Building the ultimate NOS DAC using TDA1541A

Hi EC,
I'm speechless because of the esthetic quality of your work, especially the innards :)

I have to build my own boxes :(
I name my gear "the Fred Flintstone line of products". That should describe how they look :bawling:

A couple of things bother me:

a) Have you made test to evaluate how your different power supplies sections perform?

b) Can you remind us what is your final re-clocking strategy for I2S?

Sorry to be a P... (approaching 100K views :D )

Cheers,
M
 
Dear John,


I am also speechless about the quality of the work, the casing etc. etc. In Holland we have a saying with the right tools it is half the job ...

But you have gone far beyond the borders, and apparently spent enormous amount of time & money to produce everything down to the last detail perfect without compromises.

I am deeply impressed, Chapeau !

I am very anxious to hear about the kit prices when they are avail. I am wondering whether I should use the tda1541a, or tda1541a double crown. I have been reading about it on the internet and all very promising. But I have also seen prices of about us$90 per chip...

I presume most of you guys have 'reference' speakers and 'reference' amps to make the whole just the best. I have b&W CM6 floorstanding speakers and probably will be forced to make a good amp, currently I use a Denon4306. Heard good things about the hypex amps.

Finally, Many Thanks for letting me in on this trip and look beyond borders in DIY I thought were impossible, which revives many memories of a 16yrOld soldering my first Amp and Woodworking my first Speakers and listening to 'Crime of the Century' and feeling happy.:D
 
Amazing

John,
I work with electronics equipment for almost 25 years but I never see some one like you great attitude for any critics, keen in all aspect ( tubes,digital machinery) the best thing you also have a artis taste difficult combination to find, I dont care if your design do not sound like stardivari violin I will duplicate your desing any way.
Bravo......
 
Power supply / (re) clocking

Hi maxlorenz,

Thanks for the compliments [post #1082]

Sorry for the delay, but I am working very hard to get both the new octal D-I DAC PCB's and housing design ready.

As for the housing, some new sketches of the octal D-I DAC and the TDA1543 based DAC housing can be found on my website, this housing has a significantly easier construction, using standard aluminum parts (less milling necessary) this makes it much more affordable, it has the same quality and appearance as the housing on the photographs.

a) Power supply performance:

Yes, the power supplies perform well. No audible hum, noise or mains interference. LM137 / 337 could be used, but requires more components as the octal D-I DAC core uses 29 fixed voltage regulators. Because each single TDA1541A and the I/V diff amp modules have individual voltage regulators, a lot of power supply related problems can be avoided.

b) Final BCK (re) clocking scheme:

The Differential I2S inputs are not reclocked. Differential interlinks with stranded twisted wire pairs up to 3 meters cause no audible signal degradation.

The source (CD player) has to provide a low jitter I2S signal. The DS8921 and DS8922 were used as they had the highest noise immunity and provided the lowest jitter at the end of the differential interlink. The octal D-I DAC can toggle between 3 digital input sources (USB included).

In order to connect SPDIF sources, or convert formats, a external universal interface module can be used. Using synchronous reclocking (if possible) can reduce jitter significantly.

The USB interface can be connected by using the maximum specified USB cable length (5m).

The USB interface I2S BCK signal is asynchronously reclocked by a shiftregister reclocker. The reclocker is driven by a 48 MHz low jitter master clock. The reclocked BCK is then buffered by a DS8921 to suppress common noise. So basically with the USB interface, the master clock is located close to the DAC chip's.

- The differential I2S signals (also the one from the USB interface) first enter the differential I2S input selector, based on the DS8922, selection is accomplished by the differential to single ended buffers that are necessary anyway. A separate chip is used for BCK to avoid crosstalk with DATA and WS.

- BCK is then buffered by 3 low skew UHS buffers (now integrated in the timing-chain), one for the timing-chain and logic, one for DAC1 and one for DAC2...8. DAC2...8 use inverted BCK as both their delayed DATA and WS now change on the positive going edge of BCK.

The new smaller timing-chain only supports Philips format. CDPRO2 output format seems to be programmable. I noticed a keyboard / display kit on the internet that provided multiple output formats using a CDPRO2.

I want to use the Philips format exclusively as this enables 32X and 64X D-I mode. This could be useful for future upgrades.

- The 8 TDA1541A chip's receive an external 176.4 KHz Dynamic Element Matching clock signal from the new timing-chain, synchronizing them, reducing interference.


Cheers,

John
 
Thank you very much, John, for your reply.

The Differential I2S inputs are not reclocked. Differential interlinks with stranded twisted wire pairs up to 3 meters cause no audible signal degradation.

I'm slow. As I understand, one has to make a differential I2S output, perhaps copying your receiver circuit (or using pulse TX's?)

The new smaller timing-chain only supports Philips format. CDPRO2 output format seems to be programmable. I noticed a keyboard / display kit on the internet that provided multiple output formats using a CDPRO2.

I bought my CDPRO from http://www.diy-high-end.com/
Nico Thevissen, the seller, told me that my unit outputs the Phillips format. Perhaps he re-programmed it?
How can I test it? I mean before I connect it to the DI-DAC? :D
Any other users know?

Cheers,
M

PS: I'll check your site.
 
CDPRO2

Hi Max,

I saw your post on the cdpro2. I also purchased a cdpro2m from nico thevissen. I was not aware of the sony output format. But some time back i read on the internet somewhere that the output format could be changed into another format. Let's see whether there are people how exactly this can be done, I am facing the same problem when building the d-i-dac. I would presume however that nico thevissen should know more about this, did you ask him already?

Rgds
Frederik
 
Changing CDPRO2 output format

Hi maxlorenz,

Thanks for the compliments [post #1086]

Yes, when the octal D-I DAC is used as a external DAC, a small differential I2S driver has to be added to the CD player, but if the transport is close to the octal D-I DAC core, I2S can be connected directly (single ended). I will post a schematic diagram of the final universal I2S interface circuitry soon, I will also design a small PCB for it, so it's very easy to build. With this interface, almost every digital sound source can be connected to the octal D-I DAC. I alredy used all inputs, I2S1 for the CD changer (no it's not catching dust yet), I2S2 for my modded digital sattelite receiver (watching a movie now is like sitting in a cinema), and USB for the fanless iMac 500MHz containing my entire CD collection on a 160Gb HD.

You can connect the octal D-I DAC to the CDPRO without damaging anything if the format is wrong. If it outputs the Sony format, the sound will be very weak. The following website has more information about changing the CDPRO2 output format:

http://diyparadise.com/cdm12kit.html


Cheers,

John
 
Hi Frederik,
Yes I asked him:
The output is Philips format no oversampling.
See IC spec.
Regards,
Nico

But you are closer to him :D ...it doesn't hurt to ask again :angel:

Hi ecdesigns,
Thanks for your explanation.
I'm aware that it would be easier to connect directly a DI-DAC to CDPRO, through single ended I2S, as an "integrated" CD player. I'll probably do it that way, but as I said before, soon there will be 8 or 16 dac towers radiating heat :devilr:

Thanks for the link.
Well, it seems that we have to study those different I2S formats to begin whith... :(

Best regards,
M
 
new timing chain?

Hi EC,

I am finally getting around to building your timing chain.

I am modifying my TDA1543 DAC to be an octal DAC with single passive IV stage into an OPA627 buffer. I have the timing chain built out of a tower of 16 SMD logic chips. It is the same height as a tower of 8 TDA1543 chips. Looks cool. I haven't applied power to it yet. I give it a 50/50 chance of surviving me building it. I may have cooked a couple chips in the beginning of the process. If this logic tower doesn't work, the next one will :)

Speaking of the timing chain you say you have a new simplified one. Do you mind me asking but are you ready to share/post it?
I am too green to have come up with your last one let alone a simplified one :)

Cheers,
Brent
 
TDA1543 D-I 16 core almost completed

Hi Brent Welke,

Thanks for your reply [post#1090],

I guess you are one of the few people who decided to build a D-I setup, good job! I am very curious about the results. If you use passive I/V and a OPA627 buffer, make sure NOT to invert D32, D40, D48 and D56, just connect all data lines directly to the appropriate timing-chain outputs. There is not much that can go wrong with logic chip's, just make sure the power supplies are connected correctly (polarity) and the voltages stay within specifications.

Since you mentioned the TDA1543 based D-I DAC, I have been developing the small single board D-I 16 core using 16 X TDA1543 and the new timing-chain. The PCB includes power supply regulators, D-I 16 timing-chain, UHS buffers, 16 X TDA1543 in a 4 X 4 matrix, I/V converters (AD843AR) and diff amps (OPA627), all this circuitry is squeezed on a 18.8 X 11.6 cm PCB. In fact this single DAC board is pin compatible with the larger octal D-I DAC core, and can replace the octal D-I DAC core if desired. It uses very little power (approx. 7 watts) and has a unbelievable performance, that comes very close to the octal D-I DAC core. Like the octal D-I DAC core it has a I2S input. It has a Virtual sample rate of 705.6 KHz, theoretical 20 bits and 256X resolution enhancement.

Because it's pin compatible with the octal D-I DAC core, it can be used just the same (with tube output, I2S selector and USB interface).

I added a silkscreen plot of the D-I 16 core, at the bottom is the timing-chain for 16 DAC chips and the UHS buffers. At the right, the 5 power supply stabilizers, in the center, the 4 X 4 TDA1543A matrix. The large space around the TDA1543's has two functions: spreading the dissipated heat more evenly and reducing crosstalk to a minimum. At the top are the OP-amp I/V converters and diff amps. Even the output attenuator (bypassable) is integrated. All chip's are placed on sockets for easy replacement / testing. Jumper JP1 toggles between single ended and balanced mode, so it's possible to use different types of output stages.

PCB routing is completed, but I still need to optimize traces, this might take a few more days as the PCB has to be routed manually.

As for the D-I 16 core power supply, it's a very small PCB with 2 encapsulated torroid transformers (7 watts and 12 watts). The following voltages are needed: +10V, +20V and -20V.

> The new timing-chain was developed to significantly reduce the amount of logic chips needed, this required the use of a PLD (PALCE16V8). It uses Philips format exclusively. Each application requires a slightly different setup, it depends upon the amount of DAC chips used in D-I mode and if a DEM sync clock is needed or not. For example the D-I 16 core needs no DEM sync clock, but needs addittional output signals.

Cheers,

John
 

Attachments

  • di16silk.jpg
    di16silk.jpg
    95.1 KB · Views: 1,556
100000 views!

Hi all,

Seems this thread has just passed 100000 views! :cloud9: :cheers: so I want to thank you all very much for your feedback, support and (huge) compliments, it inspired me to design the "ultimate NOS DAC". The critics I received were used to constantly improve the DAC's performance, the (huge) compliments I received made me determined to design even better diy kits for you all.

I hope this thread will inspire you all.

Kind regards,

John
 
maxlorenz said:
Congratulations John!

For differential amps, have you seen the new dual opamp from National:

http://www.national.com/pf/LM/LM4562.html

And also:
http://www.diyaudio.com/forums/showthread.php?threadid=86276&perpage=25&pagenumber=1

"Free" samples available.

M


Very nice, John!

Yes, your work is very inspiring.

I also buy CD player with CDM1 transport, to be more suited for DI octal DAC instead CD PRO2 with i have already.

I tested LM4562 in CD player CD304MKII.
Before i put double OPA627 and i can say here is much much better sounding. Much more everything: clear, details, musical...
Difference is not subtle at all.
LM4562 is in A-class (4.7k neg. supply to output).
Please try it. I got samples in 2 days. Paid only shipping.

regards, Bostjan
 
SMD logic tower is working !!!

Ok I have a working quad TDA1543 DI DAC now, thanks EC.

I am waiting for more TDA1543 chips and will add them to the DAC when they arrive next week to make it an octal DAC.

The DAC has only been running for 5 minutes so it is too soon to comment on how it sounds exactly. But it is not bad sounding I can tell you that much :)

Tomorrow I will post some pictures if the sun shines and I have a feeling it will.

Cheers,
Brent
 
A glance at the D-I 16 core PCB design

Hi maxlorenz,

Thanks for the compliments [post #1093]

Before ordering huge amounts of TDA1543's, maybe it's wise to build the standard D-I 16 core first, then just listen to it. If you're not fully satisfied with the sound quality, then you might consider adding more DAC chip's or start modding.

The LM4562 looks very promising, most of the specifications seem to be similar to the OPA627. But the LM4562 is a dual op-amp and has a reasonable price, it's particulary suited for the I/V converter as the adapter socket (2 X SMD OPA627 > DIP8) is no longer needed. Seems this is the part I have been waiting for.

Then about the new TDA1543 D-I 16 DAC, I just completed optimizing the PCB design (I added a picture). The CAD program runs on a ancient operating system called DOS, anyway it has functioned flawlessly for over 19 years now. The layout shown is completely manually routed, trace by trace.




Cheers,

John
 

Attachments

  • di16pcb.jpg
    di16pcb.jpg
    95.3 KB · Views: 1,470