A Small 15 W Isolated Voltage Converter

Another isolated converter, this time stepping down 36–75 V to 12 V. It’s relatively low-power at just 15 W, but I still think it’s useful.
A look inside, a little information from the datasheet, tests, and conclusions.


This review continues my series on various voltage converters designed to run from a primary power supply or battery packs with voltages from 36 to 75 V. It’s a fairly specialized device, but it has its niche, where it can replace ordinary step-down converters while also better protecting your load, even if the converter fails.

I bought it from the same Taobao agent as the previous 50 W converter. It costs a little less, but has almost three times less power.
The agent shipped it in a plain bag. The converter has no packaging of its own; it’s simply tucked into a piece of polyethylene foam. It’s genuinely tiny and weighs just 8 grams, so, like the previous ones, it’s easy to add to a shipment—especially if your agent has weight limits for parcels.


Here it is compared with the previous converters. It’s not actually that much smaller.


The listing names the manufacturer as Lineage Power and the model as SW001A2B91Z, but searching for that designation turned up only this datasheet. It doesn’t really matter, since it’s the same module.
General information.


It comes in four output-voltage versions: 3.3, 5, 12, and 15 V. The seller only had the 12 V version, which was what I needed anyway, though 15 V would also have worked.
I only noticed the “External capacitance” line later: 1000 µF for the 3.3 and 5 V versions, and 220 µF for the 12 and 15 V versions. In my tests, I connected the load and oscilloscope directly to the board, without any extra capacitors, which of course made the results a little worse.


Here’s the information on maximum operating conditions, input-to-output isolation voltage, and service life, rated at 8,200,000 hours with an 80% load at 25°C and a little airflow.


Enough about datasheets—let’s take a closer look at the module itself :)
A little dusty, compact, and neatly made. There’s also some other small transformer on the secondary side of the board.


The module comes in two versions: with leads or for surface mounting. As you can see, I have the first version. Interestingly, the second type comes on tape, like ordinary resistors and ICs.


The modules also optionally support enable control and output-voltage adjustment.
Applying a low level to the middle input pin turns the module on; there’s also a version controlled by a high level. The voltage on this pin is about 9 V, and the short-circuit current is 25–27 mA.
The output can be adjusted over a range of 90–110%—essentially a small trim, from 11.8 to 13.2 V.


Pinout and module dimensions. I’ve always appreciated how much information manufacturers include in their datasheets.


Top view


Bottom view; the feedback optocoupler is visible at the top. Essentially, this is a miniature power supply, just designed for a lower input voltage and without an input bridge rectifier.


As with all the previous converters, this one uses a planar transformer: the windings are formed by traces in a multilayer board, which is partly why the board is so thick.


As usual, on to the tests.
Current consumption at 36 and 62 V in standby and active mode with no load.
Consumption in active mode is fairly high, but I noticed something interesting: after 10–20 seconds, the current drops by about a third—from 30 mA to 20 mA.


The module starts up at 33 V. At 36 V, it delivers 1.5 A at the output, despite being rated for 1.25 A. At 1.6 A, the voltage starts to drop.


With an input of 62 V, it delivered 1.6 A without any problems. I didn’t load it any further. Overall, voltage regulation accuracy gets top marks: the converter maintains a stable output voltage across the entire load range.


Output ripple at 36 V with no load, and at 33%, 66%, and 100% load.
Settings: 50 mV/div, 20 MHz bandwidth limit; probe connected with 0.1 µF and 1 µF capacitors as a filter.


Same measurements, but with an input voltage of 62 V.


And here's a brief explanation of why I connected the probe with two capacitors, as I did before, instead of connecting it directly to the output.
The filtered setup is on the left, with a 62 V input and 66% and 100% load; the unfiltered setup is on the right.
Since the oscilloscope is powered from the same mains supply, all sorts of noise gets into it along with the useful signal. I still can't get used to this; apparently, I'll have to add galvanic isolation to its power supply after all.
After my little DS203, taking measurements really is awkward, because the setup is already complicated: power supply, electronic load, computer, each with its own switching power supply—and on top of all that, the oscilloscope's power supply.


Now a little about efficiency. Below are a few photos at 36 V and 62 V, with load currents of 0.2, 0.6, and 1.25 A, respectively, giving output powers of 2.4, 7.2, and 15 W. Of course, a power supply isn't a measuring instrument, but I checked specifically: the current setting differed from my multimeter's reading by about 1–3 mA, so I think we can ignore that and use the readings from the power supply's wattmeter.


And now the same data, but with a few more details, in table form.
It turns out that at low input voltage, efficiency is higher at light loads; at high input voltage, it's higher closer to maximum load. On average, though, it stays around 90%, dropping noticeably only at light loads, so using it in that range isn't very efficient.


A couple of thermal images. First, I ran it for about 20 minutes with no load, and it heated up to around 55°C. Then I loaded it to the rated 15 W and ran it for another 10 minutes; the temperature rose to 72°C. In both cases, the transformer core was the hottest component. The input voltage was 48 V.
Basically, it can be used across its full power range without worrying too much about cooling. Of course, if you stuff it into a small sealed box, it'll probably overheat.


The seller's page offers a version with an enclosure, though the label says 12 V, 2 A with an input of 15–50 V. I think the photo is of a different model.


Although it has the same board inside as the one reviewed here. In any case, I don't see much point in ordering an enclosure from Taobao when it's considerably larger than the converter and adds extra weight.


A group photo for comparison, showing both the high-power models and the tiny one reviewed here.


Based on the tests, all I can say is that I like this converter, and I already know where I'll use it: in a new adjustable power supply that needs an isolated 12 V supply rated for around 5–7 W, powered from the main supply at around 36–40 V.
I also wish I'd bought more than one. I should have got at least three and kept them on hand in case they come in useful.

That's all from me. I hope you found this information useful.