Back in January 2013 (which seems a long time ago), before I started all of this powerbank testing, as someone who was none-the-wiser, I bought one of my friends a pair of “generic” unbranded power banks. The cost was about AU$25 a piece, and they arrived in a bag (bulk packaging).
Staring at this today really gives me the shivers. Remember what happened when I last tested a unit that looked like this, but was twice as big? Not only was the capacity poor, but the ripple was terrible as well.
Realizing my mistake, and wanting to make sure there’s no chance my friend was left with a substandard product, I instituted an exchange of two new genuine Xiaomi 10400mAh silver banks for his two generics. That would make sure he was receiving nothing but the best, and give me a chance to see how big my mistake was.
The unit itself looks like a generic power bank. These come in three different sizes from what I can see – the four cell unit (this one) is commonly advertised as 12000mAh, a six cell unit advertised as 20000mAh and an eight cell unit advertised as 30000mAh or 50000mAh. Of course, these claims are a little dubious, especially at the larger end.
This unit was sold without a branding, but is branded P.Power. It doesn’t sound like any serious branding, especially when their logo uses Comic Sans font. The bank itself is interestingly rated for a total of 2.1A output, but the outputs are only specified for 0.5A and 1.5A for a total of 2A. Someone failed at math.
The ports themselves seem to imply you could pull 2.1A from the top and 1A from the bottom, but it’s likely to be an “either” port deal when at maximum loading. This particular unit seems to be a bit old fashioned in that it has no auto-start and will need a push of the button to start supplying power. However, if the load is disconnected, it will power off automatically in a matter of seconds. So I suppose this one was “on the road” to fully automatic.
The charging port is an old fashioned mini-USB B, which was appropriate for the time. The unit itself weighs about 211gm.
Taking apart this power bank normally entails removing a screw from underneath the label. To my surprise, this one didn’t even have a screw installed! The power bank is still solidly shut because of the plastic clips along the edges, which can’t really be disengaged without breaking them.
So what is this power bank made out of? Well, nothing but the cheapest stuff, as it seems.
Four 18650 lithium-ion cells, with plain blue shrink wrap and no branding are found. They’re probably Chinese domestic cells, which are likely to be under capacity and potential safety hazards. Good thing I took this one back. The cells themselves are wired to pads on the PCB, using thin hookup wire – could have done a little better here.
The converter PCB itself seems to differ. There are actually many different variants of this PCB in units that look similar. It’s likely the reason is that the external casing is provided by one particular moulding manufacturer and are “assembled” into finished products by different companies. So while the exterior looks the same, the internals can be different.
The PCB itself features a single Hotchip HT4901 8-bit microcontroller controlling everything from charging to power conversion. A single-filament wound open inductor is found, with an SS34 3A Schottky Diode. The diode itself is a little closely rated for a 2.1A output, and it seems the pads are available to mount larger diodes which can handle higher currents. The PCB seems to have two questionable quality OST branded electrolytic capacitors to clean the output as well as stabilize the input, which is a decent arrangement. A small ceramic cap for the output is seen as well, but a second place to mount an electrolytic is left unpopulated.
The underside shows the PCB is a revision D board dated 21st August 2012. The product is 18-LXTP1-05/ZSY. The underside has four SMD blue LEDs to indicate charge level, the activation button and two MOSFET transistors for protection.
The pack itself is constructed by soldering 2p sets of cells to the PCB, connected by thin wire. The two packs are paralleled by the PCB traces, so it’s effectively a 4p configuration, as normal. There seems to be some attempt to use mylar tape for insulation, but with the way it has been applied, I can’t see much of a point.
There is also some double-sided flat tape applied, but the backing was never peeled off and it never really served any real purpose. Removing the backing and attempting to use it made it clear why it was never used – no adhesion occurs since the cells don’t actually touch the case well enough. Instead, it seems that foam tape should have been used to keep it snug.
The top and bottom tabbing seems to show thin tabs being used, but their lengths are about right, and while the spot welding is a little “doubled up” in places, they look decent enough.
Capacity testing was performed in line with the current test methodology. The capacity results are as follows:
|Load (mA)||Run||Capacity (mAh)|
|Load (mA)||Run||Capacity (mAh)|
|Load (mA)||Run||Capacity (mAh)|
It seems that there is a wide variance in the readings, suggesting charge termination problems and possibly end of discharge determination variation as well. At the 1A rate, the capacity increased dramatically as a function of run number, while at the 2A rate, the capacity decreased moderately as a function of run number. At 500mA, it was roughly stable. This type of behaviour is unusual, but I didn’t decide to waste more time re-doing the runs and “cherrypicking” results, because that would be wrong.
The capacity seems to drop quite markedly as a function of load current, suggesting the converter itself isn’t well designed. The capacities are 5611mAh at 500mA, 4428mAh at 1A and 3969mAh at 2A. This would correspond to an efficiency of 46.8%, 36.9% and 33.0% respectively, if the capacity was really 12000mAh.
Instead, it seems that the capacity is likely to be 7200mAh total, for a per-cell capacity of 1800mAh and approximately 77.9% efficiency at 500mA. Otherwise, it could be 6400mAh total or 1600mAh per cell with approximately 87.6% efficiency at 500mA. That’s my best guess.
The voltage profile shows the voltage output is a little on the higher side at 500mA, but still within standards, which should make for a good charging experience with even thinner cables. The voltage pretty much stays within the regular envelope limits, with significant drooping at the 2A load, suggesting the converter is having regulation difficulties. Acceptable, especially for 500mA and 1A loads, but not optimal.
For a power bank with electrolytic capacitors, the ripple waveform is actually quite good. There are no strong transient spikes to “increase” the ripple to astronomical levels. At 500mA, the ripple averages 201.6mV peak to peak, which is a little more than the 150mV typical wall chargers put out. Even at 100mV above the normal output of a 5.15v, it just hits the high-side USB output limit of 5.25V, so it’s actually “safe” within the 4.75v-5.25v envelope.
At 1A, the ripple shape stays similar, with the average ripple increasing to 320.5mV peak-to-peak. This is now twice the level of the average phone charger, but is also within the absolute USB voltage envelope. Not an optimal condition, but not immediately threatening either.
At the 2A rate, the ripple increases to 447.9mV peak to peak, which is actually likely to exceed the USB power envelope limits by dipping below the necessary voltage. This isn’t likely to cause immediate damage, but will cause some increased stress in reacting to transients and is likely to slow down or cause problems with charging at the 2A rate.
Interestingly, having known the later generic 30000mAh power bank, I had expected this earlier power bank to be just as horrendous. Ironically, this one seems to be half-decent by comparison, and offers a bigger effective capacity than the 8-cell unit. What the hell!?!
Even more interesting is to see that this unit was even better than the counterfeit Xiaomi unit, which really goes to show just how bad that counterfeit was.
It seems that continual price pressure on the power bank market has led to the quality of the units declining over time despite retailing at similar price points to keep the profits up. This is a very disappointing sign.
While this unit is half-decent, it is still under-capacity for the rating, likely being a 6400-7200mAh pack in total. The efficiency of the converter skews strongly towards lower load currents, suggesting poor design. Also while the ripple itself is high, it’s not above 500mV, which is likely to be tolerated but not optimal. The unbranded cells might still be a safety risk, and the inconsistent capacity readings suggest charge termination issues.
In short, it seems that this older generic power bank isn’t as nasty as the newer generic power banks, but it’s still not a quality item. I’m glad I took these out of service.