At the beginning of the year, I took at trip which included visiting Hong Kong where I managed to pick up a Mi Power Bank Pro and review it. At the end of the review, I expressed the futility of keeping up with technology as there was a QC 3.0 version apparently available by the time that review went up.
Fast forward and I’m back from my second trip, which also included some time in Hong Kong. Visiting their new store in Hang Lung Centre in Causeway Bay, I managed to pick up the latest version of the Mi Power Bank Pro and give it a whirl. The new one maintains the same ratings and name, but instead has a new model number of PLM03ZM.
Unfortunately, for readers, I have no box nor instruction leaflet. The reason is that I just couldn’t afford to carry the weight/volume of the box home, but it is very much the same as the previous in the sense of having the same sort of warnings and authenticity check labels. Instead, lets look at the product itself.
One of them is the older Pro, and the other is the new Pro, can you tell?
I mean, it’s pretty obvious … the differences …
… which for the most part, there are absolutely none, except for the specification panel on the bottom. (If you’re really interested, old is on the left or top of images, new is on the right or bottom of images.)
The new Pro appears to be designed for domestic market, and hence features less certification logos and lots of Chinese. However, they have improved the labelling which makes it explicitly clear as to the capacity specs. It claims to deliver 2.4A at 5V, 2A at 9V and 1.5A at 12V making this an 18W-capable unit. However, the labelling is unclear as to whether it is QC2.0 or QC3.0 capable (and I have no devices to check this out with). The website claims it is QC3.0 capable, in which case, output currents in a voltage range is expected as QC3.0 expands the ability of the output to be adjusted in 0.2V increments. An addition to the labelling is the expected charge output at different voltages – 7.2Ah at 5V, 3.8Ah at 9V or 2.9Ah at 12V which also allows a crude calculation of efficiency (due to the limited precision) – 93.5% at 5V, 88.8% at 9V and 90.3% at 12V.
The design retains the use of a USB-C port for charge input, and USB-A for output, which makes a bit of cable-adapter-juggling necessary for microUSB-B users. As a result, it now comes supplied with a micro-B cable that has a captive adapter to USB-C. This is much improved over the previous iteration where the adapter could easily be lost. That being said, the use of an adapter may not be ideal for charge speed due to increased contact resistance contributions, so two cables may be the preferred result especially as the unit claims to retain simultaneous bi-directional quick charging ability.
Having learnt from the previous unit, the way to getting in is to attack the unit from the bottom. As a result, the top faceplate was left undisturbed this time around.
Two screws secure a blocking plate to the bottom, and once removed, the whole unit can slide out of the aluminium channel chassis.
Just like the last unit, the whole device is built into a plastic frame with a thin metal plate covering one side.
The other side has foam-rubber strips to ensure a snug fit for the battery, with some space around all sides for expansion. The PCB resides in the top portion.
An NTC pack thermistor is provided for thermal protection of the pack, and sits routed in a specially enlarged channel of the plastic tray – identical to the previous version.
A close look at the PCB reveals that nothing much appears to have changed. A 15A fuse protection is provided for the battery. One of the controllers has been changed from a ZMD 1204UB in the previous unit to a ZMG 1206USBN in this unit. The S61088A and BQ25895M remains the same.
To get to the underside requires some effort to spudge-out the battery from its cradle.
The PCB has the same QPB100-MB code, however, the code beneath is substituted with 6C-160714 (presumably a date) whereas the older one has 69-160419. As a result, it seems that the new design occurred less than three months after, and may well be just a result of a change of software in a microcontroller. This particular PCB is dated Week 12 of 2017 and is again virtually identical to the older unit.
The same observations can be made about the battery which is an ATL lithium polymer pack made of two cells in parallel. Each of them are rated at 5000mAh minimum or 5090mAh typical, with a 3.85V nominal voltage. As a result, the total pack is 10000mAh minimum or 10180mAh typical – as a result, the number on the packaging is conservative and good results are to be expected.
The battery is held in with two self-adhesive tape strips, and the PCB by two small black screws. On the upside, this time the battery did not appear to have any strange dents in it.
I usually spend a very long time testing power banks on my makeshift rig due to the time needed to do the discharge/recharge, but since I’ve got my BK Precision 8600 DC Electronic Load, I have now streamlined the test procedure. It was established in the previous test that the unit produces very accurate and repeatable results, so instead of five runs at each current level, just one will be performed. In exchange, a new test is also thrown in. The same 20cm lead is used to connect from the power bank to the load, its resistance and voltage drop are included in the test results and reduces the output voltage/calculated efficiency slightly.
Output Voltage Trend
Output voltage stability was absolutely rock solid with no significant variances. Very good, and a good level of output voltage as well.
- At 500mA, the unit delivered 7248mAh at 5V for a calculated 94.1% efficiency based on a capacity of 10,000mAh at 3.8V.
- At 1A, the unit delivered 7348mAh at 5V for a calculated 95.4% efficiency based on a capacity of 10,000mAh at 3.8V.
- At 2A, the unit delivered 7316mAh at 5V for a calculated 95.0% efficiency based on a capacity of 10,000mAh at 3.8V.
All figures exceeded the claimed 7.2Ah at 5V as stated on the specification plate. For comparison, when normalized to 3.7V, the unit delivered 9794mAh, 9930mAh and 9888mAh at the three current levels respectively. A very good result that easily proves the honesty and high efficiency of the conversion circuitry used. Very similar to the results from the older unit, which is not unexpected due to their similar designs.
Output at quick-charge voltages of 9V or 12V was not determined due to a lack of equipment to simulate quick charge demand signals. Maybe I will build this one day, but this is not a great priority. However, when connected to my Mi Max, it did negotiate Quick Charging and complete charging in an expected fashion.
It was good to see the “double-click” low-current charge mode is available as with the last model, which over-rides the low current shut-off feature for charging Bluetooth headsets and wrist-bands which often don’t draw enough current to stop units from turning off automatically.
At 500mA, ripple averaged 16.37mV peak-to-peak, which is an excellent result, likely less than that of original equipment chargers.
At 1A, the ripple increased to an average of 27.44mV peak-to-peak, still very good.
At 2A, it further increased to 37.08mV, which is still exceptionally good. A new, lower periodicity ripple with larger amplitude is excited under higher current draw.
Charging utilizes an “opportunistic” peturb-observe cycle of operation where charging is interrupted slightly and a higher current is attempted, and retained if the voltage does not collapse under higher currents. As a result, the fastest possible charge (within safe parameters) for a given adapter and cable combination is obtained, although it does have a tendency to potentially cause slight overloading in some cases (15W Quick Charge adapter loaded up to 16-18W).
Under a regular 2A wall-adapter, charging was completed in 5 hours 45 minutes, with an estimated charging efficiency about 80.7% assuming 38.5Wh cells. Using a 15W QC2.0 capable charger at 12V, a 3 hour 38 minute charge was observed with a slightly lower estimated charging efficiency of 78.4% assuming 38.5Wh cells. It’s likely that a further reduction in charge times could be observed if an 18W (or greater) QC3.0 adapter was used instead. The charging efficiency could be slightly higher, as the cells may have a slightly higher capacity than the minimum capacity assumed here.
Now that I have an adjustable DC load, a good test is to see whether over-current protection exists and operates correctly. The current was stepped up, first in 200mA increments, then later in 100mA increments, every five seconds while readings were being taken.
The resulting graph showed that OCP was invoked upon reaching 3.1A, where just one reading was returned and then the voltage fell to zero and the unit shut down. The unit suffered no damage and could be restarted by pushing the button. While this is higher than the 2.4A written on the unit, it gives enough headroom to allow certain units to charge faster – potentially covering the USB-C cables that may advertise 3A capability to downstream units. It does show a functioning OCP.
A secondary result from removing the end-points seems to show that the wire resistance is about 97.113mOhms, which is not far from expectations. As a result, it seems likely the observed voltage drops are due to the cable, and the Mi Power Bank Pro actually outputs 5.1V at all tested currents, although this cannot be definitely concluded as there could be some slope-compensation or additional voltage droop which has an “ohmic” signature within the power bank.
Given Xiaomi’s track record of delivering high quality, good value power banks, this latest iteration to the Mi Power Bank Pro actually feels like a small one. On the one hand, the internals are almost identical, the exterior is the same, and the performance remains exemplary showing an “under-promise, over-deliver” philosophy. The ripple performance remains exceptional as well, and the price … hasn’t moved at all. Best of all, it’s quick charging capable and simultaneous bi-directional capable too, with a low-current override mode for charging Bluetooth earpieces and wrist-bands.
On the up-side, we do get a cable with a captive USB-C adapter, which is an improvement and claimed compatibility with QC3.0, which should improve power transfer efficiency for devices that support it. In all, it’s all upsides it seems, but if you have the older model … maybe there’s no great need to upgrade just yet.