Introduction: DIY SEPIC Converter

A while back, I decided to design and create my own DC/DC converters as DIY converters were usually much cheaper than the ones that could be purchased online. After messing around with both buck and boost converters, I started to wonder if there were any designs that were a combination of both, meaning a type of DC/DC converter that could both step-up and step-down the input voltage.

A single-ended primary inductor converter (SEPIC) is a type of DC/DC converter that allows the input to be greater than, equal to or lower than its output. Of course, it isn't the only type of DC/DC converter that can step-up or step-down, others include buck-boost, Ćuk, and flyback converters. In design, the SEPIC, Ćuk and buck-boost are similar. However, the SEPIC has a non-inverting output and uses a series capacitor. A non-inverting output means that the polarity of its output is the same as its input while the series capacitor means that if a short were to occur on either the input or output side, the opposite side would not be affected.

Supplies

All the following parts were bought from taobao, however, you are free to choose which supplier you want to obtain the following parts from. All the capacitors are MLCCs.

In case you do not think you will use or need that many resistors ever, I have provided the following links to resistor values used in this project:

Step 1: Understanding SEPICs

Firstly, some basic knowledge regarding some electrical components:

Capacitors oppose changes in voltage. For example, if the voltage across a capacitor increases, the capacitor will try to store that energy, if the voltage across the capacitor decreases, the capacitor will discharge as an attempt to keep the voltage across it constant. Similarly, inductors oppose changes in current and will do the store the extra current, or induce a voltage to try and raise the current through it. Capacitors also only allow AC to pass through it, but not DC. Diodes only allow the current to pass through it in one direction, providing that the voltage across it is above a certain threshold (usually 0.7 volts, but this value can differ).

Now, we need to understand how a SEPIC works. SEPICs are in a continuous conduction mode (CCM) when the current through both inductors never reach 0. During this steady-state operation, the following steps take place

  1. When the switch, S is first turned ON, current flows through L1. Since inductors oppose changes in current, the inductor, L1 will store the energy in the form of a magnetic field (this is also how electromagnets work). Thus the diode, D is reverse-biased, meaning it doesn't conduct and there is no voltage output.
  2. When S is turned OFF , there will no longer be current flowing through L1. This sudden change in current causes L1 to induce a flyback voltage in order to oppose said change. This voltage is not purely DC and thus can flow through the capacitor. The diode is also forward biased and can conduct, allowing capacitors C1 and C2 to charge up. The final output voltage is equivalent to the input voltage + the voltage across L1.
  3. S is turned ON again. As L1 charges up, C1 discharges into L2, thus both inductors charge up. As C1 discharges, D becomes reverse-biased while the output voltage stays the same due to the output capacitor.
  4. S is turned OFF again. This time, both inductors, L1 and L2 charge up the capacitors, C1 and C2 respectively, and the cycle repeats from step 2 onwards.

In an actual SEPIC, we will use a MOSFET in place of a switch, and a boost controller to switch that MOSFET at really high speeds. The more time it spends ON, the higher the output voltage will be, and vice versa.

As stated before, the capacitor in series can protect components in the event of a short circuit. Since capacitors block DC, if a short were to occur on the input side or output side, that would not be able to translate to the opposite side, thus protecting the components.

Step 2: Choosing Components for the SEPIC

Next, we need to decide exactly what values and components we should use.

The calculations will be done according to the formulas shown in the video.

In this design, we will assume that the input voltage ranges between 5V and 24V while the output voltage is 12V. Additionally, the max load current is 2A. Additionally, I picked the NCV887200 as the MOSFET controller, which outputs a PWM signal to rapidly switch the MOSFET ON and OFF. This will determine your switching frequency, in this case, according to its datasheet, my nominal switching frequency should be around 675kHz. You are free to choose which MOSFET controller and what switching frequency your SEPIC will operate on, although while high switching frequencies can reduce ripple current and voltage, it can result in higher power losses, and low switching frequencies result in lower power losses but also in greater ripple current and voltage, which can damage components.

Next, we need to calculate the maximum and minimum duty cycle. The duty cycle refers to the ratio of the time a signal is ON, compared to the total period of a signal, and is expressed as a percentage. In my design, I took the voltage across the inductor as 0.3V, the voltage across the capacitor to be 2V and the voltage across the diode to be 0.5V, thus the forward voltage would be 2.8V. In the end, my maximum duty cycle is 74.7%, while my minimum duty cycle is 38.1%.

Following the remaining formulas and assuming that the efficiency is 80%, the ripple current is 1.8A, and my inductor has to be at least 1.5uH. Since two of these inductors were required for the design, a pair of coupled inductors with a 1:1 turn ratio was chosen. Assuming that the voltage ripple is 100mV, and the ripple voltage across Cp is 700mV, my output capacitance would be at least 221uF and Cp would be at least 3.2uF.

Finally, ensure that there are ceramic capacitors parallel to the input voltage as well.

Step 3: Choosing Components for the Controller

Following the datasheet for the NCV887200, a 1.0uF ceramic capacitor is placed in series with VDRV, with a 15kΩ pull-down resistor in series with GDRV.

A 100kΩ pull-up resistor is placed on EN while a 100kΩ resistor is placed in series with the gate of the MOSFET to limit the gate current and voltage. Neither of these resistors need to be any specific value; I merely picked 100kΩ out of convenience.

As for the compensation components, the following values are chosen: R2 as 3kΩ, C1 as 22nF, and C2 as 1.8nF.

For the shunt resistor placed in series with the source of the MOSFET, a 9mΩ resistor was picked. When picking the shunt resistor, ensure that the value is less than 1Ω since any current through the load must pass through the shunt resistor and result in power losses.

For the feedback resistors, the following values are chosen: 90kΩ as R1 and 10kΩ as R2.

Step 4: Designing the PCB

For the PCB (printed circuit board), I designed it with Autodesk EAGLE 9.5.2

The schematic is made up of all the components decided on earlier. Almost all the components are surface-mounted devices (SMDs) in order to save as much space as possible, and preserve its compact form factor.

As for the board, any connections and traces should be as short as possible. Additionally, any traces or connections that the load current must flow through, must be thick enough to handle that much current. In this case, my maximum load current is 2A. After inputting that data into a trace width calculator, and assuming that the thickness is 1oz/ft^2 and the temperature is allowed to rise by 10-degree celsius, the traces have to be at least 0.781mm thick. In this case, I chose 1mm.

The schematic and brd files will be provided down below.

Finally, the PCBs were ordered from jlcpcb.

Attachments

Step 5: Soldering the Board

The custom PCBs should arrive in about a week or two. After arrival, immediately check for any visible defects in its design. Most of the time, JLCPCB does an amazing job at producing the PCBs according to design, however, defects can happen sometimes.

Since almost all of the components on this PCB are SMD components, soldering this board would require a bit of prior experience.

If my instructions are confusing, or you have trouble following, simply follow the video above (skip to 4:04)

Firstly, coat the soldering iron with a bit of solder. For all the SMD resistors and capacitors, add solder to the metal pad corresponding to one side of the component first. Then, simply position the respective component correctly, before placing the soldering iron on the metal pad with a thin layer of solder, melting the solder, while you push the component into place. Let the solder work its magic and "stick" the component to the PCB. Afterwards, rinse and repeat for the other side of the component.

For the controller, carefully position the component correctly (ensure that the dot on the component in on the same corner as that on the footprint), before placing the soldering iron on one of the metal leads of the controller. This should ensure that the component can now stay in place while you solder the rest of the leads.

Components such as the inductors, MOSFET and diode will be slightly harder to solder since their metal leads extend below the actual component itself. I have also linked a second video, specifically for such components. While the component in the video is not the same as the ones we are using, the same principles apply. In such cases, I usually make use of a heat gun, or a hot air gun, which melts the solder underneath a component via blasting hot air at it, as well as a bit of flux. Firstly, coat all the metal pads with a thin layer of flux. Then, coat the metal pads on the PCB with a thin layer of solder. Next, position the component correctly above the metal pad. Finally, aim the hot air gun at the component, and wait for the solder below it to melt while pressing the component down. After some time, the solder underneath should melt and the component can be gently pressed down and "stick" to the metal pad.

Lastly, the only through-hole components on our board are the screw terminals. Soldering them is pretty straightforward. Secure the screw terminals in their respective positions first. Coat the soldering iron with a little bit of solder, before touching it to the component leg and metal ring at the same time. Continuously push the solder into the iron and feed it into the joint. The solder should flow around the leg and fill the joint. If it fails to flow around the whole leg, touch the soldering iron to the unsoldered side and feed the solder into the joint there.

Just like that, with a bit of patience, and a decent amount of time and effort, you have just soldered components onto a custom PCB.

Step 6: Testing

Finally, you are ready to test your creation.

Since we are simply making a voltage converter, testing it should be quite straightforward. Attach a voltage supply to the input side of the SEPIC. This can be a battery, or a lab bench power supply. Since the main function of a SEPIC is to maintain a constant output voltage despite changes in input voltage, the latter is recommended. If everything goes well, your output voltage should be maintained at the expected level while the input voltage can vary.

If there is no output voltage at all, it is quite possible that some of the components are not soldered correctly, or that there are faulty connections created by unintentional solder bridges.

If there is an output voltage, but the output voltage is not what you expected (e.g. higher or lower than expected) it is possible that the components that you have picked out are of the wrong value, or some components are not soldered correctly.

Step 7: Resources

While this project may have been more technical and a bit more focused on theory rather than the hands-on experience, I hope that you had as much fun as me learning about how a SEPIC works, and making your own one.

Here are all the resources that I used for this project:

Understanding SEPICs:

Soldering: