XY6020L Adjustable Voltage Converter

A quick look at an inexpensive adjustable 60 V, 20 A voltage converter. It’s not exactly great, but I wouldn’t call it terrible either—in short, you get what you pay for.

A reminder: to keep up with new posts on the blog, I recommend following me on Instagram. I’ll post notifications about new articles there and might also write about something interesting from time to time - link to my account

You can also subscribe to updates and new articles on the Telegram channel - https://t.me/KirichBlog


I recently mentioned that I bought this converter for a specific purpose: as a universal charger for a battery tester, and also as an adjustable power supply, though that was secondary. Since it ended up in my hands, I decided to take a closer look at it. This information may come in handy for whoever uses it in the future.

I ordered it a long time ago, I think during the spring sale, but I put it in a desk drawer and never found the time to work on the device I’d bought it for. I also bought a 60 V, 300 W power supply, but I’ll probably cover that another time.
It came in a cardboard box and arrived undamaged.


The kit consists of a power board and a control/display module. There’s also a version sold with just the board, where you can add potentiometers or set a fixed voltage, as well as a version with an extra Wi-Fi module. That wasn’t needed for this application.


The specifications label. There’s nothing new here, apart from perhaps a few minor differences from other similar converters.


The power board is neatly assembled, and I actually like it. The only odd thing I noticed right away was the number of current-sense shunts...


On the input there’s a 1000 µF, 80 V capacitor and a decent-sized EMI suppression inductor. That’s a nice touch.


The auxiliary converters that power the board’s electronics are here. There’s a switch on the left, though it doesn’t really serve any purpose.


The power stage is built around an EG1163 controller. Since it’s a synchronous step-down DC-DC converter, there are two transistors on the heatsink rather than the usual transistor-and-diode pair. That’s good, as it improves efficiency, especially at high load currents.


The control circuitry and analog section. That’s why it’s not quite right to call this a power board: it has both the power stage and the control circuitry. Essentially, it’s a full-fledged “smart” DC-DC converter.
There’s a switch on the right, which I assume is for setting a fixed output voltage. No idea why you’d need that—it seems like overengineering, but there it is.


The display connects via two connectors, and they have exactly the same color order. To make it easier to figure out what goes where, the boards are marked R and L.


Now, about the shunts.
First, they’re not soldered very well. It looks like they skimped on the solder, so I resoldered them properly. The main issue isn’t so much the electrical connection as the thermal one: they get noticeably hot, and if a shunt isn’t soldered well, it won’t transfer heat to the board as effectively.
Second, I couldn’t figure out why there are two: one rated at 5 mΩ and another made up of three 10 mΩ shunts.
At first, I thought the first one measured the input current for some reason, though I couldn’t figure out what for...


But the datasheet says the shunt is on the output, so it measures the output current. There’s also another shunt on the board, near the output, and that one definitely measures the output current too—again...
I wondered if the input shunt was there because there’s a similar board with an MPPT controller, which might need to measure the current before the converter. Anyway, they built it this way, and the main thing is that it works.


A few more details.
1. Some kind of diode on the input, perhaps for the converter or maybe for protection. I didn’t investigate.
2. A 30 A fuse, installed in a spot that makes replacing it a nightmare—though I suppose it depends on who your enemy is...
3. One of the auxiliary power chips.
4. Output op-amps.
5. Three 470 µF, 80 V capacitors on the output.
6. Some kind of diode or TVS on the output. I didn’t check.


The display itself—there’s not much to see here. It’s just a display, nothing more, plus a few buttons and an encoder. There’s no “brain” in it.


The fan is relatively quiet, rated for 5 V, and measures 40 × 40 × 10 mm. It turns on when the load current exceeds 2 A. The heatsink is positioned oddly; I think it would have been better to rotate it 90 degrees.


The underside of the board is empty, apart from the switch label.
The board measures 110 × 72 × 45 mm.



As usual, I checked the current draw at different input voltages: 55 mA at an input of 6 V and 11 mA at 36 V. Beyond that, it doesn't change significantly.


Display and control modes.
When switched off, it displays zeros in the voltage and current fields. The bottom line periodically shows the set current and voltage. When switched on, the top lines show the actual output current and voltage, and the bottom line shows power.
The controls are extremely simple and actually quite convenient. First, to set a value, press the V/A button. It prompts you to set the voltage first; pressing it again switches to current adjustment. Pressing the encoder or the SW button changes the adjustment step.
One handy feature is that it remembers the adjustment step we used. So if we set it in 1 V increments, it will use 1 V increments next time too. Unfortunately, this only applies to voltage adjustment; for current, you have to press the button each time.
You can display the input voltage by simply pressing the SW button.


There is an additional settings menu. To open it, hold down the SW button for a few seconds.
1. Undervoltage protection. A useful feature in some situations. The minimum is 4.8 V; it is enabled by default.
2. Overvoltage protection, set to 65 V.
3. Maximum current limit, set to 22 A, although the converter lets you set a maximum of 21 A.
4. Maximum output power. A very useful option: if, for example, your power supply is rated for 300 W, set this to 280 W and you won't have to worry about accidentally overloading it.
5. Maximum Ah value, at which the converter will turn off the output.
6. The same, but for Wh.
7. The same, but for time.
8. Protection against the board itself overheating.
9. Overtemperature protection using an external sensor. There is a white connector near the output terminal block; connect a 10 kΩ NTC thermistor to it.
10. Device address on the network.
11. Option to automatically activate the output when power is applied.
12. The parameter adjusted by turning the encoder without going into any menus. Voltage is selected by default, but you can set it to adjust current instead.


A few tests.
Checking the accuracy of the output voltage setting.
And I have to say, it's not quite that straightforward. I set the voltage using the usual increments: 0.01, 0.02, 0.05, 0.1, 0.2, and so on. It turned out that:
1. At 0.01 and 0.02, it's a complete mess.
2. Below 0.5 V, it tries to regulate, but not very well.
3. It works reasonably accurately, albeit a little low, starting at 0.5 V.
4. Despite all that, it measures 0.01, 0.02, and 0.05 V correctly. Go figure.

Overall, the calibration accuracy could be better. Riden is definitely better in this respect, but it costs more and is especially larger.



I was curious at what voltage the setting accuracy would be highest. Chinese manufacturers usually calibrate at 30 or 28 V, but in this case I got the best accuracy at 40 V.


This test doesn't tell us much, but even so, with an input of 61 V, I could get only 58 V at the output with no load. That's a drop of at least 3 V. Taking the load into account, I think the power supply should provide at least 5 V more than the required output voltage.


The current results are much the same as the voltage results. It doesn't handle low currents very well. For example, below 10–50 mA, the reading starts jumping around so much that the multimeter's autoranging can't even settle on a range (first photo).
At 100–200 mA, the readings are closer to reality; see the second and third photos.
It starts working reasonably accurately above 500 mA. I then tested at 1, 2, 5, 10, and 12 A, and the error seemed acceptable to me.


When I first started checking current-setting accuracy, I noticed that there was still a tiny current at the output even when it was switched off. Just to be safe, I checked the voltage: 0.5 mV. It's just a small leakage current, nothing to worry about.


Next, I decided to check the efficiency and assess the thermal performance. For this, I used a power supply based on an RD6030 converter and one of the electronic loads. that I’m putting together.


Efficiency was measured—or, more accurately, estimated, since it was affected by voltage drops in the input and output wires, which I didn’t account for. I only used the set load current and the readings from the wattmeters on both devices. In other words, the actual efficiency will be a little higher, so the result is essentially “no worse than specified.”
I tested with an input voltage of 55 V, output voltages of 5, 12, 24, 36, and 48 V, and currents of 10 and 20 A.
1. 5 V, 10 A: input 61.6 W, output 48.3 W, efficiency 78.4%
2. 5 V, 20 A: input 124.3 W, output 94.9 W, efficiency 76.3%
3. 12 V, 10 A: input 132 W, output 118.7 W, efficiency 89.9%
4. 12 V, 20 A: input 266 W, output 235.6 W, efficiency 88.5%
5. 24 V, 10 A: input 252.4 W, output 239.3 W, efficiency 94.8%
6. 24 V, 20 A: input 509.3 W, output 476.6 W, efficiency 93.6%
7. 36 V, 10 A: input 373 W, output 359.8 W, efficiency 96.4%
8. 36 V, 20 A: input 751.8 W, output 717.4 W, efficiency 95.4%
9. 48 V, 10 A: input 493.3 W, output 480 W, efficiency 97.3%
10. 48 V, 20 A: input 994.2 W, output 957.8 W, efficiency 96.3%

Overall, the efficiency looks fine. Naturally, it drops at low voltage and high current, which is normal; otherwise, it’s quite in line with what you’d expect from a synchronous converter.


Board temperature check at 24 V and 10 and 20 A, and at 48 V and 10 and 20 A.

First, 24 V at 10 A on top and 20 A below. The only component whose temperature I wasn’t happy with was the 5 mΩ shunt near the heatsink. The temperatures of the other components were perfectly normal.


And at 48 V, 10 and 20 A.
The shunts behaved the same way, since their temperature depends on the load current. The transistors on the heatsinks got hotter, but I still didn’t notice any significant heating.
I didn’t run the load for very long, but the board’s thermal inertia isn’t very high, so the temperature didn’t change much after that. Even though I tested at close to 1 kW and the transistors were around 75°C, I doubt their temperature would have risen to anything concerning.


Ripple and transient response tests. The only additional equipment here was an oscilloscope, alongside the power supply and load.



First, the output voltage is 12 V, with load currents of 0, 7, 14, and 20 A. The probe is connected directly to the output, with no filter, and the 20 MHz bandwidth limit is enabled.
The specs claim 100 mV peak-to-peak; in reality, the ripple is a little higher. Yes, there are spikes, but they aren’t critical in this case. The main ripple is 100 mV at 14 A and about 150 mV at 20 A.


Here the current is 20 A in every case, but the voltage is 5, 20, 30, and 48 V. At 5 and 48 V, things are basically fine, but at 20 and 30 V the ripple changes character and exceeds the specified 100 mV, reaching closer to 150 mV in reality. Ideally, I’d add some kind of output filter.
The results may also be affected by the fact that each device—the bench power supply, load, and oscilloscope—has its own separate switching power supply, so there could be some interaction between them.


I was also interested in testing the transient response.
1, 2. Powering up at 5 and 25 V. The startup is veeeery slow, which is also visible on the converter’s own display. Interestingly, the display shows a slight overshoot above the set value, but the oscilloscope doesn’t detect any overshoot.
3, 4. Powering down. This is just awful: it takes about 1.5 seconds to go from 5 V to 0, and even longer—from 48 V to 0, it takes 13 seconds.

Let me explain what’s wrong here. First, startup. Some devices don’t handle a gradual rise in supply voltage well. For example, microcontroller-based devices may not process the reset signal correctly when power is applied, so a slow start isn’t ideal.
I think the designers ran into some kind of problem and made the soft-start much slower to avoid potential issues.

As for shutdown, the voltage usually needs to be removed quickly. Since switching converters have a large capacitance at the output, they typically use a special circuit that acts as a constant-current discharge. This speeds up shutdown considerably. There’s clearly no such circuit here, or it’s very low-power, because 1500 µF at the output really isn’t that much.



This is what powering up, powering down, and powering up again without a load looks like, with the voltage set to 25 V.


1, 2. Transient response at an output voltage of 12 V, with a 30 Ω resistor permanently connected and a 3.3 Ω resistor being connected and disconnected. Connecting the resistor is fine; disconnecting it causes a brief spike, which isn’t critical for most devices.
3, 4. CV-to-CC and CC-to-CV transitions. The voltage is 15 V, the current limit is set to 3 A, and a resistive load drawing 4.5 A is connected and disconnected. In general, almost everything is fine, except that when switching from CC to CV, the voltage first rises slightly above the set value, but then almost immediately returns to normal.


While running the tests, though, I noticed some strange behavior specifically during CC-CV and CV-CC transitions. I'll try to explain.
First, all the tests were done simply by connecting a resistor—literally touching a wire to the output terminal. Second, these were consecutive measurements: touch, record the result, take a screenshot, restart the oscilloscope trigger, touch again, record the result, take another screenshot.

But I got different results even when doing the same thing. I'll check this again later, once the converter is installed permanently, and add the results. After I'd got everything ready, I wondered whether something might have been making intermittent contact during the test. I wouldn't rule that out.

So, here's what we have:
1, 2. A strange step, and if you look closely, you'll also see that the final voltage on the graph is different.
3, 4. An even stranger picture: this is the CC-CV transition. The graph on the left is normal, but sometimes I got the result shown on the right: with the voltage set to 15 V, it rose to 16 V.
5, 6. An even more interesting picture emerged when I stretched the image. It turned out the transition happens in steps: first the voltage rises, then the unit switches off the output (remember, it shuts down very gradually), and then switches to CV mode.
This explains the overshoot. The load is removed, there's no discharge path, so the voltage overshoots. The feedback loop shuts off the PWM, the voltage drops, the feedback loop starts the PWM again, and the voltage settles at the set value.
I think this could be fixed—or at least improved—by adding either a simple resistor or, better yet, a basic constant-current circuit to discharge the capacitors. Of course, it would need to go before the current-sense shunt so it doesn't affect the ammeter reading.



And now for the conclusions.
I'm left with very mixed feelings. On the plus side: decent build quality, a respectable output current, low heat and therefore good efficiency, enough features for a wide range of tasks, and a low price.
Unfortunately, there are drawbacks too. First, the ripple could be lower. There's no discharge circuit, the feedback loop is relatively slow, and I can't remember what else I mentioned in the review.

Overall, it's perfectly adequate for the job I bought it for—charging batteries. It can also work as a basic adjustable power supply, but only for relatively "rough" loads.
In short, it's fine as a charger, but I wouldn't use it to power anything sensitive or "delicate" (for example, a microcontroller board without its own voltage regulator). For that kind of application, an RD60xx-based power supply is a better choice, and an RD60xxP is even better—though it costs quite a bit more. If you just need an adjustable power supply to run a motor or car stereo, or charge a battery, this one is fine: a sort of all-purpose "utility" power supply with current up to 20 A.


That's all for now. I hope this was useful, and as always, comments are welcome—especially since, as I've learned, other blog readers have this board too.