A common theme I hear from readers is “well, those cheap power banks are rubbish, so why don’t I build my own?” I suppose that is not a bad thought, although designing one properly takes a bit of skill and good component selection. Instead, what they meant was they’d go out and buy a cheap DIY power bank case and stuff it full of good cells and call it a day. I think we all can see where this is headed.
Anyway, because of a lack of free equipment to do the measurements, and an oversupply of power banks (just how many do I need?), I put this test off for quite a while. Recently, I decided to actually do it, seeing as how many products were available under AU$2 on eBay which are seemingly popular and sell in large volumes.
The Contenders and General Impressions
The three contenders came from three separate sellers, in three separate packages, but likely, from the same fulfillment company somewhere in China. The power banks, from left to right, will be known as Type 1, Type 2 and Type 3 or as Black Rectangular, Yellow/White Cylindrical and Silver Cylindrical.
This was the most expensive unit of the batch, at AU$1.65, but is also the most generous, including a keyring lanyard and USB charging-only cable. It is rectangular in shape made from somewhat sturdy plastic and is available in a range of colours.
The casing itself is designed to house the circuitry and battery snugly, and features no buttons for operation as it is automatic load-sensing. Indications are made via LEDs which shine through the white plastic lid which clips into place once the bank is assembled.
The provided circuitry uses a metal contact at one side and a spring at the other. The mechanical fit of the case is good, thus the 18650 cell remains in good contact even when moved. The PCB is branded CHD-XS V6.2 and uses an MP3401A all-in-one controller. The controller claims to have linear charging, synchronous boost discharge with up to 700mA charge current and 1A output current. Charge termination is C/10 with 4.2V charge voltage termination accuracy of 1%. It provides over-current, short-circuit, over-voltage and over-temperature protection, although the latter is probably to save the converter itself, rather than the battery. A MLCC capacitor provides the filtering, and the inductor is rated at 1.5uH. A micro-USB-B connector is used for power input to charge the device.
The other side features two LEDs for charging and discharge indication, which blinks red during charging, changing to solid red once charged and shows blue during discharge.
The cheapest of the bunch, selling at AU$1.45, this product is also available in a range of colours and comes as just the body with no cables included. The unit is quite light and slides open to reveal insides.
The circuitry is ensconced inside a plastic shell on the left, with the spring arrangement also used in this power bank, except that the mechanical fit of the 18650 is somewhat loose, resulting in intermittent contact when bumped or moved.
The circuitry is marked HXX-168 V1.3 which suggests this is a product of Shenzhen Huaxiangxin Electronic Technology Co.,Ltd. The unit is dated 12th December 2014. The bottom side has an MLCC capacitor, the Schottky rectifying diode, a 3.3uH inductor, an 8205A MOSFET and a controller IC (unidentified) marked 134R4P (probably remarked).The top side has the USB connector, where it appears the D+/D- pins are bridged for Android compatibility and the indicator LED which similarly shines blue for discharge, blinking blue for empty, blinking red for charging and solid red for charged.
The final type is priced in-between the two units at AU$1.59 and uses an aluminium shell around the batteries. It is also available in a range of colours, and is easily unscrewed to extract its innards for assembly. The shell itself is a little rough in its construction, which is to be expected for the price.
The insert is somewhat tight around the battery, allowing for a good solid fit which is further reinforced by the pressure applied by the screw cap. This unit has a PCB marked TXT-01-0907, suggesting this may be a product of Shenzhen TXT Electronics Co. Ltd. It uses an unmarked IC, an MLCC capacitor and a 1.5uH inductor. Many component positions for MOSFETs, capacitors and resistors are left unpopulated, suggesting that this may be a model relying on internal switching capacity in the IC, and a larger current model/models using other ICs may exist which requires external components.
The other side is not populated with much, with a diode indicator which functions as per the two units above. Owing to the similarity of inductor value, the IC used in this unit could be identical to the one in Type 1.
It was found that the metal spring contact wire is bent a little too far out making it protrude from the casing somewhat making the casing “negative” as well, and interfering with proper fit in the casing. This resulted in a very tight assembly and the USB connectors rotated around with reference to the plastic holder, resulting in difficult USB connections.
It seems that all three have similar compact PCB designs utilizing SOIC-8 package “all-in-one” style controllers which are an advance from the Holtek microcontroller based early products I had reviewed. However, it’s important to note that this integrated approach may have downsides especially when it comes to current capacity, as all units are only rated for 1A load at the maximum. It may also have impacts on safety. Of note is that none of the units feature secondary fail-safe protections (e.g. voltage monitors, one-time fuses, thermistor monitoring, thermal fuses) which you might expect if this were a “branded” quality product. While Type 3 had a better exterior, its interior had poor dimensional fit making it hard to use due to misalignment of connectors. Type 1 was bulkier but sturdy and solid, whereas Type 2 had a propensity to slide open if pushed in the right place and had a somewhat looser battery fit which was vulnerable to bad contacts from vibration or putting it down on a desk. All of them have similar bi-colour LED indications with automatic load sensing meaning no buttons or other interfaces but may struggle with extremely low loads (e.g. charging fitness bands, running analog hobby circuits). It also seems likely that because of the generic style of these units, they are not capable of exploiting the full capacity in extended Li-Ion cells (which need charge voltages up to 4.35V, and discharge further, whereas these are likely configured for 4.2V termination). This is where a “pre-built” unit has the upper hand, as these can be properly tailored to the cells in use.
To test the three units, the classic new test rig was used. Because there were three power banks, to streamline the testing, we needed three well matched cells. I decided to use three 2250mAh Panasonic cells scavenged from the same almost-unused Asus laptop battery assuming its available capacity is 2200mAh for efficiency calculations. As the cells were all stored and used in the same conditions, and were likely well matched from the factory, discrepancies in performance are most likely to reflect differences in the performance of the circuitry.
Charging Current Profile
The charging current profile is important as it determines how well the a wall charger’s available output is utilized and how quickly a unit can be charged. As a rule of thumb, a good Li-Ion charger can usually charge a cell within 2-3 hours if the currents are well suited, faster times possible with higher levels of sophistication.
It was found that the Type 2 unit had the highest charge current of 860mA and had a nice taper section, terminating around 2 hours and 48 minutes making it the fastest charging in the batch. Type 1 and Type 3 displayed similar performance with peak currents capped at around 700mA (give or take) and with a wavy CC section implying some poor regulation which may be due to the linear charge algorithm and thermal protection kicking in. These resulted in charge times of 4 hours and 10 minutes to 4 hours and 21 minutes which is relatively poor and under-utilizes the output available from many wall chargers. It does, however, support the hypothesis that Type 1 and Type 3 may share a similar/the same IC.
None of the units actually consumed 1A, contrary to the sellers’ claims.
Discharge Capacity, Conversion Efficiency and Voltage Profile
Utilizing the Panasonic 2250mAh cells, and assuming a capacity of 2200mAh (due to tolerances), the following results were obtained. Figures are rounded to zero decimal places, as the greater precision is not entirely meaningful.
From the figures, we can see that the differences between the power banks in terms of output power are not particularly large. At the 1A rate, the difference in power delivery between the best performing and worst performing banks was 173mAh, and at 500mA, just 94mAh. That being said, as the tests resulted in standard deviations of below 16mAh, it’s likely that differences of 32mAh represent actual differences in performance rather than differences in charge/discharge termination.
Overall, Type 3 provided the best output power and hence conversion efficiency of approximately 80 and 86% at 1A and 500mA respectively. Type 1 was close behind offering 77 and 82%, leaving Type 2 as the worst of the bunch with 72 and 81%. The actual delivered charge was about 1750mAh which means that larger phones are likely to see half-a-charge at best. The conversion efficiency is not particularly good though, as branded power banks can achieve close to 92% efficiency which results in more usable power for the same rating of cells.
Type 1 was able to maintain its voltage at both 500mA and 1A loadings, although the voltage was only about 4.835v at 1A which is half-way to the 4.75v USB limit.
Type 2 was unable to maintain voltage regulation for 1A loading, falling below 4.75v and spiralling towards 4.4v towards the end of discharge. This will cause devices to taper back their current demands, and charge slower as a result. It also implies that the design was not suitable for 1A and the chip itself may actually be 700mA rated (hence why it is okay for part of the beginning of discharge where cell voltage is high and less boosting is required).
Type 3 was most impressive, with it holding close to 5v in the 500mA test, and around 4.925v under 1A loading which makes it much closer to the 5v target. In fact, it’s so close that the difference is not actually meaningful – its performance is better than that of Type 1 and is likely to result in better device charging.
Type 1 was able to achieve good ripple performance averaging 42.96mV peak-to-peak for 500mA loading and 63.8mV peak-to-peak for 1A loading. While this isn’t strictly below 50mV (where we’d like to see it), it is still under the 120-150mV that some stock phone chargers put out and is considered “safe”.
Type 2 produced nasty spikes at both current levels, resulting in an average of 865.5mV peak-to-peak under both current loads. This implies ineffective filtering which cannot respond to the fast transients, and such a ripple means that the voltage can likely exceed USB specifications (e.g. 5V + 0.5V forward transient = 5.5V, above 5.25V maximum) and is not considered healthy for use with USB devices.
Type 3, like Type 1, achieved good ripple performance as well, averaging 43.7mV peak-to-peak for 500mA loading and 66.43mV peak-to-peak for 1A loading.
This test shows what you can get for under AU$2, which is considered insanely cheap. The results seem to vary quite a bit. Type 2 was the worst of the bunch when it comes to quality of the power output due to high ripple and undervoltage at 1A loading and weak build quality resulting in looser connections and ease of falling apart, but had the fastest charge strategy.
Type 1 and Type 3 were similar in electrical performance, with Type 3 edging out on efficiency and Type 1 edging out on output voltage and ripple. However, Type 3, despite its high quality outer appearance, was let down by its internal construction which had a negative spring contact protruding a bit too much resulting in a bad fit which distorted the fit with the case, resulting in connectors which were misaligned with the cut-outs and frustration overall.
Type 1 hence is the best of the products, all things considered, especially as it has a bit more in terms of inclusions, but its lower output voltage could be a problem in some cases.
I think the key point to remember is that there are a lot of things which go into making a good power bank. If you just stuff good cells into a box, you might have the raw Li-Ion capacity, however, it goes beyond this. Key considerations include:
- Is the output nice and clean (free of damaging transients)?
- Will the case stand up to regular usage (well designed mechanically)?
- Will the batteries hold together well even during transport or vibration?
- Is it safe? Does it feature the appropriate layered protections necessary for the Li-Ion battery? (Most do not.) Does it respect the battery paralleling requirements? (Normally no more than 4.)
- Is the charging and discharge mechanism appropriate (to ensure fast charging of power bank and devices, to ensure long life of batteries)?
- Is the output signalling compatible with my devices to optimize charging speed?
- Are the indications sufficient to understand what the power bank is doing, or its state of charge?
- Will it run my low-current devices?
Of course, I haven’t been able to address every single aspect in every single review – but it’s worth remembering that this simple device is a lot more complicated than it may seem. Buying any product is about getting the best set of compromises – if you go cheap, you should expect more compromises in return.