Building the ultimate NOS DAC using TDA1541A

[ThorstenL]

Hi,

Originally Posted by Ken Newton

The practical reality is that the 11uS temporal channel displacement of the multiplexed 44.1kHz rate L/R is inconsequential. The spatial displacement is 15/100ths of an inch. I very much doubt that your speakers are equidistant from your listening position to less than that, and even if you carefully made them so you would have to lock your head in a vise to maintain such an equidistance.

I would largely agree, plus it is trivial to compenate this, if desired (stopped bitclock operation, **** register delay to 2nd DAC etc... All it needs is a handfull of 74HCT logic.

Originally Posted by Ken Newton

I'm rather certain that the quote you included is by Thorsten Loesch.

You are correct, guilty as charged...

I had actually typed a longer comment on this but the dog ate it.

Ciao T

[qusp]

Originally Posted by brubeck

Another Dutch company, Metrum Acoustics, introduced a 24/96 pure NOS DAC named 'NOS Mini DAC': No idea what chips are being used inside? AD55xx? anyone?

I think metrum use some sort of industrial (read non-audio) high speed Vout dac, several in parallel for their other offering, one would presume they have followed a similar tack for the mini.

[regal]

Originally Posted by Ken Newton

Regal,

The practical reality is that the 11uS temporal channel displacement of the multiplexed 44.1kHz rate L/R is inconsequential. The spatial displacement is 15/100ths of an inch. I very much doubt that your speakers are equidistant from your listening position to less than that, and even if you carefully made them so you would have to lock your head in a vise to maintain such an equidistance.

I'm rather certain that the quote you included is by Thorsten Loesch. I'm not sure how you could have cut-and-pasted it without obviously seeing who the author is.

According to T.I.'s website, the PCM56 continues to be in production. Not sure where you heard otherwise.

Lastly, I must tell you that I find your attempts to play thread police by admonishing me to, in your view, "keep the thread on topic", and your continued suggestion that because YOU can't imagine any reason other than a commercial one for my question must therefore mean it is also MY reason, are rediculous and mildly offensive. In the future, if you don't know what you are talking about, which you clearly don't here, then please don't respond to questions which are not directed to you.

With headphones you think this delay in inconsequencial ?

I wasn't reffering to you personally sorry you took it that way. I was just saying to look around for an authentic TDA1541 they are very attainable no need for fear mongering.

The K rated PCM56 which is the grade to sensibly considered has been discontinued, for all PCM56 it says right on ti website "NRND: This product is not recommended for new designs. However, the device, tool, or software continues to be in production to support existing customers. TI does not recommend using this part in a new design. TI encourages designers to consider alternative products for new designs." That means its done, mouser and digikey no longer sell the PCM56k.

The idea you dropped that all TDA1541x's available are becoming counterfeit or unattainable is absolutely absurd. Thread Police? is this not a TDA5141 thread, please lets move on and no need for personal attacks.

[regal]

Originally Posted by ThorstenL

Hi,



I would largely agree, plus it is trivial to compenate this, if desired (stopped bitclock operation, **** register delay to 2nd DAC etc... All it needs is a handfull of 74HCT logic.



You are correct, guilty as charged...

I had actually typed a longer comment on this but the dog ate it.

Ciao T

Thorsten I saved that on a notepad file years ago as it was very eye-opening, sorry I didn't remember who put it together but I did say it was a quote from a tda1541 expert at least I remembered that.

[Ken Newton]

Originally Posted by regal

With headphones you think this delay in inconsequencial ?

I wasn't reffering to you personally sorry you took it that way. I was just saying to look around for an authentic TDA1541 they are very attainable no need for fear mongering.

The K rated PCM56 which is the grade to sensibly considered has been discontinued, for all PCM56 it says right on ti website "NRND: This product is not recommended for new designs. However, the device, tool, or software continues to be in production to support existing customers. TI does not recommend using this part in a new design. TI encourages designers to consider alternative products for new designs." That means its done, mouser and digikey no longer sell the PCM56k.

The idea you dropped that all TDA1541x's available are becoming counterfeit or unattainable is absolutely absurd. Thread Police? is this not a TDA5141 thread, please lets move on and no need for personal attacks.

No, I don't think the linear-phase offset of the 11uS delay is consequential for headphones, where the whole listening experience is devoid of a natural image perspective anyhow, pleasurable as it may be in it's artificiality. In addition, as Thorsten points out, correcting the delay is a trivial matter in terms of the logic gates required.

Apology accepted.

I did no such thing as to suggest all TDA1541A are becoming counterfeit. What I did was open the issue of the increasing scarcity, not impossibility, of obtaining devices. The notion that the availability of a chip last manufactured nearly 20 years ago, and what John Brown sees as the next best alternative, is somehow off topic seems an oddly myopic view.

Yes, I couldn't agree more about the personal attacks. May I suggest that you refrain from initiating them in the future.

[Ken Newton]

Originally Posted by ThorstenL

Hi,



I would largely agree, plus it is trivial to compenate this, if desired (stopped bitclock operation, **** register delay to 2nd DAC etc... All it needs is a handfull of 74HCT logic.



You are correct, guilty as charged...

I had actually typed a longer comment on this but the dog ate it.

Ciao T

Should the dog ever vomit it back up I would be interested to read any futher thoughts you had on that, or any unpublished thoughts you have on any aspect of DAC design, Thorsten. Usually is interesting reading, as are John's thoughts regarding DAC design.

[ThorstenL]

Hi,

Originally Posted by Ken Newton

Should the dog ever vomit it back up I would be interested to read any futher thoughts you had on that, or any unpublished thoughts you have on any aspect of DAC design, Thorsten. Usually is interesting reading, as are John's thoughts regarding DAC design.

I'll see if I find the time.

TBH, personally I would probably do a hybrid DAC, not dissimilar to the Ultra Analog stuff, with a discrete DAC in an FPGA plus suitable bitswitches and reference sources for the upper maybe 8 Bits and an integrated DAC like that TI DAC8580 for the lower maybe 16 Bits (and no, I see absolutely no point in 32 bit DAC's).

Ciao T

[-ecdesigns-]

Hi FRUGIVO,

My assault on the K8 version recloker

Nice solder work, but I have some remarks,

Try to keep ALL connections as short as possible, every millimeter really counts. The resistive attenuators (4K7 and 2K2) only work reliably with very short interlinks between attenuator and DAC chip (capacitive loading).

Mount all circuits on a ground plane like the copper side of a piece of PCB material.

Try to keep incoming I2S signals well separated (crosstalk), you could use flat cable with following arrangement: GND, DATA, GND, WS, GND, BCK, GND

Datasheet DEM clock configuration leads to low performance due to inter-modulating clocks (on-chip crosstalk) between free running DEM clock and I2S timing signals. It's better to synchronize the DEM clock with BCK (2.8224 MHz) when using NOS.

The DEM clock can also be synchronized on multiples of WS when fine tuning the value of the external timing cap. Locking can be checked using an oscilloscope. Keep in mind that this type of synchronization is instable and locking can be lost with varying chip temperature.

I stopped using ceramic SMD caps a while ago after I experienced problems with reliability and performance. I now use SMD film caps exclusively.

[studiostevus]

Originally Posted by -ecdesigns-

MK8 concept schematics.

I attached TDA1541A-MK8 concept schematics.

On the top there are 4 common mode voltage regulators for +5V, -5V, -15V, and +5V. They consist of cascaded filtered buffered zener diode regulators.

This works as follows, T1 and D5 form a 10V - 1.2V = 8.8V pre regulator. This pre regulator is placed in the ground reference line. T5 and D1 form a 6.2V - 1.2V = 5V post regulator that is referenced to the 8.8V GND reference provided by the pre regulator. R7 and C23 filter the zener diode noise.

The 5V post regulator is placed after the 8.8V pre regulator and re-stabilizes it to +5V. The output of the 8.8V pre regulator serves as ground reference. The output of the post regulator is decoupled to GND by a 1uF SMD film cap (C11).

The negative common mode voltage regulators function similar. The -15V regulator has a -18.8V pre regulator and a -15V post regulator.

Note that all common mode regulators have separate plus and minus inputs that are connected to the main power supply using twisted pairs. The main power supply contains rectifiers, smoothing caps and common mode capacitance multipliers.

.

As usual we need to thanks John for his generosity to share his work!!

If I understand correctly, the common-mode regs posted in 3978 have now been replaced by the ones in the main schematics plus the ones in the main power supply of the dac?

Also, I thought you were using air-coil inductors... now I see a center-tap connected to GND?

[qusp]

Originally Posted by ThorstenL

(and no, I see absolutely no point in 32 bit DAC's).

Ciao T

why is that? 32 bits used for just audio, i agree with you, but 32 bit dac chips allow filters, volume etc to be calculated with far better accuracy and be more 'out of the way'