EBC-A20, another electronic load from ZKEtech

The second electronic load from ZKEtech, bought quite recently, and the third device in my collection of measuring instruments. Though in this case, calling it a measuring instrument is a bit of a stretch, and it doesn't quite fit the definition of an "electronic load" either, since it's also partly a battery tester. Anyway, today I'm reviewing the 20 A version.


The previous version, like the very first one (EBC-A10), had a maximum load current of 10 A. This one, however, can handle up to 20 A, making it better suited to testing high-current batteries. Although the load current is twice that of the previous model, its power rating is almost half as high: just 85 W, compared with 150 W for the EBC-A10H. Since this version has a lower power rating and doesn't include a power supply, it costs noticeably less: $39 for the “economy” version and almost $47 for the full kit.

The full kit includes an interface cable and a power supply (with a mains cable). I already have a cable from the very first load, and I definitely have enough power supplies, so I wasn't about to pay for things I don't need—especially when I'd also have to pay to have them shipped to me.
After factoring in the forwarding service, it cost me about $55. More precisely, $128 covered two loads, a USB adapter, a pile of packaging, shipping, and the intermediary's fee.
On AliExpress, this load costs $84 for the full kit (load, power cables, interface cable, and power supply) - link.


Brief specs, translated from Chinese by me
Power supply: 19–24 V, 4 A
Input voltage range: 0.00–30.00 V, measurement resolution 0.01 V
Output voltage range: up to 18 V with a 22 V input
Load current: 0.1–20.00 A, setting resolution 0.01 A
Charging current: 0.1–5.00 A, setting resolution 0.01 A
Load modes: CC, CP
Charging mode: CV
Voltage measurement error:
Up to 4.5 V: measurement resolution 0.003 V, error ±0.5%
From 4.5 to 30 V: measurement resolution 0.01 V, error ±0.5%
Current measurement: 0.1–20.00 A, resolution 0.01 A, error ±0.5%
Capacity measurement
Up to 10 Ah: resolution 0.001 Ah (1 mAh)
10–100 Ah: resolution 0.01 Ah (10 mAh)
100 Ah and above: 0.1 Ah


So, the “economy” kit includes:
1. The EBC-A20 load itself
2. Power cables
3. Manual.



1. The manual. There's absolutely nothing interesting here: it's all in “understandable” Chinese, and they didn't even include any pictures.
2–3. The cables are pretty good, though: sturdy copper alligator clips, and the connectors look solid too. But I'm curious why they didn't provide four clips right away, since at these currents you'd want to use four rather than two. Oh well, that's easy enough to fix.


When I picked up this little box, my first thought was, “Why is it so small?” I'd seen it in photos many times before, and it had always looked bigger to me.
I'm glad the manufacturer replaced the bright orange color, like the paint used on “black boxes,” with a more familiar gray.


On top are the display and four buttons: power, parameter selection, and up/down adjustment. Once again, it's done “the Chinese way”: the up button is on the left and the down button on the right. But here you'd have to rewire them yourself. As I mentioned, though, I don't use the keypad.


At the front are a fan and a pair of connectors for the test leads. The finish is neat, apart from the gap between the cover and the case, and the odd cut along the top of the case.


There are ventilation grilles on the sides—logical and straightforward.


There's another ventilation grille on the back, along with the power input, a switch, and a microUSB port for connecting to a computer. A reminder: this isn't USB, it's UART TTL. So connecting a cable directly here is safe, but just as pointless—you need a converter.


As before, there's a warranty seal (though there's no warranty card, by the way) and a microUSB port nearby.
There's nothing on the bottom except the feet—not even screws holding the internals in place. The paint job is neat; you can tell they made an effort.


The internals—what would a teardown of such a specialized device be without them? :)
It's fairly easy to take apart, but you can't separate the two halves completely right away; they're connected by wires.


After a bit of fiddling, I separate the two halves.


It's odd that they didn't connect the buttons the same way as the display, using a long pin header.


It looks quite neat from the outside.



As before, the switch is connected via terminals rather than soldered in place.


The heatsink is unexpectedly small; its total surface area is even a little smaller than the one in my A10. The fan blows air out, which somewhat reduces the cooling efficiency of the heatsink itself, but helps keep the charger—and especially the current-sense shunt—from overheating. In my old unit, the shunt was cooled by hot air from the heatsink, which could in principle make the current more dependent on the power dissipated by the load.


I don't think the controller board needs an introduction. It's probably the basis for most of this company's electronic loads; the loads themselves differ in the design of the power stage and in their firmware.


Nothing interesting on the bottom.


Although most of the wires are connected with plugs, the power wires are soldered in place. The contacts on the power connectors are duplicated: two of the three are for the power connection, and one is for measurement.
The inside of the case is a little dusty and smudged, but the flash makes it hard to see in the photo.


The charger—nothing new here either.


Three diodes are visible nearby. The pair on the left protects against reverse polarity. Judging by the markings, they're either SR104 or SR1040; either way, each is rated for 10 A and 40 V. I'd recommend replacing them with higher-rated ones, since they'll get noticeably hot in use. Besides, that extra voltage drop is the last thing you want at these currents.


The shunts are paired. I haven't measured them; the A10 had a resistance of 10 mΩ, and this one is most likely the same.
To the right, you can see the relay that protects against overload when the unit is switched on with a low-voltage battery connected.


As I mentioned earlier, the old version could overload the power supply and send it into short-circuit protection. I ran into this glitch recently too, while running a battery endurance test. Good thing I was home.
So I quickly put together a simple circuit that adds a short delay, giving the power supply time to start up properly before applying voltage to the electronic load. That way, with the power supply running normally and its capacitors charged, the charger starts up correctly.


The overall build quality is quite good. I'd even say the board feels nice to hold; everything is clean and neat.
There's a 220 µF, 50 V capacitor on the input and a 220 µF, 25 V capacitor on the charger's output.


The power stage surprised me, though. I honestly expected to find a pair of transistors here, given the high current, but it turns out the power section is almost identical to the one in my 60 W A10 load.
The manufacturer clearly cut corners somewhere, because 85 W from a single TO-247 package is a hell of a lot really pushing it. What's more, the IRFP260M transistor has an on-resistance of 40 mΩ, which means that at 20 A the load physically can't maintain an operating voltage above 0.8 V. Add the voltage drops across the diodes (about 0.5 V), the shunt (0.1 V), and the wires and connectors, and it all adds up to quite a bit.

As before, the thermal switch forces the fan on if the normal (software) control fails. In this mode, the fan also runs noticeably faster.


Now it's time to put it to the test. The first thing I ran into: this one uses a 5.5×2.5 power jack, not 5.5×2.1 like the old one, so I also had to find a power supply at home with the right plug.


The software and features are more or less the same as on my old A10 (if anything, there are more of them), which is a little strange, since this one is marketed specifically for working with batteries. But it has no presets or test mode for nickel-based batteries.


Excited, I started not with measurements but with a simple battery discharge test. That’s when I realized I’d definitely have to modify the alligator clips. Otherwise, what was the point of buying the proper battery holder in the first place?
No, the clips themselves are good quality: reasonably stiff and they grip well. But at these currents, everything contributes to the measurement error, and I don’t want to ignore that.


I set the discharge current to 20 A, limited the time to two minutes, and started the test. Nothing burned out, and the power dissipation was about 60 W. After playing around with it, I started charging. No issues there either: it charges and shuts off.
The fan is noticeably noisy, though. The A10H is noisy too; only the old A10 runs more or less quietly.
The fan turns on when the power exceeds a certain level, regardless of temperature. If the unit heats up above 50 degrees, a thermal switch turns it on at full speed.


And now for the checks.
Voltage. At first, everything went well, but near the top of the range the difference reached 0.1 V. Although that’s still within the stated 0.5% error (±0.15 V), it seems like a lot to me. It can be calibrated, but I don’t plan to use it at voltages above 15–20 V.


There’s also an issue with the ADC’s low resolution, which shows up as large steps in the measurement results. For example, with certain voltage combinations, a 0.01 V change at the input is displayed as a 0.03 V change.


Before checking the current measurement error, I checked how much current the load draws from the source under test. At 5 V, it’s about 37 µA.


Load current. Unfortunately, I can only check it properly up to 12.5 A. The multimeter is rated for 10 A, but it works just fine at 12–13 A.
Across the entire range I tested, the load reads low, which means that when measuring battery capacity, the result comes out higher than the actual value. The difference isn’t very large, and calibration might help.
Update: I tried calibrating it. It can be calibrated, but the calibration steps are quite large, so getting it exactly right isn’t easy, though it is possible. The bigger problem is that calibrating the discharge current also affects the charging current, even though I tried calibrating both. In other words, if the discharge current was low and you increase it, the charging current will also increase. And if that was already high, it may end up even higher.
In general, it’s usually enough to calibrate the discharge current. But, for example, on the A10H load, the discharge and charging currents were both offset in the same direction, so calibration there would be noticeably easier.


As it warms up, the current drops slightly, so if you’re going to calibrate it, it’s best to either let it warm up first or leave a little headroom on the high side.


Charging current.
At low currents, the results are a little worse than for discharge, but at higher currents they’re better—sometimes even spot-on. During a test at 2 A, it initially showed a slightly higher value, then stabilized fairly quickly at exactly 2 A.


But here’s what’s interesting: above, I showed that at 2 A the actual current dropped slightly. At 5 A, the situation was exactly the opposite: the current was initially a little low, but after a while it rose to the set value. I suspect some kind of digital correction may be at work, but I’m not sure.


To connect it to a computer, I used the cable from my old load. I connected the old one using a regular USB-UART converter.


Interestingly, my old load has a separate connector marked BTSD. It was apparently intended for a wireless interface or some kind of adapter for recording to a memory card, but they later decided to remove it. I also happened to find out that the connector has ground, RXTX, and... 12 V, not 5 V as you might expect. In my setup, connecting a regular cable could damage the computer port rather than the load, although I have connected it that way before.


I won't describe the software; it's exactly the same as in the A10H review and almost identical to the one I showed with the A10.


But I was surprised that there's no automatic test. What really threw me was that the tab is where the automatic test used to be—which would make sense. I even wondered, “Where's the window for entering a test program?” But no, the programmable operation is still there.
Honestly, I just don't understand why they removed the automatic test.


The only operating modes listed are CC, CP, and CV (charging); there are no presets.


The battery internal resistance measurement function also behaves very strangely—or doesn't work at all. I can't say I need it, but still. You can select any current, but when you try to run a test, the software loses its connection to the device. The only way to reconnect is with the connectdisconnect button.


For the next tests, I unsoldered the test leads from the alligator clips and soldered them to some smaller ones I found at home. I have to praise the high-quality, very flexible wire with silicone insulation. I saw that the seller offered it separately and regret not buying some.


Now let's connect the holder properly, using the full four-wire setup, and measure the battery capacity at the maximum current.


I set the current to 20 A, the cutoff to 2.5 V, and started the test.


I waited for the battery to discharge completely, changed the lower cutoff to 2 V, and continued the test. The graph marker shows 19.89 A. That happened by chance: I caught the moment when the load started reducing the current according to the program. You can see this in the final capacity, both at the marker and in the bottom line. Moving the marker just one pixel to the left would have shown exactly 20 A; I only noticed this while preparing the screenshots for the review.


After the discharge finished, the battery had warmed up to 73 degrees. You may have noticed that it's next to the fan grille; that was intentional, as the airflow helped cool the battery. Although it's rated for currents up to 30 A, it's already about 3 years old.
On the right is the overall view; you can see that the wires are heating up too.


I started charging, and there were no problems here either.


Using the built-in data logging function, you can create detailed graphs with the supplied software. For example, here the green line shows a discharge at 10 A, and the blue line at 20 A. The jump at the end of the graph is because I continued the test instead of running it in one go. The matching capacity readings in mAh are just a coincidence, nothing more.
Just in case you use the data logging function and it doesn't work correctly, change the Windows decimal separator in the settings from a comma (the default) to a period.


Now, about ZKEtech itself and its products.
First, here's what I have right now. In the middle is the A10, which I use constantly and for long stretches. It has its shortcomings, but it's very reliable. On the sides are the last two, whose reviews you’ve already read.
I once considered buying their 35 W USB load, mainly because it can control chargers with QC. But after I got some fancy USB “doctors,” that idea moved to the back burner.


Next, here’s what else this company makes, along with an explanation of its device model numbers.
EBD — a basic electronic load; this also includes the EBD-USB.
EBC — an electronic load with a built-in charger.
EBC-X — a multichannel load with a built-in charger.

Next comes the letter suffix:
A — low-voltage series, from zero to 5 or 30 V.
B — high-voltage series, 9–88 V, designed for use with lead-acid batteries, such as car batteries.
I’ve also seen claims that loads with the A suffix can measure internal resistance, but today’s review suggests that’s not entirely true.

Next is the load current rating:
05 — 5 A
10 — 10 A
20 — 20 A
40 — 40 A

The multichannel models are more complicated:
0501 — 1 A charging, 1 A discharging
0510 — 5 A charging and 10 A discharging
0510 — 10 A charging and 10 A discharging. That’s not a typo: they really do use the same model number, but there’s also an X0510A version with a 10/10 configuration.

There’s another suffix at the end:
H — higher power, for example, the A10 and A10H
L — I don’t know.

At the moment, I’m considering two options to expand my toolkit:
1. The 40 A version for “heavy-duty” testing.
2. The multichannel version with a 5 A charging and 10 A discharging configuration.
3. I’m unlikely to need the 10/10 version (the photo shows the model with a built-in holder), and the 1/1 version even less so.
4. For greater convenience, it’s probably better to use separate holders, which are sold in packs of 8.

By the way, the multichannel versions are more accurate. I suspect they use separate ADCs.


I’m currently leaning toward the multichannel model, which is convenient for capacity testing and comes with software tailored for that kind of testing. This load costs around $270 on Taobao (the 5/10 version), plus you’ll need holders. A “budget” setup adds another $45; for something a bit neater, expect to pay closer to $55–60.


While using it, I noticed a few things I hadn’t spotted before.
First, let’s look at when the fan turns on.
During discharge, it’s controlled by the software when power exceeds 6 W, and “in hardware” by a thermal switch when the temperature exceeds 50 °C.
During charging, it’s controlled only by the software, when power exceeds 5 W.

I also noticed something else: when the fan switches off during charging, there’s a little plateau in the current readings. I stretched the image a bit to make it easier to see.


Stretch it further and you can see that the plateau starts right at the fan shutoff threshold. I think there’s a shared circuit somewhere between current measurement and fan control. Why do I think it’s related specifically to the fan control? Because the plateau doesn’t appear when the thermal switch trips.



That’s about it. I’ll try to sum up what I liked and didn’t like about these two loads (A10H and A20).
The good: the manufacturer is clearly making an effort. Everything is assembled fairly neatly—not perfectly, but well enough. They’re also making improvements to the design, circuitry, and software. Technically, everything works and even meets the stated specifications, which is a plus.
The bad: using a cheap microcontroller and no separate ADCs means the accuracy is lower than it could be. In my opinion, the power stage is being pushed to its limits. In the 20 A version, I’d use two transistors instead of one, make the heatsink a little bigger, and get 120 W for sure, if not 150 W. The cost of that upgrade would be low.
The A10H does have two transistors, but there’s no circuit to balance the load between them, which isn’t very elegant—or, more accurately, it’s downright ugly. Fortunately, you can fix it yourself with a small board I wrote about some time ago.

Overall, I liked it. I’ll try writing to the manufacturer with my observations and suggestions for improvements; maybe it’ll help. Another option would be to sell kits for DIYers, with tested boards but no enclosure or heatsink. That’s all for today. I hope you found this useful.