Building the ultimate NOS DAC using TDA1541A

Hi,

How will you, please, limit the input swing ?
At least what's your prefered method here ?

With the AMR CD-77 I used CMOS to PECL "translator" IC's (MC100EPT20) with a TL431 based supply that had a diode as additional temperature compensation, to undo the temperature compensation build into the IC (TDA1541 is not temperature compensated).

I think MC10EPT20 without temperature compensation might work better, I didn't get around to trying.

Temperature tracking must be quite good, the voltage swing is just enough to make the TDA1541 trigger, so if the temperature drift is out of sync the signal may drift outside the reliable trigger window.

More on the topic here:

https://www.mvaudiolabs.com/digital/tda1541-digital-input-attenuation/

Thor
 
Last edited:
  • Like
  • Thank You
Reactions: 1 users
With the AMR CD-77 I used CMOS to PECL "translator" IC's (MC100EPT20) with a TL431 based supply that had a diode as additional temperature compensation, to undo the temperature compensation build into the IC (TDA1541 is not temperature compensated).

For illustration, here more or less the insides of one of these CMOS to PECL translator IC's:

1702286529034.png


The outputs are basically differential Emitter followers.

Here the full TDA1541 schematic from the CD-77, it's long discontinued:

1702287606387.png


It includes the DEM reclocking per HTP. Each TL431 shunt is fed from a CCS which is off board.

And this is what it looks like on a PCB:

1702287325084.png


On the left hand side are all the CD drive electronics,

Thor
 
  • Like
  • Thank You
Reactions: 8 users
Disabled Account
Joined 2019
:love:
This TDA1541A comunauty is a great sharing comunity : John (ECdesigns), Thorsten Loesch (AMR, Ifi,...), Pedja Rogic (Audial), not forgiven the diyers here that gave all their little tips.

Makes me meet people around this chip I love since Philips CD371 (my first cd player). Some of these guys a little everywhere became good audio buddies as well, from Australia, to Hong-Kong, Singapoor, Serbia, Norway, Sweeden, USA, Italy...

Thanks to you all again.

Big Thanks for the Christmass gift Thorsten ! I have no doubt some here will heat the verroboard machine to experiment and offer some spare given pcbs around that with your permission in the sharing spirit behavior.Or let will 3D assembling ;) !

Wonderfull end of the years celebrations :)

Damien from France
 
Last edited:
  • Like
Reactions: 4 users
Here is where we do "Ripley's, believe it or not!"

I am currently listening to a Marantz PM-75 from 1989, with build in TDA1541A Single Crown Chip DAC that is essentially a DA-12, including the FIFO buffer and secondary PLL based on an LC Oscillator. The Amp is non-switching Class AB, using the same control Chip as JVC.

This thing sounds amazing.

Thor
 
  • Like
Reactions: 2 users
Yes, the interesting point is the bias ic.

I got OPA828 to replace the OPA and I think I will reduce the loop gain a bit (lower loading resistors on 2nd VAS).

But the DAC is the killer part. Have a look.

Thor

I did take a look at it, yamaha dir, saa7220 filter, quite complex secondary pll (not easy to figure out).

What about the analog section after the dac, are you still using all those opamps? Opamp I/V, and apparently a high order low pass filter with GIC cells (opamps). And a servo, to source a part of the bias current of the tda1541 and keep the opamps more or less centered (but the final output is cap coupled).
 
It acts as one.

If you look, it is a serial to parallel converter, 16bit. It is followed by a 16 bit parallel to serial converter.

So first one audio sample is loaded into the first shift Register (16bck delay) and then converter back to serial. This reliably bridges the two clock domains.

The secondary PLL is complex, because it locks on the preamble (slow) as opposed to the bit lock or word clock.

Thor
 
I don't think it's available anymore.

The 673 was a 16 Bit device. It's also latching, so you would need a pair of 595 to replace it, not sure what to do on the output side, where the 75HC674 is essentially a latching parallel to serial converter.

I would not try to copy the design.

For SPDIF in, the WM8805 under software control has low enough jitter, with a USB Interface we get asynchronous clocks and can use local crystals.

So just stick a set of 74AHC74 in series into each line and use a high frequency MCK (minimum 2xBCK) to reclock everything before sending the signal to whatever Attenuator you use.

I guess you could use ECL Flip Flop's for reclocking and an ECL clock, just to be contrarian.

Thor
 
  • Like
Reactions: 1 user