YS-65SCB power supply, 24 V, 65 W

Another power supply I was asked to test. I was interested in it myself, so I ordered one for a review.
As usual: inspection, schematic, tests, conclusions.


In a way, this power supply competes with the “people’s choice” model, although it differs in both power and price.

It comes in a cardboard box with some store stickers on it.


According to the datasheet, there are seven versions of this power supply, with different output voltages and corresponding maximum load currents. All versions except the 5 V model are rated at 65 W; the 5 V version is rated at just 50 W.
This review covers the 24 V version, rated for a load current of up to 2.7 (3) A.


It looks neat. At least in my opinion, it looks better than the “people’s choice” series. The input uses a two-pin connector, and the output a four-pin one, although the board has space for six pins..


The usual potting compound is missing from the board, which is not ideal.


For comparison, I put a Sanmim next to it, which I’ve already reviewed, and the 24 V people’s choice model. The power supplies are almost the same size.


The manufacturer’s datasheet includes the full dimensions.


A little about the circuit design and components.
1. A proper mains filter at the input, and an input capacitor rated at 100 µF.
2. A 3.15 A fuse, thermistor, and varistor.
3. The power supply has overtemperature protection; the thermistor is pressed against the high-voltage transistor’s heatsink.
4. The output has four 470 µF, 35 V capacitors and an inductor to reduce ripple.


The capacitors connecting the output common to the mounting-hole pads are installed in an unusual way. There are two capacitors, one for each pad, so the total capacitance depends on how many contact points are connected.
For some reason, at least one of the pads isn’t tinned, so the connection may be unreliable. The primary side has the same arrangement, but the corresponding capacitor isn’t installed there.


1. The CS8N65F high-voltage transistor is in a fully insulated package.
2. The FHF20150 output diode assembly is also in a fully insulated package. Both power components are mounted with thermal paste.
3, 4. Since I had to desolder the capacitors to take the photos, I measured their characteristics while I was at it. Well, the capacitance is noticeably below the stated value.


The clearance between the primary and secondary sides is wide enough, but there are no isolation slots.


Overall, the soldering and circuit board quality are good. The SMD components are also secured with glue.


The datasheet includes a block diagram, but more interestingly, it also recommends ways to reduce interference and even lists component options for doing so.


To make it easier to understand, I drew the power supply schematic. There’s nothing unusual about it, apart from a couple of details.


The odd details are two circuits, each consisting of a Zener diode and a 1 kΩ resistor. The first is connected in parallel with the PWM controller’s supply circuit, and the second in parallel with the output.
Neither circuit works as configured; I’ve written about this before. I get the impression that the designers are adding these circuits more or less at random. To make them work, you need to:
1. Reduce the value of R20 to about 10–50 Ω.
2. Connect the left lead of resistor R29 to the optocoupler’s lower lead.

The first change adds overvoltage protection to the PWM supply circuit. The second adds output overvoltage protection that doesn’t depend on the feedback circuit.


As I mentioned, I’d already reviewed a similar power supply from Sanmim, and at a quick glance I even mistook it for the one I’m reviewing here. But if you take a closer look, it becomes clear that these power supplies are quite different.
They look similar and use the same PWM controller, but the Sanmim has larger capacitors, an additional overvoltage protection circuit with its own optocoupler, safety slots in the circuit board, and generally much better build quality.


Although looking at the circuit board, you get the impression that it may have served in part as the prototype for the one under review.


That’s it for the inspection; time to move on to the tests.
With no load, the output is just over the specified 24 V, while power consumption is 0.4 W.
The product page incorrectly says that the power supply has an adjustable output voltage. It doesn’t, although there is a spot for a trimmer resistor.


All tests were performed using an EBC-A10H electronic load and a four-wire connection.


The load-capacity test showed good voltage regulation and a current limit of 3.7 A, after which the protection kicked in. Once the overload was removed, the power supply returned to normal operation.


The manufacturer specifies an efficiency of around 85–86% for this model and provides a graph showing how it varies with output voltage.
There are also graphs showing how the maximum load depends on input voltage and ambient temperature.


The test was conducted at a mains voltage of 225–230 V, and the measured efficiency was around 86–87% over the 10–90 W power range.
The power supply ran reliably at nearly one and a half times its rated power.


Ripple was measured in four modes with the power supply cold: no load, and at 30, 60, and 80 W. The probe was connected through a C+C filter made up of 1 µF and 0.1 µF capacitors.
Output ripple is very low, around 5–7 mV peak-to-peak, which is excellent.


Even connecting the probe directly doesn’t change the picture much: it just increases the spikes during transients, and the total ripple still stays below 20 mV in normal operation and 27 mV under overload. As a reminder, the datasheet specifies up to 130 mV peak-to-peak.


I also tried measuring at 90 W, but things started getting “ugly” at the output. That’s why I showed results at 30, 60, and 80 W above, rather than 30, 60, and 90 W.


Measuring ripple at a lower frequency showed that things are basically fine, although I’d increase the input capacitor’s capacitance—it would look even better.


As usual, the thermal test was run in 20-minute stages. I started at 20 W, then went up to 40, 60, and 80 W. I wanted to try 90 W next, but the power supply shut down, apparently because the protection threshold had dropped slightly.
I accidentally left the 20 W test stage out of the screenshot, but that stage lasted 1 hour and 20 minutes in total.


Temperatures at the end of the 40, 60, and 80 W test stages.


Almost immediately after it shut down, I set the output current to a level just below the protection trip point: 3.5 A. I let it run like that for another 40 minutes, and the power supply handled it without any trouble.


After two hours, I still couldn’t get the power supply hot enough for even one component to exceed 100 degrees, which is pretty good.


The classic test of output voltage versus temperature.
1. No load, power supply cold
2. 3.5 A load, power supply hot
3. No load, power supply hot.

The feedback circuit components clearly aren’t very temperature-stable; the voltage shifted by about 70 mV. For most loads, this usually isn’t critical, but some power supplies do better.



Conclusions.
Well, after comparing it with the prettier Sanmim, this one managed to surprise me with its test results: it handled a significant overload, had low ripple, and didn’t overheat—even though it ran at 80–85 W for about an hour in total, despite a rated maximum of 65 W. The power supply also stayed quiet across the entire power range.

Any drawbacks? Of course there are some. For example, I personally didn't like the fact that the input capacitor's actual capacitance is lower than required for operation across the full input voltage range, even though the unit is rated to operate at 115 V at full power.

The obvious question is: which one should you choose? If I were choosing among the ones reviewed here, Sanmim GPT70A and the “people's choice,” I'd pick... Mean Well :)
Seriously, though, the one reviewed here and the Sanmim GPT70A are roughly on par overall. The GPT70A has more ripple, but better circuit design and components. The “people's choice” would come in last.

In short, it's a decent power supply, although they cut corners on the components. If the price works for you, it's worth buying.
The store provided coupon BG2f546e, which brings the price down to $11.99 through September 15.

That's all for now. I hope you found this useful, and as always, I'll be glad to hear your comments and questions.