TDA1541S1 demystified?

[tritosine]

...I was looking for something else, but....

tda1541s1 - Google Patent Search

TDA1541S1 of Philips is known as a DAC performing digital/analog conversion with high accuracy and have therefore frequently been employed in various digital audio instruments such as a compact disk (CD) player and digital audio tape (DAT) recorder. It has been recognized that this DAC employs high-speed switching circuits to realize glitchless operation, requiring no deglitching circuit. However, the applicant has found by experiment that this DAC of TDA1541S1 still generates glitches of minute level according to certain rules related to input data.

Maybe we can improve other chips with intentional glitch ?

It won't be easy, code dependant ,channel dependant, yet inherently pleasant.

 

[abraxalito]

Interesting - I've never seen a patent's text so specifically directed towards fixing perceived flaws in one vendor's chip before. Looks to me to be a fairly general patent on a feed-forward error correction system applied to a DAC.

 

[lcsaszar]

Interesting - in theory for a pure R-2R converter, which the TDA1541A is not. Other thing, it may work for NOS, but the patent refers to a digital filter, and sometimes to 8x OS. In that case there are no rectangular steps any more, just ringing.

I tried to visualize the switching artifacts, descibed here

http://www.diyaudio.com/forums/digital-source/113620-dac-linearity-test-cd.html

but it is extremely hard to clearly identify any glitches, not to speak about measuring the amplitude thereof. More apparent is the unevenness of the steps, AKA linearity error.

 

[SoNic_real_one]

More apparent is the unevenness of the steps, AKA linearity error.

Correct. And that's what Philips engineers (and not only them) saw when the decided to go for delta-sigma DAC's.

 

[tritosine]

nah, it was supposed to autocalibrate the MSB & its MSB errors that ruin linearity. ADI and Burr Brown had MSB trim for this purpose, and held the line with r2r for long.

 

[Bernhard]

I also remember that in the original thread we found that the TDA fails to reproduce the steps accurately, which would be the primary task...
So why worry about glitches ?