Universal battery tester based on the Atorch DL24EW controller

Not long ago, I posted a couple of reviews: one of a voltage converter and one of a power supply. At the time, I mentioned that I’d bought them specifically for a new version of one of my products—more precisely, a battery tester. Today I’ll tell you a little about it.


As a reminder, to keep up with new posts on the blog, I recommend following me on Instagram. I’ll post notifications about all new articles there, and might also write about something interesting from time to time: link to my account

You can also follow updates and new articles on the Telegram channel - https://t.me/KirichBlog


Let me start with some background. The devices I make can be roughly divided into several groups:
“Stock” models are electronic loads based on the DL24MP. I make these all the time, so I usually have some ready-made units at home.
“Made to order” models are also things I build, but because demand is lower and for various “reasons,” I don’t particularly enjoy making them. This group includes loads based on the DL24EW. The controller is pretty badly designed, and it needs modifications to make it not just work, but work safely. I only build these for people who understand exactly what they are and why they need one.
Loads based on the TEC-80k also fall into the made-to-order group, since they give me much more room to experiment—for example, discharge currents of 120–160 A and power of 2400 W or more. In fact, just today I ordered three enclosures for 2400 W loads. I’ll probably build them right away and move them into the “stock” category.
“Custom” models are less common devices, or devices built to the customer’s specifications. Usually that means a different connector configuration, the option to connect an external battery for backup power, and so on. This group also includes fully specialized devices unrelated to either Atorch or TEC80—for example, units for small-scale production, designed to perform one specific function.

You can find some information about the current models in one of my OLX listings; the “Shop” link also takes you there. I should also mention that my prices are currently the lowest, both among Ukrainian sellers and Chinese ones. For example, the 600 W, 40 A Atorch version costs around $180–190, compared with about $140–145 for mine with the same specs (and a few improvements). But the Chinese one has no fewer than six fans, while mine is cooled by a single fan with automatic speed control. The design was simply thought through and selected better from the start, whereas the Chinese makers just took stock modules, put them in a box, and blew the heat out of it. Not a great solution, to put it mildly....


What’s more, for some reason the Chinese version even has an external power supply, while mine is built in.


It’s hard to put this device into any one of those groups, because I originally built it simply because I wanted to see if I could. Last spring, I put one together. Essentially, it’s half of a 1200 W load, with a power supply and voltage converter in place of the second power module.


The internals are relatively simple: the load, power supply, converter, battery switching board, and reverse-polarity protection. But since this was the first time I’d built the device, a whole bunch of issues came up and took a lot of time to sort out. You can also see in the photo that the power-stage boards are homemade; these days I use factory-made ones.


In the end, one of the blog’s readers bought this load almost immediately. After receiving it, he asked an interesting question: could the power supply be used separately from the load? Hmm, I thought. That’s actually a pretty good idea, and entirely doable. But I wanted to take it a little further, and that’s how the idea for a new device came about.

I don’t usually take many photos while I’m assembling things, but I did take a few this time.
1. Test fitting
2. Assembly
3, 4. Almost finished.

You can see that the layout is quite compact, but reasonably well optimized. And, by the way, I’m already planning to expand its capabilities. I have a 25 A converter at home, and now I’m thinking about what power supply to pair with it: a Chinese 300 W unit, or a Mean Well rated for 500 or 600 W. Mean Well is certainly more appealing, but the price stings—something like $80–120 just for the power supply. I think I’ll only offer that as a made-to-order option.

Not long ago, someone replied to one of my YouTube comments: “So you just take Chinese Atorch boards and put them in an enclosure?” Well, yes—Chinese boards, in an enclosure. The key word is “just” :)



Here it is assembled. As you can see, it turned out very similar to the first version, except for a couple of terminals.


1. The front panel has the power supply output terminals, an input for an external temperature sensor, a charging-mode indicator, the load’s power and measurement terminals, an encoder, a start/stop button, and a power switch. And, of course, the converter module and the load’s display.
2. On the back are ventilation slots, a power connector, USB-C for connecting to a PC (with galvanic isolation), and a WiFi-USB mode switch.


Of course, its main function is as an electronic load. Since I added control for the intake fan, I was able to raise the power a little, to about 650 W. I should point out that the module's maximum power is 800–1000 W; the thermal protection will simply kick in after a few minutes. The stated 500–600 W is the continuous rating.

So, the load can handle up to 40 A and 650 W.


It can now do more than just use a power supply: it can also test itself. This required modifying the switching board, which added an extra 2 mΩ of parasitic resistance, but it means the unit can be used to measure converter efficiency and for similar tasks.
I had to modify the switch because the power supply and the load have different grounds, and they can't be connected together.

And of course, you can simply use it as an adjustable power supply, with an output voltage of up to 52–54 V, current up to 20 A, and power up to 290–300 W.


The third function is, of course, battery testing. To get ahead of myself, parasitic voltage drop in the wires and switch makes the end of charging take a little longer, especially at low voltages. So the tester is better suited to batteries rated at 10 V or higher, where the effect is less noticeable. That said, there's nothing stopping you from using it with individual cells, which is exactly what I'll show now.


This mode may seem complicated at first, but after using it once or twice, the principle becomes clear and you can do everything on autopilot.
First, set the charge termination voltage and charging current on the charger, then activate the output—that is, apply voltage to the output.

In my example, that's 4.2 V and 4 A. The converter does lie a little, though, so at low voltages it's worth taking that into account. The current measurement error doesn't matter; the load does all the counting anyway.



1. Next, choose the test mode: CDC (charge-discharge-charge) or CDCDC (charge-discharge-charge-discharge-charge). I chose the shorter mode.
Set the discharge current. In my example, it's 15 A, since this is a high-current battery.
And don't forget the minimum voltage, i.e. the discharge termination voltage. In my case, it's 2.5 V; for a 6S pack, it would be around 15–17 V, depending on the cell type.

Press and hold the Start button to open the settings menu, then set the charge cutoff current. This is usually about 1/10 of the charging current (item 10).
If you're worried the charger might accidentally output more than it should, set item 9 to a voltage slightly above the required value—about 50–100 mV per cell. Then, if the voltage goes too high, charging will stop based on this setting.

You can also change the discharge cutoff voltage (item 8) here, but it's more convenient to do that from the main screen.
Adjusting any parameter through the menu is quite the ordeal. Anyone who's used it will know what I mean. The Chinese programmers managed to turn the trivial task of changing, say, 500 to 1000 into a fairly complicated process. Interestingly, adjusting the current from the main screen works almost correctly, apart from a little "slippage," so there's no point in turning the encoder quickly.


Press and hold Start to exit the menu. Then connect the battery and press Start again, this time briefly. By the way, the controller behaves a little strangely when you try to force it to stop: you can press the button and get no response. It's not a problem with the button; sometimes the controller only registers the press on the second try.

Anyway, press Start and you'll get one of two prompts:
1. If you've already run other tests, it will offer to erase the old data and start a new test.
2. If you just ran a test but stopped it, it will offer to resume or start over.
The first case is straightforward: start a new test. In the second, it depends on the situation.



That's it—the test is underway. When it finishes, you'll get the results for each stage. Ignore the stop trigger; for some reason, the Chinese programmers decided to display an overvoltage, even though there definitely wasn't one.
In any case, at the end you'll see what was charged or discharged during the test. You can also display the results as a graph; to do that, open the menu and select Curve mode.

As you can see, the full test took 50 + 13 + 77 minutes, for a total of 140 minutes—almost 2.5 hours.




So that's the finished product. I'd really like to hear your comments and opinions on whether it's worth developing this product line further. By the way, one possible use is testing standard 6S batteries. The device can test one automatically, charging at up to 10 A and discharging at up to 25 A, which makes it suitable for all configurations, from 6S1P to 6S3P.

A question while we're at it: this version costs around 12,500. The next version, with a 25 A converter module, will be a little more expensive if I use a 300 W Chinese power supply (probably a 48 V one, since 60 V models are harder to find, although I do have one). It will be noticeably more expensive—by about $60–100—if I use a Mean Well EPP-500-54 or LOP-600-54, but then the power supply can deliver 500–600 W instead of 300 W.

Anyway, as always, I’d be glad to hear your comments, even if they’re off topic—and especially if they’re on topic :)