GOPHERT CPS-3205 bench power supply, or that rare case....

Sometimes I feel like no matter what I review, I always end up reviewing a power supply.
But in this case, reviewing a power supply was the plan from the start. I’d seen this one before, so I was even more interested in getting my hands on it and sharing my impressions with you.
Tests, teardown, analysis—the usual, all in the usual place :)

Sometimes you might think I leave overly positive reviews of products. That’s partly not because I’m trying to praise the product, but because I always make a point of choosing products for reviews carefully, and sometimes I ask the store to add a product to its range myself.
This product is one of the ones I asked them to add, so naturally I decided to review it :)
But that’s enough of a digression. Let’s get down to business: taking a look, testing it, and all the other interesting stuff.

As usual, I’ll tuck the unboxing under a spoiler, so it’s there for anyone who wants it without getting in everyone else’s way :)

Here’s everything that comes in the box. Bench power supplies don’t usually come with a pile of accessories, and this one includes everything you need.
Power cord
Load connection cable
Manual
Power supply


The manual is in English. It’s not packed with information, but it does cover the essentials. That said, people who buy this kind of thing usually know what it is, and often only need the manual at first, as a quick reference.


This power supply comes in a bunch of versions (there should be about six in total).
There are three versions with different output voltages and maximum currents, plus two versions with either a three- or four-digit current display.
Mine is the CPS-3205, with a maximum output voltage of 32 V and a maximum current of 5 A.
In the manual, that’s the middle column in the specifications table. A reminder: all the photos in this review are clickable.


The package also includes a cable for connecting a load. Honestly, the cable is junk. When I got this power supply, naturally I wanted to try it out right away :)
For a load, I used a 12 V, 45 W car bulb. Since it was just sitting on the table, I didn’t leave it on for long—I’ll need that table again :) But when I was done playing around, the wire was only a little cooler than the bulb. To be fair, I hadn’t unwound it; I tested it as it was.
The cable’s cross-section is far too small, so you’d be better off replacing it with something decent right away.
The connectors and alligator clips are perfectly usable, though, so you only need to replace the cable itself.


I had no complaints about the power cable, apart from one small thing: I was accidentally sent (or accidentally ordered) a cable with a US plug :(
The good thing is that a standard computer power cable fits, so it wasn't a problem. I just grabbed the first one I found in the drawer.


All right, I realize nobody is interested in looking at cables, so let's move on to the power supply.
The first thing that stands out is its compact size. I think everyone understands that a power supply with this much power in such a small package is usually a switching type.
Switching power supplies have their advantages and disadvantages. For a bench power supply, the advantages are its compact size despite its high power output, and the fact that this unit uses passive cooling (but we'll come back to that during testing).
The usual drawbacks are higher ripple and a large output capacitance, but we'll come back to those as well.
The power supply measures 192x120x58 mm, including all protruding parts.


The front panel layout is quite convenient, though not without its drawbacks. I wasn't too keen on the output terminals and power switch being on the back. You can get used to having the output terminals there, but a rear-mounted power switch is rather inconvenient.
Along with the controls, the front panel has indicators for the set and actual output voltage and current. This model has a four-digit voltage display and a three-digit current display.
There is a model with a four-digit current display. It's a little more convenient, but also more expensive.
On this model, currents below 1 A are displayed as xxx mA, while currents above 1 A are shown as x.xx A. The decimal point on the right indicates when the display switches to the lower range.


The front panel also has indicators for:
Constant-voltage mode
Constant-current mode. This power supply doesn't disconnect the load when the set current is exceeded; instead, it switches to current limiting at the set value.
Control lock mode.

Current and voltage are set using the encoder. Turn it to decrease or increase the value; press it to select the adjustment step.
There is only one encoder, so a selector switch lets you choose whether to adjust current or voltage.
Two buttons are located at the bottom right: one locks the controls, and the other enables the output terminals.


The back is simple and understated: two output terminals, an input voltage selector, a power switch, and a power inlet.
Pay close attention to the position of the input voltage selector. If your mains supply is 230 V and you set the selector to 110 V, the consequences will be unpleasant.
The output terminals are perfectly adequate; they're more than enough for 5 A.


The entire case is made of anodized aluminum and serves as a large heatsink.
This design not only provides decent heat dissipation, but is also quite rigid; the power supply feels like a solid block.


There are no cooling fins on the bottom, but there are four small feet. They reminded me a bit of old Soviet checkers: they look the same and are just as slippery. I’d recommend sticking on some pieces of rubber.
The feet are made of either soft plastic or hard rubber. You can press them in a little with a fingernail, but they still slide on a smooth table.


When switched on, the power supply starts in the Off state; it doesn’t automatically apply voltage to the output. That’s a good thing in its own way, but it would be even better if you could choose whether it turns on after power is applied.
Correction: it turns out you can choose whether or not to apply voltage to the output after switching on.
To do this, hold down the on/off button and turn the encoder until dON appears (it doesn’t show up immediately). To switch back, do the same until dOF appears.


You can set it slightly above the stated specs: 32.3 V and 5.1 A.
When you select the adjustment step, the corresponding digit on the display lights up more brightly. So, if you’re adjusting tenths of a volt, the rightmost digit lights up.
Once you stop adjusting, the display lights evenly and switches to showing the current output voltage. If no voltage is being applied to the output, it shows 00.00.
During testing, I noticed a slight parasitic glow on one digit from its neighbor; you can see it in the photo. It’s not a big deal, but it’s there.


First, I decided to check how accurately the power supply’s voltmeter displays the output voltage. I tested it across the entire range, from 1 V to 32.30 V, in 5 V increments.
The only issues were at the ends of the range, at 1 V and 32.30 V. At all the other test points, the voltage was displayed very accurately.




For each measurement, I set the corresponding voltage, after which the power supply switched to displaying the output voltage. To keep it from running with no load at all (a difficult operating condition for a switching power supply), I connected a small “load” across the output: a 1 kΩ resistor. By the way, the photo also shows how current measurement works at low values: the output current in mA is equal to the output voltage in volts.
The resistor got a little scorched during the measurements because 32 V is a bit much for it.


After checking the accuracy of the output voltage display, I wanted to test what is probably the most critical parameter for a switching power supply: output voltage ripple.
I decided to test it at the maximum rated current right away—5 A—and at various voltages, from 5 V up to the maximum.
The oscilloscope probe was set to 1:1, the scale to 50 mV per division, and the input coupling to AC.
An electronic load was used as the load.
This is what the setup looked like.


At low voltages, I didn't notice any significant ripple at all. It was only at 10 V that the waveform started to show signs of life.
The photo also shows how well the power supply holds its voltage at almost the maximum load current (the voltmeter on the left).


Increasing the output voltage to 15 and 20 V made practically no difference; the ripple remained at exactly the same level as at 10 V.


I tried 25 and 30 V, and got the impression that I wasn't changing anything at all. I even started checking the oscilloscope connection, but everything was connected properly.


Finally, I set the output to almost the maximum the power supply can deliver: 32.3 V and 5.08 A. I didn't set it any higher because the electronic load regulates the current, and when the power supply switched to constant-current mode, its output voltage dropped to zero and the load shut off automatically. This isn't a fault; it's simply how this combination behaves.


But I felt like I was taking the same screenshot over and over. Good thing the waveforms were numbered, or I would definitely have gotten confused.
Not only is there absolutely no difference in ripple between 10 V and 32.3 V, but the ripple level is also very respectable. The specified ripple is 30 mV peak-to-peak, while the power supply actually delivers around 10 mV—an excellent result, three times better than the stated specification.


After checking the output voltage reading and its ripple level, I moved on to another parameter: the accuracy of the output current reading.
At low currents, the measurement accuracy leaves a little to be desired. This is because the unit's ADC has limited resolution, so at low currents the reading starts changing in steps of 3–4 least significant digits rather than one.
You can see what this looks like in the photo below.


At higher load currents, everything returns to normal, and above 1 A the readings are impressively accurate.


The readings match digit for digit all the way up to the maximum output current.
The tests were performed at an output voltage of 15 V, chosen as roughly halfway between the minimum and maximum.


I once saw someone ask in the comments: why have so many measuring instruments?
The photo below is a good example of several instruments being used at the same time.


I was just as curious about the second part of getting to know this power supply, perhaps even more so than the first. Of course, it's the teardown.
I'd already seen photos of the internals, but seeing them is one thing; getting to handle everything yourself is another.
The front panel is secured with decorative hex-socket screws, while the rear panel uses ordinary Phillips-head screws.


At first I picked up a hex bit and got ready to unscrew the front panel, but for a partial teardown, it's actually easier to remove the rear panel.


Remove the four screws securing the rear panel, then slide the bottom toward the rear panel.


Behold, a marvel of Chinese engineering. Although experience shows that many devices like this are clones of products from better-known manufacturers, I won't make any claims about this particular unit.
First impression: neat. Second impression: very neat!
The board doesn't take up all the space inside; there's even some room to spare, which is good.


There are quite a lot of photos of the internals coming up. Some may think they're unnecessary, but honestly, I already cut the number down as much as I could while preparing this article. When something is put together this well, though, you want to show it from every angle. Consider it a little weakness of mine :)
On the other hand, I wanted to show it the way I'd want to see a review of a device like this myself.


You can clearly see the large number of transformers and inductors.
The device has seven wound components in total. Here's the breakdown:
Two input inductors
Electronics power transformer
Isolation transformer for driving the high-voltage transistors
Power transformer
Inverter output inductor
Output inductor for suppressing interference.


The build quality is good; I'd even say it's almost on par with industrial equipment.


Of course, there are a few minor flaws. The first thing I noticed was that the wire to the input voltage range switch runs dangerously close to the output current measurement shunt. We'll deal with that later. Otherwise, my impressions so far are entirely positive; the layout is quite spacious.
By the way, I measured the voltage drop across the shunt at 5 A: exactly 75 mV, which gives us a shunt resistance of 15 mΩ.


The input uses an assembly of four capacitors: two 100 µF and two 180 µF. Interestingly, one pair is made by Capxon and the other by Samxon. I wouldn't call them top-of-the-line, but they're certainly not no-name capacitors either—solid mid-range parts.
The developers chose this combination of capacitors because they wanted to fit everything into a small enclosure, with the enclosure height being the key constraint.
Higher-capacitance capacitors are usually taller, too. I recently tried to find low-profile capacitors with a high capacitance, something like 680 µF, 200 V, but haven't found any yet.
Still, the developers could easily have used four 180 µF capacitors; that would only have been better.
The capacitors have their rated capacitance, giving a combined equivalent of 140 µF (100+180)/2.
That capacitance is quite sufficient for an output power of 160 W.
By the way, the power supply's own power consumption is very low. After the input power is disconnected and the load is removed, it keeps running for about 10 seconds on the capacitors.


The output uses a similar mix of manufacturers: a 2200 µF Capxon and a 1000 µF Samxon.
What disappointed me is that the capacitors are rated for 35 V. In my opinion, that's not enough; 50 V capacitors would have been a better choice.
The only thing I can say in the developers' defense is that power supplies built using the topology reviewed here (this is a half-bridge power supply) have lower voltage spikes, and the power supply doesn't always operate at its maximum output voltage.
The photo also shows an EMI suppression inductor.


Photos of individual sections.
The high-voltage transistors have insulated cases, and there's also an electrically insulating, thermally conductive pad between the cases and the transistors. The bridge rectifier is screwed to the heatsink as well, although it didn't really need to be.
The output diode assemblies are fitted with special insulators that cover the entire case.
The device's enclosure is connected to the ground contact on the circuit board and, in turn, to the ground contact on the power connector.
What's more, when I took it apart further, I saw that they hadn't forgotten to fit a special washer either. And not only did they fit one, they put it in the right place—between the tab and the enclosure!

Some might ask what's so special about having a washer or an extra insulator. My answer is that any engineer will tell you this kind of attention to detail reflects the quality of the manufacturing process. When manufacturers start cutting corners, it's usually on little things like these: deciding an insulator is unnecessary, leaving out a washer, or fitting it incorrectly.
So, even from a visual inspection, I can tell that production quality control is in place and working.


After undoing all the screws holding the components against the enclosure, I was able to remove the circuit board (it comes out in much the same way as the bottom panel).


Now we can take a closer look at the details.
But first, I'll fix one of the minor design flaws: the wire running near the shunt.
As I mentioned above, a wire to the input voltage range selector runs near the shunt.
Where I live, this switch is not only useless but can sometimes be dangerous. Even accidentally switching it can burn out the power supply in seconds, so I simply remove this switch from the circuit.
This modification isn't suitable for people in countries where the mains voltage is 100–120 V; they'll need to insulate the wire.


Modifying the power supply is very simple. First, find where the wires from the switch are connected.


And we simply desolder them :)
That's it, problem solved at the global level: no wire, no problem.


The input filter deserves special attention.
I've reviewed quite a few different power supplies, but in this case the input filter is built exactly as it should be and could serve as a reference.


There's a fuse at the input, followed by a varistor to protect against excessive input voltage and surges. The varistor is rated for 470 V, which is standard for varistors in this circuit. 430 V versions are also available, but they're less common.
1. One possible criticism is that the varistor won't work with a 100–120 V mains supply, and it would be more appropriate to use two, connected in parallel with the high-voltage electrolytic capacitors. But our mains voltage is 230 V, so the installed configuration is more than adequate.
2. An X2-class MKP capacitor is installed at the input, with a small two-winding inductor next to it.
3. The device uses proper Y1-class capacitors. That's something I usually criticize in cheap power supplies. Everything is excellent here.
4. The power filter uses two two-winding inductors; you can also see a thermistor to limit the capacitor charging current.
The manufacturer deserves a big plus for this power filter. A single filtering stage is often used, but here we have a proper filter with two two-winding inductors. I've only seen this in branded equipment.


Input power filter circuit.


1. The inverter uses high-voltage MOSFETs f13nk50z with a built-in Zener diode that protects the gate from voltage spikes.
2. The input has a diode bridge GBU407, rated for 4 A and 700 V (1000 V peak).
3. The output diode assemblies are MUR1640CT, rated for 16 A and 400 V. They’re connected in a slightly unusual, but perfectly correct way. The key is that they aren’t paralleled, as is usually the case: one assembly serves as one rectifier diode, and the other as the second. This arrangement improves diode performance because the diodes paralleled together are inside the same component, on the same die.
4. They even tried to impregnate the wound components with varnish. Maybe they didn’t actually impregnate them, but they definitely tried to coat them with varnish.


The output rectifier and filter circuit. Note the correct arrangement of the interference filter and current-sense shunt.


1. The PWM controller is a classic TL494. The designers took the usual, sensible approach and used a hardware PWM controller, which is controlled by the processor. This provides good voltage and current regulation accuracy, as well as a faster response to changes in load.
The PWM controller is on the secondary side and is powered by a separate power supply.
2. All the power supply’s electronics are powered by a separate supply built around a TNY274.
I was glad to see that the designers used readily available components that you can buy locally, rather than Chinese equivalents that you can only get on eBay or AliExpress.


I didn't draw the schematic, as that would have taken an inordinate amount of time, but I did sketch a rough block diagram of the device.


I had a few comments about the circuit board. Flux left behind after some of the capacitors were installed hadn't been cleaned off.
That surprised me, especially given how neatly everything else was assembled. I wouldn't call it critical, but it's unpleasant.
Maybe the process-control inspector stepped out for a smoke just as my board was being soldered. :(
A general view of the board.


Since the general photo doesn't show the markings on all the components, I'll show a few close-ups of individual sections next. Maybe they'll help someone with a repair.
Since the inverter uses MOSFETs, while the circuit was originally designed to drive bipolar transistors, the engineers made a few changes to the drive circuit, adding small transistors. As I understand it, these are there to ensure the MOSFETs turn off reliably.


This power supply uses an isolation transformer that differs from the standard ATX power supply type, although the circuit is partly similar. A standard transformer for use with bipolar transistors has more windings: two primary and three secondary. Here, it's a straightforward isolation transformer.
The primary winding is connected to a bridge of four small transistors driven by a TL494. The transformer has two output windings to drive the upper and lower high-voltage transistors.
This is all because the inverter uses MOSFETs, and it also makes it easier for the inverter to operate over a wide PWM duty-cycle range, right down to fully off.


The TL494 support circuitry (right) and some of the high-voltage circuit snubber components (right).


The secondary output power section (top) and the shunt signal amplifier section.
The power supply for the internal electronics provides two voltages. One powers the PWM controller, and the other powers the control and display electronics.


Oddly enough, the manufacturer coated this section with lacquer, so I had to clean it off the IC to find out what they'd used here.
It's a plain old LM358. Its specs are far from ideal, but perfectly adequate for this application. It could be replaced, but I don't think that would improve anything.


After inspecting the board, I cleaned off some of the flux and coated it with a layer of lacquer.
Partly to restore the coating where I'd removed it, and partly just to give it some extra protection overall.


That’s it for the power board; let’s move on to the control board.
It’s mounted on the front panel. Some of the flux hasn’t been cleaned off either (the controller must have been stepping out for smoke breaks a lot—it’s time to quit), but that’s not critical here.


Remove the three screws holding the control board to the front panel.


Well... they could have installed the display a bit straighter. I noticed it right away, which is why I took the whole thing apart partly to straighten it. Fortunately, getting to the display is easy: just four screws on the front panel and three screws securing the circuit board.
Most people probably wouldn’t even notice, and it definitely has no effect whatsoever on the power supply’s operation or specs, but as they say, “it leaves a bad taste in your mouth.”
I didn’t spot any other issues; everything looks pretty neat.




1. An 8-bit microcontroller, STM8S105K4T6, serves as the “brain.”
2. It drives the display through a pair of 74HC595 shift registers, which boost the current and expand the number of outputs.
3. This whole “contraption” is powered by a 3.3 V linear regulator.
4. There’s another op-amp on this board, also an LM358.


I also spotted a temperature sensor on the board. According to the manual, the unit has overtemperature protection, but I couldn’t get it to overheat under normal operating conditions.


That’s the front panel inspected; nothing else interesting turned up :)
I remembered the rear panel just as I was about to put everything back together.
There are a few components on the rear panel too :)
But one essential and useful component is missing—the output fuse. If you connect a battery to the power supply with the polarity reversed, at the very least the output wires will burn out. It’s best to add a fuse rated for 6–7 A.


A small circuit board is mounted right on the output terminals, with one of the output capacitors on it.
The second circuit board is soldered to the power switch contacts. What’s noteworthy is that they not only added a separate small board, but the switch also disconnects both mains wires, which is the right way to do it, especially with switching power supplies.


I also measured the actual capacitance of the input filter capacitors.
They weren't lying; it all adds up. Above, I explained why the capacitance should be 140 µF. Here's the calculation again: 100 µF + 180 µF = 280 µF with the capacitors connected in parallel.
Since the capacitors are rated for 200 V, they are also connected in series, giving 280/2 = 140 µF.


After assembling everything, I decided to check how hot the components inside the power supply got.
Since I'd already measured the other parameters, there were far fewer “participants” in this test.
Power supply
Electronic load
Non-contact thermometer
Pen and paper.


The test ran for 1 hour and 40 minutes. I checked the temperature every 20 minutes, increasing the current in 1.25 A increments.
At the end, I ran an additional 20-minute test at maximum current and voltage.


I put all the measurement results into a table.
Don't be surprised that the diodes started out a little warmer: the power supply was sitting on the table, with a desk lamp fitted with a halogen bulb above it, which warmed that side of the case a bit.
Overall, the results are pretty good, but the additional test at maximum current showed that the transformer could overheat when the ambient temperature is high. For example, if it reaches 88 °C at 25 °C ambient, it will be 98 °C at 35 °C, which is close to the critical limit. I wouldn't recommend running the power supply at maximum power for extended periods in high ambient temperatures.
You could improve things a little by placing a piece of thermally conductive rubber between the bottom panel and the transformer, but sliding the bottom panel back into place with it there would be very difficult.
All these issues stem from the power supply's passive cooling—as they say, it's a double-edged sword.
The temperatures of the other components were within acceptable limits. Under maximum load, the case heated up to about 55 °C. You can hold it, but it's uncomfortable.
I also checked the temperature of the PWM controller powering the electronics (TNY274) at the end; it was 67 °C, which is perfectly fine.


As one possible use, I decided to charge a battery. Obviously, you can use a bench power supply as a charger, although it's not exactly the right way to do it :)
I connected the battery and.... got an output overvoltage protection trip. The output was set to 10 V, while the battery voltage was noticeably higher, so the power supply treated it as a fault. I hadn't even turned the output on.
To clear the error, disconnect the battery, toggle the current/voltage selector switch, and set the output voltage higher than the battery voltage.


That's it, charging is underway. The charging current is set to 1 A, and the power supply is operating in constant-current mode.
By the way, I tried shorting the terminals (without the battery, of course) with the output on. There was a quiet click, and the power supply switched to constant-current mode.


If you increase the current, the voltage at the battery terminals rises to the set limit (the battery is getting old), and the power supply switches to constant-voltage mode.
You can press the control lock button and leave the battery to charge.


A few comments on the controls.
I used this power supply in normal operation for about a week and a half, so I can give you a pretty good idea of what it’s like to use.
Power switch — convenient if the PSU is on a desk, but awkward if it’s on a shelf.
Output terminals — not really an issue; you connect them and forget about them.
Heat — when powering various devices under normal conditions, it got only slightly warm. It got hottest during testing.
Display — it looks worse in the photos than it does in person. The LED indicators are easier to read, though I’d add a tinted filter.
The controls are a whole topic in themselves, so I’ll try to describe them.
At first, I found the control and display logic a little unfamiliar. The PSU has only two indicators, when ideally it should have four or five (the fifth would show power).
The adjustment controls are both convenient and inconvenient.
They’re convenient because in practice you usually adjust only one parameter, so one encoder is enough. Switching between modes is easy, too.
The downside is that if the PSU exits adjustment mode, you have to turn the encoder one extra click to get back (to switch to displaying the set value); the next click adjusts it.
Also, while you’re adjusting, both indicators switch to displaying the set values. But there’s no getting around that—it’s the trade-off for keeping things simple.
The most inconvenient thing, though, was that after switching back to displaying the output parameters, trying to adjust them again puts the PSU into adjustment mode at the finest step size.
For example, I was adjusting in tenths of a volt, then let go of the knob and the PSU returned to normal mode. If I then wanted to increase or decrease the voltage further, I’d be adjusting in hundredths of a volt—or I’d have to press the encoder. This annoyed me, though I could get used to it. I think the designers could easily have made it remember the last adjustment mode; that would be more convenient.

So, to sum up.
Pros
Quite good build quality. There are a few minor issues, such as poorly cleaned flux and a crookedly installed display, but they don’t affect reliability or performance.
Well-thought-out design overall.
Quality components
Good circuit design
Very low ripple
The PSU delivers excellent performance across its entire output voltage and current range
One of the few PSUs where the manufacturer didn’t cut corners—or hardly any corners, anyway :)

Cons
Power switch on the rear panel
No output fuse
A few minor build flaws, as mentioned above.
The controls can be a little awkward, though that may just be a matter of getting used to them.
The included output wires are thin; definitely replace them.
There’s a large capacitor on the output, so you shouldn’t connect a device under test to the PSU if it already has voltage on its input (for example, low-power LEDs and similar devices). This is a drawback of almost all switching PSUs. Connect the load first, then turn on the output.

My verdict: a solid, well-made power supply with good technical specifications.
You could certainly criticize it for the flux and so on, but those are minor issues compared with the fact that the PSU is genuinely well designed. The output ripple was impressively low—I wasn’t expecting that.
Current and voltage measurements are also quite accurate. I have a few minor reservations about voltage accuracy at the ends of the range, and about current accuracy only at the low end, but in my opinion it’s quite good for a device in this class.
Overall, I liked this PSU. It’s a genuinely interesting, high-quality power supply for electronics hobbyists. Of course, I could find faults with anything, but here the pros clearly outweigh the cons.
All in all, this is one of those rare cases... where they got almost everything right and made a device that performs better than advertised.

I hope this review was useful and helps you make the right choice.
If you have any questions or additions, I’ll be happy to answer or try out some more tests.

The product was provided for review by the store.