Building the ultimate NOS DAC using TDA1541A

Jump to Latest
fabrice63

fabrice63

Member
Joined 2015
Yes he did some correction but only on the I2S version , he never modified the left justified ( post 6116 ) version and then I thought this was a good version , it seems I was wrong , this is it

I will stand by for a moment and decide what to do , build an I2S version seems to be the way to go , at least for my skills

Thanks again for your help jpk73 (y)

.
 
J

jpk73

Member
Joined 2020
You can see John's comment in the schematic "0.5bit delay" and "1.5bit delay". Hence you can make a version for I2S and add a bridge solder jumper to link out U5. Also you might want to add 0R22 mems resistors near each ic output to avoid signal reflection.
 
Last edited:
A

Alexandre

Member
Joined 2004
TDA154X.png
 
A

Alexandre

Member
Joined 2004
I was taking a look at it, and posted above, the figure from page 142 of the dissertation of Rudy J. van de Plassche
(HIGH-SPEED AND HIGH-RESOLUTION ANALOG-TO-DIGITAL AND DIGITAL-TO-ANALOG CONVERTERS)

Unrelated question: I am curious to know how well the channels of the TDA1541A match, in amplitude. If any of you fellows looked into it please share the results.

Thanks,
Alex
 
R

rfbrw

Member
Joined 2001
jpk73 said:
It's possible to add DACs just like batteries on top of each other regardless of internal architecture, as long as the analog outputs are added correctly together and the digital data is prepared properly. This is discussed in this thread since a couple of thousands posts ;-)

Of course MSB has a value: either one or zero, as all other bits in the stream. MSB is just the first piece of information. In some formats it's used to define the polarity, but not in all formats...
Click to expand...
How is the magical thinking working out?
 
  • Like
Reactions: 1 users
S

SSerg

Member
Joined 2012
Alexandre said:
Unrelated question: I am curious to know how well the channels of the TDA1541A match, in amplitude. If any of you fellows looked into it please share the results.

Thanks,
Alex
Click to expand...
Разброс значений выходного тока полной шкалы не превышает 15% (даташит, страница 6). На эту цифру и следует ориентироваться.
 
S

SSerg

Member
Joined 2012
dddac said:
в любом случае 15% выглядят глупо высокими. Это вариация на 1,4 дБ. Я очень сомневаюсь, что это когда-либо будет так
Click to expand...
Firstly, this is not my imagination, but data from the datasheet. Secondly, this is a limiting value and on average the deviation will be much less.
 
  • Like
Reactions: 1 user
A

Alexandre

Member
Joined 2004
I don't want to brag, but I am pretty much proven right at this point. Do you fellows remember what I said? It is possible to make it work, even if there is 1 lsb error in the midscale (this error is probably unimportant).
Usually, sign magnitude dac chips implement a "technically perfect" solution that corrects the lsb error, this is described in the ad1862 and pcm1702 datasheets (as mentioned already).
With the 1541, the error due to mismatch of the channels will be larger than the missing lsb at midscale (lsb equals 60 nA in the tda1541). Not to mention noise in the I/V converter (resolving the lsb is not easy).
My conclusion is: carry on. This should work well. Actually, it is proven that it's possible to make it work well, see:
https://www.diyaudio.com/community/...on-of-sound-quality-between-some-dacs.386815/
 
fabrice63

fabrice63

Member
Joined 2015
Any useful help still welcome for this setup , the dual mono sign magnitude is not the issue here , I / we know that it work , troubles are in the logic section , left justified for sure , but also in the I2S one , a lot has been publisched here for those who really whant to help :)


.
 
  • Like
Reactions: 1 user
A

Alexandre

Member
Joined 2004
Actually, I have helped with this sort of issue here before, for example I presented a working scheme for stopped clock tda1543, dual mono, that I used with very good results.

The thing is, I prefer to use small fpga boards, I can deal with the noise they generate.
If you're not able to make it work with the discrete logic maybe you should consider this other option.


fabrice63 said:
Any useful help still welcome for this setup , the dual mono sign magnitude is not the issue here , I / we know that it work , troubles are in the logic section , left justified for sure , but also in the I2S one , a lot has been publisched here for those who really whant to help :)
Click to expand...
 
Last edited:
You must log in or register to reply here.
Top Bottom