A couple more multi-standard DC-DC voltage converter modules from YZXstudio

I got my hands on two more converters from the well-known Chinese company YZXstudio. One is simpler and cheaper, while the other is correspondingly more expensive and comes in a "top-of-the-line" configuration. The second one isn’t just a DC-DC converter, but a universal bidirectional converter. But I think it’s best to read all about them in the review.


First, thanks to my regular reader Vladimir, who sent me this pair of modules to test and review.
They were bought from the official YZXstudio DIY store. The modules differ considerably from each other: one is the cheapest in the YZXstudio lineup, while the other is the most expensive. That makes them all the more interesting to compare and see what accounts for the fourfold price difference.



The modules are very similar in design: a pair of connection terminals and USB ports, all in a very compact size.


Of course, if you look closely, the difference is obvious even to the naked eye. The ZC822 is on top and costs about $5; the ZC826 below costs $20.


The first module has two ports, USB-A and USB-C, while the second has only USB-C. Yes, it’s a bit unfair: you can’t plug the usual cables into the more expensive one.


These two modules also complement my previous review, which featured the ZC823 model.


Before moving on to the description and tests, I’ll say a little about what this company makes besides its well-known testers.

The boards I’ll be discussing today are part of a kit for assembling a universal charger consisting of:
1. Enclosure
2. Backplane. There’s also some kind of controller installed on it; I couldn’t figure out what it does.
3. Front and rear panels
4. An optional additional connector for an external battery.


The backplane can take one of four converter boards:
1. ZC822
2. ZC826
3. ZC823
4. ZC823P

The first two are covered in today’s review. I tested the third one earlier, and more recently a higher-power ZC823P model rated for up to 100 W came out. The ZC823 delivers up to 60 W, and there’s very little difference in price between them, so if you need a ZC823, it’s better to get the P version right away.


Basically, choose the configuration you need, solder the boards onto the backplane, put it in the enclosure, and you have a finished product. There’s a ready-made unit in this enclosure on AliExpress — link.


The same manufacturer also offers alternatives, already assembled. For example, the 100+60 W version, a more powerful take on the previous 100+27 W model. It’s available on AliExpress for a little more than on Taobao, at around $45 — link.


The same manufacturer makes another handy device in the form of a cable. It can automatically tell the converter to output 15 or 20 V — link.


Now, on to the boards themselves.
First up is the ZC822, priced at $5.45, Taobao link,
Specifications:
DC input voltage: 10–24 V
PD output: 27 W; USB output: 27 W.
Maximum output power: 12 V, 2.5 A, 27 W
The input voltage must always be at least a few volts higher than the required output voltage.

The board is based on the IP6528 controller. The entire power stage is “hidden” inside the controller; the only external components are the power inductor and current-sense shunts. You can see a pair of power transistors on the board, but they have nothing to do with the actual conversion—they simply switch power to the corresponding outputs. There are two shunts for the same reason: each output is monitored independently, but only one output can be active at a time. A self-resetting fuse rated for 3 A is installed on the input.


Almost all protocols are supported, but while I didn't have high hopes for QC at 20 V, the lack of Huawei SCP was a bit surprising. This standard is usually supported without any problems.


Auto-detection can sometimes be inaccurate or fail to reflect what's really going on, so just to be safe, I check each protocol separately.
1. QC2.0: supported up to and including 12 V.
2, 3. QC3.0: supported in the range of 3.7–12.3 V.
4. Huawei FCP: supported; allows setting 9 V.
5. Huawei SCP: not supported; the voltage doesn't change.
6. Samsung AFC: 9 V mode is supported.
7. PD: PD 3.0 is supported, but the actual power is limited to 27 W.
8, 9. PD 3.0 lets you adjust the output voltage in 20 mV increments. For example, I had no trouble setting it to 7.2 V, then raising it to 7.26 V. Too bad the power is rather low.


As I mentioned, the board has two outputs, USB-A and USB-C. The USB-A port supports all the protocols identified earlier except PD. It also specifically states that Samsung AFC at 12 V isn't supported; the previous tester can't work with this protocol.
When connected via a USB-C cable, Power Delivery support becomes available, but for some reason it's shown as 9 W...


Next, I connect the converter to an adjustable power supply and attach an electronic load to the output. Then I test it under load and determine its actual output power.


The input voltage was 20 V in all tests.
First, QC with an output voltage of 5 V. As the current increases, the voltage rises slightly too, reaching 5.2 V at the maximum current of 3.8 A.


At an output voltage of 9 V, the results are similar, but shutdown occurs at a current of about 3.1 A.


At the maximum output voltage, the results are again similar, but the protection trips at a current of 2.4 A.

In all the tests, shutdown is instantaneous, the voltage stays within normal limits, and the current is above the rated value in every case.


The FCP, AFC, and Power Delivery tests produced practically the same results, so I'll hide them under a spoiler.


Efficiency measurement results with an input voltage of 24 V and output voltages of 5, 9, and 12 V. The horizontal-axis labels are in 0.5 A increments; at 12 V, the last value is 2.4 A.


Temperature rise at 5 V and 3 A, and at 12 V and 2 A. Heating is moderate; the inductor temperature (on the back of the board) is around 55–60 degrees.


Ripple at an output voltage of 5 V, with no load and at currents of 1, 2, and 3 A. In the worst-case operating mode, the main ripple amplitude is about 40 mV, but there are very short spikes with an amplitude of up to 160 mV. Ripple was measured directly across the converter's output capacitor; after the cable to the load, the amplitude of the spike component should be noticeably lower.


Test at an output voltage of 12 V, again with no load and at currents of 1, 2, and 2.4 A.
Here, the main ripple amplitude reaches 50 mV, and the spikes reach 250 mV.



And here's the second board, the ZC826. It costs $21.50, link to Taobao. It's also available on AliExpress, but costs $43 there.

The board's input voltage range isn't specified, but it's claimed to work properly from 12 V and up. I tested it up to 20 V.
Output: up to 20 V at 3 A, i.e. up to 60 W in QC and PD modes.

This board is much more complex: you can see a controller, an additional chip, four power transistors, high-capacitance ceramic capacitors, and a one-time fuse rated for 6 A.


The whole thing is controlled by the IP2716 controller, whose output is connected to the universal PWM controller SC8802, which in turn controls four MOSFETs.
Current is monitored by a pair of 7 mΩ shunts, one on the converter input and the other on its output. On the other side of the board, in addition to the inductor, there are two 220 µF, 35 V capacitors, which are bypassed by high-capacitance ceramic capacitors.


I didn't mention the universal PWM controller above for nothing. More precisely, this controller uses the topology of a universal converter: it doesn't matter whether it steps the voltage up or down. More importantly, this converter is bidirectional, since it can operate in either direction, and the designer made use of this feature—though I'll get to that a little later.
To simplify things a lot, the power stage looks like this block diagram.


Those who read my reviews may say, “Haven't we seen this somewhere before?” And you'd be right: the universal converter in my review of the LTC3780-based board works on exactly this principle.


I don't have a complete list of the supported protocols, but it's available in the controller's datasheet.


One of those times when the USB tester lit everything up green.


And a more detailed check of protocol support.
1. QC2.0, fully supported
2, 3. QC3.0, fully supported
4. Huawei FCP, fully supported
5, 6. Huawei SCP, fully supported
7. Samsung AFC, 9 V mode supported.
8, 9. PD. Protocol 2.0 is supported, with output power up to 60 W.
There's something odd about the last two items: according to the datasheet, the controller supports PD 3.0, but the tester detected only mode 2.0. The difference between these modes was shown above. As with QC 2.0 and 3.0, PD 3.0 lets you adjust the output voltage in 20 mV steps, while 2.0 offers only 5/9/12/15/20 V.


My old tester also showed that only PD 2.0 is supported, with power up to 60 W. It also showed that Samsung AFC 12 V mode isn't supported; unfortunately, it's hard to see in the photo, but this is marked in red.

By the way, I finally bought a USB Type-C/Type-C cable specifically for tests like these, since my old tester has only Type-C sockets. While I was at it, I checked that the cable has all the necessary data lines. The 1.5 m cable cost just a dollar and a half at a local store :)


A significant advantage of the board is that it switches to low-power mode when there's no load, drawing less than 1 mA at the input. If you connect a load, even a basic USB tester, the board starts up reliably and stays on until you remove the load.


The same test setup as for the previous board, but since its expected current draw is higher, the power cable was shorter. Also, for the efficiency test, the voltage was measured directly at the converter board's terminals.


Testing operation in different modes. The input voltage is 12 V unless otherwise specified. First, QC mode.
The output voltage is 5 V. The board went into protection when the load current exceeded 4 A. The output wasn't completely shut off, though; the converter kept running even as the input current dropped almost to zero.
As the load current increases, the voltage rises from an initial 5.07 V to 5.12 V at the maximum current, which is still well within the standard 5% tolerance.


At 9 V and 12 V, it's more or less the same, so the results are under the spoiler. The protection also trips at 4 A, and the voltage rises slightly as the load current increases.


As I mentioned above, the board has a universal converter, so it can output 20 V even with a 12 V input. However, to run the test I had to raise the input voltage slightly to 15 V because the power supply could deliver only 5.1 A, while the board needed more. Raising the input voltage reduces the input current.
In this mode, the board delivered 3.4 A without any trouble, but I think it could easily have managed 4 A as well.







The 15 V output voltage was tested with a 13 V input because of the limitations mentioned above. Again, nothing new: 4–4.1 A at the output, but the shutdown is abrupt.


When I ran the tests above and hit the input current limit, I was still curious about the maximum output current at an output voltage of 20 V. So I applied 18 V to the input and ran a test to measure it. As it turned out, even in this mode the protection trips at 4.2 A, with output power exceeding 80 W.


The only test I didn't run was to see what happens if you “request” 20 V at 3 A with an input voltage of 12 V. But I did a similar test with a 13 V input. The load current was about 3 A, and the input current was 5 A. Since that works out to about 3.25 V per cell when powered by batteries, the converter simply won't have time to overheat: the discharge curve won't let it run in this mode for long.


The board's main efficiency measurements were taken in four modes: 12 V input with 5, 9, and 12 V output, and 13 V input with 20 V output, at various load currents—from 0.5 to 4 A in the first three modes and up to 3.2 A at 20 V output.


I also ran two more tests: 20 V input with 5 V output, and 8 V input with 15 V output. The board operated stably in all modes.


Board temperature measurements:
1. 13 V input, 20 V output, 3 A.
2. 12 V input, 5 V output, 4 A.

In my opinion, the board didn't get very hot, even in the most demanding mode, with 13 V at the input and 60 W of output power.


Ripple with a 15 V input and 5 V output, with no load and at 1, 2, and 3 A: the maximum was about 20 mV, which is very good.


Step-up mode: the input is still 12 V, with 20 V at the output, both with no load and at 0.5, 1.0, and 1.5 A. Ripple is significantly higher here: about 120 mV of regular ripple and up to 700 mV of spikes. That's quite a lot, but as I mentioned above, the cable to the load does a decent job of suppressing the spikes—or rather, the cable and the capacitance at the device input form a parasitic RC filter.


24 V input, with 5, 9, 12, and 20 V output at 3 A. Overall, my only complaint is about the 5 V mode; the rest is quite acceptable.



I've mentioned powering this board from a battery several times. What's more, the manufacturer even recommends its own 4S protection board—and that's not just for show, “by the way.”
The product page says the board has no protection against operating with low input voltage. In practice, it runs just fine from 5–7 V, so when powered by a battery, the protection board is responsible for handling that.
But that's not all. As I mentioned above, the converter is bidirectional: you can apply 5 V to the board's output, and it will charge a connected battery.
I didn't have a suitable battery on hand, so I used an old 12 V lead-acid battery and applied 5 V to the board's output.


Power was supplied by an adjustable power supply through another board that provided the required 5 V. The current in the 5 V line was 2.6 A, but because of the large voltage drop across the power cable, the voltage at the ZC826 input was only 4.48 V.
The problem was the cable itself, since I got similar results when powering the board from a pair of USB chargers.


But then I remembered that the description said connecting this board to a Power Delivery charger would automatically request 20 V at 3 A, so I had to get out the ZC823 charging board and test it with that.
This is the setup I ended up with: a 24 V power supply, a ZC823 board that can operate in PD mode at 20 V and 3 A, a cable, and the ZC826 board.


And sure enough, the board under review requested 20 V and started drawing 2.7 A. At 20 V, that works out to about 50 W going into the charging process. At 15–16 V, the current in the battery circuit is about 3.2 A. At lower voltages (with a discharged battery), the current will be higher because the limit is on the input, not the output.


Once charging is complete, the board should cut off the current and switch the LED from blue to red. But since it detects the end of charging by the drop in current, and this clearly isn't the battery it was designed for, it simply brought the voltage up to 16.73 V and then did nothing else.


Now for my conclusions, starting with the simpler ZC822 board.
It's a simple, inexpensive board that works well and supports a large number of protocols. The only downside is its fairly modest power output of just 27 W. Still, it's a perfectly good addition to other boards for charging a tablet, phone, or something else. It also supports PD as well as QC, unlike ordinary boards.

And now the "top-of-the-line" ZC826.
Here, of course, I'll start with support for almost all known protocols, high power output, and a universal power supply. But the main feature is probably its ability to work as a power bank: with a 4S battery connected, the board can not only draw power from it but also charge it, at up to 50 W. I'd recommend using high-capacity cells in a 4S2P configuration. That way, the battery would operate under an optimal load and still have a decent capacity.
In my opinion, the only drawbacks are the price and the puzzling lack of PD 3.0 support.

In short, both boards are interesting, well made, and use quality components. Both support a large number of protocols, but they belong to completely different price categories. And thanks to their design, both can be used as part of the same device.

If you're planning to order from Taobao through the Yuibei agent, which I mentioned briefly in my article "How to Buy on Taobao," I'd appreciate it if you signed up using my link.

That's all from me. I hope you found the review useful.