Kit for building a 60 V, 20 A adjustable linear power supply

Today's review is about something quite familiar to electronics hobbyists. There are plenty of reviews of adjustable power supplies, as well as kits for building them. I've posted a couple of similar reviews myself. But today's kit is a little unusual, both in terms of power and circuit design. Of course, hobbyists have used all these approaches many times before, but this is the first time I've come across them all together in a kit, so that's what I'm going to talk about.


First, I should probably mention that this is actually the first of at least three reviews. I plan to continue the series and eventually build a powerful linear power supply with digital control. I can only guess what the final version will look like; the finished design will probably be influenced not only by my own ideas, but also by suggestions in the comments.
To make it easier to understand what this "kit" is all about, I had to do some reverse engineering again and redraw the schematic.
But I'll take things in order and show you everything in due course. For now, let's take a look at the product.

I ordered this kit from Taobao. I stumbled across it completely by chance, and haven't seen it on any other marketplaces.
The seller offers several configurations, but today's review covers the set with all three boards.

They arrived in individual bags, but the bags got a little... battered in transit. The main board had some extra packaging, and in any case, everything arrived intact.


As I said, the kit consists of three boards. They're all the same size—100 × 71 mm (excluding protruding components)—but differ in height. They're designed to be stacked like a "sandwich," but there's nothing stopping you from installing them separately or even using them almost independently.

So, from left to right:
1. Transformer winding switching board.
2. Power stage of the voltage regulator.
3. Control and measurement board.

The first and third boards come with standoffs, but they're only long enough to mount the boards on the enclosure surface.


Since I've only seen this kit on Taobao, I weighed it just in case that helps with estimating shipping costs.
It weighs 360 grams, which seems pretty light to me.


The kit also includes a set of interconnects and wires for connecting the indicators and controls.
The kit also included:
1. 13 mica insulating pads (8 required)
2. Six cable lugs with insulators.
3. 14 screws with washers (8 required)
4. Two red LEDs.


I'll start with the switching board, since it's first in the circuit.
The board's width and length are given above. It's about 35 mm high, excluding the standoffs but including the leads.


The board looks a little different on the product page, mainly because different component types are used.


The power inputs and outputs are connected via screw terminals.


The windings are switched using four electromechanical relays.


A closer inspection revealed a problem—or, more precisely, an error. Let me explain.
First, you can see the relays in the photo. They're pretty decent relays, but their contacts are rated for a maximum of 16 A.
There's a fuse next to the relays, also a decent-looking, seemingly good-quality part, but it's rated for 15 A, even though the board is marked for 20 A.

For those who haven't figured it out yet, here's the issue. The specifications claim 20 A, and that's where the main "pitfall" lies. Many of you probably know that if you connect a diode bridge to a transformer and then add a capacitor, the voltage across the capacitor will be about 1.4 times higher than the voltage across the transformer winding.
The transformer terminals carry the RMS voltage (say, 10 V), while the capacitor has the peak voltage (about 14 V).
But it turns out not everyone knows that the current in the transformer circuit will also be 1.4 times higher than the current after the filter capacitor. So, to get 1 A at the output, the current in the transformer circuit has to be 1.4 A.
I've simplified things quite a bit, but in any case, the current before the diode bridge will be higher than after the filter capacitor.

Now let's get back to our board. It's rated for 20 A, which means the current before the rectifier can reach 28 A—perhaps a little less. But in any case, it's not 20 A, let alone 15 A (the fuse rating) or 16 A (the relay contact rating). So with DC, the maximum load is only 11–12 A instead of 20.


The board is powered by its own transformer winding, so it has a bridge rectifier, a filter capacitor, and a 12 V regulator mounted on a heatsink. Basically, this heatsink determines the board's height.
A separate transformer winding is required to power it.
The board also has a standard three-pin connector for a fan. The speed isn't regulated; the fan is always powered by the 12 V regulator.
There's space next to it for another identical connector. The idea is that you can connect a voltmeter here, since the three pins provide ground, 12 V, and a 0–60 V output. But you'd need to install a different type of connector, because it's very easy to burn out the fan (the tachometer pin carries 0–60 V). In any case, the circuit is quite well thought out.


And this is what actually controls the relay: a four-channel comparator, with four trimmer potentiometers next to it for adjusting the switching thresholds.

The principle is very simple. The board measures the power supply's output voltage and connects additional transformer windings as needed. Initially, power comes from a 12 V winding; you can connect up to four more windings with the same voltage, giving you 12 to 60 V in 12 V steps. As a result, even at an output current of 20 A, the transistors dissipate a maximum of about 300 W.
But I'd wind the last few windings for 10–11 V rather than 12 V, since the voltage drop across the bridge rectifier has less effect at higher voltages.


All the high-current traces have been reinforced with plenty of solder, but I'd remove the solder and attach copper wire to them for reliability. Though in any case, without replacing the relay, you won't get more than 16 A through this circuit.


Board connection diagram.
The board outputs AC, so you'll need a bridge rectifier followed by several capacitors with a decent capacitance (30,000–50,000 µF) rated for 100 V.


The second board is the regulator. It has eight transistors, low-resistance resistors, and other small components.


The board matches the seller's photo, but the diode on mine is noticeably less powerful.


Basically, the whole board is one powerful transistor with high gain. Its only purpose is to amplify the control board's output while dissipating all the excess heat.
On the left side of the board, there's a spot for a 55 °C thermostat in a TO-220 package, which should also be pressed against the heatsink. Below that is a spot for a fan connector. But there's no thermostat, so if you want to modify the board, you'll need to install one, solder on a connector, and supply 12 V to the board (a shortened connector without this pin is installed).
It's easiest to install a standard thermostat in a flat package, rated for 55–60 °C.


Since the transistors are bipolar, each emitter circuit has a 0.1 Ω current-balancing resistor. But there are seven such resistors, not eight; the eighth position has a 100 Ω resistor instead.


The eighth resistor is connected to the first transistor, which controls the other seven. In other words, seven transistors handle the power, while the eighth drives them, increasing the gain of the entire module.


All the transistors are identical, TIP35C, each rated for a maximum power dissipation of 125 W (at 25 °C) and a current of up to 25 A (40 A for short periods). So, theoretically, the module can deliver up to 175 A and dissipate up to 875 W. The transistors don't have very high gain, around 50, so there's another transistor, the first one, to drive them. With it, the gain rises to 2500 (in theory).


The board is designed so that all the transistors lie in the same plane and can be clamped to a common heatsink. That's what the included screws and mica insulators are for.
The collectors of all the transistors are connected together, so the insulation is mainly for safety, since the collector can be at up to 100 V. But if the heatsink itself is properly insulated and nobody goes poking around inside the power supply, you can leave out the insulation (though it's not recommended). I'd use it—your health is worth more.


The seller has lots of different power modules, and this is one option. As far as I understand, they’re compatible, but shipping will be quite expensive because of the weight.


Control board.
This board is also universal, as it can work with different modules, and I’ll most likely show that in my next review.


There are quite a few more components here—six IC packages alone. But it’s all analog; no microcontrollers :)
There are also more connectors, but I’ll get to those later.


The connectors are located on one of the short sides:
1. Board power. Two 15 V windings; the board uses a split supply. As a last resort, you can power it from a single winding, in which case the bridge rectifier and capacitors will act as a voltage doubler, but 100 Hz ripple will increase.
2. 0–60 V input, which also serves as the output to the relay board, since the two connectors are wired in parallel. The board supports a four-wire output connection. For the full setup, connect the power supply output wires to one terminal block, preferably close to the load. Take the signal for controlling the winding switching from the other terminal block.
3. Three terminals: ground, input, and output. The shunt is installed in the positive (adjustable) rail of the power supply, which is very useful.

The power supply output has a 100 µF, 100 V capacitor plus a 0.1 µF capacitor.


As I already mentioned, the board uses a stabilized split supply, so you can see a pair of 12 V regulators mounted on heatsinks.


The control and measurement sections use a curious assortment of op-amps, rather than all the same type: TL072, TL082, OP07, LM258.


But I suspect there’s a good reason for this “op-amp zoo,” since the board also has a precision voltage reference, AD586L. It looks very much like a counterfeit, but its specs are pretty good: 5 ppm over a temperature range of -40 to +85, and it’s also low-noise. A special capacitor is installed next to it, as specified in the datasheet.


Output stage. This section controls the power board. The control board’s name is printed here too, but I couldn’t find any information about it.


Connectors at the other end of the board.
1. CCCV indicator LED. Two LEDs were included, connected in antiparallel. Alternatively, you can use a two-lead, two-color LED.
2. Current adjustment potentiometer
3. Voltage adjustment potentiometer. Both potentiometers are 10 kΩ; the value is printed on the board.
4. Output to the power board. Some of the pins aren’t soldered, and I didn’t investigate what they’re for.

A 2.7 kΩ resistor is installed to the left of the connectors, between the control board ground and the power supply output ground (they’re different).


The soldering and assembly are generally pretty good. The only thing that annoyed me was that the components on the board aren’t numbered.



Schematic. I know it looks intimidating, but I tried to redraw it as close to the original as possible. The board layout made the process a nightmare at times, even though the circuit itself isn’t very complicated.
I simplified it a little by removing the ±12 V regulators and their bridge rectifier and capacitors.
As you can see, the power module uses floating control, which is why a separate transformer winding is needed to power the control board. The board ground is connected to the power supply output.


Block diagram showing how the modules are connected. It’s all very straightforward and goes together like a kit.


You’ve probably noticed the mysterious switch in the schematic. At first I didn’t quite understand what it was for, but when I did, I was pleasantly surprised.
The thing is, this power supply can also work as an electronic load. The green arrows in the block diagram show the current flow during normal operation as a power supply, while the red arrows show it operating as an electronic load.
In this mode, the board sets the load current up to 20 A, with the power limited by the rating of the power stage—more precisely, by its cooling capacity. Since this power supply needs cooling capable of dissipating around 200–300 W, we get a load with the same specifications. The ammeter works normally and displays the load current.

Overall, it’s a simple, elegant, and functional solution. The only downside is that it requires separate terminals on the front panel. The positive output terminal of the power supply serves as the negative input terminal for the electronic load.


Although I’m not planning to build anything right now, I’ll still run a quick test. Although in this case my main goal is really to put together a sort of assembly guide.
First, I gathered everything I might need.
1. Transformer. I have three in this case, but even so, they can't cover the full range of current and voltage.
I recommend using two transformers: a main one with five 12 V windings rated for 10–20 A, and an auxiliary one with three 15 V windings—or preferably four, so there's a winding available to power the ammeter as well.
2. Diode bridge. I'll show it later.
3. Filter capacitor. For the experiment, I used a 2200 µF, 50 V capacitor, though I later added another 1000 µF. But that's nowhere near enough. Plan on at least 20000 µF; the seller recommends more than 40000 µF.
4. Potentiometers. I used ordinary ones, but of course multiturn pots would be better, and digital control would be better still—but that's for another time.


I bought the KBPC3510 diode bridge on AliExpress as an add-on to some other item, and also just to test this board.
I temporarily soldered on the resistors and LED. The LED needs replacing, and I'll deal with the resistors separately.


I actually liked the seller: they sell not only boards and kits, but also a more suitable diode bridge option and a board for the potentiometers (the potentiometers are sold separately). With all this, the assembly really does start to feel like a kit.


First, connect the power section and rectifier. Since I ended up with only three 12 V windings instead of five, two terminals were left unused.


I mounted the power module on a “toy” heatsink :) In general, you'll need a fairly substantial heatsink, since it may have to dissipate up to 100–200 W, depending on the operating mode. And if you plan to use this power supply as an electronic load, its power rating will be determined by the size of the heatsink.


Connect all the power wiring; I think it's pretty self-explanatory. The main thing is to make sure the ground terminal on the control board has a good connection. If contact is lost there, the full voltage will most likely appear at the output.


Next, connect all three boards so they work together. The three-wire cable runs from the control board to the regulator's power board, while the two-wire cable goes to the relay board. In fact, only two wires are used in each cable; the middle wire in the three-wire cable is cut near one of the connectors. Since all the cables have connectors, wiring it up is very straightforward.
When you're done, you should have three cables left, each with one connector.


Here's a closer look, just in case.


Connect the auxiliary transformer. The best one I could find had 15 V, 9.5 V, and 19 V windings. I used the 15 V winding to power the control board and the 9.5 V winding for the relay board. Yes, it's a bit of a compromise: the control board really needs two 15 V windings, while 9.5 V is a little low for the relay board, so I didn't get a regulated 12 V. But it's more than enough for testing.
A reminder: the control board needs two 15 V windings connected in series, and the relay board needs one 15 V winding. The boards must be powered from separate windings!

In principle, you can put all the windings on a single transformer, but if you plan to use the electronic load function, I'd use two transformers and switch on the high-power one only when operating as a power supply. You can combine the power and operating-mode controls in a single switch.

And that's basically it: power is on, and the LED lights up. I accidentally wired it so it indicates CV mode, although it would make more sense to use red for CC mode.
I also connected a fan to the designated connector, but in this configuration it's always on, which is annoying.

You'll be left with one wire, which is needed for a four-wire load connection. Everything will work without it, but if you want to improve the accuracy of the output voltage regulation, it's best to use it.


A small tip: keep the wires from the rectifier as short as possible. At the initial stage, I connected my adjustable power supply instead of the rectifier, using a long cable, and got oscillation over a small range of output voltage (if I remember correctly, 20–23 V). Connecting even a 1000 µF capacitor to the control board's input terminals completely fixed the problem.


For the first test, I simply connected a car bulb to the power supply output and immediately ran into a small problem.
The thing is, the relay board isn't configured when it arrives, so my power supply wasn't switching the windings.

Use the trimmer potentiometers to set the switching thresholds. To do this, set a specific voltage with the resistor, then turn the trimmer until the relay switches. Adjust them from the bottom up: first relay 1 with resistor 1 at the minimum voltage, then relay 2 with resistor 2. Turning clockwise increases the voltage.
The product page lists the recommended thresholds:



A few more tests. If you'd like to see other tests, let me know. I'm planning at least two more reviews of these boards, so I'll include the test results in the next one.

1. With three 12 V windings, I got a maximum of 46 V. But that was with no load.
2. The maximum current with the output terminals shorted was 17 A. My transformer is very underpowered, and the filter capacitor was only 3200 µF (2200 + 1000).

On the other hand, the output voltage stability is simply excellent, especially for this price :)
3. Set the voltage to 10.747 V with no load.
4. Apply a load of about 4 A, and we still get 10.747 V. The last digit occasionally fluctuated by +/-1, but I don't think that's significant.
Keep in mind, though:
1. The filter capacitor is only 3200 µF.
2. The control board is powered from one winding, not two.


Video version of the review


Now I can offer a few preliminary conclusions. The advantages include:
1. A decent design and circuit topology.
2. Almost everything you need is included. Well, everything except the transformer, heatsinks, and filter capacitors, which are cheaper to buy locally.
3. Excellent voltage regulation accuracy.
4. Four-wire connection.
5. A relay board that significantly reduces heating in the power module.
6. Can also be used as an electronic load.

There are some drawbacks, too.
1. The voltage is fine, but to get the advertised 20 A, you'll have to replace the relay and fuse.
2. Switching the windings reduces heating, but there can be small spikes during switching.
3. The transformer needs a large number of windings: at least 8 in total, ideally 9.

I sometimes noticed that the windings didn't switch cleanly under load. This was caused by the very low capacitance of the filter capacitor, so I had to lower the switching thresholds a little.

Even with these drawbacks, I can say this is a very interesting kit. It may not be particularly cheap, but building something like this from scratch is expensive too, even just in terms of components. I really like how easy it is to assemble; there's practically no soldering to do. The power supply also performed well in terms of stability. By the way, the seller recommends using wirewound resistors for adjustment, since ordinary resistors have poorer long-term stability.

In the next part, I'll cover an alternative power module. After that, I'll prepare a review of a module with digital control. That's all for today; as usual, I look forward to your questions and comments.

The order was placed through the agent yoybuy.com. This is a referral link; it gives you a 10-off-50 coupon, and I don't get anything from it.
The kit costs approximately $40–45 including shipping.