Universal Voltage Converter, or a Few Words About What SEPIC Is

In today's review, I want to tell you about a rather useful device: a universal voltage converter.
What it is, how it works, and what it can do—as always, read on.

Some time ago, in one of my reviews, I mentioned this type of converter and even built one as an example. Today it's time to review a ready-made converter of this type.

First, just a few words about what this clever converter actually is.
Converters usually come in three types.
1. Step-up
2. Step-down
3. Inverting

But none of them can produce an output voltage higher or lower than the source voltage.
For example, a step-down converter can't turn 10 V into 12 V, and a step-up converter can't turn 20 V into 5 V.
But sometimes the input voltage can vary above and below the required output voltage during operation.
For example, you need 12 V (say, to power a hard drive or monitor), but the power comes from a car's electrical system, where the voltage can be anywhere from 10 V to 14.5 V.
This problem is usually solved in one of two ways.
1. Step the voltage up to 15–20 V, then step it down to the required level.
2. Use a step-up/step-down converter, also known as a Buck-Boost or SEPIC converter.

First, a little background information.
The converter arrived with another item and was simply packed in a zip-lock bag.


The store's website lists the following specs:

Input voltage — 4 V–35 V
Output voltage — 1.23 V–32 V
Output current — 3 A max
Maximum power — 25 W
Dimensions: 50 x 25 x 12 mm

Here's what those specs mean.
The output current cannot exceed 3 A, and the output power cannot exceed 25 W.
In other words, you have to stay within whichever limit you hit first.
You can get 10 V at 2.5 A (25 W), or 5 V at 3 A (15 W).
In fact, the specs differ from the advertised ones, but more on that later.

The board looks quite neat. You can see the trimmer for adjusting the output voltage (the current is neither adjustable nor limited).


You can also see two inductors on the board, one of the telltale signs of a SEPIC converter, though not an essential one. Sometimes a single inductor with two windings is used instead, but it looks different too.


And here's the circuit board from above :)


The underside is bare. You can see the vias that could conduct heat to the bottom of the board, but their placement doesn't make much sense. Most likely, they serve mainly as electrical connections.
It's a shame—they could have improved heat dissipation, but this is better than nothing.


I think this photo makes the board's dimensions easier to judge :)


Okay, we’re done with the appearance. Now let’s take a closer look at what this thing actually is.
I’d really like to explain in detail what it is and how it works. The problem is that this type of converter is not easy to describe simply. What’s more, even while preparing material for this review, I came across conflicting explanations.

First, here’s the block diagram for this type of converter. It’s worth noting that there are two topology variants for this converter; I’ll show the one used by the board under review.


Next, I’ll try to “hand over the mic” to specialists with more experience.
While searching, I came across an explanation that I think is the most accurate. Here’s a link to the original article; below, I’ll quote a brief description of how it works.

In the diagram, the power switch is closed. When the switch is closed, the input inductor charges from the source, while the second inductor charges from the capacitor. Meanwhile, the output capacitor supplies current to the load.
No energy is delivered to the load during this time; the current directions in the inductors and the capacitor voltage polarities are shown in the diagram. The fact that both inductors, L1 and L2, are disconnected from the load while the switch is closed makes the control characteristics more complicated, as we’ll see later.


Once the switch opens, the circuit takes on a somewhat different “look.”
When the switch is open, the first inductor charges capacitor C1 and also maintains current to the load, as shown in the diagram. The second inductor is also connected to the load at this point.


Put simply, the circuit works by transferring energy back and forth between its components, allowing it to step the voltage up or down.
To make things clearer, I’ll show where all these components are on the board.
By the way, one of the telltale signs of a SEPIC converter is that it has one switching element (whether a transistor or a power PWM controller) and one diode.


I drew the schematic for this board. The values of a couple of components may differ slightly from the actual ones, but otherwise it’s accurate.
One drawback I’ll point out right away is that the trimmer resistor is connected to the output rather than ground. This connection is strongly discouraged, because if contact is lost while adjusting it, the maximum output voltage will be applied to the output.


The heart of this board is a small PWM controller, which drives a power MOSFET and regulates the output voltage.
The PWM controller is an FP5139; here’s a link to the datasheet.
This PWM controller operates at 500 kHz, which is pretty good. Its input voltage range is 1.8–15 V, which is also nice, especially the low end of 1.8 V. I think I’ll buy a few of these chips separately.
The controller drives an 088N04L MOSFET. It’s a 40 V, 50 A transistor with an on-resistance of 8.8 mΩ, and it can be driven by a logic-level signal (usually 5 V).


Another distinguishing feature of a SEPIC converter is its high-capacitance ceramic capacitor.
In general, a SEPIC differs from other converters in that it contains more components.
Classic boost, buck, and inverting converters have three main components, connected in different configurations: an inductor, a transistor, and a diode.
Here, another inductor and a capacitor have been added to the mix.
The output diode on the board is SK86, a pretty decent diode rated for a maximum current of up to 8 A.


Next, I moved on to testing.
When I put together this kind of “test setup,” I almost felt sorry for the converter.
“They’re going to tear it apart like a dog with a hot-water bottle,” I thought. As it turned out, I wasn’t far off the mark.


First power-up.
Let me explain what all those numbers on the displays mean.
The power supply is on the left.
Top row: output voltage and output current.
Bottom row: output power and the amount of charge delivered to the load in mAh (not important in this case).

The electronic load is on the right.
1. Set current, cutoff voltage (not important in this case).
2. Measured load current, measured input voltage (the converter's output voltage).
3. Accumulated capacity (not important in this case), load power (current × voltage).
4. Not important.


Next, I ran the converter through various operating conditions. I chose them somewhat spontaneously, while also measuring the temperature of the main components and recording the results in a table.


I didn't raise the input voltage above 14 V. I'll explain why below.


Based on the temperature measurements, I can say that the board doesn't deliver the advertised performance.
There's one caveat, though: it falls short because it overheats. The power components can handle the rated output for a short time, but they overheat during prolonged operation.
You could add a heatsink, of course, but you'd need to cool the transistor, both inductors, and the diode. That's tricky :(
I also noticed a slight drop in output voltage as the converter warmed up. This is often because non-precision resistors are used, and their resistance “drifts” as they heat up. The change isn't very large, though, so it can be ignored.


Since this type of converter is said to have higher efficiency than other designs, I decided to check that too.
For a demonstration, I ran a simple experiment. To make the results easier to see, I set the operating conditions so that the input power was always 10 W (or thereabouts). In this setup, the output power is equal to the converter's efficiency.
The actual efficiency will be higher, since this setup doesn't account for losses in the wires. But they're short, so the measurement error is unlikely to exceed a couple of percent.


A few more photos in different modes: step-up, step-down, and with different voltage settings.
By the way, you can also calculate the efficiency from the previous photos. Just divide the measured load power (on the right) by the measured power from the source (on the left).
For example, the power supply reads 15.45, and the load reads 12.3. 12.3 / 15.45 = 0.796
But even this is enough to say that its efficiency is higher than that of a step-up converter combined with a step-down converter.




Above, I mentioned that I limited the input voltage to 14 V.
There was a reason for that. I started testing first and only redrew the schematic afterward.
At first, I thought the manufacturer had simply followed the datasheet schematic, with a transistor on the board to control switching the converter on and off (by the way, one advantage of SEPIC is that you can turn off the output; for example, you can't turn off a step-up converter), and that the input voltage shouldn't exceed 15 V (according to the controller datasheet). I was about to complain that the specified input range was 35 V.
But when I started tracing the circuit, I realized the manufacturer had been cleverer: they added a voltage regulator on the board that supplies about 9.5 V. I wouldn't assume all boards are built this way, so be careful.
What threw me off was the regulator's pass transistor, since the datasheet schematic also includes a transistor.
By the way, the jumper on the board controls the converter's on/off state.


After figuring out the circuit, I decided to continue testing, but I burned out the board before I even got started.
The power transistor shorted out, and I didn't even understand how it happened.
I rummaged through my stash and found some motherboard, from which I desoldered a MOSFET in the same package. The difference is that this one is rated for only 30 V :(
I quickly soldered it in, and luckily nothing else failed.
By the way, this converter is somewhat “safe”: if the power transistor fails, it won't put the full supply voltage on the output, unlike a step-down converter.
One more thing: this type of converter has more output ripple than other types, but much less ripple on the input, which is an advantage when running from batteries.


Next, I wanted not only to continue testing, but also to figure out why the transistor had failed.
During testing, I noticed that efficiency dropped as the input voltage increased.
For example, with a 15 V output, efficiency was 80% at an input of 20 V, but only 62% at 26 V.
And the higher the output voltage, the lower the efficiency. At a 20 V output, I could easily get an input current above 2 A and efficiency below 40%.
Then I remembered there had been a small blob of solder near the transistor that wasn't there before it failed. The output voltage after my last experiment was 25 V, and I'd cranked the input up to almost 30 V. It didn't even have time to squeak.
So it looks like the transistor literally “cooked.” This was most likely caused by the inductors starting to saturate.
Of course, SEPIC can operate over a wide voltage range, but the optimal range still depends on the components used, so you can't cover everything.


Experiments showed that the lower the output voltage, the higher I can raise the input voltage.
With 10 V at the output, I easily cranked the input up to 27 V. I didn't go any higher because the transistor's maximum voltage is only 30 V.
This is perfectly normal; you just need to keep it in mind when using the converter. In other words, it's a characteristic rather than a fault.


I won't go through the pros and cons; I think they're clear from the review, but I'll briefly sum up what we've learned.
1. The converter works and delivers higher efficiency than a step-up + step-down converter combination.
2. The board's specs are overstated, but with the right input/output voltage combination, you can get 3 A and 25 W out of it.
3. The components used are quite good. However, the inductors should be rated for higher current, and the transistor needs additional cooling.
4. The board has a power supply regulator for the PWM controller, allowing the input voltage to be raised above 15 V.
5. With certain input and output voltage combinations, the power transistor breaks down. :(

Overall, the board works well enough, but it has some limitations, as described above.
It's suitable for powering devices with low current consumption over a wide input voltage range, but it won't work for high-power devices because it overheats.
I saw a short review of this board online. The results were a little different, but what they said about the protection was rather unclear. It tripped once for me: there was no output voltage until I disconnected power to the board. But I don't understand how it detects an overload, since there are no current sensors, even though the datasheet claims short-circuit protection, and it did trip…

I hope you found the review interesting and useful. If you'd like, I can test it with other input/output voltage combinations.


The product was provided by the store for review.