I must be crazy to build my DAC this way

[Peter Daniel]

I finally decided to build TDA1541S2 based DAC. As usually I'm looking for best ways to implement the mechanical aspects of the design, so this is what I think could be a proper method to dress up S2 chip

 

[Peter Daniel]

Initially I was considering PCB, but somehow didn't like the restrictions dictated by 2 dimentional plane. So last evening I took a piece of brass and played with it a bit. I mounted the chip on top of 1/2 square rod and attached to it on top a piece of PCB forming ground plane. The rest is pretty self explanatory. I wonder myself if this is a good move or not?

Any comments on 560p Wima? I didn't know if this a good choice, but that's what seemed to fit there.

 

[Peter Daniel]

Worth noting is the way I attached SMD cap bypasses (X5U 0.1u and X7R 0.01u). This is technically the shortest path possible. Look under 2 BG caps on the right.

 

[Pedja]

Hi Peter,

Yes, it is OK I'd say (excellent actually), but try to divide digital and analog grounds before you joint them. Decoupling caps (pins 26 and 28) belong to the digital, those 14 caps to the analog part.

Pedja

 

[Pedja]

Also, though it doesn't hurt, that jumper between pins 2 and 4 shouldn't be necessary.

Pedja

 

[Peter Daniel]

I was thinking about splitting grounds, but checked the PCB board on my Marantz CD 94 and both grounds were connected at the chip (same ground plane). Doing it this way definitely simplifies things, but I'm not sure how this affects sound. If they are separated at the chip, were should they meet then?

 

[Peter Daniel]

Hi Peter,

Yes, it is OK I'd say (excellent actually), but try to divide digital and analog grounds before you joint them. Decoupling caps (pins 26 and 28) belong to the digital, those 14 caps to the analog part.

Pedja

So 5v supplies are digital and -15 is analog? I could use a center ground strip to route grounds from pins 26 and 28 to the common point on the other side.

 

[Pedja]

If they are separated at the chip, were should they meet then?

Somewhere below/above the chip. Just make sure that the analog and digital ground paths are not messed, i.e. provide that the digital part doesn't polute an analog part.

So 5v supplies are digital and -15 is analog?

Yes.

Pedja

 

[Pedja]

So 5v supplies are digital and -15 is analog? I could use a center ground strip to route grounds from pins 26 and 28 to the common point on the other side.

Yes, it seems like you have three nice ground buses.

 

[Peter Daniel]

Pedja,

I just checked your site and the info is good as usually (on TDA1541 Dac)

 

[Richard C]

Nice work, Peter!

I've found that these DACs benefit enormously from good PSU decoupling and screening and it isn't possible to optimise this using a PCB alone.

As regards connecting analogue and digital grounds: ignore the method used by Marantz, on my CD80 they're daisy-chained together by a thin track. They should meet at a low impedance point like a big decoupling capacitor.

 

[Pedja]

Pedja,
I just checked your site and the info is good as usually (on TDA1541 Dac)

Thanks.

I've found that these DACs benefit enormously from good PSU decoupling and screening and it isn't possible to optimise this using a PCB alone.

No doubt, it benefits. I screened the chips by the copper sheets later but still have to upload newer pics. And yes, the way the Peter is doing this might be better than usage of PCB, regardless of how careful it is designed. Though it is certainly not the easier way.

As regards connecting analogue and digital grounds: ignore the method used by Marantz, on my CD80 they're daisy-chained together by a thin track. They should meet at a low impedance point like a big decoupling capacitor.

Two pages of extremely useful text on the topic:
http://www.sigcon.com/Pubs/edn/adcgrounding.htm
http://www.sigcon.com/Pubs/edn/multipleadc.htm

Pedja

 

[Peter Daniel]

Also, though it doesn't hurt, that jumper between pins 2 and 4 shouldn't be necessary.

Pedja

I see it is both on Marantz schematic and Thorsten DAC. What is the reason for it?

 

[Pedja]

In the I2S mode the TDA1541’s pin 4 should receive the signal clock and TDA1541A’s (AFAIK there is not a crowned version that is not A) pin 4 is not used.

Pedja

 

[agent.5]

Pretty as always. Have you also consider to mount the receiver chip in the same manner, but face it inverted and mount it underneath the DAC chip. This way, the 3 I2S pins (FSYNC, SCK, SDATA) will be P2P also. This will solve your ground plane problem too, by sandwiching a ground plane (say digital ground under the chips) between the two chips. Analogue ground will be on the top of both chips.

 

[Peter Daniel]

That's a good suggestion, but with this one I will go directly from transport to DAC (I2S) with reclocking in between.

Finally my CD pro will get a proper DAC. Everything in one box.

 

[tone]

S2

hi peter,

use 470pf 'Silvermica's for Pin 16 & 17.
BG NXHiQ works best without Bypass caps.
Use BG PK Series 220uf/50v for Analog -15V supply..much better than Oscons.
hope this helps.

tone.

 

[rbroer]

Peter,
You are crazy

I would suggest not using any cap between pin 16 & 17, but instead implement the DEM reclocking circuit.
You'll be suprised how much it does.
Thread about DEM is somewhere in the archives.

BTW TDA1543 board is finished, just need to install it in a cd player. Probably start a thread tonight.

 

[Variac]

Hey Peter,
what DAC circuit design are you using? I haven't been following the
digital forum much. With everything globbed together (in a good way) it looks pretty simple.
Thanks, Mark

 

[Peter Daniel]

TDA 1541 circuit is mostly based on this http://www.fortunecity.com/rivendell/xentar/1179/projects/adagio/Adagio.html

The reast will be based on this http://www.diyaudio.com/forums/showthread.php?s=&threadid=23044

because it sounds good I won't be using input receiver, I2S from transport goes directly to DAC (through reclocking circuit)