What do you not want to see in a power bank that you intend to carry in your pockets, luggage, or backpack? Loose metallic objects likely to cause the lithium cell to spontaneously dump its ~100kJ charge. Here, we have a small solder bead, which could break loose from the flux it is stuck into and find a new home where it could short out the cell.
Did the manufacturer use cotton balls to clean this board? Whatever it was, it left a lot of fibers stuck in flux that didn’t get cleaned out by other means.
At first sight, you’d think that someone forgot to solder the 5mm LED’s leads from the lack of solder around them. For some strange reason, they were soldered from the top instead.
Does the micro-B port have mechanical tabs poking into the holes? If it does, they are too short to be visible at the bottom of the dimples formed by the solder’s surface tension when it reflowed within them.
As far as blobs on a pad go, the one on the switch’s top-right lead is about as big as it can be before spilling solder around the pad. Also, with the switch so close to the board’s edge, I’d be a little worried about excessive force on the switch cracking the PCB and breaking a trace.
In one of my e-mails with GETIHU, a representative told me that many people complained about the switch protruding too much and frequently getting activated by accident. This was allegedly addressed by a revision, which hopefully backs the switch off by about 2mm from the edge.
What are the two integrated circuits on this side of the board? The 9926A in SO-8 packaging on the left appears to be a dual independent N-channel 6.5A/20V MOSFET, while the 8205A in SOT-26 packaging to the right looks like a dual N-channel MOSFET with common drain rated at a slightly more modest 6A and the same voltage.
A typical battery protection circuit only requires one dual-N package to operate, which makes me wonder why there are two of them here. But wait: there’s more!
The Other Side
The top is quite a bit more crowded, with a Legendary LDR5108 special-purpose chip doing practically everything apart from cell protection, a 2.2µH inductor for the DC-DC boost converter, and a handful of additional support components. Overall soldering quality looks much better on this side, suggesting that everything except the 5mm LED was oven-reflowed.
Soldering on the top side looks far more typical of reflow soldering. All of the surface-mount components show a smooth fillet between board pads and component leads, and just the right amount of solder paste. This is particularly evident on the inductor pad and C20, near the bottom.
Here’s what soldering component leads from the top looks like. With long leads pulling heat and solder away from the pad and into that bend, you end up with oddly shaped blobs.
Both battery pads show the same inverted doughnut shape visible to the left of B+. This is caused by solder pooling on top of the jumper wire's tip when it was soldered from the back, and the rest of the pad not getting hot enough to reflow with it.
Also of interest are the unpopulated pads for C13, to the inductor’s right. No input filter capacitor on a boost regulator may have some interesting effects on output waveforms.
The Real Battery Protection
If you have ever seen tear-downs of gadgets powered by a single rechargeable lithium cell, you won’t be surprised to find a DW01A next to a dual N-FET (the textbook combo). What’s a DW01, for those who don’t know? Possibly the most common single-cell over-charge and over-discharge protection IC in the business. All it does is enable the charging path’s MOSFET for as long as the cell voltage is below the over-charge threshold, doing likewise with the discharge MOSFET and the over-discharge threshold. This is an awfully simple, yet important responsibility when the goal is to prevent lithium cells from spontaneously combusting.
Why So Many FETs?
Seeing three FETs where I was only expecting one made me pull the LDR5108’s datasheet to explore further. Page nine of the PDF provides Legendary's reference design, which shows its DW01 implementation in the top-left corner using a pair of 8205 dual-N FETs in parallel for reduced losses. That's one mystery solved. The bottom-left schematic reveals that the LDR5108 is a synchronous rectification controller using one half of the 9926 dual-N FET as its boost converter switch and a separate FET for the synchronous rectifier, while the 9926’s other half turns off the outputs by disconnecting them from the low (ground) side.
GETIHU or its ODM went for a seemingly verbatim implementation of Legendary's reference design.
One Size Doesn’t Fit All
There are two problems here. First, the connector’s mechanical tab is half-way off of its pad’s location. Second, the clearly visible seam between the tab and solder puddle is usually a clear sign of a cold (weak) solder joint. A good joint wets the surfaces it adheres to, leaving no visible seam. Another possible issue is that, since the mechanical pads aren’t connected to anything, and aren’t staked to the board with solder-filled vias or slots, the only thing preventing them from ripping off the board is the glue between the three pads and PCB core.
What happened to throw the positioning off? Did GETIHU under-size or misplace its pads? We’ll never know.