Drive NOS AD1865/62,PCM1704/02/63,TDA1541 from FIFO: Universal I2S-PCM driver board

Hi Ian

Thanks for making this board available to the aficionados of old Dac chips! Allow me to make reference to the discussion surrounding the implementation of the PCM protocol for the TDA1541a.

The way it is implemented now does sound well, as you have tested for yourself. However; it is true that the TDA1541a shifts data to the register when the LE signal goes positive. In the case of the I2S-to-PCM board that means that the data is shifted after 15bits have been recieved; turning the TDA1541a in fact into a 15bit Dac.

Don't get me wrong; it sounds lovey like that. It even sounds lovely when being used as a 14bit Dac, see here.

However, it does sound better as a 16bit Dac, especially when being fed with a 44.1khz signal.

I would like to motivate all those who use an I2S-to-PCM board with a TDA1541a to try the following:
(1) Feed your Dac with a 44.1 khz signal (non-oversampled) and listen carefully to get an impression.
(2) Disconnect the LLLR out of the I2S-to-PCM board.
(3) Split the WS signal that goes into the I2S-to-PCM board and feed it both into the I2S-to-PCM (WS in) and also into the Latch Enable input of the TDA1541a. If you have a McFIFO just take a second u-fl cable and connect another WS out of the clock board to the LE in of the TDA1541a.
(4) Listen to some music and see if you notice the difference; specifically at 44.1khz.

In the case of my AYA Dac the difference was not subtle. Before the mod described above the heights sounded so recessed and muffled when feeding a 44.1khz signal that I contacted Pedja because I thought that my DAC was faulty. Oversampling in SOX to 88.2 or more solved that and clearly improved the sound. After the mod the music sounds brilliant at 44.1khz and with 44.1khz material I cannot hear a difference between a 44.1 and 88.2 or higher signal.

Without the mod feeding a 384khz signal to the TDA1541a produced silence - after the mod it produces heavily distorted sound - which is a kind of an improvement. It plays music with a 336khz signal in both setups. The fact that it does not work with a 384khz signal might be a hint that the timing of the BCK's rising/falling edge could still be optimized; since it has been shown that the TDA1541a can handle 384khz signals.

Ian, I understand that your efforts are currently focused on the battery/capacitor power supplies (which I am very much looking forward to). However, if you have some time and interest to look at the I2S-to-PCM board again, I would like to kindly ask you if you could change the TDA1541a settings according to the data sheet (DATA change on the rising edge of the BCLK and clocking it in at the falling edge of BCLK; then giving it a short pause before shifting the data to the register on the rising edge of LE). It should then be possible to max out the TDA1541a with a 384khz signal using your board.

Kind regards
ElEsido

Exactly what I suggested here:
https://www.diyaudio.com/forums/dig...e-nos-dac-using-tda1541a-667.html#post5570744
 
Member
Joined 2005
Paid Member
Long time ago i got the TDA1541 working with a PMD-100 digital filter (with additional glue logic) in simultaneous mode. Also outside the timing restrictions shown in the 1541 datasheet. It works.

Don't ask me how it "sounds", never completely finished the project (still on my to-do list, which is long).

What a coincidence! I just cleaned out the lab and found some PMD100s and a few TDA1541As. I remembered your project from 10+ years ago and I plan to breathe new life into that project tomorrow! Hit me up if you’re interested, would love to compare notes! Thanks!
 
Ian, I understand that your efforts are currently focused on the battery/capacitor power supplies (which I am very much looking forward to). However, if you have some time and interest to look at the I2S-to-PCM board again, I would like to kindly ask you if you could change the TDA1541a settings according to the data sheet (DATA change on the rising edge of the BCLK and clocking it in at the falling edge of BCLK; then giving it a short pause before shifting the data to the register on the rising edge of LE). It should then be possible to max out the TDA1541a with a 384khz signal using your board.

+1
Although it seems that Ian has addressed the timing issue to his satisfaction at #6712:
Building the ultimate NOS DAC using TDA1541A
HK
 
Last edited:
I would like to motivate all those who use an I2S-to-PCM board with a TDA1541a to try the following:
(1) Feed your Dac with a 44.1 khz signal (non-oversampled) and listen carefully to get an impression.
(2) Disconnect the LLLR out of the I2S-to-PCM board.
(3) Split the WS signal that goes into the I2S-to-PCM board and feed it both into the I2S-to-PCM (WS in) and also into the Latch Enable input of the TDA1541a. If you have a McFIFO just take a second u-fl cable and connect another WS out of the clock board to the LE in of the TDA1541a.
(4) Listen to some music and see if you notice the difference; specifically at 44.1khz.


Hi ElEsido,
To confirm that you have reconfigured as per images.
If WS from re-clock board is okay driving both PCM board WS and 1541A LE without buffering, a simple cut and shut using the input header on the PCM board to jumper off to 1541A.
Thank you!!
HK
 

Attachments

  • PCM_1541A_Mod.jpg
    PCM_1541A_Mod.jpg
    59.7 KB · Views: 324
  • PCM_1.JPG
    PCM_1.JPG
    118.1 KB · Views: 326
  • PCM_2.jpg
    PCM_2.jpg
    156.8 KB · Views: 331
Last edited:
Hi Hanze

+1
Although it seems that Ian has addressed the timing issue to his satisfaction at #6712:
Building the ultimate NOS DAC using TDA1541A
HK

I am no engineer so I can't comment on the specifics, but my i2s-to-pcm board can drive the Aya dac "only" up to 336khz and does not work at 384khz. My un-educated guess is that it might be a timing issue.

The wiring diagram you drew is correct and your pics show the right points to steal the signal and feed it back in.

I built a quick-and-dirty setup to check whether my McFifo can drive both the PCM board and the DAC (see this picture), and this works. The yellow wire is connected to a u.fl cable using Ian's Raspberry adapter. (My permanent setup is to take the WS signal directly from the McFifo with the u.fl cable that you see connected to the yellow cable).

You seem to have the first generation clock board, so you have to try for yourself whether the signal is strong enough. Please describe your listening impression.

Best
ElEsido
 
Last edited:
I am no engineer so I can't comment on the specifics, but my i2s-to-pcm board can drive the Aya dac "only" up to 336khz and does not work at 384khz. My un-educated guess is that it might be a timing issue.

Thanks ElEsido,

WaveIO > complete Ian first gen setup > earliest AYA OPA861. Never rolled off HF, difficult to tell difference between 44 and 176.

Ian I2S to PCM block diagram shows PCM outputs are re-clocked (presumably together), this will then be different. Maybe someone with some more knowledge can provide insight, and also to the additional loading of WS from the re-clock board... Mr Ian ??.

Kind regards,
HK
 
Last edited:
No you don't; it works up to 336kHz with just feeding the WS signal to the DAC.

I want to thanks all who contribute and discover timing issues with Ian's I2StoPCM board, regarding TDA1541.

At the moment I choice Nakamichi DAC-101for my next project.
This is known and kind of unique realization of two crown TDA1541.
It uses each TDA c L and R channels individually, balanced mode.
The first is to make it NOS and I2S from RPi capable. Here, I need to convert i2S to PCM. As a result of this issue, can we consider proposed fix (WS rewiring) as a sollution for Ian's board to make it correctly drive TDA1541 at 16bit resolution? Is the any other alternative to realise correctly i2s to PCM convertion for TDA1541? Please advice before oredl placed...
Thank all you again!