In my last post I went through the steps for selecting and designing a linear regulator power supply. This post will go through the steps for a switching power supply. I have a lot of experience with linear regulators and only a little bit with switchers, but I’ve picked up a little knowledge.


Switching Power Supply: The Simple Method

Step 1. Go to your favorite supplier (I’ve used Texas Instruments and Linear Technology, but there are many others).

Step 2. Enter your values in their selection tool and select a part from the results.

Step 3. Follow the guidelines in the datasheet for part selection.

Step 4. Follow the layout guidelines as closely as possible.


That’s it. IC suppliers produce some great parts that make it easy to put a switcher in your design. 10 years ago circuit designers needed to know all about switching power supplies, but these days it is almost drag-and-drop unless the design has extreme requirements or operating conditions.


Switching Power Supply: A Little More Detail

The requirements I am using are here (see my last post for details).

Input Voltage: 2.75 to 4.2V

Output Voltage: 3.0 or 3.3V

Supply Current: 100mA


Secondary requirements are a small and simple design with few parts that are commonly available and packages that are easily soldered at home (SOP, DIP, SOIC, QFP). Any grid array packages (LGA, BGA) are essentially impossible to solder at home and any no lead packages (QFN, DFN) are more difficult, especially if there is an exposed pad on the bottom.


The selection guides led me to Linear Tech LTC3534 in an SSOP package and available in small quantities from Newark for $6.80 and Texas Instruments LM3668 in a DIP package and also available from Newark for $4.35. A quick review of the datasheets indicates that both are suitable for this application. The TI part shows a 3.3V Typical Application circuit on the first page and the LT datasheet on page 17 shows a circuit specifically for a 3.3V supply from a Li-Ion battery.


I am choosing the LT part, mainly because the SSOP part will make layout smaller and easier and the TI data sheet doesn’t list the DIP version, so if there are any issues I might be in trouble. I suspect it is old stock that TI doesn’t support any more and I could find an old datasheet out on the web, but I think I will be happy with the LT part.


Datasheet Review

The Absolute Maximum Ratings (datasheet page 2) table is all suitable. I need to remember to never apply more than 8V, which is ok since highest voltage on the board will be 5V when the USB is powering the board. I doubt there will be any thermal issues since it is rated for 500mA but will supply less than 100mA.


There are a few items of interest in the Electrical Characteristics (page 3). Startup can be an issue with switchers. They are much more complicated than simple linear regulators. The startup voltage is 2.4V max which is below my minimum battery voltage of 2.75V. Input and output voltage will work across the full range of operation. None of the other parameters in the table look like trouble. The feedback voltage is 1V which will be important when setting the output voltage. The current limit is 1A minimum which needs to be checked, but the current in the application is well below the parts capability so it is unlikely to be an issue. It is more likely that low current will be an issue. Some switchers don’t like supplying very light loads.


There are a few items of interest in the Pin Functions (page 7). LT recommends an RC filter on the RUN/SS pin to ensure good soft start behavior. SW1 and SW2 layout traces should be as short as possible to minimize EMI issues.


The PWM pin controls regular PWM mode vs. Burst mode which is better for light loads. I suspect I will want to operate in burst mode all the time, but I will provision for both options and for controlling it from the microprocessor. Note that the TI part would automatically switch between modes as needed.


The equation in the Buck-Boost section indicate it will spend most of the time in Buck-Boost mode, entering Buck-Boost at 3.77V and switch to Boost mode at 2.89V.


The Burst Mode Operation section is important since the current draw will be low so it will probably operate in burst mode for max efficiency. The first point of interest is 2% ripple on the output supply. This could be an accuracy issue when taking measurements if the circuit is ratiometric to the supply voltage. This can be reduced with increased capacitance and adding a small capacitor across the upper feedback resistor. This will be added to the schematic in case it is needed.


Using the efficiency equation on page 11, the efficiency in burst mode at 89.6% at 10mA and 90.0% at 100mA, so the circuit will have good efficiency across the normal operating range. If the current is at the higher end it will probably be a little more efficient in normal PWM mode.


Starting with the Inductor Selection in the Component Selection (page 12-13) section, the equations on page 12 results in:

               LBOOST = 4.91 uH

               LBUCK   = 7.07 uH

This design will spend most of the time in Buck-Boost and Boost modes. The datasheet doesn’t offer any guidance as to whether it is better to err on the side of higher or lower inductance. LT shows 3.3uH in the circuit diagram on page 17. The Newark selector page for Power Inductors returns lots of choices for pretty low cost. I chose the Taiyo Yuden NRS50xx series for no better reasons than they are cheap, it is a family of parts in the values I might need and I have the Eagle geometry.


Looking at the Output Capacitor Selection equations on page 13 and using 3.3uH and 4.7uF we can calculate:

               VP-PBoost = 0.0038 or 0.12%

               VP-PBuck = 0.0057 or 0.17%

Those values look very good and shouldn’t cause any trouble, but remember that in Burst mode the ripple can be higher. They recommend 22uF for transient response, so that is a good starting point for the design. They recommend low Equivalent Series Resistance (ESR) capacitors, which to me means ceramic. The data sheet offers guidance for specific tantalum caps that are suitable, but I’ll use ceramic caps unless there is a need for higher ESR (some regulators need resistance for stability). I have some 4.7u and 10u caps in my parts stock which should do nicely for both input and output capacitors.


Beyond that I am following the recommended circuit on page 17.