There are two connectors with two front mechanical tabs each. Shouldn’t that add up to four pads? Yes, but the second connector is missing a pad altogether due to the fiducial mark’s location. Having only one side of the connector correctly tacked down should be sufficient to prevent the port from wiggling and eventually breaking connections at the back. Unfortunately, there are none of those to be found in my sample.
While re-assembling the power bank to re-take some photographs. I noticed something potentially dangerous that I initially missed: the lithium cell’s jumper wires connect to the board's unprotected back where their respective solder blobs could potentially short to the power bank’s aluminum housing if you apply enough pressure to it.
From the way the plastic shroud over the board's top side gives largely unimpeded access to the two battery pads, it looks like the jumpers were meant to be soldered after assembly on the inside, a safe distance away from the aluminum body.
Dumb “Smart” Detection
GETIHU’s manual claims that the PB-01 detects how much power attached devices need, but I didn't see a charging protocol detection chip on either side of the board to read data lines. Putting the data pins under a microscope shows that they are simply tied together. So much for smarts. The power bank's DC-DC converter simply provides whatever the load draws, as long as it is within its limits.
What does the output look like? From 500mA, where the output enters steady operation, until it enters overload at 2.1-2.5A (depending on cell charge), the output maintains roughly the same noise waveform, progressing from 85mVPP to 160mVPP mainly due to switching transient peaks. RMS noise is much milder at 35mV.
At a switching frequency of 250 kHz, I was expecting better output filtering than this.
Once the regulator enters cut-off mode, presumably due to low cell voltage before hitting the over-current shutdown limit, the output voltage drops 0.4V below the cell’s voltage with sporadic peaks from the boost converter attempting to restart. This causes the cell voltage to droop, and the converter to shut down again.
I was expecting the power bank to simply shut down its output, rather than lingering one diode drop below cell voltage between converter restarts.
At 0.5A and 1A, the output quality is acceptable. Then, noise picks up starting from 1.5A. While 160mVPP may sound like a lot by PC power supply standards, the USB Battery Charging specification allows any amount of noise from 4.1V undershoots to 6V overshoots as long as the 250ms average is between 4.75V and 5.25V. By that standard, 160mVPP should be a non-issue for devices that follow the BC specifications.
Voltage does hold up quite well, starting at 5.04V until hitting 4.89V at 2A. Beyond that, it collapses from the boost circuit hitting its duty cycle limit until it outright shuts down at 2.5-2.6A, allowing straight cell voltage through.
Output Waveform With C13 Populated
Remember that missing C13 I pointed out earlier? In previous tear-downs, whenever I attempted to improve an adapter or power bank by fitting something onto vacant capacitor pads, I was rewarded with little to no improvement. Not the case here: adding the capacitor reduced peak-to-peak noise at 2A output from 150mVPP to 26mVPP. That's much better than my expectations.
This is a perfect example of how shaving $0.01 off the bill of materials (BoM) can drastically affect a product’s performance.
In the QQC and SilverStone power banks, cell charging is done by a dedicated switching regulator easily identified by the presence of separate inductors for input and output. To confirm that GETIHU’s design doesn’t use a clever bidirectional switching arrangement to share its inductor between charging and discharging, I measured current coming out of my UC-01 and current going into the cell. A 30mA difference between the two (1.27A vs 1.24A) can be attributed to GETIHU’s LEDs and DC-DC converter, which means we are definitely dealing with a linear regulator. That explains why it took a substantial 39Wh to charge a 29Wh cell.
Calling A Spade A Spade
After getting more intimate with GETIHU’s PB-01 than most people who weren’t involved in its design, I can say that the board’s manufacturing quality is lacking. I am displeased with its fake rating and peeved about seemingly being lied to about it. I wish I could call the real 8Ah cell a saving grace, but it is in no way, shape, or form an excuse for the clearly fake 10Ah marketing claim. Cell jumpers soldered on the exposed side of the PCB, where they could potentially short against the enclosure, are cause for concern. It also irks me that GETIHU sacrificed clean output to save a $0.01 part. Lastly, the lack of an efficient charging circuit is fair enough considering the price.
This is an unfortunately long laundry list of things to be disappointed about. For what it is worth, I do like the form factor, the overall construction, and the convenience of having a basic LED lamp built-in.
What do you think are the chances that GETIHU was honest when it told me it will investigate that 8Ah cell?