TDA1541A reducing DNL

I checked the DNL error of various TDA1541A grades: S1, Taiwan, no mark, and R1. From bit #0 to bit #7 the error is <0.1 LSB for all grades. These bits are not included in the table.

The full scale THD shows some minor difference:
S1: 0.00026% / 0.00030% (L/R)
TW: 0.00039% / 0.00040%
no mark: 0.00037% / 0.00032%
R1: 0.00033% / 0.00037%
 

Attachments

  • DNL_survey.xlsx
    DNL_survey.xlsx
    9.4 KB · Views: 10
  • DNL_survey.jpg
    DNL_survey.jpg
    35.8 KB · Views: 17
  • Like
Reactions: diyiggy and Brijac
Seems worthy of implementing, so any chip can be deduced to ~same spec. There are any many non standard resistor values in smd, so trimpot can be used to figure out values, then replace. This way, it's less noise, and no worries that values will drift too much, at a much much lower cost than.
 
Honestly, costs almost nothing, and thd improvement's are great. Implementation is easy with using socket for trimpot, measure and replace with resistors. And use just that one trimpot to measure all, so quite cheap. Now, only caveat is that, one making this needs to have knowledge and gear to measure. This would especially work great in a temp controlled enviroment, like with mvaudio approach using temperature probe and noiseless noctua fan. So no deviations no matter the summer/winter.
 
I agree that the SMD resistor realization is more stable. But it is not possible to tune for minimum THD each bit (only bit 10 to bit 15) separately, because they interact.
The correct procedure is: tune bit 10 for lowest THD, then keep the variable resistor at this position, then move to bit 11 and do the same with the next variable resistor, then move to bit 12, etc. until bit 15.
The alternative method using +/-1 LSB step around each bit can be used independently for each bit 1 to 15 (whereas only bit 10 to bit 15 can be adjusted for exactly 1 LSB step).
After you have done, you can measure and replace all 2x6 variable resistors to SMD.
If tuning the upper 6 bits is too difficult, I recommend tuning only the MSB i.e. bit 15. This is the bit that flips at analog zero crossing, that has most influence on the THD at very low signal level.
 
  • Like
Reactions: diyiggy and Brijac
Imo it should be done on a pcb just made for that with the trim that stay in lieu. Then since you have the value put the chip in the final pcb with the according value with the hope there is a close value in the resistors catalog.

I have made the pcb à la Thorten, there is really no place left nowhere if you want small loops.

Will be interresting to see the correlation with the sound to check the quantity of improvment that can be hear beyond the numbers.