

The 8 x OS filter has it's own low pass brick wall filter, and the SD modulator has it's internal, in the Sabres case, 5th order
LPF. The modulators filter is working much higher and as you can see from the vid in a different way.

In both Sabre and 1794 you can disable the OS filter and just use PCM straight into the modulator - I think.
Yeah, I came to that realisation while thinking and typing last night (note the quote below).  

If there is a way to bypass that oversampling module, and get straight to the dac modulator,  then it's undocumented.  

If that modulator handles the re-sampling from 16/24bit into whatever multibit wordlength the chip converts at (?), then maybe the modulator also has its own noise-shaping/dithering algorithms?  

I just dont know how Doede is bypassing the OS filter, as I cant find that documented into the datasheet anywhere??   There must be some pins that you can tap into on the 1794 that take you straight to the modulator.

Quote
However I'm not sure of the real consequences of running a DS DAC at say 44.1 or even 96 / 192 without
upsampling to say 352kHz. There must be some impact WRT the OOB noise.

That's why I asked the question WRT wide band plots of 1794 running 'zero OS'.

Well my view is that NOS is a purist thing.  If the dac chip can take the data and simply convert it, with no other messing around,  then this is a purist NOS implementation  (in my naive view).      If the 1794 is a sigma delta chip,  then it's messing with the data anyway.   This NOS business on a SD chip, is something I'm sceptical about.    R2R chips are the way to go, IF simplicity is what people are chasing.   But that said, I havent heard this dddac,  and I remain open minded about it.  
  
Quote
I'll just have to do all this myself in the usual zen - no BS fashion, this is what it does, this what I
understand, this is what I don't  etc etc.
  Yeah, I know.    But there was an email address offered in the video,  for people to use,  who have questions they want answered and cant find the answers online.    The speaker even explained how ESS simply dont publish alot of information, but it is available.

Quote
The video was most interesting WRT a) how modulators react to non steady signals, b) humans perception
of music through various DAC / electronics.
Absolutely,  the speaker also talked a number of times about all the things that cant be measured.  And relying on those with trained ears, to judge whether their "theory" translates into a better sound.   I liked that.    

Quote
Once you consider all these things, the bog simple R2R looks like a complete doddle to get right in comparison. :)
Yeah.   How big a hill do you want to climb?    
« Last Edit: December 01, 2012, 02:42:17 PM by ozmillsy »
It's all about the music,, not the equipment.

Offline ozmillsy

  • Hero Member
  • *****
  • Posts: 2163
  • Liked: 245
Re: DDDAC 1794-NOS 24-192
« Reply #93 on: December 01, 2012, 02:41:03 PM »
Here's some more. The first section is on digital vol control versus analog.

The second half is on jitter. This is worth persevering with as it explains various clocking systems,
PLL's, ASRC's etc.
Might watch it tonight, if I cant sleep again.  haha.  ;)  

Thanks, busy doing family stuff today, but will check it out for sure at some point.
It's all about the music,, not the equipment.

Offline ozmillsy

  • Hero Member
  • *****
  • Posts: 2163
  • Liked: 245
Re: DDDAC 1794-NOS 24-192
« Reply #94 on: December 02, 2012, 09:30:13 AM »
The other thing Sabre has is an inbuilt ASRC. This recalculates the samples based on a separate and independent
clock at the DAC. So you can feed anything into Sabre and the OP will always be locked to a fixed clock.
These (ASRC's) are notoriously bad for sound. The Sabres is reportedly very good but it is still a question.
This makes me wonder what that fixed clock is,  and if it inherently favours (works better) with certain data types to others?

Quote
The secret is you can also disable this feature and run the chip synchronously.
That should give you greater control,  but I still have this irrepressible feeling that the OP wants a certain speed.    So if you bypass it,  if you are giving it what it wants/likes, all is fine.   If you're feeding it a speed it doesn't like, it could compromise the performance of the OP step.

I dunno, it's speculation, but it's these types of questions that ESS would be able to answer.

I have a tendency to over simplify complex stuff.  But for the Sabre, my starting point would be,   Learn what that fixed clock speed is (you might already know?),  then decide on the most appropriate source format to match the end clock,   then optimise your middleman to suit the source/end points.   So we have as simple as possible reading, transporting and processing chain. 

Then get everything else around the dac chip optimised,  power supplies,  output electronics, etc.

Once you're hitting home runs.   Then start to experiment with other source formats/reclocking the different sample rates.

Just thinking out loud.
It's all about the music,, not the equipment.

Offline zenelectro

  • Hero Member
  • *****
  • Posts: 776
  • Liked: 144
Re: DDDAC 1794-NOS 24-192
« Reply #95 on: December 02, 2012, 10:47:16 AM »
The other thing Sabre has is an inbuilt ASRC. This recalculates the samples based on a separate and independent
clock at the DAC. So you can feed anything into Sabre and the OP will always be locked to a fixed clock.
These (ASRC's) are notoriously bad for sound. The Sabres is reportedly very good but it is still a question.
This makes me wonder what that fixed clock is,  and if it inherently favours (works better) with certain data types to others?

Quote
The secret is you can also disable this feature and run the chip synchronously.
That should give you greater control,  but I still have this irrepressible feeling that the OP wants a certain speed.    So if you bypass it,  if you are giving it what it wants/likes, all is fine.   If you're feeding it a speed it doesn't like, it could compromise the performance of the OP step.

I dunno, it's speculation, but it's these types of questions that ESS would be able to answer.



I have a tendency to over simplify complex stuff.  But for the Sabre, my starting point would be,   Learn what that fixed clock speed is (you might already know?),  then decide on the most appropriate source format to match the end clock,   then optimise your middleman to suit the source/end points.   So we have as simple as possible reading, transporting and processing chain. 

Then get everything else around the dac chip optimised,  power supplies,  output electronics, etc.

Once you're hitting home runs.   Then start to experiment with other source formats/reclocking the different sample rates.

Just thinking out loud.

Oz, I have already looked into all of this and for a USB connected DAC it can be all done really well by virtue of the
fact that with asynchronous USB, you can have fixed clock(s) in the DAC itself. 

I'll PM you.

cheers



Offline ozmillsy

  • Hero Member
  • *****
  • Posts: 2163
  • Liked: 245
Re: DDDAC 1794-NOS 24-192
« Reply #96 on: December 02, 2012, 11:13:45 AM »
No worries.   My issue with hi-res solutions is they try to be all things to all digital formats, and as a result, somewhere in the chain a compromise is made.

I guess what I'm saying is,  somewhere a frequency conversion has to happen,  if the source material doesn't evenly translate to the end point (44.1 or 48 derivatives).

I reckon it would be ok to put that aside altogether for the moment,  and be focussing on 1 or the other (and also be focussing on PCM or DVD).    [shrugs]

It's all about the music,, not the equipment.

Offline zenelectro

  • Hero Member
  • *****
  • Posts: 776
  • Liked: 144
Re: DDDAC 1794-NOS 24-192
« Reply #97 on: December 02, 2012, 01:30:42 PM »
No worries.   My issue with hi-res solutions is they try to be all things to all digital formats, and as a result, somewhere in the chain a compromise is made.

I guess what I'm saying is,  somewhere a frequency conversion has to happen,  if the source material doesn't evenly translate to the end point (44.1 or 48 derivatives).


With async USB -> I2S that is not the case.

If file on PC is 44.1 then DAC will work at 44.1, if file is 192 DAC will work at 192, an so on.

So ideally the dac should be run 0 x OS and any OS would be done at PC end.

If the USB -> I2S converter is designed properly, the appropriate clock will be engaged and the
-DAC- can be run completely synchronously without any PLL, or ASRC required. This is the most
ideal configuration WRT low jitter.

Offline ozmillsy

  • Hero Member
  • *****
  • Posts: 2163
  • Liked: 245
Re: DDDAC 1794-NOS 24-192
« Reply #98 on: December 02, 2012, 08:37:47 PM »
The fixed clock in the chip (that you want to bypass), what frequency does it run at?
It's all about the music,, not the equipment.

Offline dddac

  • Newbie
  • *
  • Posts: 4
  • Liked: 0
Re: DDDAC 1794-NOS 24-192
« Reply #99 on: December 05, 2012, 03:17:33 PM »
I am not sure I want to get dragged into too much of a techno-philosophical discussion on what might be the best way to do things, so I leave it up to the ones who actually build the DDDAC1794 NOS if they like what they hear coming out of the DAC. That is what Reinhard wanted to show here in this thread, namely his listening experiences and joy he founds in it.

nevertheless, I picked up some ideas for further testing, by means of listening and measuring, so there is value in discussions like this to be very honest  ;D

One thing I like to answer is about avoiding the digital filter and oversampling in the PCM1794. Actually it is very well documented in the data sheet. it tells you exactly what to do and what selection pins need to pulled high or set low. Also timing diagrams are provided. The trick I used to overcome the need for up sampling the world clock, by just paralleling the system and bit clock is not mentioned. But basically the rest is the same. Did you miss that, or did I did not understand the question right?

just let me know if things are cleared ?