Epilot E4 power bank for 4×18650 cells with pass-through charging

How many times have I started with the words, “I’ve already reviewed a similar product,” only to find myself doing it again? About a year and a half ago, I posted a review of a rather unusual power bank for four 18650 batteries, with independent charging and pass-through power. Today’s review is of a very similar device, though it does have a few differences.


By and large, all power banks of this type are clones, to one degree or another, of TOMO power banks. Since I’ve already reviewed one of the clones, I’ll refer back to that review from time to time—especially since the model here has some differences, and even a few improvements.

The power bank comes in a black-and-yellow box that brings another brand to mind: Nitecore.
The package also included an English-language manual with the specifications and a brief description of how to use the device.


The power bank is available in two case colors, black and white. I got the black one.


The manufacturer lists the following specs:
Input: 5 V, 2 A
Output: 5 V, 2 A
Weight: 85 g
Dimensions: 100 × 84 × 27 mm


As before, there’s just one control button, but this time it’s on the top of the case. The display has been moved there too.


One end of the case has:
1. One microUSB power input
2. Two full-size USB ports for connecting devices to be powered.
3. As befits any Chinese device, the power bank now has a flashlight :)

The USB output ports are labeled 2 A and 1 A.

The other end just has the device name and specifications.


The power bank is easy to operate:
Press and hold the button to turn it on or off.
Press briefly to turn on the flashlight. It starts with a steady beam; press again and it flashes SOS.


The overall dimensions are slightly smaller, probably in part because the display has been moved to the top cover. The batteries are installed just as before: slide the case cover open to access the four battery slots. The cover is a tighter fit than on the previous version.


Batteries up to 69 mm long will fit. I tried a protected battery and it went in without any trouble, though I’m not sure why you’d use one here, since the device has its own protection circuits.
Both long and standard batteries stay in place and don’t fall out, even with the case cover open.


The display looks similar to the one on the previous model, but the graphics have changed (it’s a graphical display). Although the display is now in a more convenient location, the information is harder to read because the brightness is lower than before. In the light of a flash, the indicators are almost impossible to see; the same will happen in bright sunlight.
On the plus side, it now shows the temperature. I’m not entirely sure why that’s useful here, but later we’ll see where the temperature sensor is and what it measures.


Examples of the display. The power bank can show the approximate battery charge, charging status, batteries that are missing or incorrectly installed in their slots, temperature, output voltage, output current, and when protection trips on each output independently.
While charging, and while waiting to charge, the status of each battery is shown independently. In discharge mode, only the overall status is displayed.


When two devices are connected, the display shows the load current on each channel, the overall output voltage, and the approximate power bank charge.


The battery compartment cover has a latch, but if you really want to, you can remove it completely—which is exactly what I did. It’s difficult to remove: you have to bend it in the middle to get past the latch.


But I was curious about what was inside—probably even more curious than I was about the outside—so I opened the case completely. There are fairly tight clips along the sides and at one end. The two case halves fit together quite well; everything looks neat and feels very solid.


Compared with the previous version, the circuit board has clearly been significantly revised. The power traces are a little wider, and the circuit design has changed too—you can see that at a glance.


Interestingly, the product page clearly shows an older version of the board, as you can tell from the six power diodes (4+2), which aren't present in the version under review.


Last time I complained about how hot these diodes got. Here they've been replaced with MOSFETs, which should improve efficiency and reduce heat.


Here's a closer look at the modified section. The protection circuit uses a pair of APM4953 P-channel MOSFET arrays and an LM393 dual comparator. Each such set handles two power bank batteries.

In this case, the manufacturer made a useful change: each transistor in the array has an on-resistance of 53 mΩ, which means that with two fully discharged batteries and an output current of 2 A, the voltage drop is about 0.1 V. With all four batteries installed, the drop falls to 0.05 V. In my previous review, I used high-power Schottky diodes, but the MOSFET option is even better—and it's built in from the start.


Using a MOSFET to protect against reverse battery polarity isn't a new idea; I showed this circuit last time. But I also noted then that it can't be used in a power bank in this form, because the voltage present after the protection circuit would prevent the transistor from turning off.


This device uses a modified version of the circuit, with the transistor controlled by a comparator. It monitors the voltage on both sides of the transistor and disconnects the battery if its voltage drops below that of the rest of the circuit. This can happen if the batteries are badly out of balance (some discharged, others charged), or if a battery is installed backwards. However, this type of protection is slightly less reliable than the diode-based version, because reversing a battery that's already installed can cause a current surge.
The circuit also receives a signal that disconnects the batteries from the power supply while charging. The comparator is powered from the output of the protection circuit; initially, power is supplied through the MOSFET's internal body diode (shown in parallel with the transistor in the schematic).


The step-up converter is an MT5033, and the overload protection circuit for the outputs uses a 9926A N-channel MOSFET array.
There have also been changes to the protection circuit, but this time for the worse. Previously, there were two arrays, one for each output. The 2 A channel had a 0.05 Ω shunt, while the 1 A channel had a 0.1 Ω shunt. Now a single array handles both channels, and both use identical 0.1 Ω shunt resistors.
As a result, the voltage drop at the USB outputs is higher. Previously, the total circuit resistance was 0.03/2+0.1=0.115 Ω for the 1 A channel and 0.065 Ω for the 2 A channel.
Now it's 0.03+0.1=0.13 Ω on each channel. There's almost no difference on the 1 A channel, but the resistance on the 2 A channel is twice what it was before.
So at 1 A, the voltage drop is 0.13 V; at 2 A, it's 0.26 V, which is quite a lot.


The power bank uses a pretty good MT5033 step-up converter, with an operating frequency of 800 kHz and an output current of up to 2.5 A, even with an input voltage of 3 V (fully discharged batteries).


1. The whole thing is controlled by an Atmega168 microcontroller; the previous version used a PIC16F1933. Subjectively, startup and shutdown, as well as screen updates, are now a little slower.
2. Charging is handled by four controllers marked 55b6. I didn't look up the full part number, but essentially they're the same TP4055 chips, with an expected charging current of 0.4–0.5 A per channel.
3. Voltage regulator for the display and microcontroller.
4. A transistor that serves as a temperature sensor.


The device supports pass-through charging. The batteries are disconnected from the converter by the blocking circuit with diode D2, while power is supplied to the load through diode D3.
That is, The load does not interfere with charging, but the converter is always running, either from the batteries or from the external input.

The reverse-polarity and charging protection circuit is shown for one channel; the others are identical.


As I wrote above, there is a temperature sensor inside the power bank. I just don't understand why it's there, since it doesn't monitor the temperature of either the converter or the batteries, and basically just shows how hot the circuit board is....
Out of curiosity, I heated it with a soldering iron. Above 99 degrees, some incomprehensible gibberish started appearing :)


The back of the circuit board has only the connectors, the LED, and the converter inductor.



You can see a phosphor spot inside the LED. It doesn't shine very brightly—just enough to illuminate something small on the desk, or serve as a small night light.


For the test, I connected the power bank to an electronic load, bypassing the connectors. In practice, the voltage will be a little lower because of contact resistance, but since different cables have connectors of varying quality, I decided to leave that out of the test and see what the power bank itself can do.


Testing the output rated for 2 A: at 2.1 A, the voltage dropped to 4.83 V, and when I tried to increase the current further, the device went into protection mode.


The results for the second channel were exactly the same, which was predictable since the channel protection circuits are identical. The only difference between the connectors is that on the 2 A channel, the data lines are connected to resistor dividers, while on the 1 A channel they are connected to each other.


Voltage on the data lines for the 2 A and 1 A outputs, respectively.


But I was more curious to see what would happen if I loaded both channels at the same time. To do that, I soldered the protection outputs together and set the maximum load current to 4 A, with a current step of 0.1 A. At 3.9 A, the voltage dropped to 4.63 V, and at 4 A the power bank's protection kicked in. The converter itself was working normally, but I wouldn't recommend doing this.


The device displays the output current. I ran one of the channels to see how accurate the reading was; below are the results at load currents of 0.5, 1.0, 1.5, and 2.0 A. In every case, the displayed value is slightly lower than the actual load current.
The displayed output voltage barely changes, since it's measured before the protection circuit and doesn't account for the voltage drop across the transistors. If I remember correctly, it was the same last time, so in this case it's more of a "look-what-I-can-do" meter.


There wasn't much point in measuring the ripple waveforms here, but I decided to check anyway. The oscilloscope was connected in parallel with the load, with no additional filters in front of the probe.


Here are the ripple amplitudes at load currents of 0.5, 1.0, 1.5, and 2.0 A, respectively. I wouldn't call the ripple small, but in my opinion even the 100 mV measured at maximum load is perfectly tolerable.


Last time, I checked the efficiency—that is, the batteries' capacity in Wh and how much energy could be drawn from them at the output. This time I ran the same test with the same load and the same batteries.
The result: the two batteries have a capacity of 14.6 Wh, and I got 11.95 Wh at the output, giving an efficiency of 81.8%. The previous one managed 78.5%. The average output voltage was 4.91 V, compared with 4.97 V for the previous one, thanks to its better-designed protection circuit.


The main "culprit" behind the lower efficiency is the same: steel springs. Nothing has changed in that regard, and probably never will.


These are the springs I'll be modifying. This time, I took apart some kind of relay to get some extra-flexible wire. It also has a silicone jacket, though that doesn't matter here.
The spring's resistance is almost 0.1 Ω, which gives a 0.1 V drop at 1 A.
Meanwhile, a fairly long piece of copper wire has a resistance of just 6 mΩ, or 0.006 Ω. As they say, feel the difference.


Since the wire seemed too long, and I only had three pieces, I cut one in half to get even shorter pieces. I stripped a short section on one side and about 5 mm on the other.


I tinned the springs closer to the part that contacts the battery. It’s best not to touch that part itself.
The coating takes solder very well. I didn’t even use any flux apart from the flux in the solder. Then I bent the long section of wire into a hook, caught it on the tinned spot, and soldered it in place. This is what the “tails” looked like.


So I soldered the short sections to the flat part of the springs and tucked the wires inside so they wouldn’t stick out or get in the way when installing the board in the case.
By the way, installing the board in the case is much harder than removing it. There’s a rib near the LED that prevents it from being pushed inward, and that’s the part you’ll have to wrestle with.


After the modification, the total resistance of the springs dropped almost 20-fold, from 0.1 Ω to 0.0055 Ω.


As an optional extra, I decided to reinforce a few traces with wire.
1. From the middle battery contacts; these traces are the longest and run around the display
2. Between the protection transistors
3. From the protection circuit to the converter itself.

There’s no point reinforcing the traces from the converter: the voltage drop there is noticeably lower. Besides, because of the way many tablet/phone chargers are designed, this wouldn’t make any difference.


But I wouldn’t say the change had a major effect on the results. In the end, I was able to get about 0.6 Wh more out of it, and the overall efficiency came to 86%, up from 81.8% before the modification. I managed to get 87.6% from the previous power bank. I think efficiency could also be improved here by reducing the resistance of the current-sense shunts, but unfortunately that isn’t possible.

I tested it with two batteries. With four installed, efficiency should increase a little more.


Interestingly, the boost converter ran cooler after the modification, even though I measured the temperature right at the end of the discharge, when the voltage and current were at their lowest and therefore the same in both tests.
1. Before the modification
2. After the modification
3. Temperature while charging four channels at a total current of 2 A.



One of the key features of these power banks is pass-through charging: they can charge and power devices at the same time. In other words, they essentially act as a kind of uninterruptible power supply.
The feature works fully and correctly, but the current drawn from the power supply will be quite high.
1. The power bank itself draws about 1.8 A while charging four batteries, so the charging current is 0.45 A per channel.
2. If you connect a load, its current is added to the charging current, and the total can easily reach around 3.5 A.


With a different cable, the charging current was almost 2 A, so the total current can reach 4 A.


The device also has automatic on/off. It switches off at a current of around 0.05–0.08 A. I couldn’t determine the current at which it starts up, but I can say it’s quite sensitive.
But as you can imagine, nothing comes for free, and the device draws about 2 mA in standby.
I wouldn’t call that a lot: a pack of four 2.5 Ah batteries would take about six months to discharge completely. Still, it’s certainly not a good thing.


Overall, this is essentially the same TOMO as the previous version, just slightly improved.
I later found out that this is actually a TOMO. Here’s the manufacturer’s website; there’s also a short video review there.




Compared with the version I tested earlier, the device reviewed here has some clear improvements: higher efficiency, an improved reverse-polarity protection circuit, a smaller form factor, and, once again, a flashlight :)))
But there are also some drawbacks: for some reason, one of the output transistor pairs was removed, and the shunt value on the 2 A channel was increased.
On the neutral side, they added a thermometer but left the spring design unchanged.

The power bank itself works just fine. Its output voltage is within spec, though it could be higher if the manufacturer hadn’t “optimized” the output stage. I liked the more efficient protection circuit and the decent output converter. It works fine even without modifications; the modification gives efficiency a small boost. The pass-through charging function is useful, as is the ability to swap the batteries quickly and use the power bank as a charger for 18650 batteries.
I don't need the flashlight or thermometer at all, though the flashlight might come in handy someday.

What it definitely lacks is fast charging, both on the input and output. In my opinion, fast charging on the output is especially important. I think the device would be noticeably more useful if the manufacturer added it. There’s room inside for it, and the microcontroller could handle the additional service functions.


The store provided a coupon for the review, IT$LEDE4, which should bring the price down to $11.99. That’s all from me; I hope you found the review useful.