An unusual power bank with a QC/PD output of up to 22 W and a few quirks

A couple of months ago I was browsing AliExpress and came across some interesting power bank cases. What was interesting is that I kept seeing them in lots of different versions, but they all shared the same style: a colored cassette inside a transparent case. So I ordered one and it arrived.

As usual, a reminder: to keep up with new posts on the blog, I recommend following me on Instagram, where I'll post notifications about every new topic and maybe write about something interesting too — link to account


The variety of case options is truly impressive. There's another common feature, too: none of the listings use photos, only renders. Here are the ones I found at a glance:
8 × 21700 cells, rated at 22.5 W, but no 12 V output listed — link.
2 × 21700 cells, rated at 15 W — link.
8 × 18650 cells, rated at 22.5 W, but no 12 V output listed — link.
2 × 21700 cells, rated at 22.5 W, but no 12 V output listed — link.
1 × 1376074 battery, rated at 22.5 W PD — link.
2 × 21700 cells, rated at 22.5 W, but no 12 V output listed — link
8 × 18650 cells, rated at 22.5 W, but no 12 V output listed — link.
1 × 1260110 battery, rated at 22.5 W PD — link.
4 × 21700 cells, rated for 22.5 W output and a whopping 100 W input; the latter is definitely out of the question — link

That's 9 models right off the bat, not counting the one in this review. Some are listed at 66 W, and almost all at 22.5 W. I don't know how realistic those figures are, but I have a suspicion they're all built around the same chip. I ordered the version that explicitly listed 12 V output.



When I showed a friend what had arrived, he said, “Hey, it's Bubble Gum Storm 2.” And yes, the design really does have a lot in common with it, but in a much simpler version, at a far lower price, and of course with different specs.

For context, this series of power banks includes models with 100–130 W (100+30 W) output. The price isn't exactly appealing, but as I understand it, you're paying for the case with the batteries. Though honestly, I don't understand 100 W models with 8 × 21700 cells, let alone ones with 4 × 21700 cells.
4 × 21700 version — link
8 × 21700 version — link.



It's definitely a cool power bank, no question about that. It could do with a slightly lower price and a bit more capacity, but as a geeky gadget, it's more than good enough.


A middle-ground option is a modular power bank. They come in different sizes, for 8, 16, or 24 cells, and for different cell formats: 18650 or 21700.
Model for 16 or 24 18650 cells — link
Model for 16 21700 cells — link
Model for 24 21700 cells — link
Model for 8 cells — link.



The modular design starts with 100 W output, but adding extra modules lets you scale it up to 100+100+60 W, or even 100+120+120 W.
Also, you don't have to solder or spot-weld the batteries; they use holders. Personally, though, I'd rather spot-weld them.



There are also some “original” options in the “I made it out of whatever I had lying around” category — link.




Now, about the one I ordered and received. There's a link to the same power bank in the title. I ordered it here, but it's no longer in stock there.
The seller had three options, differing only in case color. I couldn't decide which color I liked best, so I ordered all three. It came to $15.77 after a small discount. I ordered them because this was the only listing that mentioned 12 V output in the description. One buyer even got a refund without any trouble because there were no instructions on how to assemble it all.



It comes as a DIY kit, but the first thing that surprised me was that the batteries connect directly here, without holders, i.e. you'll have to solder or weld them. The seller didn't mention this at all, so it came as a surprise to me.
There were also complaints that something arrived broken because of poor packaging. Everything was fine with mine, possibly because I ordered several and the package was bulkier.


Another surprise was that there were only 8 small self-tapping screws, although the design seemed to call for more. But first, it's surprising they didn't get lost at all, and second, it turned out that was how it was supposed to be: after a while, a photo showing the contents of the package appeared on the product page. There was something odd about that too, though: one of the plastic inserts in the photo is yellow, while mine is gray...


The case came in three colors: yellow, blue, and gray. They all look fine, so choosing was difficult. Later, though, I realized that a power bank should be bright, so it doesn't get lost and is easier to find when the power suddenly goes out at home.


The board runs the full length of the case. Apparently there wasn't enough room for a couple of traces, so they were replaced with a pair of wires. The board is very recent; it was designed shortly before I ordered the power bank, which explains some of the oddities I'll get to later.


There are four ports: an "Apple" port and Type-C on one side, and the usual pair of USB-A ports on the other.


The power bank is built around a pretty decent IP5356 controller.


The board also has:
1. A chip near the ports that I don't recognize. I assume it's a transistor array for switching the inputs and outputs, but I'm not sure.
2. A pair of transistors. I'd guess they're in the battery protection circuit, but the chip seems to handle that itself.
3. An MC32P7031 microcontroller (?), apparently for controlling the display. I can't think of what else it could be for.
4. And, of course, the display: a large one showing the operating modes, charge percentage, voltage, and current.


The first oddity came up when I tried to fit the board into the case. First, there were no screws included to secure it, and second, the plastic case has ribs that get in the way of the wires. Apparently they were in such a hurry during development that they just threw in a couple of wires, while the case was designed for a flat board.


There's not much else to describe. To test the power bank, I need to connect batteries to it, so I used four Vapcell F52 cells from my recent review. Why not? I had four cells, they're great cells, and I'd even say they're overkill in terms of current for this power bank, so they'll work reliably.

And here I ran into the second oddity. To be safe, I stuck cardboard onto the positive contacts, but then the cells wouldn't fit lengthwise in the case, so I had to trim the edges. Don't mind the odd-looking cardboard; I laser-cut it for larger 21700 packs, but because I was cutting it on the balcony and the cable had gone stiff, the mechanism slipped and the second pass was offset. It wouldn't fit in the intended device anymore, but it's more than good enough for this.



1. These protrusions here.
2. Judging by the case design, spring contacts were originally planned, but that's another oddity. Clearly something went wrong during development.


Then it turned out that it wasn't clear how to connect the batteries at all. The configuration is obviously 1S4P, but how do you route the contacts...?


Later, I saw their connection method in the description for another model, but in my opinion, it's complete nonsense.


I'm test-fitting the board. Looks okay, at least at first glance.


Yeah, right. Looks okay? Here's problem number three (or four).
See the connector housing?
See the battery's positive contact?

Well, they're really, really close together, and one is essentially the battery positive and the other its negative...


There's clearly space inside for contacts, but never mind that there aren't any. How were they planning to connect the negative contacts in that case?


The other side seems fine. Seems.


But no, the problem showed up here too. Let me explain.
This slot is on the USB-A side, while the chip is near USB-C, so on the other side of the board. And as you can imagine, if the battery terminals are only routed here, the current has to flow through the entire board—and that can easily be around 7 A.


Anyway, I looked at this from an engineer’s perspective and realized I wasn’t happy with this arrangement at all. So, here’s what I did:
1. I assembled the battery pack with two positive leads and one negative lead. The board has three terminals too: positive/negative/positive. I also added a jumper between the positive leads and slipped insulating tubing over the wire for safety.
2. I added cardboard insulation around the positive terminals.
3. I drilled a hole in the case on the chip side, where there’s also a battery connection. I also cut away some of the plastic to make room for the wires.
4. I carefully soldered the battery wires and routed them so they wouldn’t get in the way of anything, including the latches on the display housing.


I turned it on. It first showed 0%. I connected it to a charger, and the reading changed to 95%, even though the batteries were fully charged. But this was the first time it had been switched on, so anything could happen.


Now for the rest of the kit.
It came with a plastic display housing, a button, a light filter, and even a trim piece for the filter. It’s genuinely quite nice and neatly made. In one of my kits, though, the light filter had somehow come unstuck from its backing paper and stuck to the housing instead. That’s not a big deal, since it doesn’t really need glue anyway—the housing holds it in place. By the way, the light filter has a protective film, and it’s best to peel it off right away; it’ll be harder to remove later.


I put the display housing back in place. It’s held by four latches, but there’s a problem: the clear plastic case won’t fit over it. So I took it off again.


The correct assembly order is:
1. Take the side panels and clear cover, and peel the film off the cover.
2. Insert the power bank into the cover.
3. Fit the side panels and secure them with the included screws. Hopefully all eight are there and none have gone missing. Be careful when tightening them—the screws are short and thin.
4. Only after all that, insert the display housing into its opening.


And that’s the power bank assembled and ready to go. It still has the protective film on the display here; I didn’t even notice it at first.



For a quick first check, I plugged my phone into the power bank and checked which charging mode it negotiated. My phone turned out to be picky: it won’t charge at 12 V with every power adapter (charger). The power output doesn’t really change, but this makes for a good test.


My first impression wasn’t great: the current and voltage readings are both pretty inaccurate, and both read high. That said, it depends on the load current. If I remember correctly, at low current it actually reads low instead.


1, 2. At high current, the discrepancy is even greater—around 20% at 2 A.
3, 4. When charging from a PD power adapter, it charges at 12 V and around 1.5 A, giving a charging power of 18 W. Pretty good, in my opinion.


The protocols are all good. Sure, there’s no 20 V, but nobody promised that. The vast majority of devices will charge in the fastest mode supported at the advertised power, so it passes this test too.
1, 2, 3. USB-A, USB-C, PD
4, 5, 6. The same outputs, tested with a different tester.
7, 8. FCP and AFC support 12 V. I checked with my wife’s Samsung, and it charges perfectly at 12 V.


It handles the load without any issues. I didn’t test every mode, just the three main ones.
At 5 V, the output current goes up to 3.6 A, then the output shuts off. Once the load is removed, the voltage recovers.


At 9 V, the current goes up to 2.4 A, after which the voltage drops to around 4.2–4.2 V. If the load current keeps increasing, the output shuts off.


The 12 V test passed at 1.7 A. After that, just as at 9 V, the output voltage dropped; if the load current keeps increasing, the output shuts off.
In all three cases, when the load current exceeds 1 A, the output voltage rises slightly to compensate for the voltage drop in the cable.


Now, about capacity.
To start with, I fully discharged the power bank, but in an uncontrolled way: I simply used it to charge other devices, so it discharged at varying voltages and currents. I noticed that even after the power bank reached 0%, it kept charging devices for quite a while.

I checked the next charge from time to time and noticed the same oddity with the displayed charge percentage. For example, the power bank stayed at 98% for quite a while, even though charging was still in progress. In this case, it’s more accurate to compare the displayed percentage with the amount of energy “put in” using Wh rather than Ah.


I wasn’t too happy with this discrepancy, so I decided to see if I could fix it. Modern controllers usually have an option for setting the battery capacity, since batteries can have a high capacity and different discharge curves. It’s more accurate to use the “coulomb-counting” method, which tracks the capacity delivered or received.
And yes, this chip supports it. The parameter is set with a resistor, and the datasheet includes a corresponding table.


I opened up the power bank, swore a blue streak, and took the opportunity to remember all the developers and their relatives.

The datasheet says the resistor can be connected to either the chip’s LED5 or LED6 pin, depending on the firmware and other details. In my case, it was connected to LED5 and had a value of 40.1 kΩ. According to the table, that gives a capacity of 40.1 × 448 = 17,964 mAh, while my battery’s actual capacity is closer to 20,200 mAh—12% higher.
Unfortunately, the resistance needs to be increased, so adding a correction resistor in parallel isn’t an option. I found a 4.91 kΩ resistor at home, connected it in series, and ended up with a total resistance of about 45 kΩ, which corresponds to 20,160 mAh according to the table.


I connected the power bank to an electronic load and started testing.
After fully charging it, I measured the energy delivered. Again, Wh is the more useful figure here, since the power bank discharged at both 12 V and 5 V.
First, I set the load current to 1.5 A at 12 V. After a while, when the power bank had delivered 50 Wh, the output voltage started fluctuating. I suspect the chip began reducing power due to temperature, so I lowered the load current to 1 A.


I also checked the thermal performance, though with the power bank still in its case. It does get warm—noticeably so—but there’s no obvious overheating.
Overall view.
Left side. At first I thought something around the display was heating up, but no—it was the circuit board.
Right side, showing the case heating up. It’s awkward to check the chip temperature while it’s running.


After a while, the power bank switched down to 5 V. Apparently, delivering full power at the end of the discharge was too much for it.
The results were as follows:
50 Wh delivered before the output voltage started fluctuating
59 Wh delivered before it switched down to 5 V
60 Wh delivered in total before it shut off completely

I also periodically recorded the delivered energy alongside the displayed charge percentage:
27.5 Wh — 48% (should actually be 54%)
42 Wh — 21% (should actually be 30%)
45 Wh — 14% (should actually be 25%)
51.5 Wh — 0% (should actually be 14%)

As you can see, by the end the capacity measurement error had reached 14%, which is quite a lot. At the time, it seemed to me that the changes hadn’t made any difference at all.


I kept poring over the datasheet and found another table near the very beginning. It lists different chip variants, probably with different firmware. It turned out they differ not only in which pin the setting resistor connects to: some variants have a fixed capacity of 10,000 or 20,000 mAh. I still couldn’t figure out exactly which chip I had.
But if we assume the resistor is connected to the LED5 pin, then it must be variant 5, 6, or 8.
You can also see that in all three of the options listed, the battery voltage is set by an external resistor. In my case, this pin is connected to an input on an external microcontroller and a 510 kΩ resistor, which would seem to qualify as NC (not connected), although that's debatable.
Also, all three options allow the LED6 pin to be used, but in two cases it's a fast-charging indicator output, while in the other it's the CL signal from the PD output. However, continuity testing showed that this pin isn't connected to anything at all.

Anyway, decoding the chip marking would certainly make things a lot easier. Has anyone come across it?


But when I put the power bank on charge again, things turned out completely differently this time. The tester reading in the first photo is hard to make out, but it was about 6 Wh.
On the second charge, it seemed like everything had changed in exactly the opposite direction. For example, it was at 98% charge after 65 Wh, and the total capacity measured was the same: 80 Wh.


As for efficiency, in 12 V, 1.5 A mode the power bank needs 80 Wh for a full charge, but delivers 60 Wh. The battery capacity is 73 Wh when discharged to 2.5 V.
The datasheet has graphs showing efficiency versus output voltage and load. At 12 V, it's around 93–94%. I got less, but it probably doesn't shut down at 2.5 V. Later, I found a note in the datasheet saying that the indicator flashes when the voltage drops below 3 V. More precisely, it says the indicator flashes below 3.2 V and turns off below 3 V.
According to my tests, when discharged to 3 V, each cell delivered about 17.37 Wh, or around 69.5 Wh for the whole battery. We got 60 Wh at the output, which works out to an efficiency of about 86.5%—quite a bit short of the stated 94%. But there's a small caveat: I measured after the connector and the QC trigger board, so I think the actual figure was closer to 90%.


But I wasn't satisfied with that. To double-check my theory about the capacity-setting function being disabled, I opened the power bank again, swore at it all over again, and removed the extra resistor, restoring the resistor to its original 40.1 kΩ value.
Then I charged it again and repeated the discharge test, but everything went exactly the same as before.
10 Wh - 82%
20 Wh - 64%
30 Wh - 44% (27.5 Wh - 48%)
40 Wh - 24%
42 Wh - 20% (21%)
45 Wh - 14% (14%)
50 Wh - 3%
51.3 Wh - 0% (it reached 0% at 51.5 Wh)

Unfortunately, I didn't keep precise records across the whole range the first time, but the points I was able to check matched almost exactly. The figures in parentheses are from the first test.


And, just like last time, I noticed that before going into overheat protection, the chip raises the output voltage. You can see it in the graph. The graph also shows some fluctuation at first, then a steady section, then more fluctuation. That's when I cooled the board with the chip a little.


The datasheet has a graph of output voltage versus load current and battery voltage (and it's a really detailed datasheet, by the way), but it doesn't show this kind of increase at low battery voltage.


Since I had to take the power bank apart again anyway, I also checked how hot the chip itself got.
The thermal image on the left was taken after about 2 hours and 40 minutes under an 18 W load, so close to the end of the discharge.
The image on the right shows the moment the overheat protection kicked in. Even a slight drop in temperature brings the power bank back to normal operation, so if you want to get full power right to the end of the discharge, you could stick at least a small heatsink on the chip. That should help.


So, what can I say in the end?
On the one hand, I liked the power bank, partly for its original design, large and fairly informative display, support for 21700 cells, and compatibility with both QC and PD. Overall, I really have no complaints about its electrical performance or ease of use.
But I do have major complaints about two things. First, the construction—or rather, the assembly (and subsequent disassembly)—is a complete mess. It really feels like two teams worked on it without communicating, and both were in a huge hurry, because the case and the board clearly have a life of their own.
Second, the charge indicator. I couldn't get it to display the charge level correctly. But if you're using cells that aren't 5000–5100 mAh, but instead some used Tesla cells from a teardown with a capacity of 4800 mAh, things will probably work a bit better.

So overall, this is an interesting power bank, but it has some teething problems. I also bought three because after another project I'll have some spare new, name-brand, good-quality 21700 cells—not enough to make up the number of batteries I need. I wanted to use them with these cases and probably sell them, but of course only after assembling and testing them. What's more, a power bank like this, together with a PD trigger, can be used to power the same router and optical terminal for around 5–6 hours.


That's all for now. I really hope this was useful. I originally planned to write a short review, but somehow it ended up much longer than expected. Sorry about that—I’ll try to be more concise next time.