Voltage converter + control module with a nice design

Not long ago, I came across an interesting kit on Taobao—or, more accurately, not a kit but a converter control board, since it’s also sold separately. Curiosity got the better of me, so I decided to order one and give it a try. There was also a useful story involving the seller, though. Anyway, it’s best to read about it in the review.


I really like converters, partly because I’ve had a lot of them over the years. This time I came across a model I hadn’t seen before, and it caught my interest not so much because of the description—which was pretty sparse—as because of the video demo posted by the seller. It works quickly, looks good, and has a large display, unlike the various DPS models. But, as always, "there was one catch": everything was in Chinese.
So I decided to message the seller, and he replied almost immediately. I asked him this: “Hey, I saw this great device in your store. There’s just one problem: I’d like one too, but with mother-of-pearl buttons an English menu.” At first, the seller said something like, “No, probably not. This is just a hobby of mine; I only sell on Taobao on the side.” I thought, “Wow, great!” and told him I work on similar things too, and that maybe we could make it happen—I believed in him!
In the end, the seller, who I understand is also the developer, translated the entire menu into English over a couple of free evenings. What’s more, he demonstrated everything in a video. He did mention that the Chinese labels are shorter, and the English ones don’t fit very well :(


I told him that worked for me. Great. I also explained that I’d be buying through an agent rather than directly, and we agreed that I’d include the "code phrase" “for kirichblog” with the order, and he’d send the right version. After that, everything went smoothly: the agent placed the order, the seller shipped it, and some time later I received it. Now I’m sharing my impressions.

The packaging is standard: antistatic bag plus a regular bag. There are two packages because these are actually two separate listings: the converter and the control board. But first, of course, I checked with the seller to make sure they were compatible, so there wouldn’t be any surprises later.


Here’s the complete set: the converter, control board, and wires to connect everything together.


I’ll start with the converter. It’s a small, neat board and costs just over $5. Shipping to the agent costs another $3, but that’s for both items together — link.
Specifications (Google-translated in places):
Input voltage: 12–50 V, 36–45 V recommended
Output voltage: 0–30 V (provided the input voltage is higher than the required output)
Output current: 0–7.5 A
Topology: Buck
Switching frequency: 58 kHz
Current/voltage adjustment input: 0–5 V.
Independent overcurrent protection: if the current exceeds the set value by more than 2.3 times, the protection should shut down within no more than 0.1 µs.
Soft-start function
Initial output offset: -0.5 V
Output ripple: <50 mV pp (input 36 V, output 30 V, 7 A)
Output noise level: <80 mV pp; test conditions: 0.5 m twisted pair, 20 MHz bandwidth
100 Hz sine-wave transfer ratio: less than 1/10000
Output voltage variation with changes in input voltage: less than 1% with a DC input of 20–50 V and an output of 15 V, 3 A
Output voltage variation with changes in load: less than 10 mV with a 36 V DC input and a 30 V output, from no load to a 6 A load
Constant-current stability: less than 1.5% + 8 mA
Typical efficiency:
Input 36 V, output 30 V, 7.0 A: 96%
Input 36 V, output 12 V, 7.0 A: 90%
Constant-voltage mode—response time with constant current: less than 0.7 ms. Test conditions: input voltage 20 V, output set to 6 V, 5 A load; 1 Ω resistor, load switched to a steady state.
Constant-current mode—voltage response time: less than 0.4 ms. Test conditions: input voltage 20 V, output adjusted to 6 V, 5 A load with a resistance of 1 Ω; after the load is disconnected, the instantaneous power overshoot is equivalent to the calculated time.
Dimensions: length × width × height, 6.8 × 5.5 × 5.3 cm


The connection wires are really meant for the control board, but since there are so many connectors, the seller connected them to the converter board to avoid mixing them up. The control board connects via a single wide connector; the red and black wires supply power to the control board.
The wires were in my way, so I had to disconnect them temporarily. To avoid mixing them up, I marked the connectors with a marker.


The converter can be used entirely on its own—in fact, that’s how it was originally designed. So if you want knobs, you can use it that way too. This is an image from the product page.


There are two decent input terminals with proper wire clamping. I wasn’t too impressed with the shunt, though; I’ll come back to that later.
The capacitors are 470 µF, 50 V: one on the input and two on the output. I recommend replacing them with something from a more reputable brand and of better quality.


The converter is simple and based on a classic design: a TL494. There’s also an LM358 op-amp on the board. But what I liked wasn’t that—it was the beautiful PCB layout, with the components packed in tightly. Class.
There’s another small bonus: the board has two LEDs indicating CC/CV mode, something sorely missing from the DPS/DPH boards. Of course, this indication is also shown on the display, but in practice a simple LED is much more convenient in some situations.


The inductor is relatively small, but the converter’s output is only about 250 W. The power components are mounted on a heatsink. I didn’t remove it because it’s soldered to the board, but the board itself tells us what’s installed there:
MOSFET - P55NF06
Diode assembly - 2560CT
Bipolar transistor - H669.

The key difference is that the transistor is N-channel. Since there’s no high-side driver IC on the board, the driver is most likely built from discrete components. Of course, I know how to do this without an IC—I’ve built drivers like that myself many times—but it still feels a bit unusual.
The second transistor isn’t attached to the heatsink, although it could easily be. The transistor tabs are connected together in the circuit, so there’s no need to insulate them from the heatsink. This transistor is part of the 9 V regulator that powers the PWM controller and the rest of the circuit.


On the underside, as usual, there’s a little flux and some not-so-pretty soldering, but otherwise it looks fine.


The control board, on the other hand, costs quite a bit more—about $32: link.
Brief specs from the product page.
Adjustment range: 0–30 V, 0–7.5 A (MAX. 32 V, 8 A)
Setting accuracy: 0.003 (depending on calibration and operating conditions, the actual control error is less than this value)
Measurement accuracy: 0.002


The controls consist of a pair of buttons and a push-button encoder.


The display isn’t screwed down; underneath it there are just a few tantalum capacitors.


All the other components are on the back of the board, and there are quite a lot of them.


The analog section: a pair of op-amps and a negative supply rail generator. The op-amps are powered from a split supply.


Everything is controlled by a 32-bit STM32F103RGT6 microcontroller, running at up to 72 MHz and packed with useful features, including three 12-bit ADCs and two DACs.
A lithium battery is located nearby. Its purpose remains a mystery to me, but I was pleased to see that the designer used not only a controller with decent (relatively speaking) DACs and ADCs, but also seems to have added a couple of separate ADCs! You can see them to the right of the microcontroller.

Later, I asked the seller about the battery. He replied that it was needed to store data (as I understand it, Ah and Wh readings). Naturally, I asked whether it wouldn’t be simpler to store them in nonvolatile memory. He said that memory has a limited number of write cycles, and this way is more reliable. Well, there you go!


The microcontroller also has COM and USB interfaces, but they aren’t used here—at least not yet, according to the seller.


There’s a slight complication when connecting it: the board needs a separate power supply, galvanically isolated from the main one. The supply voltage is 12 V, and its negative terminal is connected to the negative output of the power board. That means you can’t power it from the main input; besides the lack of isolation, the power board’s supply voltage is clearly higher than 12 V.
There is a small advantage, though: you can use a 36-to-12 V converter, then add some kind of 12-to-5 V regulator to its output and get an extra output voltage.

I connected it to two power supplies, and everything fired up right away.


The main screen and control menu.
This is what I liked about it from the start: a large screen and crisp digits. Compared with the DPS, it’s a real treat for the eyes.
1. The main screen shows:
Top bar: fan control, tracking mode, control lock
Current voltage/current/power
Set voltage/current
Ah/Wh counter—the readings are saved.
Operating modes and output status.
Time since power-on and output activation time, protection mode, and board temperature.

2. Clicking the encoder (or pressing a button, more on that later) brings up quick-select options for output voltage/current. Really handy.
3. Pressing "Menu" takes us to the settings.
4. Mode Set - current cutoff mode: standard CC mode or trigger mode.
5. The trigger mode lets you set the overload threshold relative to the set value, as well as the restart delay.
6. Double-clicking the encoder saves the setting and displays a nice animation. Pressing "menu" simply takes you back.


1. Menu - Switch takes you to a choice between two modes: a regular power supply and a battery charger.
2. After you confirm, it returns to the main screen, which now looks different.
3. Go into the menu again and the first icon has changed; it now opens the charger operating mode settings.
4. There are two fixed modes plus automatic selection, allowing you to charge lithium or lead-acid batteries.
5, 6. Set the charging voltage and current. The left image shows the default settings for lithium batteries, and the right one for lead-acid batteries.

Unfortunately, this mode is still under development. The seller said they would provide new firmware once they’ve "finished polishing it". But I think it’s a useful idea.


Settings menu.
1. There are two screens here. The first contains the main settings, calibration, appearance options, and limit settings.
2. The second has fan control, memory clearing, factory reset, and developer information.
3. The main menu lets you configure:
Auto-lock
Control mode: this changes the button layout. There are four options; the first and third are the most convenient (in my opinion). In mode 1, pressing the encoder opens the presets, while you switch between current and voltage by turning it while pressed. In mode 3, presets are accessed with a short press of the exit button; pressing the encoder switches between current and voltage, and turning it while pressed changes the step size. Overall, it’s also quite an original setup.
Buzzer on/off
Screen refresh rate, with ten levels. I didn’t time it precisely, but it seems to range from about 1 to 6–7 times per second.
Follow - tracking mode; I’ll show it later.
4. Calibration. To enter this menu, turn the encoder until the squares line up. This prevents accidental entry.
5. Appearance options: enable borders, choose a color, and so on.
6. You can set the maximum voltage and current. I’d also like to be able to set the power limit; the seller had a version with that option, so I’ll have to ask them to add it. The maximum settings are 32 V and 8 A.


1. Fan settings. You can choose continuous operation or set the turn-on temperature. The fan switches off when the temperature drops 5 degrees below the threshold. When the output is enabled, a fan icon appears on the display.
2. Clear the accumulated capacity readings in Ah and Wh.
3. Save data. I didn’t quite understand this one; I’ll have to look into it further.
4. Software info: firmware release date and version, plus an operating-time counter in minutes, which, as far as I can tell, cannot be reset.


Different appearance options: two without borders and two with borders. The "cube-shaped" one was the default; there’s also a "rounded" option. You can also change the background color in the menu.


The output status and current mode are shown in the bottom-right corner.
1. CV - constant-voltage regulation
2. CC - constant-current regulation
3. ER - an attempt to set the output voltage higher than the input voltage, as far as I understand.
4. And this is how the trigger protection works. I had the overload threshold set to 0%. If the output is shorted, it shuts off and displays a countdown window (the delay is configurable), then the output turns back on. If the overload hasn’t cleared, the cycle repeats.


And of course, the tests.
Output voltage accuracy.
1. With the output disabled, there’s a small negative voltage at the output terminals.
2. When the output is enabled and the set voltage is zero, the actual voltage is almost zero.
3–9. At low voltages, up to about 1 V, the reading is a little low. It then levels out and is practically spot-on up to about 30–31 V, though it reads slightly high right at the top of the range.


Current accuracy is a little worse.
1, 2. With the output enabled and the set current at zero, the output is in CC mode, with negative voltage and a current of around -1 mA.
3–9. The output current is slightly below the set value across the entire range. The built-in ammeter shows the same readings, except at the very top of the range: there, the current is still low, but the ammeter shows it as higher.


As it turned out, the last tests took a long time, mainly because I photographed the steps at much smaller intervals than shown above. Part of the delay was also because, as usual, the camera flash overheated and I had to wait a long time. But toward the end, I noticed that the power board was getting quite hot. The shunt got hottest, but the thermal camera can’t see it. It did show the inductor heating up to 68 degrees and the transistor to 75. It was more or less acceptable: even though the shunt got hot, the readings didn’t drift.


In any case, I got curious about how much power was being dissipated by the shunt, so I decided to work it out.
The shunt resistance is 27.9 mΩ. At 8 A, the voltage drop across it is 223 mV, and it dissipates 1.78 W. I’d say that’s right on the limit, since the shunt has a small surface area due to its wire construction.


Above, I mentioned the Follow option. I asked the seller what it does. He said it’s a voltage-tracking function for something more accurate, though I never quite figured out exactly what, so I decided to test it.
Go into the menu and turn it on; an icon for the function appears in the upper-right corner of the screen.

Yes, the function does something, but honestly, I expected a little more. For example, my electronic load has a similar function: the DAC sets a certain voltage using PWM according to its calibration, then measures the current with its ammeter and adjusts the voltage slightly if the current differs from the set value. A kind of real-time correction.
I compared the two and found that the actual output voltage doesn’t change at all, whether the function is off (left) or on (right); only the displayed value changes. That’s a shame, because the controller’s voltmeter and ammeter are calibrated almost as well as my Uni-T, so it could actually use them to make corrections. If you switch the board to CV mode, i.e. let the voltage “drift,” it still remains slightly “off” from the actual value.


I also noticed that the current in CC mode depends slightly on the input voltage. Not by much, but there is a difference: 35 V input at the top, 45 V at the bottom.


Voltage regulation accuracy under different loads. I tested at three voltages—3.3, 12, and 24 V—and three current levels: 55 mA (effectively no load), 4 A, and 7.9 A. In almost every case, the maximum difference was 20 mV, which is very good. Voltage was measured at the terminal block screws.


Efficiency measurement. I used two power supplies, an electronic load, and a multimeter. The multimeter measured the current upstream of the converter. In reality, the efficiency is slightly higher, since the voltage at the board’s input was a little lower.


Test results.


It also turned out that with a 36 V input and a load current above 4.5 A, the board can’t deliver 32 V at the output and reports insufficient voltage. It can still deliver 31 V at 8 A, though. With a 45 V input, everything worked fine.
This all shows that 36 V is really pushing it, and it’s best to use a power supply with an output closer to 40 V.


Let’s add an oscilloscope to the setup and see what’s happening at the output.
I also ran a lot of tests in different modes here, but set the input voltage to 40 V.


Output voltage of 3.3 V (left) and 12 V (right), with load currents of 1, 4, and 7.9 A.
The board didn’t meet its claimed 50–80 mV peak-to-peak ripple specification. I measured up to 100 mV at 3.3 V and up to 140–150 mV at 12 V.


Output voltage of 24 V (left) and 30 V (right), at 1 A and 7.9 A.
The results are about the same, although the ripple is noticeably lower with a 30 V output.


1, 2. No-load ripple, with the output set to 3.3 V.
3, 4. You may have noticed the “spikes” on the waveform at maximum current. At first I zoomed in to see what they were; they looked like some kind of resonance. But then I simply touched the oscilloscope probe, and they dropped by half (right). I suspect that a significant part of the measurement error came from everything being powered from the mains, which could have caused ground loops and mutual interference.


Earlier, you could see a spot on the board for a USB connector, as well as a separate connector marked RX/TX. After talking with the seller, who is also the developer of the control board, I found out that there’s already a beta version of the software for it, though it’s currently in Chinese and very basic. I also asked him to write it in English as well as Chinese from the start.


The seller also offered me a ready-made version of the device, which isn’t listed for sale on his page. I don’t remember the price—something around $75, if I recall correctly, but you’d need to check. It looks great, and I’d probably have bought one if it weren’t for the 1.5 kg weight, which would have driven up the shipping cost considerably.
But it looks really good: an aluminum case, machined finish. The photo isn't very high quality, but this unit appears in the seller's video, which you can watch on the product page.


And now, as always, my conclusions.
I've tried many different converter boards, including ZXY, DPS/DPH, and the relatively new DPX, not to mention all sorts of smaller models. I can say that the board reviewed here is in a league of its own in terms of hardware compared with all of those.
Leaving the converter board itself aside, since it could be a different one, here are the key points.
1. Large, highly readable display
2. Quite well-designed circuitry
3. Simple but convenient controls.
4. Settings that are just as simple, yet convenient and flexible
5. Fan control
6. Good setting accuracy and high measurement accuracy.
7. Current limiting not only in CC mode, but also with an adjustable trigger mode.

But that's not the whole story. The project is also under active development, and the seller (who is also the developer) is open to feedback. One example is the English menu translation, which he made at my request.
Battery charging is also expected to be added, a feature you won't find on other converter boards.

I'm now planning to put together a list of software features for the developer to consider.
1. A maximum power limit, but implemented like on the ZXY, not the DPS/DPH—in other words, without switching off the output.
2. Fan speed control.
3. Overtemperature protection with an external temperature sensor.
4. A function to correct the output voltage based on the measured value.

I'd like to see a wider adjustment range for both current and voltage. I don't think that would be difficult to implement in software. On the hardware side, you could simply use a more powerful converter, or even control a switching power supply directly, without an intermediate conversion stage.

There are downsides too. In my view, there are two:
1. Price. At $30, not including shipping, it's not exactly budget-friendly. And that's just for the controller.
2. It requires a second power supply, galvanically isolated from the main one. It's not particularly expensive, just inconvenient.


That's all from me. I hope you found the review useful, and I'd be very happy to hear your questions and suggestions for the project's development.