MAX541/MAX542
+5V, Serial-Input, Voltage-Output, 16-Bit DACs
_______________________________________________________________________________________ 9
Applications Information
Reference and Analog Ground Inputs
The MAX541/MAX542 operate with external voltage ref-
erences from 2V to 3V, and maintain 16-bit performance
if certain guidelines are followed when selecting and
applying the reference. Ideally, the reference’s
temperature coefficient should be less than
0.4ppm/°C to maintain 16-bit accuracy to within 1LSB
over the 0°C to +70°C commercial temperature range.
Since this converter is designed as an inverted R-2R
voltage-mode DAC, the input resistance seen by the volt-
age reference is code-dependent. The worst-case input-
resistance variation is from 11.5kΩ (at code 8555 hex) to
200kΩ (at code 0000 hex). The maximum change in load
current for a 2.5V reference is 2.5V / 11.5k Ω = 217µA;
therefore, the required load regulation is 7ppm/mA for a
maximum error of 0.1LSB. This implies a reference
output impedance of less than 18mΩ. In addition, the
impedance of the signal path from the voltage
reference to the reference input must be kept low
because it contributes directly to the load-regulation
error.
The requirement for a low-impedance voltage reference
is met with capacitor bypassing at the reference inputs
and ground. A 0.1µF ceramic capacitor with short leads
between REFF and AGNDF (MAX542), or REF and
AGND (MAX541), provides high-frequency bypassing.
A surface-mount ceramic chip capacitor is preferred
because it has the lowest inductance. An additional
10µF between REFF and AGNDF (MAX542), or REF
and AGND (MAX541), provides low-frequency bypass-
ing. A low-ESR tantalum, film, or organic semiconductor
capacitor works well. Leaded capacitors are accept-
able because impedance is not as critical at lower fre-
quencies. The circuit can benefit from even larger
bypassing capacitors, depending on the stability of the
external reference with capacitive loading. If separate
force and sense lines are not used, tie the appropriate
force and sense pins together close to the package.
AGND must also be low impedance, as load-regulation
errors will be introduced by excessive AGND resis-
tance. As in all high-resolution, high-accuracy applica-
tions, separate analog and digital ground planes yield
the best results. Tie DGND to AGND at the AGND pin to
form the “star” ground for the DAC system. Always refer
remote DAC loads to this system ground for the best
possible performance.
Unbuffered Operation
Unbuffered operation reduces power consumption as
well as offset error contributed by the external output
buffer. The R-2R DAC output is available directly at
OUT, allowing 16-bit performance from +V
REF
to AGND
without degradation at zero scale. The DAC’s output
impedance is also low enough to drive medium loads
(R
L
> 60kΩ) without degradation of INL or DNL; only
the gain error is increased by externally loading the
DAC output.
External Output Buffer Amplifier
The requirements on the external output buffer amplifier
change whether the DAC is used in the unipolar or
bipolar mode of operation. In unipolar mode, the output
amplifier is used in a voltage-follower connection. In
bipolar mode (MAX542 only), the amplifier operates
with the internal scaling resistors (Figure 2b). In each
mode, the DAC’s output resistance is constant and is
independent of input code; however, the output amplifi-
er’s input impedance should still be as high as possible
to minimize gain errors. The DAC’s output capacitance
is also independent of input code, thus simplifying sta-
bility requirements on the external amplifier.
In bipolar mode, a precision amplifier operating with
dual power supplies (such as the MAX400) provides
the ±V
REF
output range. In single-supply applications,
precision amplifiers with input common-mode ranges
including AGND are available; however, their output
swings do not normally include the negative rail
(AGND) without significant degradation of performance.
A single-supply op amp, such as the MAX495, is suit-
able if the application does not use codes near zero.
Since the LSBs for a 16-bit DAC are extremely small
(38.15µV for V
REF
= 2.5V), pay close attention to the
external amplifier’s input specification. The input offset
voltage can degrade the zero-scale error and might
require an output offset trim to maintain full accuracy if
the offset voltage is greater than 1/2LSB. Similarly, the
input bias current multiplied by the DAC output resis-
tance (typically 6.25kΩ) contributes to the zero-scale
error. Temperature effects also must be taken into con-
sideration. Over the 0°C to +70°C commercial tempera-
ture range, the offset voltage temperature coefficient
(referenced to +25°C) must be less than 0.42µV/°C to
add less than 1/2LSB of zero-scale error. The external
amplifier’s input resistance forms a resistive divider with
the DAC output resistance, which results in a gain error.
