Atorch DL24, an inexpensive electronic load with four-wire connection

Some time ago, I announced a new product from Atorch, the DL24 electronic load. But it’s hard to call this device new; more accurately, it’s a budget version, with no heatsink included and a maximum power limit of 1000 W.


Around the same time, someone told me an even newer version had come out, with four-wire connections. I wasn’t planning to order the first version, but I couldn’t resist the second one, so I ordered three sets (just like in the joke) :)

I placed the order and received it fairly quickly. There was no outer packaging; they simply taped the three boxes together, and that was that.
Everything inside was packed very neatly. I tried to put it all back the same way later, but I wouldn’t have managed without the photo.


The package contents are pretty generous, although I think they could have left out a few items. That would have lowered the price and made it more popular.
1. Electronic load board
2. Power supply
3. Two sets of power cables
4. Adapter board
5. USB cable
6. External temperature sensor
7. Thermal paste
8. Plastic standoffs
9. Power supply adapter
10. Manuals



There’s a slightly more expensive version that includes a battery holder, but I didn’t need one since I already have universal four-wire holders.


Specifications
Test voltage: 2–200 V
Operating current: 0.2–20 A
DL24 discharge power: up to 999 W

There’s a warning about the maximum power:
1. You need to add an IRFP260N (150 W) discharge tube
2. You need to add a high-power heatsink (supports 1000 W)
3. You need to be good at modifications!

1, 2, 3. Three manuals are included: a general one, one explaining the connection methods, and a third with recommendations for installing a fan.
4, 5. The power supply is very budget-oriented, and I don’t think it needs to be included. For all my tests, I used my own 12 V supply.
6. The thermal paste and standoffs are nothing unusual.


Since this load has four-wire connections, it comes with two sets of power cables: one pair with a larger cross-section, around 2.5–3 mm² by the looks of it, and a thinner pair. One drawback, though: the cables are very stiff.
There’s also an external 10 kΩ thermistor and a USB-to-microUSB cable for connecting to a computer.


An adapter board is also included for testing power supplies with a 5.5 mm plug or miniUSB, microUSB, and USB-C connectors. There’s also an LED on the board, though I still haven’t figured out why it’s there—you can check whether voltage is present by looking at the display.


The boards are identical, and even the power transistors are scuffed to about the same degree.


In fact, more than half the board is taken up by the space for the heatsink; all the electronics are concentrated at the bottom.
The displays are held in place by tight rubber bands. It later turned out that these bands are almost pointless, since removing the display is quite difficult: it’s stuck down with double-sided tape underneath.


I checked that each board powered on. Two behaved normally, but the third went straight into the menu for some reason. The explanation is simple: the rubber band was pressing one of the buttons. Watch out for that :)


Since we’re talking about the display, I should mention that it isn’t very bright. It’s adequate on its own, until you compare it with another display. I may have to figure out how to increase the backlight brightness, if that’s even possible.


Overall view of the circuit board. All the components are mounted on top, and a bunch of holes have been made in the area for the heatsink.


The underside is empty. There’s a hole under the transistor that lets you bolt it to the heatsink—or the heatsink to the transistor—which is definitely useful at high power levels.


To the left of the display is a terminal block for connecting the equipment under test. Below it is a connector for the external temperature sensor (for some reason labeled “jakc” instead of “jack”). On the right are four control buttons and a microUSB port for connecting to a computer.
What I disliked most in this case is that the display is glued in place, which isn’t always convenient.


1. A quality terminal block, of the “elevator” type rather than a simple clamp
2. The CH340G chip and a Bluetooth module are used to connect to a PC. There is no galvanic isolation on the USB connection. To the right is an interesting connector: it carries the positive input terminal and the negative terminal after the shunt. In other words, you can connect a load to it and use the board as an ammeter. Nearby are four LEDs: two indicate UART communication, and the other two indicate Bluetooth communication.
3. At the top are the power jack and fan connector. In the middle is a footprint for a different type of power connector, simply wired in parallel with the one already installed. The specified supply voltage is 6–12 V, but I think they just didn't update the labeling: there's no step-up converter on the board, so I'd recommend a range of 8–12 V, or 8–15 V at most.
4. The board also has a diode to protect the measurement input against reverse polarity. By the way, if you move the power transistor to a separate heatsink, you'll need to move the diode there too, since it needs airflow from the fan when mounted on the board. One thing that makes this easier is that, if you use the common-cathode diode assemblies, you can mount one on the same heatsink as the transistor without insulating pads.


Current is measured using two 10 mΩ current-sense resistors connected in parallel. At 20 A, that gives a voltage drop of 0.1 V, or 2 W of power dissipation. I'm concerned about the high power dissipated by these resistors. If the power components are moved to an external heatsink, we'll need to figure out how to cool the resistors, or possibly replace them with higher-power ones. But splitting the power stage across several transistors could help here; more on that later.


Some chips have had their markings removed, but I can say that the one on the left appears to be an ADC, while the one on the right is a microcontroller.
At the top are two LM321 op-amps. One amplifies the signal from the current-sense shunt, while the other acts as a filter for the PWM control signal.
We can already say that all modes except CC are software-controlled. I was hoping that at least CV mode would be hardware-controlled too. A software implementation requires a high ADC sampling rate and isn't very accurate under dynamic conditions.

Two traces run to the shunt. At first I thought it was a four-wire connection, but it turns out one trace goes to the feedback op-amp and the other to the ADC for current measurement.


The transistor is marked IRFP260. It looks old and even dusty, and its package is slightly different from the ones sold today. It also lacks the usual marking area, which is typically glossy and laser-marked. It looks used. Some product comments said it was counterfeit, but I won't claim that's the case. It seems like an awful lot of trouble to source used transistors or counterfeit them just to put them in finished products.



All right, let's move on to the built-in software and testing.
When powered on, the manufacturer's logo appears briefly, then the device switches to the main screen, which displays:
1. Operating mode and setpoint
2. Input voltage
3. Load current
4. Load power
5. Equivalent load resistance
6. Energy consumed in Wh
7. Capacity used in mAh
8. Operating time since the counter was reset, and the external temperature sensor reading
9. Operating time limit timer
10. Minimum-voltage cutoff threshold: 0.0 when there is no cutoff, then from 0.7 V upward. You can't set a value between 0.0 and 0.7.
11. Internal temperature sensor reading
12. Equivalent internal resistance of the source
13. Load status: on/off.

Hold the bottom button to open the settings menu.
1. Language: Chinese/English
2. Reset the accumulated counters (time, capacity, energy)
3. Zero the voltage and current readings. Do this with no test sources connected.
4. Voltage calibration
5. Current calibration
6. Brightness in normal mode
7. Brightness in standby mode
8. Time before switching to standby mode
9. Internal temperature sensor reading correction
10. External temperature sensor reading correction
11. Maximum power limit, set to 185 W by default, with a maximum of 999 W
12. Restore factory settings
13. Saving settings and exiting.

As for the first part, I think the main screen is overloaded with information. It would probably be better to have two screens, one for the main information and one for the secondary information.
I have no real complaints about the settings menu. There isn't much to do there anyway, since most settings are configured once, and in practice you'll probably only use the option to reset the accumulated results.




More or less the same description is on the product page, but in English


Operating modes.
Four modes are supported. As far as I can tell, only CC is implemented in hardware; the others are software-based.
Overall, things look pretty good here, at least at first glance, but I quickly ran into a glitch: CV mode behaves strangely and doesn't work correctly with all sources.

To change the operating mode, press and hold the top button. Once the mode field starts flashing, use the left/right buttons to select a mode. After a while it will stop flashing; then briefly press the top button to change the adjustment step, and use the left/right buttons to change the value.
I have to say I don't like the controls. For now, I find the controls on the Gandf load the most convenient, but that one has an encoder.



I was a little surprised that CV mode is also implemented in software, since it can be implemented in hardware relatively easily. The same goes for CR mode, though that's a little more complicated to do in hardware. The hardest mode to implement in hardware is CP; there's even a dedicated chip for it, but it comes with its own "quirks".

So, what are "operating modes"?
CC, or Constant Current: current regulation.
CV, or Constant Voltage: voltage regulation. The load prevents the voltage from rising above the set value by increasing the current.
CP (CW), or Constant Power: power regulation. The device varies the current to maintain the set power consumption.
CR, or Constant Resistance: constant resistance, with the load current automatically adjusted according to the input voltage.



When the test ends due to the time limit, a voltage drop, etc., the results screen is displayed.
In CV mode, after I removed the input voltage, the device didn't turn off for some reason and displayed the message “UREG,” even though it should have shut down when the voltage dropped below the set limit.



Since I needed to install a cooling system before running the tests, I decided to see what other people were using. I came across a photo in the reviews showing an Intel CPU cooler, with a note saying that one from an older processor fit perfectly.


And this is where I ran into a problem. I had two types of coolers at home: an old one for Socket 775, given to me by a friend, and a relatively new one for Socket 1155.
But it turned out they have different mounting-hole spacing. I'd never really thought about it before; I just installed the one that came with the CPU and didn't worry about it.
The "old" one has 72 mm spacing, while the "new" one has 75 mm. Naturally, I wanted to use the old one because it's larger, but the board is designed for the "new" ones, with 75 mm spacing. Of course, I could try bending or filing something down, but it would be very awkward.
AMD coolers might fit, but because of the nature of my work, it's even harder to find one at home: almost everything I've ever built has been Intel-based.


All right, I applied thermal paste and installed the cooler, only to be in for a huge disappointment: the heatsink doesn't reach the transistor by about 2 mm. How it fit for the guy from Poland remains a mystery to me. And note that the Polish reviewer used the standard Intel mounting hardware.


I didn't want to break the "new" cooler, especially since it came with a processor, so I used the old one instead. I removed the plastic mounts and simply screwed it in with M3 screws and washers, since the holes in the mounts are oversized.


It worked great, except that the board bent a little. If you install it the way I showed above, it's best to add an extra plate underneath.
For a quick test, I connected it to some cheap USB charger and loaded it to 3 A. It works great. At power levels above 10 W, the fan starts automatically; it also spins up briefly when power is applied to the board.


But I'll start not with the tests, but by demonstrating the problems with CV mode. I connect it to an adjustable power supply, set its output to 5 V with a current limit of 2 A, set the load's voltage limit to 3 V, and start it. As a result, the power supply switches back and forth between CC and CV modes for a while, then the load board displays a warning that the voltage is below the limit. If you set this voltage to zero, it will keep flickering like this.


The process looked something like this in real time


I tried reducing the current from 2 A to 200 mA, but nothing changed. I also tried voltages of 10, 20, and 30 V, and a different power supply, with virtually no change.


Judging by the load's display, it was shorting the input, so the power supply was dropping its output voltage. I also tried it with a regular power supply (photo above), which is completely unsuitable, but the load worked in this mode with it...


Everything worked perfectly with the same power supply in CC, CR, and CP modes.


I tried one of my two loads that can operate in CV mode. Everything was fine, except that its somewhat odd potentiometer behavior means it can't be set below 7–8 V; it drops to zero instead.


Before the tests, a quick note about the operating modes and parameters
The CR mode is shown above. It lets you set resistance from 0.01 to 9999.99 Ω in 0.01 Ω increments.
1. CC mode. I was a little surprised here: it was rated for 20 A, but in practice the load lets you set up to 30 A in 0.01 A increments.
2. CP mode: 0.1–999.9 W range, in 0.1 W increments.
3. CV mode: 0.1–150 V range, in 0.1 V increments.
4. When the voltage drops abruptly, the load may sometimes fail to switch to auto-off mode while in CC mode. I haven't seen this happen when the voltage drops gradually, so testing batteries shouldn't be a problem.

As I mentioned, I was a little surprised that the load can handle currents up to 30 A. That's a good thing, but it also imposes some limitations and creates a few problems. First, the protection diode would dissipate around 15–20 W at 30 A, which is a lot; it would simply overheat in this mode. Then there's the current-sense shunt: at 20 A, the voltage drop across it is 0.1 V and its power dissipation is 2 W. At 30 A, that's 0.15 V and 4.5 W, which is too much for resistors like these. It would also need to be replaced with a higher-power one, since otherwise its reliability would suffer and its parameters would start to drift.
And third, the PCB traces. I don't think they'd take kindly to currents like that, so they'd need to be reinforced with wire.


I checked voltage measurement accuracy over the 0.1–50 V range. The photos are mainly for reference, but I can say that the accuracy looks very good to me.


Current setting and measurement accuracy. The electronic load has a DAC that sets the current and an ADC that measures it, so accuracy in one doesn't necessarily mean accuracy in the other. By the way, my Gandf load can apply a correction: after startup, it adjusts the current if there's a large difference between the set and measured values.

The only thing you could really fault is the accuracy at low currents, somewhere below 100–200 mA, which is still very good considering the range goes up to 30 A.
The current was set to 10, 20, 50, 100, 200, and 500 mA, then 1, 2, and 5 A.


There is one small drawback: at low currents, the current fluctuates, and the lower the current, the larger the fluctuations as a percentage of the set value. They're most noticeable in the 10–100 mA range.


Above 100–200 mA, they almost disappear. The fluctuations are within a fairly narrow range, though, so I wouldn't worry about them.


The highest current I tested was 12 A. The difference between the set and actual current was 0.25–0.3%, which is excellent for a device in this price range. Using a separate ADC is a really good idea. The load can measure current from 0.01 to 30 A with 1 mA resolution (not to be confused with measurement accuracy). Tell me, what stopped power supply manufacturers—RuiDeng, for example—from using an ADC like this in their products?

The power stage's own consumption when switched off is close to zero. The input resistance of the voltage measurement circuit is about 150 kΩ.


The load can measure a source's equivalent series resistance. This function is called BatRes here, but in my opinion it has little to do with battery testing (though not nothing). Here's why I think so.
I ran a quick test:
1. I connected the probes so they wouldn't include the voltage drop across the wires to the power supply. I set the power supply output to 5 V.
2. I applied a 10 A load. The voltage at the load input dropped, and it showed the combined resistance of the wires and the power supply output as 12 mΩ.
3. I lowered the power supply voltage until the load input was exactly 1 V lower than before, i.e. 4 V. The load then showed a resistance of 100 mΩ. The calculation is simple: the load “sees” that at a current of 10 A, the voltage has dropped by 1 V. You’d get the same result if the wires had a resistance of 100 mΩ, so everything is correct.

But keep in mind, a battery isn’t a power supply. As it discharges, its voltage drops, and the measured resistance will therefore increase, since it’s measured continuously. Also, if you raise the power supply’s output voltage, the display will show dashes, because that doesn’t make sense for this type of measurement.

Overall, it’s useful and might come in handy, but this is how you need to test batteries: apply the load and take a reading immediately. Even then, the accuracy will be very rough.


I ran one more test mostly out of curiosity. I set the power supply output to 10 V and applied a load of 10 A, so the load was dissipating about 100 W. I ran it like that for about 25 minutes. Over the last 10 minutes, the temperature reading on the load barely changed, settling at 50 degrees.


The heat was concentrated around the protection diode and shunt. After warming up, the diode assembly reached about 80 degrees, and its temperature barely changed after 10 minutes of operation. The device had reached thermal equilibrium.


Since everything was working normally, I increased the load current to 13 A, for a dissipation of about 130 W, and ran it for a while longer.


The protection diode’s temperature rose to almost 100 degrees. I think its realistic maximum in this mode is around 15–16 A.
I slipped the thermal camera under the heatsink. The highest temperature I could see on the transistor case was 71–72 degrees, though the accuracy here isn’t great.

To be honest, I was a little surprised. I expected it to get hotter at this power, although the transistor was actually dissipating about 120 W.
This is a critical operating mode, and it’s worth remembering that 10 V at 13 A is one thing, while 130 V at 1 A is quite another. In the second case, the transistor is operating under much more critical conditions.


Now for the main reason I ordered these boards. In the announcement, I wrote that the board can handle up to 1000 W, but that’s obviously unrealistic with just the included transistor. I also suggested a modification there—and, as they say, I put my money where my mouth is.
But let me make this clear upfront: everything described and shown below is just an example of what you can do, and I’ll also offer some recommendations. This isn’t a final design, since it depends on what you want to achieve.

First, here are some relatively simple ways to increase the power.

1. Install a powerful MOSFET that can operate properly in linear mode. That would be great, except genuine parts are very expensive, and my attempt to buy used ones on Taobao ended in failure: I was sent counterfeits.
Advantages: a simple replacement can handle up to 500–600 W with the transistor at 75 degrees.
Disadvantages: cost.

For comparison, here’s an IRFP250N, which I wouldn’t recommend loading beyond 50–70 W.


2. A bipolar transistor. Yes, you could install something like a TK265 and get decent power, but there are some drawbacks.
Advantages: high power in linear mode, since that’s a normal operating mode for bipolar transistors.
Disadvantages: with the TK265 and similar parts, control is difficult because they have low gain. You also won’t be able to load it down to almost zero because of the fixed collector-emitter voltage drop.

The first drawback can be addressed by using the transistor in a Darlington configuration, or by installing a more convenient KT827, which already has this configuration built in. Unfortunately, this will only make the second drawback worse.
Another option is to beef up the transistor’s control circuit so you can use the same TK265 or equivalent parts.


3. IGBT transistors are essentially hybrids of MOSFETs and bipolar transistors. Since the “input” is a MOSFET, they’re easy to control. But the “output” is bipolar, so they have the same drawback as other bipolar transistors: a voltage drop across the collector-emitter junction.
However, if you don’t need to work with low voltages (around 1.5–2 V), this doesn’t matter much.


There’s one more problem, in addition to all those mentioned: continuously dissipating a lot of power is difficult even with a powerful transistor if there’s only one, because the die will simply overheat.
Here I’ll also explain how to get higher power using multiple low-power MOSFETs.

First, one option is to connect them in parallel. For example, a relatively simple approach is to


I also came across a more complex option, but unfortunately both of these designs only work at low voltages (around 5–10 V). The higher the voltage, the more transistor parameter variations will affect the results.



To get the best of both worlds, a circuit that balances the load between transistors is used. I used this approach in my review of the electronic load power module.


The operating principle is very simple, so I won't describe it here; I already covered it in my review of another electronic load.
In the end, I used the power module circuit from my main electronic load, which I built many years ago. It proved very reliable: at least, I haven't managed to burn it out in 5 years.
I only made a few minor changes and added an input divider (R14, R15), which is needed here to improve the circuit's stability, although it works fine without it.

The only thing you need to calculate is the current-sense shunts, so they don't overheat but also don't have too low a resistance. You also need to consider which transistors you're using and how much power you need. The number of channels is essentially unlimited; I use 8, and the diagram below shows two channels.


I quickly sketched a PCB in Sprint, gave it a go with the iron, and ended up with four boards since I do plan to use them all. I soldered one up right away for testing and debugging.


Everything is soldered onto the board except for the pair of power transistors and the capacitor, which is on the other side. I used PSMN-series transistors.




I can already hear the question: can you do all this without etching PCBs and so on? Yes. For example, you can buy a kit for building a simple electronic load, modify the circuit slightly—or, more precisely, leave out the potentiometer. Then add four power transistors to get four parallel channels.


The idea behind the experiment was to keep the original shunt and replace the MOSFET with this board. You only need to add one more wire for the board's power supply; the other three are soldered where the transistor connects.

The advantage of this solution is that the accuracy of the shunts on the boards doesn't matter at all: current is measured using the same shunt as before. Wire gauge doesn't matter much either, since everything will be compensated for. Still, I'd recommend mounting the transistor board on the electronic load's board and running wires from it to the transistors themselves.
There are some drawbacks, but fortunately they're minor. I'll explain them below and show you how to fix them.


1. The problem is that, with the connection shown above, current drawn by the power board starts flowing through the load board's shunt—in this case, 11 mA.
2, 3. The second problem is that when power is applied to the load input, the power module may partially turn on the transistors, allowing a small current to flow—in this case, 22 mA (11 + 22 = 33 mA). This current doesn't depend on the input voltage.


In practice, the problem isn't all that serious, especially since it doesn't affect how accurately you can set the current.
For example, the tests at 500 mA and 1 A are unchanged from before.


Still, it's better to do it properly, since parasitic current won't do you any favors when testing batteries.
The first problem, the board's current draw, is easy to fix: go to the menu and zero the ammeter.

After applying power, only the current from the partially turned-on transistors remains.
There's a catch: you could also zero out the second problem, the current from the partially turned-on transistors, but that would be the wrong thing to do. The board's current draw doesn't flow through the source under test, but the current from the partially turned-on transistors will keep flowing even after you zero the meter.


The circuit includes a bias-voltage generator to solve this problem, but sometimes it isn't enough, and you need to reduce the values of resistors R2 and/or R7. I said “and/or” for a reason: the problem doesn't necessarily affect both channels. In my case, for example, only one channel had the issue. You can find out which one either by trial and error or by measuring the voltage across the shunts: the one with the higher voltage is the problem channel. I had to add another 3 MΩ resistor in parallel with the existing one.


In the end, everything worked as it should: the current is zero when the unit is off, nothing changed in normal operation, and the set current still matches the actual current.


Unfortunately, this introduced a slight imbalance between the channels. It can be corrected by adjusting the resistors mentioned above, but the difference is small, so I just left it as is. Another option is to modify the second channel's circuit as well.

Below are photos of the voltage across the shunts at different currents. The first channel (the one I adjusted) is on the left, and the second is on the right. The total load current is 200 mA, 1, 5, and 10 A.
At first, you can clearly see that more current is flowing through the second channel. Then the currents become roughly equal, but in reality, the voltage drop across the board traces and at the probe connection points starts to have a significant effect at higher currents, so I'll repeat the test differently below.


To check, I first measured the transistor temperatures without heatsinks under different operating conditions:
1. 2 V, 2 A
2. 10 V, 0.4 A
3. 20 V, 0.2 A.

As you can see, as the voltage increases and the current decreases, the right-hand transistor gets hotter, but this is not at all critical under these conditions.


For the next tests, I mounted the transistors on a heatsink.


The operating conditions here are:
1. 5 V, 6 A
2. 20 V, 1.5 A
3. 60 V, 0.5 A.

There is a slight imbalance, but it's much smaller now. The imbalance will decrease as the load current increases, so we can consider everything to be working properly. In other words, roughly speaking, the closer the circuit gets to its critical operating conditions—where the imbalance starts to matter—the smaller the imbalance becomes.


Here's the actual wiring. Everything is held together "with a bit of string," since this is just an experiment. That's also why the component values in the circuit weren't really calculated, but chosen "by eye." As it turned out, though, it works pretty well.
The voltage divider is wired up with point-to-point connections, since it probably won't be needed in the device where I plan to install these boards.


And one last experiment: operation at high power. This is only for a short time, since the cooling is clearly inadequate.
Voltage: 38 V, current: 8 A, power: 304 W.

Based on everything shown above, I can say that the board works. Using several boards, you can easily scale the power up to the advertised 1000 W. There are no particular difficulties in increasing the power, and inexpensive transistors can be used. The biggest challenge is dissipating the heat and keeping the transistors below 100°C.
Basically, you can simply reproduce the circuit I built; this version is easy to mount on heatsinks. You can make a board with 4–8 channels or more—there are quad op-amps available. It would be better to use slightly higher-value shunts than I did, for example, 40–100 mΩ; it will work better.

Depending on the maximum load current and whether you use an external power module, you'll need to:
Up to 8 A — simply install an external power module.
Up to 17–20 A — move the protection diode onto the same heatsink as the transistors.
Up to 30 A — replace the current-sense shunt with a higher-power one.
Also, with any external module, you'll need to relocate the temperature sensor. You could try using the original remote sensor, though, if you don't plan to use it for anything else.


Someone posted a photo in the reviews showing a 300 W load at 33 V and 9 A. Honestly, I was surprised. Yes, some of the voltage drops across the diode and wires, so the transistor is actually dissipating about 290 W, but that's still a lot. The transistor won't last long under these conditions, but on the other hand, this speaks to the good quality of the transistor installed.


The device can be connected to a PC or a tablet/smartphone. There's a Windows program for the former and an Android app for the latter.
Let's start with the PC software. What can I say about it? Well, it works.
You can set a parameter value, but only for the parameter currently active; you can't choose between current, voltage, power, and resistance. You can turn it on and off. That's it. There's no graph export or access to the settings, and some buttons have ended up somewhere off-screen. On the plus side, the device's keypad isn't locked while using the software.


The Android app isn't much better. First, connect to the device over Bluetooth; its name is DL24-BLE. Then launch the app and connect to the device from there.
Here we have a single graph for current, voltage, and power.


One drawback is that the vertical scale defaults to a range of up to 300 V, so you have to stretch it, which is rather inconvenient. There are some settings, including a rather pointless electricity cost setting and a display brightness adjustment for the load...



Conclusions.
Overall, this is a pretty good device: wide current, voltage, and power ranges, easy to modify, high measurement accuracy thanks to a dedicated ADC, good setpoint accuracy, and a four-wire connection. Because of that last point, I recommend this version over the standard one, which costs 1–2 dollars less.
And considering that this little board costs just 20 dollars, I’d say it’s excellent—if not for a few minor shortcomings and, in my opinion, the biggest one: the CV mode barely works. I think the manufacturer could have added a hardware CV mode, especially since it wouldn’t be very difficult. I’d also like an autotest mode, where the current automatically increases up to a set limit or until the input voltage drops to a threshold value.

So, all in all, I’d say it’s very good and has great potential. It would make an excellent basis for building your own high-power load. There are a few minor shortcomings, which I’ve covered in more detail in the review.

Update: at your request, I’ve uploaded the Sprint-Layout files in two versions: the one I made, and one with the adjustment signal divider placed directly on the board. When printing the toner-transfer layout, select mirror mode.

That’s all for today. I hope the review was useful, and as usual, I’ll be glad to answer any questions.