500mA of Current with a 1% Voltage Output Accuracy: A Low Noise LDO from Maxim Integrated

February 07, 2018 by Nick Davis

Maxim Integrated recently announced their new series of linear regulators that maintain ±1% output accuracy over a wide operating temperature range with only 12µVrms of noise.

Maxim Integrated recently announced their new series of linear regulators that maintain ±1% output accuracy over a wide operating temperature range with only 12µVrms of noise.

Maxim has released a physically small LDO that offers low output noise and a ±1% voltage output accuracy (despite significant load, line, and temperature variations), provides configurable output voltage, and comes with many protection features. The MAX38902x series of LDOs offers four regulator types to choose from (they all share the same datasheet):

Offered in Small Packages

Two of the devices from this series, namely the MAX38902A and MAX38902B, are available in a 2 × 2mm, 8-pin TDFN, or thin DFN, package. The other two devices—the MAX38902C and MAX38902D—come in a much smaller package, specifically a 2 × 3 bump, 1.22 × 0.82mm, 0.4mm pitch wafer-level package (WLP).

The figure below shows an overview of these two package types. You can find more detailed information in their app note (Application Note 1891) "Wafer-Level Packaging (WLP) and Its Applications" (PDF).


Figure 1. An overview of the two package options (TDFN and WLP) and their pinouts, from the datasheet (PDF).

The Major Differences between the Devices in This Series

Apart from the package type, the primary difference between the various members of the MAX38902 family is the manner in which VOUT is set.



Part number MAX38902A has a very novel approach for setting its VOUT, and one that definitely has the "cool" factor in my eyes. Personally, I haven't seen this approach used before, but if it's rather common, or at least not so unique, please let me know in the comments section.

VOUT is determined by configuring two of the IC's pins—appropriately labeled SELA and SELB—in one of three possible arrangements: left unconnected (floating), tied to IN, or connected to GND. These nine possible configurations allow for nine predetermined VOUT settings (see the table below). What's even more surprising (AKA cool) is that a new regulation voltage can be selected by simply changing the pin states and then “rebooting” the regulator by either cycling power or toggling the enable pin. Pretty neat!


Figure 2. MAX38902A VOUT is set by use of the SELA and SELB pins. Table taken from the datasheet (PDF).



The MAX38902B and MAX38902C use two external feedback resistors for settings the VOUT value, which can range from 0.6V to 5.3V. And to help minimize the feedback (FB) input bias current error, Maxim recommends that R2 should be 300kΩ or less. Here is the equation for choosing the R3 value, along with an example:


Figure 3. MAX38902B/C R3 equation and example, from the datasheet (PDF).



Setting the output voltage of the MAX38902D is the easiest since you don’t really have to set it at all. The device comes pre-programmed according to the part number (see table below). The table shows the "standard" VOUT values, and any voltage between 0.7V and 5.3V, in 50mV steps, is available by special order.


Figure 4. MAX38902D standard VOUT options, from the datasheet (PDF).


A final comment on the four different part numbers: Although Maxim has provided ample information on the various parts within this series of LDOs, be mindful that (as of this writing) only the "A" version is available for purchase (see the figure below).


Figure 5. Only the "A" version is currently available for purchase. Table taken from the datasheet (PDF).


I suspect, however, that it won't be too long before all the versions are available. If you get your hands on one of the other versions, please let us know about your experiences in the comments.

Device Vout Accuracy

Regarding part numbers MAX38902A and MAX38902D, the stated output accuracy of ±1% is straightforward. For the MAX38902B and MAX38902C, the output accuracy is influenced by the FB voltage accuracy (±1%) and the precision of the voltage-setting resistors R2 and R3, so keep the tolerance in mind when selecting these two resistors.


Figure 6. VOUT accuracy specifications, from the datasheet (PDF).
Figure 7. For the “B” and “C” versions, remember that the precision of the voltage-setting resistors influences the accuracy of the output voltage.

Low Output Noise

Although the datasheet states that these LDOs can provide up to 500mA of output current with only 12µVRMS of output noise, the associated specifications table lists an IOUT condition of only 100mA (not 500mA) when presenting the 12µVRMS output noise guarantee. So if your design requires more than 100mA of output current, it would be best to consult Maxim about the details of their 12µVRMS output noise specification.

An Evaluation Kit Is Available

If you're interested in testing this series of LDOs, Maxim has released the MAX38902EVKIT evaluation kit. As you can see in the datasheet, Maxim has created a single eval kit that includes all four devices in this series. Nice work, Maxim!


Image used courtesy of Digi-Key.


Have you had a chance to use any of these low-noise LDOs from Maxim, or the evaluation kit? If so, leave a comment and tell us what you think.

  • A
    Andref25 February 08, 2018

    As far has i know that method to setting the output voltage isn’t very common but I’ve found this Linear Technology regulator which uses a similar approach

    Like. Reply
  • A
    a.genchev May 21, 2019

    What I miss in these datasheets is the graph of output voltage at load (say 20 mA) vs temperature (-25..80 C) . Because the main function of this is to give stable voltage and the temperature we usually can not guarantee.
    The noise is low for the 500 mA class, but usually if I need low noise, I go in the 150 mA (max) class, where I have used MIC5205 for it’s a bit lower noise (looks like half of these 12µV) which I reduce further by increasing the bandgap bypass capacitor to 3.3 nF. Comparing the noise p-p is a bit difficult, because other datasheets specify only spectral density as a graph…

    Like. Reply