True24 DAC

Home Schematics Applications Forum Impressum

First  Previous  Next

Part Selection

Let's start to select the parts, circuits, we could potentially use ...

First of all, let's define some constraints, needs we want to get provided.

  • the DAC should have a noise spectrum of: <10nV/sqrt(Hz)
  • the DAC should have a settling time of: <1microSec (so that higher sample rate is possible)
  • the DAC should have a very small and short glitch period
  • it can be a single DAC, but a Dual or Quad DAC can give us benefit if all in one package (drift, same temperature, similar parameters ...)
  • if DAC has built-in feedback resistors - would be nice (drift, temperature, same effect on all channels)
  • DAC should be able to accept LVTTL/CMOS (3.3V) digital signals (for direct connection of FPGA)
  • DAC should work with very high reference voltage: +/- 15V or more
  • DAC must have an LDAC signal, used to synchronize the data words, to activate in synchronous way all DACs
  • DAC should have a high Slew Rate: >30V/microSec (settling time and full Vref range)
  • the OpAmps for further processing should fit to DAC:
  • at least the same Slew Rate: >30V/microSec
  • OpAmp must run with same power suppy voltage: >+/-15V, Rail-to-Rail would be nice
  • OpAmp should have a low voltage input spectrum: <6nV/sqrt(Hz), smaller as DAC
  • the distortion, THD, THD+N, should be small: <-100dBc, even better: -120dB would be great
  • the OpAmp must have low offsets and bias (necessary for this type of DAC)

16bit DACs

There seem to be two candidates for the DAC chip:

  • Texas Instruments, TI DAC8871B : this is a 16bit single DAC, supporting +/-18Vref, 10nV/sqrt(Hz), 1microSec - OK
  • Analog Devices, AD5544 : as quad DAC (nice), 16bit, noise is better: 7nV/sqrt(Hz), 0.9microSec, just Vref is a bit smaller: +/-15V, and: it is a current DAC, DAC8871B is voltage output, current output is way better - looks a bit more promissing


There are several options which have to be investigate further:

  • Texas Instruments, TI OPA211 - prefered
  • Texas Instruments, TI OPA827 - FET
  • Texas Instruments, TI OPA1611
  • Texas Instruments, TI OPA634
  • Texas Instruments, TI THS3201
  • Analog Devices, ADA4891-1 - quite nice, prefered
  • Analog Devices, ADA4627-1 - not Rail-to-Rail
  • Analog Devices, AD744 - BiFET
  • Analog Devices, AD844B-1 - very fast, but bad offsets

Voltage Reference

There are several options which have to be investigate further:

  • Texas Instruments, TI OPA2277 - Dual (nice for + and -)
  • Analog Devices AD688 - Dual (nice for + and -)
  • Maxim, MAX6143
  • Texas Instruments, TI REF50101
  • Texas Instruments, TI REF102

OpAmp Selection

OpAmp Selection

First  Previous  Next

First Conclusion

Let's assume we follow also the concept of a Bi-DAC - why not using also a PCM1794A DAC for it? We could combine the DualMono concept with the Bi-DAC concept.

We know this DAC quite well, e.g. T-DAC, RPiDAC. And potentially we can run it instead of +6V (already used) with an even higher voltage, e.g. +7 or +8V which gives us also a bit more headroom for a nice SNR. Just to check how far away official specs we can operate.

The use of a real NOS DAC and running with a quite high Vref, e.g. +-17V, seems to have the benefit of an easier management to provide a high SNR. But we do not know what happens on the cross-over, the sum functions, what is the non-linearity we create? OK, we had to verify and try (and measure).

Which OpAmp is quite obvious from the investigation: OPA211 seems to be a great candidate. Also, the ADA4627B seems to be a nice one.

Regarding the other options: even I like the LME49990 due to its amazing noise and THD parameters, I have used it and the OPA827 as well in RPi-DAC, T-DAC. And it sounds to me, the OPA827 is doing a bit better but this could be due to fact a FET OpAmp might work better on an I/V stage. But hard to judge just on personal experience. OK, maybe let's try the OPA211 also in a RPi-DAC setup first and measure there the performance (to measure is always a great idea).

Nevertheless, let's try to continue with the concept, actually extend it already by a Bi-DAC approach and let's check how we can handle the sum function. Just to bear in mind: I check the quality always with a headphone, using another power amp and loudspeakers might complicate the things. So, we need also a well selected OpAmp to sum the Bi-DAC channel output and to provide a nice headphone or line driver OpAmp.

Let the bird sing our favorite songs!