What Should We Expect?
In terms of output quality, I have no doubt we’ll witness the whole gamut, from horrible to decent. Moreover, I suspect most of our test subjects will be basic fly-back converters, with pennies separating their respective bills of material. Likely, they'll be differentiated mainly between those using a discrete two- to three-transistor oscillator and those using an integrated flyback controller-driver.
Since it takes very little extra care to design transformers capable of providing well over 2000VAC of isolation, and the other boundary-crossing components are typically rated in excess of 3000VAC, there shouldn’t be any excuse for component failures on the 1500VAC hipot test I plan to use. Clearance and creepage issues due to poor circuit layout is where I expect to find the worst safety disappointments, even though those are equally trivial to avoid with a little bit of cleverness.
As a teaser of things to come, let’s have a peek at what might be the cheapest adapter in my lot: one of those ubiquitous and usually dreadful Apple 5W charger knock-offs. I scored a colorful four-pack as sacrificial lambs for $6.
Here we have the innards of what I suspect is going to be a typical cube charger. From the top, we immediately see that it is one of those two-transistor oscillator designs using optoisolator feedback. There is one probable red flag in this picture: the optoisolator’s presence on the output board means that mains-referenced signals are going over the ribbon cable to the isolator. Given how tightly packed things are, there is an extremely high likelihood of clearance issues between mains and low voltage on this board.
This adapter meets my expectations in the Y-class capacitor department by using a regular blue 1nF cap lacking any sort of safety approval marks or manufacturer brand. Should it ever fail, expect to see full AC line voltage appear between the output and ground if the non-polarized prong it is referenced to happens to be plugged in on the AC outlet's hot side. Yup, this is definitely shaping up to be one of those potential death-trap adapters.
No, it does not get any better on the back: there is next to no separation on the mains board between the mains and low-voltage sides around the resistor pad. On the output board, separation between the optoisolator feedback traces referenced to mains voltage, the output voltage traces, and the USB connector shield are equally inadequate. At this point, there is no use in analyzing the design any further. This adapter's design clearly falls in the unsafe-to-use category. Only one safety-related question remains: what fails first?
No, that’s not a blue LED you see shining there. At approximately 1kV, an arc appeared between the optoisolator feedback trace and USB connector shield, 500V short of what I was looking for and possibly unpleasant (or worse) news for you if you happen to touch anything connected to the USB shield when such a fault occurs.
Is this adapter a worst-case scenario? It does tick most of the “death trap” checkboxes I could think of up-front and verify via visual inspection. That leaves little room for other adapters to do much worse. We’ll have to see what the other adapters in my collection have in store. Also, I’m far from done exploring this one's issues. It may very well have more unpleasant surprises for us to uncover.
In this case, exceeding my expectations for the worst is actually a good thing. I would have felt a little cheated if adapters I had purposely purchased to showcase one of the worst designs possible exceeded my expectations in the opposite direction. Will this turn into a horror show? Time will tell.
MORE: Best Deals