Adjustable Power Supply Kit Board, or a Proper Power Supply Should Be Heavy, Part 2

I think some readers still remember my review of the kit for building a linear bench power supply. Despite a few shortcomings, the board turned out to be quite interesting.
When a friend sent me a link to the kit being reviewed, my first thought was, “That’s the same board.” But in reality, things turned out a little differently.
Anyway, if you’re interested in this kind of thing and want to find out how to build a small bench power supply, read on.

I’ve drifted away from my usual power supply topics a bit while reviewing various consumer electronics. I’ve already reviewed many adjustable and fixed-output power supplies.
But I couldn’t pass up this kit. I think you’ll understand.
This review covers not only the power supply board, but also a few additional products, which I’ll take a look at, show you, and give my opinion on. Some were provided by Banggood for this review; others I bought locally.

Since many kits started out as projects by enthusiasts before being copied by Chinese manufacturers, I did a little digging to find out where this kit came from—and uncovered a few things :)

But let’s take things in order. Why break with the usual flow of a review?

As always, a few words about the packaging first.
Since the items shipped from different warehouses, they arrived in several parcels. Only one parcel contained two items.
Maybe it’s just me, but it seems the store has started packing items more carefully.

First, of course, the power supply board. It comes in its own small cardboard box, with an antistatic bag inside.


Everything is just loose inside the bag, so if you dump it all out onto the table, you get a pile like this.


Remembering that familiar picture of toys in a sandbox, I tidied everything up and laid it out more neatly :)
This photo shows the first differences from the previous version of the power supply. This one has digital control and, accordingly, an LCD.
The manufacturer has also split the design across two boards: one power board and one control board.


Let's take a closer look at the boards.
Both boards are very well made, with a solder mask and silkscreen labels showing the values of the installed components. So you don't even need a schematic—though none is included anyway. The boards are made of FR-4, not cheap phenolic paper laminate, which is a big plus for beginners and experienced electronics hobbyists alike.

Power board.
Dimensions: 84 × 67 mm.
At first, this board reminded me a lot of the one from my previous review. I even thought it was the same board with digital control added, but the two power supplies are actually completely different.




Control board.
The board is a little smaller, at 80 × 56 mm.
You can immediately see the footprints for two encoders, as well as a large number of resistors.




Since no schematic was included, I drew my own. I may have made a mistake somewhere, but I tried to be as accurate as possible. If you spot an error, let me know and I'll fix it.

I originally planned to break the schematic down into functional blocks, but I'll start with the complete version.
A few things stood out:
The output capacitor is only 100 nF.
Apart from the input capacitor, there are no electrolytic capacitors on the board.
The penny-pinching Chinese engineers put a capacitor in parallel with the thermal switch contacts.


As I mentioned at the very beginning, this isn't a new circuit, so I found the original as well.
The circuit on the board under review has been modified and tweaked, but you can still see that the two designs are fundamentally almost identical.
I found the schematic here and here, complete with source code. The original device can even communicate with a computer, although I haven't tested that feature.


The circuit design is very simple. Its “heart” is a microcontroller connected to a DAC (digital-to-analog converter) implemented as an R-2R ladder.
By changing the code at the microcontroller's output, we can vary the DC output voltage. These DACs are simple, but they require a lot of microcontroller outputs, since each bit needs its own output. The more bits, the finer the output voltage adjustment.
This device uses a 10-bit DAC, which provides 1024 voltage levels.
For this power supply, that works out to a voltage setting resolution of 0.027 V and a current setting resolution of 0.002 A.
In practice, though, the voltage can be adjusted in 0.1 V steps and the current in 0.01 A steps. That's enough for most situations.


The key difference lies not in how the voltage used for adjustment is generated, but in how the feedback works.
Most often, the microcontroller outputs a reference voltage, which is then compared with the actual voltage by an op-amp, resulting in a regulated voltage or current.
The reference voltage is usually generated using PWM, then integrated (averaged) by a capacitor.
But this approach requires two DACs, one for current and one for voltage. Since the designer chose a different generation method, two DACs with an R-2R ladder simply wouldn't have worked. That's why the microcontroller also handles the comparison.


This control method is usually slower than the more familiar approach using an op-amp. But the designer implemented a software solution with two operating loops: a fast one and a slow one.
As I understand it, the fast loop works more roughly to ensure speed, while the slow loop then sets the voltage more precisely.
I'm not a programmer, so this is just my understanding. Perhaps those in the know can make more of the code and description provided here — link.


The voltage from the DAC is fed to the power stage.
In reality, the power stage on the board we're looking at is implemented a little differently: the current amplifier uses a power transistor of the opposite polarity, and the circuit has been modified slightly, but the operating principle is exactly the same.
The DAC output goes to the voltage amplifier. The 0–2.5 V range isn't enough for us, so the voltage is first amplified to around 0–30 V (the left-hand side of the circuit).
But since the voltage amplifier can't supply the required current, there's a current amplifier after it. This barely changes the voltage, so the output provides the specified 0–28 V, but now with a load current of up to 2 A.
The circuit description on the designer's page gives two options: 22 V at 2.5 A and 28 V at 2 A.
A current of 2 A isn't very high for a bench power supply, but I think the firmware could be modified to provide more current if desired.


Despite its apparent complexity, the circuit is extremely simple.
For example, I've broken it down into functional blocks:
Red — current amplifier
Blue — DAC and voltage amplifier
Green — voltage feedback
Pink — current feedback.


I think we've covered the theory for now, though we'll come back to it from time to time. But it's time to start assembling the board.
First, let's find all the resistors included in the kit. I'll also try to show how to assemble this board without using a multimeter to measure the resistance of the color-banded resistors.
Up to a certain point, I just insert all the components into the board (populate it), and only then solder them. I know some people do it differently, but I'm used to doing it this way. Can't help it :)

I'll hide this part of the process under a spoiler, since there are a lot of photos.

The main part of assembling the boards is done. At the end of this stage, we should have three things left: a power transistor, a thermal switch, and a ribbon cable.
If you have anything else left over, there are two possibilities: either you have a spare part, or you forgot to solder something somewhere. The first option is preferable :)


Actually, I should have assembled everything first, mounted the transistor on the heatsink, and only then tried it out. But I couldn't resist and tested it right after assembly, simply plugging the power transistor into the holes. Better not do that, though :)
To be fair, when I first powered it on, all I got was a uniformly lit display. If everything is assembled correctly, you just need to adjust the contrast using the trimmer on the control board until the display is clearly visible.


The device works—or at least, for now, it is doing its best to look like it works. I'll move on to describing what else I ordered.


I also had to stop by the market and buy a few more things for this power supply.
As it turned out, the photo includes some things I didn’t use, and leaves out some components I did use. It’s not always possible to plan 100% in advance what you’ll need during assembly.
But in any case, I’ll do my best to explain and show everything.
Of the things I bought, I used:
1. Transformer
2. Fan with grille.
3. A pair of knobs
4. Power switch
5. Connector.


I decided to buy the transformer locally for a couple of reasons.
1. They’re cheaper here.
2. Our transformers are much better quality than the mass-produced ones from China.

I chose a transformer rated for 24 V and a little over 4 A (100 W).
Link to the transformer, about $13. The store had a pretty wide selection, but this was the only one that was a perfect fit for me.


At first, I wanted to use a 12 V, 60 W transformer I had lying around, but the power supply needed a transformer rated for at least 65–70 W, and I also wanted to add some extras that would need power too…


The 60 W transformer weighs this much compared with the 100 W one. If we subtract the weight of the extra wire wound onto the 60 W transformer and the connector, the difference comes to about 1.5 times.


As always, putting it together involved agonizing over choices—for example, how best to arrange the “guts” inside the case.
At first I had four perfectly workable options. Each had its pros and cons, but I kept feeling that the result was a mess.


In the end, I settled on this arrangement for the heatsink, transformer, and boards. It seemed like the best option to me.


Originally, the designer intended it to be powered by a switching power supply, and even recommended a spot for installing a noise-suppression capacitor.
But in my opinion, why spoil a linear power supply by powering it from a switching source? If you're going to do it, do it properly :)


By the way, you can see above that the heatsink is tiny—clearly not up to this power level.
In a short-circuit condition at maximum current, about 60 W is dissipated by the heatsink, while the one in the photo is rated for 10 W at most.
The store listing shows a heatsink with a fan, but it's still laughable—and it's inside the case, too.
What I did like were the output terminals. I should get some of those somewhere.
By the way, the case in the photo is the same as the one in the review, for comparison :)


Then another problem cropped up. The heatsinks had vertical fins, which just wouldn't fit my design, so I had to cut them. At that point, I was glad the heatsink body was thin. As they say, every cloud has a silver lining.
For some reason, heatsinks with fins running along the long side seem to be much harder to find than the one in the review.


Next, I took a short M3 threaded rod, which is basically a screw with its head cut off.


I tapped the outer holes in the heatsink to a depth just over half the rod's length, so that even in the worst case it couldn't be screwed all the way into one half.


After that, I screwed the stud halfway into one of the heatsinks. To avoid stripping the threads, you can thread on two nuts and tighten them against each other near the end of the stud, then turn the outer nut.


We tightened everything together. By the very end, the heatsinks were already touching, but we still hadn't reached the required position, so I had to use some force. The threads were right on the verge of stripping, but in this case that wasn't a big deal.
The stud was there to keep the heatsinks in the same plane and make the assembly more rigid.


Finally, I took a small piece of the aluminum plate that had originally served as a heatsink, drilled four holes in it, and applied thermal paste.
At first, I had to choose between a thicker, more rigid plate and a thinner, more flexible one. I went with the second option: the stud added rigidity, and I had no use for the extra thickness.


Let's see how it fits. I think the idea is clear by now.


Now let's make the rear panel.
Mark out all the required holes. The fan opening is cut short at the top and bottom so as not to weaken the rear panel.


A quick digression.
People sometimes ask me which fans are best.
I use Sunon fans because I consider them the best available on our market. For this review, I bought a 50 mm, 12 V fan.
Just in case, here's a link to the fan and to the grille. The fan costs about $3.5, and the grille about $0.4.
The grilles come in 50 and 52 mm sizes; I bought the 50 mm one.


I originally wanted to use a 60 mm fan, but it was a very tight fit, and I didn't want to risk it.


The fan uses maglev technology, which reduces noise and extends its service life.
I got one of them wrong: I wanted a high airflow, but overdid it. At full power, the fan is very noisy; I should have chosen a different model.


As I mentioned above, the fan is rated for 12 V. That caused a few problems, since the board is designed for 24 V. So I had to step the voltage down, and it would be best to know the current, since that determines how much power the regulator has to dissipate.
I was surprised that the fan starts at just 2.7 V and already moves a decent amount of air. At 12–15 V, it gets blown away, so I had to brace it against a wire.
All in all, if you're choosing a fan, I don't think you'll find a better one here. They're more expensive than generic computer fans, but worth it.


I mounted the transistor on the heatsink. It's electrically isolated from the heatsink with a thin mica sheet, since the screws will protrude through the case and I don't want to accidentally short anything later.
This power supply uses a transistor with the opposite polarity (PNP) to the one in the previous power supply (NPN).
Interestingly, the original schematic uses an NPN transistor.
The transistor has a maximum continuous collector current of 10 A, a voltage rating of up to 140 V, and a power dissipation rating of up to 100 W, which is more than enough for this power supply.


For the assembly, I needed M4×50 mm screws and M3×10 countersunk screws.
At first I planned to use shorter black M3 screws (the ones in the photo), but unfortunately they were 1 mm too short. They would have looked neater.


This is what the assembly looks like. The M4×50 screws hold the whole thing together like a sandwich.
I had to install the fan so that it blows air out of the case. It's certainly far from ideal in terms of efficiency, but I couldn't figure out how to assemble everything so the fan would blow onto the heatsink.




Unfortunately, the store doesn't offer brackets for mounting the transformer (that would have been much more convenient), so I had to improvise. I drilled six holes and stuck on strips of thick double-sided tape.


Then I strapped everything down with wide cable ties. The ties hold the transformer in place, while the tape keeps it from sliding. I should have put electrical tape or thin rubber under the ties—I remembered only after tightening one. Since I had four ties left and one eventually snapped, I had to leave it as it was for now.
The blue wire from the transformer is there for a reason; I'll explain later.


Front panel.
To mark the holes, I use a rough draft of the front-panel template. I didn't save the photo, but I think the idea is clear.



Now for the final assembly and testing.
Here's the partially assembled unit—the board has found its place :) Two of the mounting holes in the case line up with the heatsink mounting holes on the board.


First power-up in the case.
There's a trimmer on the power board for calibration.
Use the encoder to set any fixed voltage, say 5, 10, or 15 V. Connect a multimeter to the output, then adjust the trimmer until the meter reads the set voltage. That's it.


I'll test it using an electronic load, but it's already clear that the power supply can handle its rated power.


Tests.

I cobbled in an extra capacitor (I mentioned it above). With it, the voltage will be a little higher under full load, and I already had it lying around anyway :)


And here’s the finished result.
You can see a couple of resistors near the fan connector in the photo. They’re connected in parallel with the thermal switch.
I almost forgot. The kit includes a 50-degree thermal switch to turn on the fan. I mounted it near the power transistor and pressed it against the heatsink with a clip. The thermal switch came in a silicone sleeve; leave it on, since one of the terminals is in contact with the case. The retaining clip should be shaped so it doesn’t cut through the insulation.
The resistors are connected in parallel so the fan runs continuously at low speed and is less likely to switch to full speed.


Remember to insulate any exposed contacts, especially those connected to mains.
For safety, connect the incoming wire to the outer terminal of the fuse.


Attentive readers have probably spotted a relay in the previous photos.
The thing is, a linear power supply dissipates all its “extra” energy as heat. To reduce that heat, I made a small modification.
I added a tap to the transformer at about 16 V (the total voltage is 27 V with no load), and the relay switches between the windings depending on the output voltage. To find the right spot for the tap, I used a sharp needle and a multimeter. Connect one multimeter probe to any lead of the secondary winding, then use the other probe and the needle to find the right voltage by piercing the wire’s enamel. It’s worth measuring the difference, as that’s easier. For example, if the transformer outputs 27 V and you need 17 V, look for 17 V or 10 V. In the second case, you’ll also get 17 V, but relative to the other lead of the secondary winding.
Be very careful when piercing the enamel with the needle: it’s tough, and the needle can slip and pierce the primary winding.

At first I thought I’d use a comparator—I even bought one and may still install it—but I decided to keep things simple.
To stop the relay from constantly clicking back and forth, hysteresis is used: switch on at one voltage and off at another. But an electromagnetic relay has hysteresis of its own.
For example, I used a 24 V Finder relay. It switches on at 13 V and off at 6 V. Those thresholds aren’t very convenient, of course; 12–15 V would be much better. But this is what I had. I’ll add a comparator later, which will let me set the thresholds as needed. Even so, the power dissipated at low output voltages is lower, which is already a good thing.
The relay is connected to the positive output and the negative terminal of the diode bridge, so its current draw doesn’t affect the display.


I also connected a switch to turn the load on, but instead of putting it in series with the power wire, I decided to do it more elegantly.
The control board provides a control voltage; shorting it to ground brings the output to zero.
This way, you can set the desired output parameters first, then simply flip the switch to get the settings you want.


And here's how it turned out.
For the encoders, I used these knobs, but I had to shorten them a little because the encoders sit very deep and the knobs wouldn't stay on properly. Otherwise, there were no problems.






There weren't many changes to the circuit, especially since adding the 12 V regulator was simply dictated by the fan I used. Still, I added them to the schematic just in case.
All additions and changes are shown in red.


This review turned out to be really long. I kept expecting a notification that I'd exceeded the character limit. But for those who don't like long reviews, I made a mini version in four photos :)))


A little about future improvements. I want to add proper winding switching, as well as a couple of fixed-voltage outputs and a load-current indicator for them. That's what the free space on the front panel is for. The outputs will need to be switchable, though, since the voltage will most likely be stepped down using a PWM converter, which could cause interference.
I may add a fan speed controller, but I'm happy with it as it is for now.
As supplementary materials, I've included everything I put together for this power supply, along with several datasheets, my version of the schematic, and the front panel — link.

Now let's try to sum up this long review. I won't go over the pros and cons of the additional items; I covered those in the review itself. I'll focus only on the power supply board.
Pros
Very good PCB manufacturing quality
Decent-quality components
Works properly when assembled correctly
Minimal, very straightforward setup
Current can be set without connecting a load
Precise adjustment of output voltage and current
No need to buy a voltmeter and ammeter
Thermal relay for automatic fan control

Cons
Output current is only 2 A
The transition between fine and coarse adjustment of output voltage and current is touchy
No schematic included; assembly instructions are available electronically.

My take: this is definitely an interesting kit. It has pretty much everything you need to build a power supply; all you need to add is a transformer, heatsink, and enclosure. Last time, people often asked how to use a switching power supply instead of a transformer. That's not an issue here—the power source can be anything. There were also questions about adding current and voltage readouts; here, they're already built in. And there's the nice bonus of encoders, so you don't need multiturn potentiometers. A big plus for me is being able to set the required current before connecting the load. That wasn't possible with my previous power supply, especially with multiturn potentiometers.
And I should mention that the kit comes with the original program source (though without encoder support), which you can modify to suit your needs. In theory, after modifying it, you could connect it to a computer too, but I think that would be overkill in this case.
The only downside I can think of is that digital feedback is still slower than analog feedback. There's no getting around that, at least not cheaply.

Of course, there'll be comments along the lines of, “For xx bucks, you can buy a ready-made power supply.” Sure, that's true; I won't argue—you can buy one. But not everything can be bought with money. What about the satisfaction of building it, seeing the finished result, and simply spending your time enjoyably? How much is that worth?

Who is this kit for? I think it's aimed primarily at beginner radio enthusiasts. It could make a good gift for a teenager interested in electronics—you definitely won't be embarrassed to give it. More experienced hobbyists might enjoy it too, simply as a useful project and a pleasant way to spend some time.

That's probably it. As always, I look forward to your comments and questions. I hope you found the review useful and interesting.

The product was provided by the store for review.