How many TDA1541A are needed in parallel to achieve double crown standard ?

[Blitz]

I have now searched for a while, but have not found any comments about how many TDA1541A I would have to parallel to come at a performance level of the double-crown version. In theory I understand that paralleling DACs helps to cancel out errors between them, achieving less distortion, higher resolution etc. Additionally I would get a higher mA-Output.

Economically it could make sense as well: For the 75$ a double-cronw-Dac costs these days, I can get 25 TDA1541A ! With the benefits of higher OUtput (about 100mA?), I should be even able to come up with outputstages like designed for the TDA1543, meaning not additional outoutstage required.

Why has nobody went into that direction yet ? Is it to difficult to drive so many TDA1541A in parallel ?

Best Regards

[Bricolo]

Use 25

In theory, I think that 4 paralleled devices would do the job to have a 0.5LSB error from a 1LSB error device.
Paralleling n devices divides the error by sqrt(n). At least in theory

[zygibajt]

I think one S2 and many NON-CROWN are not equal.

Bartek

[Elso Kwak]

Originally posted by Blitz
I have now searched for a while, but have not found any comments about how many TDA1541A I would have to parallel to come at a performance level of the double-crown version. In theory I understand that paralleling DACs helps to cancel out errors between them, achieving less distortion, higher resolution etc. Additionally I would get a higher mA-Output.

Economically it could make sense as well: For the 75$ a double-cronw-Dac costs these days, I can get 25 TDA1541A ! With the benefits of higher OUtput (about 100mA?), I should be even able to come up with outputstages like designed for the TDA1543, meaning not additional outoutstage required.

Why has nobody went into that direction yet ? Is it to difficult to drive so many TDA1541A in parallel ?

Best Regards

Hi Blitz, The average error is much greater than the error of a selected one.
Sounds almost like a quote from EE J. O' Mega.

[Peter Daniel]

Originally posted by Blitz

Economically it could make sense as well: For the 75$ a double-cronw-Dac costs these days, I can get 25 TDA1541A ! With the benefits of higher OUtput (about 100mA?), I should be even able to come up with outputstages like designed for the TDA1543, meaning not additional outoutstage required.

Don't forget about 14 decoupling caps per chip. This brings the total to 350. Even if not economically, when it comes to layout it would be a disaster. And I have serious doubts if it would sound as good as a single TDA1541A chip, not to mention double crown version

[Blitz]

Hi Peter, from where do the 14 couplings caps from ? Can't I simply stack the chips like with the PCM63K ?

Apropos PCM63K: Is the TDA 1541AS2 really superior to the PCM63K ? I think you had DACs with both chips, right ?

[ThorstenL]

Konnichiwa,

Originally posted by Blitz
Hi Peter, from where do the 14 couplings caps from ? Can't I simply stack the chips like with the PCM63K ?

Of course not.

Originally posted by Blitz
Apropos PCM63K: Is the TDA 1541AS2 really superior to the PCM63K ? I think you had DACs with both chips, right ?

I had "digital engines" in non-oversampling with both TDA1541 single crown and PCM63K. TDA1541 sounded much better, to my ears, but even PCM63 sounded noatbly better than PCM1702.

Sayonara

[Elso Kwak]

Originally posted by Elso Kwak


Hi Blitz, The average error is much greater than the error of a selected one.
Sounds almost like a quote from EE J. O' Mega.

Which reminds me of the latest Krell space heaters: if you take 25 pieces of s*** in parallel you get by some miracle a wonderfull part or 25 times a piece of s***? Remember 1+1 = 3?

[Blitz]

Guys, in what mood are you now in ? Sorry to have asked the question, but when reading all the comments that parallel 1543 - a significant worse DAC than the simplest 1541A - creates a wonderfull sounding DAC, the question what happens if you take a much superior DAC and do more or less the same is a valid one, or do I miss something here ?

[Peter Daniel]

I beg your pardon? 1541A much superior to 1543? You must be kidding.

without taking care of proper PS and output stage (I had OPA627 there) I still preffered TDA1543 to double crown TDA1541 S2.

I'm not sure if when properly implemented TDA1541S2 will be much better than 1543 chip, probably not, as Rudolf recently commented on that.

I also tried paralleling 2 and 4 TDA1543 and the results where much inferior to using a single chip.

If you don't want to spend so much money on S2 version, why don't you listen to A version and choose the one that sounds best.

I recently checked few of TDA1543 chips in listening comparisons and there was definitely a difference between those that had different markings. The one without white line (Thailand is written instead) sounds darker than the other chip. The one with white line is preffered by me and it has definitely much more high frequency extention. I was also trying to pick up some better sounding chips from among "white line" group, and although I was under the impression that some sounded better, I couldn't be 100% sure, but I did my choice