Whipping a cheap LED PSU into usable shape
As part of the LEDification of my house, I’m designing some wall-lighting sconces. The plan is to place an array of LED strips on the walls, covered with translucent (or in some cases, opaque!) Plexiglas plates mounted a few centimetres away. That’s the plan anyway (yes…future blog).
As before (see My kingdom for a 12V adapter), providing 12VDC power has been a stumbling block, but in this instance, I think, not insurmountable.
Thus we return to my continuing love-hate relationship with Cheap Chinese Chotchkes – in this case, 18W power supplies small enough to fit into an electrical box.
I say “18W” (well, the website says 18W), but those are Chinese specs, not to be confused with actual specs. The first batch of PSUs came a year ago, and believe it or not, could provide a full 1.5A without collapsing. However, upon opening the cases, the component temperatures after running at full-throttle for a few minutes suggested otherwise. The switching transistor, output rectifier, and output capacitor(!) were all far too hot for comfort. I imagine the 18W spec is only valid for one day of use.
Figure 1 The 12V 18W PSU. Notice the 100% efficiency (IN & OUT are both 18W), and the dire warning not to touch the surface of the plastic case because of high temperature!
I recently purchased a second batch of these PSUs, this time from AliExpress. They behave differently than the first batch. From an external vantage point, they’ll only supply 1.3A as opposed to the former 1.5A. The output voltage collapses to 8V at 1.5A.
Take it apart, and further differences appear. Most noticeably, the 12V output capacitor does not egregiously overheat at, say, 1.3A. Sure enough, the part has “Low ESR” printed on the case. The previous caps don’t.
Figure 2 PSUs from AliExpress (top) and another, forgotten source (bottom). Note the convenient dates at the bottom of the boards. Manufacture has transitioned from phenolic to fiberglass.
The design appears to be a simple self-oscillating circuit. I measured the switching frequency to be about 100 kHz. The 12V output caps are at the upper-right of each board. Though the legend says “1000µF 25V”, the installed caps are 470µF.
After discovering the output cap heat problem (but before getting the second PSU batch), I sourced a bagful of quality capacitors – 270µF @ 35V, still more than enough capacitance for this circuit, but with a high ripple-current rating and low ESR. Both of the boards above have these new parts installed. They run cool as a cucumber, versus the slight temperature rise of the second PSU caps, and of course, the extreme rise of the first ones.
Figure 3 The PCB bottoms reveal other minor design changes in the newer boards – mainly exposed copper to pick up current-fortifying solder.
I’m constantly struck by the strange state of affairs at this level of Chinese manufacture. Clearly, there is some thought and skill put into design and production, yet we still end up with stupidities, like unsuitable parts, or wishful-thinking specs.
As I mentioned, other parts get hot too. The switching transistor can get toasty at higher loads, but it was the output rectifier I focused my measurements on. At 1.33A, free-air TC registered 90°C. At 1.2A, 86°C.
Figure 4 My test setup here at EDN Labs. Note the many safety protocols employed on the bench.
Enclosed in its case, in an electrical box, I don’t think I’ll want to pull more than 1.1A from these PSUs. Hopefully, that will be enough for my LED lighting. Other options: Squeeze two PSUs into a box (possibly swapping the case for some shrink wrap or electrical tape), or, cut a hole in the case so I can bend the rectifier out and heat-sink it to the electrical box! Hmm. We’ll see.
You may have also noticed the absence of an AC line filter on the board. Well, what did you expect for US $3.15? I would much prefer to have used this 2A bare-board PSU, or this enclosed one. The former, at least, has a line filter, and even if they also don’t meet their output-current spec, they will certainly be better than the PSUs I’m using. But…they don’t fit into an electrical box.
Stay tuned for further field reports as I attempt to actually build these luminaires.
—Michael Dunn is Editor in Chief at EDN with several decades of electronic design experience in various areas.