Power Bank + charger for four 18650 batteries

By and large, this device has been on the market for quite some time, but in this case it is sold under the Tlife brand.
There are four versions of the device: black or white, and with or without batteries included.
I received the version without batteries, since the store has some difficulty shipping batteries.

Specifications as stated:
Replaceable batteries
LCD
Microcontroller control
Overcharge protection
Short-circuit protection

Capacity - 53.3 Wh
Input - 5 V, 1 A
Output - 5 V, 1 A and 2 A
Self-discharge - 2 mAh per day.


The package includes:
1. Tlife power bank
2. USB-microUSB cable
3. Manual


The manual is in English, but it is very detailed. Though in my opinion, a device like this is easy enough to figure out intuitively.


The USB-microUSB cable is short, about 30 cm. There are no data lines, so the cable is for charging only. The connectors fit very snugly.


As I mentioned above, the case comes in two colors; I chose white.
Dimensions: 130 x 83 x 25 mm. In my opinion, this is quite a bulky device.


The name of the Tmart store is also printed on the bottom, since these devices are sold under its brand.


The empty case is fairly light at 116 g; with batteries, it will weigh about 300 g (4 batteries x 45 g).


There is a display on the front, and it is a dot-matrix display, which is quite unexpected. Usually, the screen just has pre-made icons. I suspect this display adds a noticeable amount to the device's price.
The connectors and power button are on the back.
1. Power button. The device can turn on automatically when a load is connected, but sometimes the load cannot immediately draw charging current, so the button can come in handy. At least one device of mine required me to press it.
2. 2 A and 1 A outputs; the teardown will show how they differ.
3. microUSB input for charging the power bank itself.


The top cover slides back to reveal the battery compartment. It locks securely in both positions.


The positive terminals are recessed slightly into the case, but they make good contact even with batteries that have flat positive terminals. Those are the batteries I used for the test.
The negative contacts are springs, which is both an advantage and a drawback.
The advantage is that the springs accommodate batteries of different lengths, with or without protection circuits.
The drawback is their high resistance; I'll measure it later.
The springs hold the batteries securely. During testing, I often had the power bank upside down, and the batteries never even tried to fall out.


For the test, I used Sony VTC4 batteries, which also featured in my battery review.


A little about the operating modes and indicators.
1. When switched on, either with the button or automatically, the message "Power-on" is displayed.
2. During charging, an animation shows the charging process, but it is not tied to the voltage or state of charge. Animation means the batteries are charging; no animation means they are charged. Each battery has its own animation.
3. When there is no load, the battery charge level is displayed.
4. The charge level is shown separately for each battery. Since the batteries are essentially connected in parallel, though, the point of this feature is questionable. If you remove one of the batteries, its icon disappears from the corresponding position.
5. While powering connected devices, an animation of the connected port is displayed, along with the charging current. The battery charge level is shown briefly from time to time. The display is on continuously by default; a short press of the power button turns off the backlight.
Since the power bank has automatic turn-on and turn-off functions, I checked how it behaves with low-power loads. In my case, everything worked at a charging current of 0.05 A, but at a lower current the device will turn off, i.e. Charging very low-power devices will be difficult; the power bank may shut off before they reach 100% charge.


Now let’s move on to the tests and other experiments.
First, I’ll show how the displayed charging current compares with the actual current.
Using an electronic load, I tested the device at currents from 0.5 to 2.5 A in 0.5 A increments.
As you can see in the photo, the meter reads high.


The output voltage dropped to 4.92 V at maximum load current, although the power bank’s display showed 5.05–5.06 V. This is due to the device’s circuit design, but in any case, both 5.06 V and 4.92 V are within the acceptable range of 4.75–5.25 V.


At load currents above 2 A, the overload protection trips and the power bank’s output shuts off. It happens so quickly that the load didn’t even register it, but thanks to the inertia of the power bank’s display, I managed to photograph the current just before shutdown.


Here are a few tests using scenarios closer to real life, showing the current depending on which output was used.
1, 2. Smartphone. I made a small mistake here: I used a “full-featured” cable, but the current was the same with the included cable. That’s because the phone’s charger reduced the charging current when I used the full-featured cable. This effect didn’t occur with the other loads.
3, 4. PIPO tablet. The current was about 0.9 A with either output; with its original charger, it draws about 1.8–2 A.
5, 6. CUBE Android tablet. This tablet can easily draw 2 A from a charger, but on the first port (1 A) the current exceeded the limit, while on the second port (2 A) the tablet itself limited it to 0.5 A.


1, 2. Powerful Teclast tablet. In this case, the tablet was completely discharged, and the power bank went into protection when connected to either port. Many of you will know what happens when you connect a completely discharged tablet to a charger: it tries to turn on the screen, then starts repeatedly shutting down. More or less the same thing happened here, except it was the power bank that shut off.
3, 4. But if you leave it like that for a while, the tablet eventually gets tired of shutting down and starts charging at 0.45 A. After charging in this mode for a few minutes, I switched it to the second port (2 A), and charging continued normally at 2.24 A.
By the way, the ports are easy to remember: the first is 1 A, the second is 2 A—that is, 1-1, 2-2.
5, 6. A more modern smartphone that can charge at up to 2 A. In the first case, the protection trips; in the second, the charging current is just 0.91 A. I couldn’t use the cable included with the power bank here, because the smartphone has a USB-C port and charges through its own cable. I think that with a regular (non-data) cable, the charging current on the second port would be 2 A.


That’s the first round of tests done. Now let’s take it apart and look at what’s inside.
The device is very easy to disassemble. First, lift the top cover slightly and remove it. Make sure the cover is fully open when you do this.
Then insert something sharp between the two halves of the case and separate them. The plastic is very good and quite flexible, so it’s hard to break.


The case consists of three parts plus the circuit board. There’s also a button that keeps trying to fall out and get lost, so be careful.


At first glance, everything looks very neat.


But on closer inspection, you can see flux residue and some rather untidy soldering. This doesn’t really affect how it works, of course, but it does give you a general idea of the manufacturing quality.
A reminder: all photos are clickable.



There are quite a lot of components on the board. You can see a microcontroller, separate charging ICs, and so on. There are also a lot of passive components.


1. PIC16F1933 microcontroller made by Microchip.
2. Separate charging circuits built around LTC4054 equivalents/clones. This is straightforward: there’s a 2.2 kΩ resistor nearby, so the charging current is about 0.5 A. At the very beginning, I listed the power bank’s specifications, which stated an input current of up to 1 A. In reality, it’s up to 2 A, so keep that in mind when choosing a power supply.
3. DC-DC step-up converter IC. It uses the fairly well-known G5177C. According to the datasheet, it provides a continuous output current of 2.1 A and a guaranteed short-term current of 3 A. The operating frequency is 500 kHz.
4. USB output ports. This is where the difference in output current comes from. The data pins are wired differently, so connected devices set different charging currents.
5,6. But besides the different wiring of the data pins, there are also "hardware" differences. There is a MOSFET next to each port that switches it on, but there is also a current-sense resistor. The resistor for the first port is 0.1 Ω, and for the second, 0.05 Ω.

So the microcontroller monitors the output current, but the actual current is determined by the port wiring and the way the connected device works. That's why you sometimes get a situation where the power bank can supply the current, but the device doesn't want to draw it, or vice versa: the device is ready to draw 2 A, but the power bank can't provide that much through the first port. A cable without data wires can partly solve this problem, but in my opinion, what's really missing are "smart" controllers that can set the appropriate signals on the data lines themselves. It doesn't even matter whether the power bank supports QC mode; controllers would be very useful even without that feature.

I should also point out that even though the DC-DC converter can be switched off "in software," the device can also disconnect the load using separate transistors, independently of the converter.


The converter IC is wired as shown in its datasheet, so I used the datasheet values to determine the maximum output current.


There are almost no components on the back of the board; most of it is empty.


There is an inductor and a diode next to the USB ports. The inductor has the same specifications as the one in the converter IC's datasheet. I'll show what the diode does later.
I really don't like the springs, though; the wire is much too long.


I didn't draw the complete schematic because I didn't see the point, but I'll show how this power bank is put together.
First, power goes to four identical chargers, which charge the batteries independently of the rest of the circuit.
Four diodes (D2-D5) protect against inserting a battery with the wrong polarity. These diodes also let the microcontroller monitor the voltage of each battery independently.
After the diodes, the voltage goes to a step-up converter, which provides 5 V at up to 2.1(3) A. The microcontroller controls the converter, switching it on briefly at intervals. This is needed to detect when a load is connected and turn the power bank on automatically.
The negative contacts of the USB ports are connected through a switching transistor and a current-sense resistor. The microcontroller measures the voltage across the resistor and displays the current draw. The transistor disconnects the load if the current exceeds the limit, and is controlled by the microcontroller.
The output voltage is also measured by the microcontroller. But since there's a current-sense resistor and a transistor between the USB port's negative contact and the negative rail of the rest of the circuit, the display shows a steady 5.06 V, while in reality the output is only 4.92 V at 2 A. The difference is dropped across those components. This is an inherent limitation, so the power bank's display is more of a rough indicator than an accurate measurement.

One more thing: besides charging the batteries independently, controlling the outputs independently, and providing protection, the device also has a UPS function. In other words, it can charge the batteries and power a load at the same time. When power is supplied to the power bank, the input 5 V goes through a separate diode (D1) to the converter input. Since this voltage is higher than the battery voltage, the device runs from the external power supply.

People often ask me how to make a simple 5 V UPS. Here's a typical example: a charger based on the TP4054 (4056), two diodes (for one battery), and a step-up converter.

Overall, the circuit design is straightforward, with no clever tricks; everything is simple and transparent. The downside is low efficiency, since the diode isolation wastes energy.


To estimate the device's efficiency, I decided to compare the batteries' actual capacity with the energy delivered to the load. I loaded the device at 1.5 A and installed two batteries.
According to my measurements, the batteries have a capacity of about 7.3 Wh each, so two batteries provide 14.6 Wh.
I'll say right away that efficiency depends on the load current and the number of batteries installed, so it will be a little higher at a lower current and with four batteries installed.


I got 11.46 Wh at the output, which gives an efficiency of about 78.5%.


As you can imagine, I wanted better performance, so I bought more efficient 30BQ015 diodes. The original ones had a voltage drop of about 0.34 V; with the new ones, it was 0.24 V at the same current. I wanted to buy diodes with a voltage drop of around 0.15 V, but they weren't available for sale :(


For the experiment, I first soldered in a pair of diodes. They're noticeably larger, and the pads weren't designed for them, so I had to tin the diode leads first, then heat them and "shrink" them into place.


The result was better, though not by much. Instead of 11.46, I "got" 12.03; efficiency was 82.4%.


During the test, I monitored the component temperatures. The old diodes heated up to 80 degrees (all tests were done near the end of the discharge, at the highest current), while the new ones only reached 54.4. The converter IC heated up to just 56.6 degrees, which is very good in my opinion.
By the way, unlike the MOSFETs installed in the converter IC and at the output, diodes have a negative temperature coefficient for their voltage drop. In other words, the hotter a diode gets, the lower its voltage drop, so a high diode temperature isn't always a bad thing.


Replacing the diodes improved efficiency a little, but as always, I wanted more.
First, I found that the traces leading to the diodes had a voltage drop of about 0.02 V, with another 0.02 V across the trace from the far diode to the converter. That makes a total drop of 0.04 V.
I doubled the traces with 0.5 mm wire.
While I was at it, I also added a capacitor before the converter (simply because there was an empty spot for it), an input protection diode for the 5 V supply (there was room for that too, and I had a diode left over after the replacement), and made a small adjustment to the second channel's current-sense shunt by soldering a 1 Ω resistor in parallel.


But I was even more surprised when I measured the voltage drop across the battery springs: 0.17 V! That's more than I gained by replacing the diodes and reinforcing the traces.
I used some wire to bypass the springs. Of course, flexible wire is best for this, like the kind used for power-tool brushes, but I had to use ordinary wire.


The effect was quite significant: efficiency rose to 87.6%.
Here are the voltage drops across the different sections before and after the modifications:
Springs - trace before diodes - diodes - trace after diodes
0.17 - 0.02 - 0.34 - 0.02 = 0.55 V!!!
0.05 - 0.003 - 0.24 - 0.004 = 0.3 V.

The efficiency improvement is due not so much to the device shutting off at a lower battery voltage as to the higher voltage now reaching the converter input. As the input voltage rises, the input current drops, so the energy is used more efficiently.


I can already hear the next question: what else can be improved? Not much, really. At most, I could use 95SQ015 diodes, but they're more expensive, and the efficiency gain wouldn't be very large. I could also replace the output switching transistors, but in some situations that wouldn't make any difference at all.
The IC itself is about 94% efficient with a 3.5 V supply and a 1.5 A load current. In my case, efficiency came out to about 87.6%—a 6.5% difference, which is quite a lot, of course. Unfortunately, the remaining options are expensive and not very effective.


Why didn't I use the well-known MOSFET protection circuit instead of diodes? Because that circuit isn't suitable for two reasons:
1. It only works with one battery. If you connect even two of these circuits, then as long as at least one of the two batteries is inserted correctly, the second transistor will also be on.
2. Since the transistor turns on based on the output voltage, it wouldn't be possible to implement the "uninterruptible power supply" function.


People sometimes ask me how to make a simple 5 V "uninterruptible power supply." One option is to use the simplified solution found in this power bank.
As they say, let's throw out everything unnecessary and keep only what we need.


That's about it. The only thing I didn't check was the standby current. But since the current draw is pulsed, it's not very convenient to measure.

Pros.
Convenient design
Independent battery charging
Supports the "uninterruptible power supply" function
Protection against overloads and output short circuits
Battery reverse-polarity protection
Display

Cons.
Not a very efficient circuit, with significant losses in the isolation diodes.
Significant losses in the springs connected to the batteries' negative terminals.
Automatically shuts off when the current draw falls below 0.05 A, so it won't work with all devices. Alternatively, you have to charge two devices at once.
Some standby power consumption.

My opinion. As far as I can tell, this power bank is a pretty good platform for modifications, but it isn't very efficient as-is because of its low efficiency.
The ability to charge the batteries while powering a connected device is a nice feature. However, keep in mind that the power bank's own current draw will be slightly higher than the combined charging current and load current. For example: 2 × 500 mA charging + a 2 A load + conversion losses = about 4.5 A. So I wouldn't recommend using this mode with heavy loads, and I'd also choose the power supply carefully.
This power bank can also be used simply as a charger for 18650 batteries.
For a quick and effective modification, all you need to do is reinforce the springs and power traces with wire. Replacing the diodes will also improve the results, but by half as much as soldering the traces and springs.

That's all. I hope you found the review useful.
The store offered coupon TC01, which should bring the price down to $12.99. The coupon is valid until February 14.