0
\$\begingroup\$

No doubt that if I put in parallel n DACs, multibit (i.e. 16 bit) current output R2R ladder chips, I get n-times higher current at the summing node and the output impedance is also reduced by "n".

But what about linearity and precision at low signal levels? Can we apply the general law for "uncorrelated events", for which in summing configuration noise and tolerance (and, hence, DAC linearity and LSB -least significant bit- precision) may improve by SQRT(n) times?

\$\endgroup\$
  • \$\begingroup\$ If you turn off all but one DAC, there is no averaging-benefit. \$\endgroup\$ – analogsystemsrf Apr 15 at 12:46
  • \$\begingroup\$ digital inputs are in parallel too, so all n-DAC are always ON \$\endgroup\$ – Gianluca G Apr 15 at 12:53
  • \$\begingroup\$ It works for paralleled op-amps so why not. \$\endgroup\$ – Andy aka Apr 15 at 13:02
  • \$\begingroup\$ that is true for noise, but nothing is said about precision and linearity \$\endgroup\$ – Gianluca G Apr 15 at 13:11
  • \$\begingroup\$ If you have N k% current sources summed together, the equivalent current source will have k% error too. That's your circumstance. You can read more here. \$\endgroup\$ – Long Pham Apr 15 at 13:51
0
\$\begingroup\$

This approach would work, barring any eccentricities in the particular DAC you're using. Another approach is to run a single DAC faster than needed and low-pass it, which will potentially reduce noise but won't improve offsets or other errors inherent to the part.

\$\endgroup\$
  • \$\begingroup\$ Don't understand "barring eccentricity". Does linearity improve ( by SQRT ( n) ) if I get n miscellaneus parts in parallel coming from different lots? \$\endgroup\$ – Gianluca G Apr 16 at 13:19

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.