Electronic load and battery tester: ZKEtech EBC-A40L

About a year ago, in one of my reviews of ZKEtech electronic loads, I wrote that I wanted to pick up a couple more. Recently, I bought and received one of them—the model with the highest current rating in the company's lineup.
A look around, a few tests, and of course, my conclusions.

Many of you know that I test all kinds of batteries. For the most part, I do it just for fun, as a hobby. At first, a load rated for up to 10 A was enough, but eventually I needed more. I added a couple more: the more versatile EBC-A10H and the relatively high-current EBC-A20. But even then, to run tests at 30 A, I still had to go back to my main load, which I built many years ago. I like it, and it's convenient in its own way, but it has two major drawbacks: it doesn't measure capacity in Wh, and it has no software, so you can't plot a discharge curve.

Reviews
EBC-A10
EBC-A10H
EBC-A20
USB adapter

I bought this load so I could test at higher currents and, at the same time, generate detailed graphs.
A quick clarification: this device, as well as the EBC-A10, EBC-A10H, and EBC-A20 listed above, is a battery tester. It can work not only as a load but also as a charger, so it can run a complete test cycle, including cycle-life tests with configurable charge/discharge cycles. I call them loads simply for convenience.

I bought the load on Taobao, with help from one of my regular readers—many thanks to him. The total cost, including all expenses, came to about 140 dollars. The same load is available on AliExpress

When I picked up my little box at the Meest Express office, I was a bit surprised. I expected it to be about half the size, so I had to carry it in the bag they gave me.
The packaging was excellent—five stars. It had layer upon layer of cushioning, and it turned out the box was from the manufacturer, not the shipping company, complete with the manufacturer's name on it.


The specifications are only in Chinese, but in brief:
Maximum input voltage: 5 V
Voltage measurement accuracy: 0.2% +/-0.003 V
Load current range: 0.1–40 A
Charging current range: 0.1–40 A
Current measurement accuracy: 0.2% +/-0.01 A
Maximum power dissipation: 200 W
Operating modes: CC, CP, CV
Weight: 5.4 kg

The specs list three operating modes, but CV isn't a load mode—it's a battery charging mode.



Honestly, even after looking at photos of it on sellers' pages many times, I never thought it was this big. In every photo, it looked about one and a half times smaller.
The design is a little unusual, though. It reminds me of equipment from the 1950s or 60s.


The package also included several bags of accessories and a small sheet of paper on top, which added to the weight too.


1. Pink sheet: a list of the package contents with the serial number (as far as I can tell).
2. Manual and warranty card, both completely useless in my case.
3. Power cable, no more useful than the manual, but fortunately it can be replaced with a standard computer power cable.
4. Cable with a USB-UART converter, nothing unusual; the same one came with the EBC-A10H.


They also included a set of battery connection leads, and it looks pretty formidable.


1. The alligator clips for the power connections look like the kind used to connect to car batteries, which has its drawbacks: you can't use them to connect to a regular 18650 battery.
2. The alligator clips for voltage sensing are much smaller and, unlike the lead set supplied with the EBC-A20, come separate from the start.
3. The leads are very flexible, with silicone insulation. They're rated at 10 AWG (5.26 mm²) and for temperatures up to 200 degrees.
4. Connection is made using five-pin “aviation” connectors: one pin for voltage measurement and four for the power circuit.


The leads are designed to connect to batteries with terminals, but I’ll be using a holder, so I’ll most likely buy a couple of the same connectors and wire it up without the included leads.


Five stars for the case, too. Not only is it made of very thick metal, it’s also finished to a high standard.
There’s a carrying handle on top, which is definitely useful.


1, 2. The connections, controls, and indicators are identical to those on the EBC-A20 load: four buttons, a 1602 display, and a pair of connectors. There’s no point describing them, since I’ve already covered that.
3. On the back there’s a 120 mm fan, a power connector combined with a switch, and a connector for connecting to a computer.
4. There’s also a warranty seal, but as I mentioned above, it’s completely useless in my case.


There are plenty of ventilation openings, and the fan itself moves quite a lot of air.


There’s a small label on the side with the model name, date of manufacture, serial number, and so on.
There’s another label on top, but I photographed it because of the original spelling of the word “temperature” :)


But inside, it’s unexpectedly empty. I think they could easily have made the load twice as powerful in such a huge case—for example, 10 V 40 A instead of 5 V 40 A.


The 5 V 60 A power supply deserves a separate mention. I initially asked to buy it without the power supply, partly because it adds unnecessary weight and cost, and partly because I already have two similar PSUs at home, but they’re Mean Well. The seller wouldn’t agree. Too bad.
I saw a YouTube video where someone received a similar load with a broken fan. Honestly, I can’t imagine how it could have been broken during shipping; maybe he just got “lucky.” The fan is a standard 120 × 25 mm model.


All the wires are neatly crimped and color-coded, and the wires to the PC connector are insulated as well.


The heatsink is made from the same kind of extrusion as the one in the EBC-A10H, just a little larger. A step-up converter board is mounted on the heatsink. It runs from 5 V and outputs 12 V to power the relay and fan.
Given the “reliability” of boards like this, I recommend putting a fuse in front of it.


Two relays switch the 5 V power supply to the charger. The first version of the EBC-A10 that I had didn’t have these relays. When powered up, it put voltage on the output, and if a battery was connected, it couldn’t start properly and tried to charge the battery to 5–6 V. This issue was fixed in later versions by adding relays.
The manufacturer also installed two relays here, splitting the load between them. They switch simultaneously, and the split was most likely intended to balance the current across the relay contacts. Another plus is that the relays themselves are generously rated at 40 A.

The relays switch on during both charging and discharging. The fan appears to be connected in parallel with their power supply, since it switches on at the same time. It switches off depending on the temperature: after a few seconds if the thermal switch hasn’t tripped, or once it cools down if it has.


The power module. The four load transistors and four charger transistors are connected here. Yes, you read that right: the charger is linear, not switching!
At first I thought it used active current balancing across the transistors, but it turned out that the many shunts installed next to each transistor are simply for passive balancing. Each transistor also has its own wire, which acts as another passive “balancer.”


1. The load transistors are the same as before—IRFP250M. The charger transistors are P-channel IRF4905s. Overall, this is a perfectly reasonable approach, but at maximum charging current, batteries with very low voltage could put them under a lot of stress. Though it’s unlikely anyone uses batteries like that these days.
2. Accordingly, there are eight low-resistance balancing resistors.
3. There’s a thermal relay on the left that switches the fan to full power when the heatsink temperature exceeds 50 degrees. In normal operation, the fan starts at 50% power when charging or discharging begins.
4. What I initially took to be op-amps for the active current-balancing system turned out to be optocouplers providing galvanic isolation for the computer connection.


At first, using galvanic isolation may seem strange, especially since it is more often needed when testing not only batteries but also power supplies. But there is a logical explanation.
The manufacturer makes an interface board that lets you connect up to 8 EBC-A40L and EBC-B20H loads into a group controlled by the EBC-X software. That’s probably where galvanic isolation between the control channels comes in handy.


A test setup like this looks roughly like this.


And this is what the control software looks like.


But I want to buy the eight-channel EBC-X load, which already has the corresponding board built in.


The 40 A, 75 mV current-sense shunt looks seriously beefy. It’s marked as accuracy class 0.5, which doesn’t quite match the stated 0.2% in the description.


There’s no point describing the control board; it’s identical to the control boards in all the other loads.


The contacts in the connectors are also wired separately, which helps balance the current between them a little. But the connectors aren’t tightened enough; I had to tighten one, which is extremely awkward with the unit assembled.


Now for a few tests. First, I checked how accurately it sets the load current. The source is a battery with a maximum continuous discharge current of 55 A, so even my new load can’t quite keep up with it...


I tested nine points: 0.1, 0.2, 0.5, 1, 2, 5, 7.5, 10, and 12.5 A. I would have liked to test higher currents, but measuring them with a clamp meter wouldn’t make sense, since its measurement error would inherently be greater than a direct measurement across the shunt. My multimeter tops out at just 10 A, and even 12.5 A is already an overload.

The specified accuracy is up to 0.2% ±0.01 A. So even 0.2% of 40 A is ±80 mA; adding the extra 0.01 A error gives an allowable deviation of ±90 mA.

In practice, everything was within the specified limits. The largest difference was at 12.5 A, about 70 mA, but I think the result could be improved with some additional calibration.


In charging mode, the accuracy was slightly better. Of course, you could say it’s the same shunt—but no: it is the same shunt, but the polarity is reversed, so it could easily have had near-zero error during discharge and twice the error during charging.


To check voltage measurement accuracy, I used a bench power supply as the source and compared the readings with my main multimeter.
The specified accuracy is 0.2% ±0.003 V. At 5 V, that allows a deviation of ±0.01 + 0.003 = 0.013 V, or 13 mV.
In practice, I got deviations ranging from 2 mV at low voltages to 7 mV at the maximum. I also found that the maximum voltage is 4.9 V, not 5 V. If you exceed it even slightly, a warning appears and the load is disabled until the voltage is reduced.


But that wasn’t enough for me, so I went into the calibration menu and calibrated it in just a few clicks. I wouldn’t recommend doing this unless you have an instrument with the appropriate accuracy class.
The calibration process is simple: set a voltage at the input, measure it with an accurate instrument, then enter the actual measured voltage in the field. Calibration uses two points, for example 1 and 4.5 V.


After that, the voltage measurement and setting accuracy was comparable to my multimeter.


I wasn’t so interested in the temperature behavior itself as in the power balance between channels (if that’s the right way to put it), i.e. how the current is distributed between the load and charging transistors.
Testing was done with the case open, which reduced cooling, so I consider the test reasonably representative, including in terms of temperature.

The test subject was a battery that can be both discharged and charged at 30 A.


I started discharging at a load current of 30 A. Power was 86 W at the beginning, and the test took about 11 minutes. I accidentally stopped it a little early; the battery can actually deliver more.


The very first thermal images showed that most of the load was on the third transistor, while the first was the coolest. It may have been getting some cooling from the fan, which was running even with the case open.
After 11 minutes of testing, the third transistor was still the hottest, reaching 65 degrees.


The thermal sensor tripped at 50–52 degrees. By the end of the test, the wires were only at 34 degrees, which suggests they really are 5 mm². The battery temperature was 66 degrees, which is pretty good for this discharge rate.


I was worried that the long wires, all the parasitic resistance and the contacts would prevent the load from delivering 40 A at 2 V, but my concerns were unfounded. Even with the wires, holder, and so on, it works without any problems at 40 A and 2 V across the cell.


The second test was charging. The battery had been fully discharged beforehand, and I set the charging current to 30 A.
1. At the beginning, the power delivered to the battery was 105 W.
2. It delivered about 109–110 W at most, then charging switched to the CV phase.
3. Ten minutes after charging began, the current had dropped to 25 A.
4. Another 7 minutes later, charging was complete, so the whole process took about 17 minutes.


The second transistor got the hottest here, reaching almost 60 degrees. Then the charging current started to drop as charging entered the CV phase, and the temperature fell, so there was little point in continuing the test.


When I looked up the specifications for this load, I found conflicting information online: some sources said the maximum charging current was 30 A, others said 35 A, while the manufacturer only specifies the discharge current.
My test showed that it can actually charge at up to 40 A. The test was brief because the battery isn't rated for that, but it was enough to confirm the figure.
One caveat, though: the charge termination voltage depends on the charging current and wire length, so I think that at this current you should only charge to 4–4.2 V; beyond that, the charging current will drop.


The load uses the same software as all the others, so there's no point describing it. Essentially, it's the same EBC-A20, but with a higher current and a narrower input voltage range. Accordingly, it also lacks an automatic test mode; you can only run tests manually or use a program.


As always, the pros and cons are easy to separate.
I'll start with the good: it works great and meets its specifications, and it can also be calibrated. The well-made case is another plus, even if the design is a little old-fashioned. Unlike the other models, it can be linked into a group if you buy the appropriate adapter.

There are some drawbacks, too. There's no reverse-polarity protection, so it's best to add a fuse in the power circuit. There's also no fuse in the power supply circuit for the step-up converter. Current isn't balanced between the channels, but the manufacturer has used passive balancing, which helps somewhat.
The design seems pretty good, but the case could easily have been made one and a half to two times smaller. It's also very long—350 mm without the connectors—so it's not always convenient to use.
The front-panel connectors aren't secured well and are difficult to tighten. Also, after a couple of days, the power button on mine started sticking. It eventually went back to normal, but, as they say, the bad taste lingered.

As for buying one, I found this load on AliExpress for $134 plus $20 shipping, which is a pretty good deal. The same seller also has the EBC-A10H and EBC-A20 models mentioned above. The prices seem almost suspiciously low for AliExpress.


That's all for now. I hope you found this interesting. As for me, I'm already saving up for an EBC-X :)