XH-M403 Adjustable Voltage Converter

Today I'm reviewing a somewhat unusual device. Due to a few circumstances, this review will be short and won't include everything you're used to seeing here. That's just how things worked out—and maybe that's for the best.


I actually saw this product from another seller about a year ago and suggested writing a review, but for some reason the seller declined. Now I suspect there was a reason for that.

What first caught my interest about this converter was its design: it's simple and would be a good fit for beginner electronics hobbyists precisely because of that simplicity. As it turned out, it's even simpler than I initially thought. But I'll take things in order.

The converter comes in a small box with a piece of polyethylene foam inside. You can see that the box took a bit of a beating during shipping.


Description from the product page
Product model: XH-M403
Input voltage: DC 5–36 V
Output voltage: DC 1.3–32 V
Maximum current: 8 A (5 A recommended for continuous operation)
Maximum power: 200 W
Conversion efficiency: 0.94
Switching frequency: 180 kHz
Other features: Current measurement, short-circuit protection
Size: 78 × 63 × 40 mm
Real-time output voltage display, no detection.
Low-voltage display resolution: 0.1 V

I was interested in its design: everything is very clean and simple, with just one knob and two displays.


When I first saw this converter about a year ago, I assumed it used an encoder for control. Encoders are quite fashionable these days and are common because they're genuinely practical. But in fact, the entire control is just one potentiometer.
The displays are large, with 14 mm-high characters. That's an advantage in some situations, especially if your eyesight isn't great.
The board measures 78 × 63 mm. It's 35 mm high, excluding the protruding knob, and 55 mm overall.


The potentiometer caught my attention too. In addition to its three terminals, it has four wide tabs for securing it to the board. I don't think I've come across this type before.
The construction is sturdy, which is another plus in my opinion.


The assembly is reasonably neat. The board isn't overcrowded and is clean; flux is present only around the potentiometer terminals, so it looks like that was soldered last.


1, 2. The input and output each have a quality terminal block with a lift-style clamp. The terminal blocks are marked with Chinese characters on the board, but it's easier to tell which is which by looking for the current-sense shunt, which is near the board's output.
3. The power stage consists of a PWM controller and an MBR20100CT diode assembly, each mounted on a separate heatsink, plus an inductor and a pair of 470 µF, 50 V capacitors.
4. The PWM controller is an XL4016, the same one used on a well-known board rated for a maximum current of 10 A. Here, the stated ratings are realistic: 5 A continuous and 8 A for short periods.
The diode assembly and controller are electrically isolated from the heatsinks by mica sheets; no thermal paste is used.


But about half the board is also taken up by the measurement circuitry.


The ammeter-voltmeter is built around an N76E003AT20, with an LM358 op-amp nearby. Further to the left, you can also see an HC164 driving the displays and a CD4051 multiplexer. I couldn't figure out what the multiplexer is for, but I got the impression that the meter chip has a single measurement input, and the multiplexer alternately switches the voltage from the shunt (after the op-amp) and the board output to it.


The board also has a second converter to power the measurement circuitry and displays. Overall, it's done pretty well and even seems thoughtfully designed in some respects.


Since there's a DC-DC converter on the input, the current draw depends on the supply voltage. At 12 V it's 30 mA; at 36 V it's noticeably lower, just 14 mA.


I immediately decided to see how the converter behaved under load—and as it turned out, that was a good idea. The thing is, its output voltage fluctuates very strangely under different loads. At first I thought the variable resistor was "crackling" or that I’d somehow bumped the board, so I repeated the test three times and got roughly the same results each time.

I deliberately zoomed in on the graph to make the fluctuations easier to see. The output voltage is 5 V, with current up to 8 A.


I tested it at 12 V, and the results were almost exactly the same.


I tested it up to 19 V, since the load is rated for 150 W. That was the highest voltage I could use to test the current up to 8 A.


At 19 V, I had to lower the electronic load’s protection cutoff to 18 V; otherwise, it would trip because of the voltage sag.


So, voltage regulation is pretty disappointing. Next, I decided to simply check the voltage and current measurement accuracy.
I ran into another problem here: setting the output voltage precisely turned out to be very difficult. And it wasn’t just because it uses a regular single-turn potentiometer rather than a multi-turn one—the adjustment itself is the problem. For example, I turn the knob to the right, the voltage rises, and at some point it goes higher than I want. Naturally, I start turning it back, but I have to turn it quite a bit, and then I overshoot again. So instead of setting it to 5 V, I’d end up with 4.9 V, 5.2 V, or something else, but never exactly 5 V. The sluggish response and low voltmeter sampling rate made things even harder. In short, I really wouldn’t recommend using this if you have a short fuse.

I checked the voltage across its entire range, from the minimum of 1.25 V up to the maximum it could manage with a 36 V input. Above 7–8 V, the measurement error became noticeable, and at higher voltages it became downright obvious.


But since the circuit is primitive, the board has two equally basic trimmers for calibration: one for current and one for voltage.
They’re set to the middle position by default, so I doubt anyone even tried to calibrate it.


I grabbed a small screwdriver, set the output to the maximum voltage, and adjusted the voltmeter reading to match the multimeter. One more thing: the adjustment is very coarse. Even a fairly small turn of the trimmer changes the reading by 0.5–1 V.

Even so, I managed to calibrate it fairly accurately.


The current readings were much the same as the voltage readings.


Calibration here was much easier: the adjustment was smoother, so I got a good result in about half a minute. I calibrated it at the maximum current.


While I was fiddling with the voltmeter and ammeter calibration, the converter warmed up noticeably. I thought it would be a good idea to check how much its output voltage drifted with temperature. I let it cool down, set the output to about 12 V, loaded it to 5 A (the maximum continuous current), and waited.
After almost 40 minutes, I got this amusing graph. What’s amusing is that the voltage didn’t just drop as the unit warmed up—which I would have understood—but then it started rising, and after that it dropped again...


After letting it heat up for almost 40 minutes, I set the output to 20 V at the same current (about 100 W) and ran it for another 10 minutes or so, measuring the temperature along the way.
With a 36 V input, a 5 A load, and output voltages of 12 V and 20 V, the PWM controller got the hottest. The heatsink temperature was about 95 and 105 degrees, respectively. The controller itself was of course considerably hotter, but it was very awkward to reach.


And that’s where the story ends: the PWM controller eventually shorted out. It happened under very innocent circumstances. I was fiddling with the multimeter and the board, measuring current, then switched the multimeter to measure across the board’s output—but forgot to move the probes on the multimeter itself. I powered up the board (it was off when I connected the probes), and the controller went off to the great hunting grounds in the sky.
Now the board's output is always the same as its input.


Conclusion.
The converter has a ton of drawbacks. Besides the PWM controller having blown, it has very coarse voltage adjustment, no adjustable current limit, and the voltage drifts with both load current and temperature. As you can imagine, I didn't get as far as measuring the ripple.
I don't see any point in repairing it, and I'll most likely use it for parts.

Still, I think this design could be turned into something really good, and I don't understand why none of the manufacturers have thought of it.
I often come across situations where people don't need a computer connection or any of the service features, current and power measurements, and so on. They just want two large displays with no extra information, showing only current and voltage, and a minimum of controls.
The solution is obvious, and at first I thought that's more or less what they'd done here. Take a board like this, add a microcontroller, replace the potentiometer with an encoder, and implement this logic:
1. Turn the encoder to change the selected parameter
2. Short press to select current or voltage adjustment
3. Long press to turn the output on or off
4. Use progressive adjustment for setting current/voltage, where the step size depends on how fast you turn the encoder. Make the parameter being adjusted brighter.
That's it. Often, you don't need anything more. And it probably wouldn't affect the price at all, since the voltmeter/ammeter chip and switch could be removed.

Actually, one more thing. Sometimes I just want to say: dear manufacturers, please include a display filter with boards like these. It costs next to nothing, and it's not always possible to find one or figure out which one to get.
As it is now, it's not worth the money. It would be a different story if it at least had current adjustment.

That's definitely all now. I hope this information was useful.