noise performance between 1Hz and 100Hz is very critical.
The “Output Noise Spectral Density” they can cope with should be
<20nV/sqrt(Hz) @ 10Hz –100Hz and < 30nV/sqrt(Hz) @ 1Hz –10Hz.
In the AD5449 datasheet, page 4, The Output Noise Spectral density has been
specified to 25nV/sqrt(Hz) @ 1kHz, but on page 13, figure 35, the value is
~140nV/sqrt(Hz). Which value is correct or does the used AD8038 make the
They also would like to know what are the AD5449’s maximum noise values at
frequencies 1Hz -- 10 Hz and 10Hz – 100 Hz.
Is there any other DAC which could fulfill the above mentioned requirements ?
The AD5449 (and the AD5415) are dual CMOS multiplying DACs (MDACs). They are
made with silicon/chromium thin film resistors and CMOS switches. Since the
resistance of the CMOS switches is much lower than that of the resistors in the
network the noise of such devices is quite close to the thermal noise of the
The resistance is ≈10kΩ, from which we calculate noise of 13 nV/√Hz the data
sheet allows a factor of two safety margin and specifies 25 nV/√Hz . Note that
if we have logic noise on the data lines or device Vdd then that will couple
capacitively and make things worse, but the basic noise of the device,without
external interference, is as specified. It should meet Neuromag's needs, as
should the AD5415.
The reason that the noise in the graph on page 13 is so much worse is that this
performance is measured with an AD8038, which has a 1/f corner frequency (for
current noise) approaching 10 KHz.
If you compare Fig 35 on the AD5449 DS with Fig TPC20 on the AD8038 DS you will
see what I mean. The problem is the current noise of the amplifier, not the