/       /       /    50 V 10 A Digital Linear Power Supply Board from Gandf
Support the project on Patreon


50 V 10 A Digital Linear Power Supply Board from Gandf

$3.17
Go to store
Not long ago, I promised a review explaining how to add digital control to a linear power supply board. To keep that promise, today I'm posting a review of the board itself, along with some drawings, the schematic, firmware, and, for the real enthusiasts, the source code.


I should probably start by saying that this board—or, more precisely, the device you can build around it—is essentially an add-on for the electronic load by the same author. I've published a whole series of reviews of the load, and it also features in almost all my power supply reviews, and sometimes in battery reviews too.
The two devices are very similar not only in their specifications—the power supply has a maximum voltage of 50 V and current of 10 A, while the load handles 5–15 A—but also in their circuit design.
First of all, both devices have simple circuits and are accessible even to relatively inexperienced users who want to build them. I know for a fact that after my electronic load review, several readers built their own, and one even managed to modify the firmware (many thanks to him for that).

But I'll cover all that in the review. For now, a few words about the packaging.
I bought the board together with the controller for the Sousim load, which I wrote about earlier. The box turned out to be really big—I honestly wasn't expecting that.
The total weight was 222 g, the two orders weighed 114 g, and the kit featured in this review weighs a tiny 20 g. I think you could easily add a kit like this to another order :)


The kit came in a small bag; everything you saw above was the store's packaging.


The kit includes:
Printed circuit board
Assorted fixed resistors
Trimmer resistor
Inductor.
Jumper?


Yes, it's a very modest kit. What's more, the seller's page mentioned high-quality 223 capacitors, but there weren't any in the package.
Here are the resistor values:
100 Ω - 2 pcs
220 Ω - 1 pc
1 kΩ - 4 pcs
4.7 kΩ - 7 pcs
10 kΩ - 6 pcs
39 kΩ - 1 pc
82 kΩ - 1 pc
100 kΩ - 1 pc
200 kΩ - 3 pcs
3 MΩ - 2 pcs

Almost all the resistors have 1% tolerance, as indicated both by the brown band at the end of their markings and by my measurements. The only standard-tolerance value was 3 MΩ.

20 kΩ trimmer resistor and 100 µH inductor


I also listed a jumper as part of the kit above. In fact, it's a resistor—or, more precisely, a wire shunt.
The product page has a photo showing that the kit should also include a 0.1 Ω resistor. The wire's resistance is about 90 mΩ, which is roughly in line with that.
But there are one—or, more precisely, two—problems:
1. The schematic and the board specify a 20 mΩ shunt.
2. Even with a 20 mΩ shunt, it dissipates up to 2 W; with a 0.1 Ω shunt, it would be a full 10 W.

In the end, I set the wire aside and will deal with the problem another way, but more on that later.


The printed circuit board is made quite neatly. The only thing I didn't really like was the small pads in a few places.


Another issue came up during assembly: the current-sense shunt isn't connected quite correctly. The voltage isn't sensed directly across the shunt; some of the PCB traces are included too. Copper has a fairly high temperature coefficient of resistance, so it can affect the reading as it heats up, especially at high currents.
It would be better either to reinforce the power traces with thick wire or to modify the measurement circuit connection slightly.


The board has a silkscreen with both the component reference designators and their values. The latter are much more important, since I mostly relied on them.
The board also identifies its designer, the Gandf mentioned above.


I looked for the PCB layout in some standard format for this review, but unfortunately could only find it in this form. Still, if you have the time and inclination, I think it should be quite possible to convert it to the same lay format.



The board measures 87 × 60 mm, and that's no coincidence: these dimensions match those of a 1604 display exactly, including the mounting holes. The electronic load board was designed the same way, except that it used a 2004 display and was slightly larger.


The schematic turned out to be both easier and harder to find.
Easier because there are more of them available, and harder because there are several versions, and none of them looks much like the one in the review. For example, there was a version with a simplified power transistor connection.
Or rather, that's not quite right: the schematic matches the board, but not all component designators and op-amp pin numbers match.
I found the closest match, and it should be good enough to build from.


Next come several tedious assembly steps.
As usual, I start by soldering the components there are the most of, since that makes the rest easier to find.
First, the 4.7 kΩ and 10 kΩ resistors—13 in total, almost half of all the components.


Next came all the other resistors, plus the inductor. The only minor difficulty was finding the spot for a 200 kΩ resistor: at a quick glance, the value on the board looks like 20 kΩ.


After that, I was practically out of the components supplied with the kit; only the trimmer resistor was left. So I had to look for everything else at home, and buy whatever I couldn't find.
I also decided during assembly to replace some capacitor values with others.
The board originally calls for:
22 pF - 2 pcs, 0806
1 nF - 5 pcs
10 nF - 2 pcs
22 nF - 1 pc
100 nF - 2 pcs.

I replaced two of the 1 nF capacitors with 220 nF ones, since they are downstream of the voltage regulator.
One of the 100 nF capacitors was also replaced with a 220 nF one, so the list now looks like this:
22 pF - 2 pcs, 0806
1 nF - 3 pcs
10 nF - 2 pcs
22 nF - 1 pc
100 nF - 2 pcs
220 nF - 3 pcs.

I had to make the 22 nF capacitor from two capacitors—15 + 5.1 nF—since I didn't have a 22 nF one at home.

Also needed:
8 MHz crystal resonator
10 µF, 16 V polarized capacitors - 7 pcs (I used tantalum capacitors desoldered from an old motherboard)
1 µF, 50 V polarized capacitor - 1 pc (63 V would be better; I'm trusting the quality of Matsushita)
1–10 Ω resistor; I used 1.6 Ω (it wasn't included in the kit). In the schematic, this resistor is in series with the RS232 output.



Solder everything in place. The arrows mark the 220 nF capacitors.


The 22 pF capacitors go on the back of the board; I installed the 15 nF capacitor there too.


More components, this time:
1N4007 - 4 pcs. The board specifies 1N5819 Schottky diodes, but I used ordinary ones.
1N4148 - 1 pc, the gray one in the photo.
TL431 - 1 pc, used to stabilize the reference voltage. The B version is preferable; I only had an L on hand.
TL084 - 1 pc. Op-amp.
Narrow 28-pin socket for the microcontroller. I decided not to use a socket for the op-amp.


Install everything according to the orientation marks on the board.


Next up was the current-sense shunt, but first, an important note.
The board has pads for connecting the power transistor, but they are labeled incorrectly: read B-C-E instead of E-C-B—in other words, the pin order is mirrored.
This is a bug on all these boards; even the PCB layout shown above has the same error. What's more, it was mentioned on a Chinese forum by someone who bought the same board.
It's strange that the seller doesn't mention this at all.


I've now reached the low-resistance resistor. The description says it should be rated for 3 W, but it actually dissipates 2 W (at 10 A).
I used five 0.1 Ω, 1% 2512 resistors. According to the datasheet, each is rated for 1 W, but since they're installed right next to each other, their power rating should be derated.
At first I wanted to stack them, but then I arranged three in a row and soldered them together using leads left over from the 1N4007 diodes. Then I placed two more resistors on top, giving me a 20 mΩ resistor.
Of course, it would have been better to buy four 20 mΩ resistors and arrange them lengthwise in a 2S2P configuration, but I had to use what I could find for sale.


More components, this time:
470 µF, 16 V capacitor - 2 pcs
100 µF, 16 V capacitor - 1 pc
100 µF, 100 V capacitor - 1 pc
7805 - 1 pc
7905 - 1 pc.

The biggest challenge was finding 470 µF, 16 V capacitors, since they tend to be either too wide or too tall. In the end, I used some Teapo capacitors. I used a 100 µF Samwha RD series capacitor.
The 100 µF, 100 V capacitor is the input capacitor for the power supply. According to the schematic, it should be rated for 60 V (odd that it isn't 63 V), so I chose a 100 V one for extra headroom, also from the Samwha RD series.


The 100 µF, 100 V input capacitor had to be installed horizontally. The 470 µF and 100 µF, 16 V capacitors were still too tall, sticking up about 2 mm too far, so the display board rests on them.


The regulators are installed underneath, parallel to the board. Be careful when installing them to put each one in the right place, since they have different polarities.


There isn't much left to do now; the next step is to figure out how to connect the display. By the way, there are another couple of contacts in the same row as the display, for connecting the power indicator LED. If you aren't using it, install a 16-pin header.
I installed a female header on the board and a male header on the display, but at this stage it's best not to solder anything to the display yet.
There's also a single pin, which is apparently used for monitoring, although I didn't end up using it.


Now it's time to breathe some life into the device by programming the microcontroller.
It uses the classic Atmega8A.


I had no trouble with the firmware files—I’ll post them at the end of the review—but I got confused by the configuration bits.
At first, I just programmed it with the default settings. The board even came to life, but there were a few problems. Since I'm not a programmer, I contacted one of the readers who had modified the electronic load's firmware. He told me what settings to use, and I owe him a big thank-you for that. After that, everything worked properly, although I did find one quirk, which I'll get to later.


So, I flashed the firmware and installed the microcontroller, but I hadn't soldered the connector to the display yet. I just inserted its pins into the holes, with the display facing the opposite way.
That's because the contrast adjustment resistor is underneath the display, which would make it very awkward to adjust later. So I adjusted everything first and only then soldered the display in place.


Let's check: everything works. When powered on, the splash screen appears, showing the program's developer.

Interestingly, the preset current is above 10 A (top right), even though the board can handle only 10 A. Turning the encoder brings it back to 10.000, and it can't be increased any further.

Description of the main screen:
Left column
Set voltage
Output voltage
Output power
Load resistance.

Right column
Set current
Output current
Input voltage
Input power.

There's a small square in the middle that flashes when the device is switched on. If you turn one of the encoders in this mode, the screen switches to a second mode, where the two bottom lines are simply duplicated at the top.
There's also firmware for a 2004 display, but it's essentially the same, except that the two vertical columns of character cells on the left and right are left blank. Apparently someone asked the author to make firmware for the display they had.

The display shows some strange readings because, at this point, only +5 V is being supplied.


Here's a difference due to the way I plan to use the board. It has two positions for 2SD667 transistors, but since in my case the board will control a ready-made power module, I don't need the second transistor.
I used a 2SC945 instead, and connected the output to the power module through a 24 Ω resistor.

Warning! The 2SC945 comes in two pinout variants: the familiar C-B-E and the less common E-C-B. The device is supposed to use a 2SD667, which also has an E-C-B pinout. I used a C945 with an E-C-B pinout; if yours is different, you'll have to do a bit of fiddling.


The final assembly step is installing the connectors.
For the input and output, I used DEGSON DG128-5.0-02P-14-00A(H) connectors, rated for 10 A. Still, it would be better to use something more robust here, like the connectors on the boards I showed earlier.
The other connectors are more ordinary: power is supplied to the board through a three-pin NX2500-03SMS connector, and the encoders connect via a standard pin header.


Since I wasn't planning to solder anything else, I cleaned the board with acetone and coated it with protective varnish.
The connectors are installed on the back because the display gets in the way on top.




The finished control module. The display is held in place by the connector at the top, and I installed standoffs at the bottom.


The controls are rotary encoders with push buttons. I ordered five of them before Chinese New Year, and they were even shipped in time for me to receive them at the end of February. The seller only shipped the display after the holidays, though; that was the only thing I needed to finish the review. I could have bought one locally, of course, but that would usually cost more. Besides, it would most likely have been a standard Winstar display, whose backlight draws a lot of power.
During testing, I tried an OLED display, but for some reason it caused the controller firmware to crash...


There’s a guide to connecting the encoders on a Chinese forum.


But all I really needed to know was which wire was ground and which was for the button. If necessary, I could have resoldered a couple of the encoder contacts.
As it turned out, everything matched up right away. It was immediately clear which encoder controlled current and which controlled voltage.
You can see in the photo that each encoder has four wires.
Voltage
Black — ground
Green and yellow — encoder
Blue — button

Current
Black — ground
Green and yellow — encoder
Red — button

The encoder’s center pin is connected to one of the button terminals.


All that’s left is to connect everything to the power module. I won’t go through every detail, since the complete wiring for the power module is shown in my recent review of the 60 V, 20 A kit. In this case, I used only the power module and supplied it from an adjustable power supply.
Essentially, the power module is a high-power composite transistor, like the well-known KT827/825, for example.


That’s why I didn’t install a second transistor: there’s already one in the power module, so I simply connected its base and emitter with a resistor.
By the way, you can use the board in its standard configuration: install a second driver transistor and solder the 2SC5200 specified in the schematic, or an equivalent, in place of the power transistor. But I have some doubts about how reliable this solution is, since the driver transistor is very low-power. For example, the power boards shown here even have it mounted on a heatsink.
I’d recommend doing what I did above, but using the KT827 I mentioned instead of the power module. In that case, though, you’ll need to reduce the power supply’s output power.

Anyway, I connected the module as shown in the photo and in the schematic above.


I connected the power input to the left terminal and the load to the right one. The wire with the white stripe is positive.

The board was also powered from a single 9 V winding, although it should be powered from two.
There’s another important point here! The power connector pinout on the control board shown earlier is 15-0-15 V. The board under review uses the same connector, but its pinout is 0-9-9 V. It’s best not to supply more than 10 V, or the voltage regulators will get noticeably hot.


Power it on and—hooray!—everything works and nothing exploded :)
Later, I noticed something odd. The power supply is controlled as follows:
By default, the center digit blinks. The board also switches to this mode after 10 seconds of inactivity.
Pressing an encoder activates voltage or current setting mode; subsequent presses select the adjustment step size.
If adjustment of one parameter is active, press the corresponding encoder to adjust the other one.
In other words, you can’t accidentally change the second parameter without switching to it.

The selected digit is supposed to blink to show which parameter is being adjusted. But it didn’t always blink—or sometimes it blinked very faintly. I later figured out that it blinks only when the power supply can deliver the set values. For example, with an input voltage of 25 V, it blinks if you set the output to 24–25 V, but not if you set it to 25 V or higher.
A rather unusual indicator, then.


Let’s connect a 45 W automotive bulb and set the voltage to 12 V. We can see that the current is about 3.8 A, the lamp is drawing 45 W, and the total power is almost 75 W. That means about 30 W is being dissipated by the transistors.


As you may have gathered from the review, there are no adjustment components on the board, apart from the trimmer for adjusting the display contrast.
All adjustments are made through a separate calibration menu. To access it, power on the unit while holding down the voltage encoder.
This brings up a menu with six items:
Voltage setting adjustment
Input voltage measurement adjustment.
Current setting adjustment
Load parameters
Save parameters.
Exit.

As I understand it, the load and save parameter options are useful if you make calibration changes and then want to restore the previous settings.
For example, sometimes I calibrated the electronic load so that it set the load current to twice the value actually shown on the screen.



More about calibration.
All operations are performed using a single encoder (voltage), with three options: turn, short press, and long press.
From here on, I'll use L for a long press and S for a short press.

First, calibrate the output voltage.
1. Long-press the encoder (L) to enter the first stage. Set the lower reference voltage here; the default is 1 V.
2. Long-press (L) to move the cursor to the constant setting. Turn the encoder and switch between digits with a short press (S) until the voltmeter connected to the power supply output reads the value set at the start (1 V).
3. Long-press (L) to select the upper voltage (19 V by default), then long-press (L) again and adjust the constant until the output matches the set value (19 V in this case).
4. Long-press (L) and save the settings (L).


Voltage measurement calibration is a little different. You need to apply some voltage to the power supply input; the default is 19 V. But don't just apply it—you need to know exactly how much voltage you're applying.
Then set the same voltage as the voltmeter reading, 19.06 V in this case. The controller will now know the actual input voltage.


Current calibration. It's basically the same as calibrating the output voltage, but the multimeter is set to measure current and connected in series with the power supply output terminals.
Also, the default calibration points are 0.1 and 1 A. I'd recommend calibrating at 0.1 A and 4 or 5 A instead.


Now let's see how it turned out.
First, voltage setting across the range from 1 to 47 V; the last value was chosen arbitrarily.
The voltage setting is reasonably accurate, except perhaps at 5 V, but the built-in voltmeter sometimes gives the wrong reading.


The current readings show roughly the same pattern. At low currents, though, the built-in ammeter is way off: it read 147 instead of 100 mA, and 329 instead of 500, even though the current setting itself is quite accurate.


And here's the defect I mentioned near the very beginning of the review. A section of the PCB trace is in series with the shunt, and at high currents the output current starts to drift downward. The thing is, copper has a positive temperature coefficient of resistance, so as the trace heats up, its resistance increases. The power supply then thinks the current is rising because the voltage across the shunt is increasing. The op-amp compensates for this, and as a result the output current drops.
The fix is either to reinforce the traces with thick wire or change the shunt connection point. The first option is probably even easier.
By the way, this effect is also slightly noticeable when adjusting the voltage, but it's much less pronounced.


I found a lot of information about this power supply on a Chinese forum. For example, there a user tested a similar unit on their own power supply. I only included four photos, since in the others he was checking how the result changed when the least significant digit changed.
Also, another user posted some pretty good results, so there's something to aim for.


They also evaluated the response to a short circuit and overload.


They also measured the output noise, although it's not clear what power supply was powering this board.


There are examples of finished devices too, but they caught my interest for a slightly different reason.


They use a different board version as well.



And, as I understand it, this one uses PWM plus a linear regulator, though I could be wrong. I just can't explain the presence of a fairly powerful DC-DC converter any other way.
Link



But it also uses a different display. In principle, this design could be replicated, but I couldn't find the firmware for it. Judging by the comments on their forum, I'm not the only one with this problem, since people ask about it often.
This board can also measure heatsink temperatures and control fans.
I found a full description of both the board and its control system, but it's in Chinese (Chrome translated it), and once again there's no firmware :(


By the way, the well-known modder 100MHz is also active there, offering his own designs, this time based on well-known converters.


And there are also some extreme enthusiasts building a power supply based on the STM8S103F3P6 with ratings of 60 V and 50 A, and open-source code.


Video version of the review


A few brief conclusions.
As for the seller and the product, my only significant criticism is the missing 0.02 Ω resistor, as well as the lack of a warning about the incorrect pinout marking for the regulator's power transistor.

As for the device itself, it works, although it has quite a few drawbacks:
1. There's no output-disable function.
2. Some issues with the PCB layout.
3. The measured values displayed on screen refresh too frequently, which is annoying.
4. Calibration is easy, but you don't always get a good result. Thinking back to my electronic load, I can say that you can perform the same calibration twice and get different results, ranging from not very good to almost perfect.

Its advantages include:
1. The device is fairly easy to replicate.
2. The regulation is essentially all analog; the microcontroller only sets the reference voltage, while the feedback loop is built around an op-amp.
3. It comes with not only the firmware but also the source code, so those "in the know" can modify the program to suit their needs. I already have a few ideas of my own, for example.
4. The controller circuit contains no hard-to-find or expensive components.
5. Combined with the relay board I mentioned earlier, it can be made into a device that won't dissipate the full 500 W across the heatsinks if the output is shorted.
6. The board in this review differs from the one shown on the seller's page, and for the better: the seller shows a very old version, which is great to see :)

As a postscript: you probably thought this review would wrap up the series on high-power linear power supplies, but that's not the case. Unfortunately (or fortunately), this is only the beginning. I still need to choose a transformer and capacitors, order an enclosure and heatsinks (there are some challenges here), make the whole thing reproducible, possibly improve the software, add "smart" fan control, and so on. That’s why this review is more of a conversation starter on an interesting topic, or a source of ideas for further development.

And of course, an update:
Two power supply circuit options
Firmware
Source files
PCB for the reviewed version in PNG format
PCB for the expanded version in lay6 format
PCB for the latest electronic load version in lay6 format

It turns out that downloading PCB files in lay6 format requires not only a forum account but also virtual coins. Luckily, I’d already earned some in my forum threads :)

This review was sponsored by the purchasing agent yoybuy.com, which covered the shipping costs. But I would have written this review anyway, since linear power supplies have really caught my interest.
$3.17
Go to store
This page was found when searching for:
схема блока питания на 5 вольт 10 ампер на gan транзисторах, схема регулируемого бп 50в 5а, схема лбп 50в, схема мощного лбп 50в, схема мощного бп 50в, регулируемый блок питания на 50 вольт, схема dps5020, конструктор линейного лбп 50 вольт, 1, блок питания 50вольт 10ампер александр радиомастер, простой и доступный блок питания 0...50в печатные платы, схема эллектронного блока управлениря блоком питания, бп о-50в о-10а схема, 13, kaga components kic-125 электрическая схема, платы линейных блоков питания с алиэкспресс, линейный блок на 50в, https://www.kirich.blog/obzory/bloki-pitaniya/566-plata-dlya-cifrovogo-lineynogo-bp-50-volt-10-amper-ot-gandf, линейный лабораторный блок питания до 10 а 50в, схема трансформаторного блока питания на 50в /10а, kic 125, 11, линейный блок питания с цыфровым управлением, бп до 50вольт схема, линейный б.п 50в, плата для линейного блока питания, блок питания 50 Вольт, технический обзор, цифровое управление блока питания

Questions and answers

What are the maximum ratings of the linear power supply built around the Gandf board?

It is designed for a maximum voltage of 50 V and a current of 10 A.

Why did the author set aside the Gandf board’s included shunt?

The wire shunt measures about 90 mOhm, while the schematic and board specify 20 mOhm. At 0.1 ohm, it could dissipate up to 10 W.

What project files does the Gandf board review provide?

It includes two schematic variants, firmware, source code, and printed-circuit-board files.

The figures from this review are in the measurement database: Converters

Calculate manuallyAll calculators

Discounted products


You may be interested in


Discounted products


Support the project

All the content on the site and the advice are free, but we’d be grateful for your support of the project and the channel!

Discuss on the forumHave a question or some experience to share? Discuss the article with the author and readers: the forum is a convenient place to share schematics and photos.
Go to the forum

Only registered readers can comment. It takes a minute to sign up, and you can log in with the same account you use on the forum.

Log inRegister