It’s a bit of a shame that some of the mobile phones out there which have Qualcomm chipsets and are capable of quick charging are shipped with regular “plain old” 5V/2A chargers. As an owner of various Xiaomi phones myself, they very much are guilty of this. This could be in the name of cost-cutting, as higher end chargers could cost more to provide, but could also be in an attempt to reduce the stress on the battery and charging ICs.
Regardless, as the battery wears out and the charge capacity reduces, the phone tends to need to be charged more often. As we move to a more “mobile” lifestyle and depend more on our smartphones, having them in our hands untethered from the wall is the goal, and thus faster charging is desirable to have.
As someone who has review a plethora of power banks, I’m aware that the quality of chargers can vary significantly. Whether a given product produces stable output that is clean and “safe” for devices is something I definitely consider – there’s no point risking a $300 phone to a $10 charger.
I went in search for a low-cost Qualcomm QuickCharge 3.0 capable charger, of which there are a number of on the market. One primary concern was that it had to have an Australian mains plug. While EU/US plugs are most popular, plug adapters add bulk, safety risk as they can fall apart and reliability issues as the charger can “fall out”. This reduced the choices significantly. I ultimately chose to buy the BlitzWolf BW-S6 30W QC3.0 Dual-Port USB Charger for AU$17.81 including delivery off eBay.
As I purchased the item off eBay, there’s no way to know if it is truly genuine or not. However, it does look rather authentic, inside its shrink wrapped colour cardboard box. The front has a green stripe towards the bottom and the model BW-S6 written in the centre. The BlitzWolf logo is in the top right.
Along the side, there are a few logos which purport to signify some of the benefits of the charger, although they are slightly cryptic.
The underside of the box lists the product specifics – namely a black coloured charger with a 2.4A port providing 5V and a QC3.0 port providing 3.6-6.5V at 2.4A, 6.5-9V at 2A and 9-12V at 1.5A. The package claims an item weight of 150g, along with logos for CE and FCC approval. Note there is no Qualcomm QuickCharge logo anywhere – this product is not officially approved as per the current iteration of the Quick Charge Device List.
Opening the box, there is a pair of small leaflets including a user manual and support card …
… not that anyone really needs one for a USB charger.
The USB charger is contained inside a milky matte plastic bag with protective film covering the body of the charger itself.
The charger is not much taller than the Australian mains plug, with the pins sheathed as per current standards. However, it does not carry any SAA approval numbers, nor Australian Regulatory Compliance Mark, so it is not strictly compliant despite carrying the correct plug. Better than using an aftermarket plug adapter in my opinion.
The charger itself has two ports – the lower orange port is QC3.0 capable, while the upper port is a 2.4A port for “regular” USB devices. The ports are marked Spower, which I suppose is their “Power 3S” technology which allows compatibility with different devices.
The underside is marked with the same specifications, along with the double-insulated symbol, indoor use only symbol, and VI Efficiency Mark which in case it is considered a multiple-voltage external power supply, should mean <0.3W standby and >=0.075*In(Pout)+0.561 efficiency (which would be 0.075*2.4+0.561 = 74.1%).
The claimed weight of 150g on the box seemed to be a point of confusion. Weighing the item revealed a weight of 96.93g, while weighing the whole package was only 140.95g. It did not reach the 150g weight which is suspicious, although the website claims the weight to be 100g +/- 10g, which the charger does meet.
Readers who have seen my testing in the past have noted that I had not tested mains chargers nor have I ever tested any Quick Charge capabilities. This was due to a lack of test equipment and time – however, for this charger, I have decided to go the extra mile and build some test equipment to actually test the unit properly.
Quick Charge 3.0 Capability
The easiest way to test whether it is capable of QC 3.0 is simply to plug it into my phone which is QC 3.0 capable and observe the charging voltage and current through an appropriate charger doctor, in this case, my Keweisi KWS-V20.
To no surprise, it actually did work – starting at 8.01V, dropping to 6.51V by completion. It allowed the charging of my phone to proceed quickly and with reduced heating compared with a QC 2.0 charger.
Primary to Secondary Isolation
An insulation resistance test was conducted with the Keysight U1461A Insulation Resistance DMM. Before this could happen, I needed to rig up a USB plug that connected all pins and shield to a wire to act as one of the test poles, with the other side connected to the mains pins.
Readings into both ports at 1000V test voltage over a period of one minute showed off-scale readings of >260Gohms. The insulation is definitely healthy based on this result, but without knowing the design and transformer construction, it cannot be definitely concluded.
Output Voltage versus Current
In order to test this, I had to build myself a few tools to connect my equipment up to fairly test it. The first thing is to build a USB A plug interface that would connect the output through wires to my B&K Precision Model 8600 DC Electronic Load, while providing extra connections for the sense lines (4-wire kelvin connection to compensate for voltage drop in the leads), for my Picoscope 2205A for ripple voltage measurements and to the D+/D- and GND lines for communication with the QC 3.0 controller. The mains would be generated using a my HIP-300 Pure Sine Wave inverter through a variac for voltage stability, with consumption measured on the Tektronix PA1000 Power Analyser. Note that testing was done with 230V AC input +/- 1%.
The board was constructed out of a shred of vero board (aka strip-board). On that, I mounted a brand new, high quality Molex USB A through-hole plug from element14, two rows of header pins, a row of header sockets and two wires.
With some crafty “cutting” of adjacent holes together, I was able to slot the support “ears” on the USB A plug through the board, but cutting a little too much copper made it hard to solder it to any significant bulk of solder. The wires to the load were 22AWG (0.5mm^2) which were soldered to crimp terminals at the other end to avoid changes in resistance during testing.
In order to command the QC 3.0 controller, I used the QC3Control library by Vincent Deconinck and a spare Arduino Leonardo board. A six-resistor circuit is necessary to interface the Arduino to the D+ and D- pins, along with a shared ground – the way this was derived was quite fascinating. Thus four pins come in from the Arduino (Vcc, D+, D-, GND) and three out to the USB interface (all except Vcc). This avoids having the Arduino exposed to higher voltages and also prevents interference of the test results by the Arduino’s power consumption.
While I couldn’t find the perfect value of resistor, I substituted a 12k ohm resistor for a 10k ohm which seemed to cause no ill effect and then wrapped up the interface in white heatshrink. I wrote a very short Arduino program that takes a character and commands the power supply accordingly – 0 for 5V, 1 for 9V, 2 for 12V, 3 for increment, 4 for decrement and nothing else. I also took advantage of “hanging” the board from the A0-A5 header and powered the board by setting A0 low and A3 high for some wiring reduction.
It is only with all of this preparation that I am finally equipped to test a QC 3.0 charger properly!
At a 5V output from both ports, we can see two distinctly different behaviours. The non-QC port starts off at just above 5.2V, falling nearly linearly to 5.1V at 3.1A before falling more steeply in an unstable way, cutting out by 3.4A. The QC port reverses this, instead, starting near 5.05V and increasing output to about 5.24V as the current increases to 3.2A before shutting down at 3.3A. This behaviour is more akin to the cable-loss compensation paradigm, whereas the non-QC port is more likely “signalling” its current supply capabilities by reducing voltage as current increases.
Once QC is introduced to the mix, things are a little different. At both 9V and 12V, a slight upward voltage trend is noticed with increasing current, but here, the charger appears to follow a constant-power envelope instead as the current increases, before cutting out entirely at 3.3A. At 9V, power delivery was maintained to 2.2A and at 12V, maintained until 1.7A, exceeding the specification by a small margin. While in all cases, available current exceeded specification and no permanent damage seems to have occured, having in excess of 3A available on a single port may not be the safest result and tighter current limiting may be desirable.
Efficiency versus Current
Spot-calculations of efficiency from the collected data shows efficiency in the ballpark of about 85% when loaded, with high efficiencies gained even on modest loads. The QC port seemed to have better efficiency at higher output voltages as well.
Ripple Voltage versus Current
Measured using the Picoscope, I cannot exclude some external noise contribution and the possibility that the B&K Precision Electronic Load may have interacted with the supply to exacerbate the ripple slightly. However, the measured ripple voltages seem to be relatively consistent, increasing with higher currents especially when overloaded. On the whole, ripple voltages of up to 250mV peak to peak were experienced, with higher ripple on the QC port compared to the non-QC port. However, most of the peak-to-peak contribution was due to switching transient spikes, which suggests poor filtering, and thus the main ripple as reflected by the RMS value is much lower. This is perhaps an acceptable but not meritorious performance.
In most cases, the RMS ripple is under 30mV, which is a much nicer looking figure, however, better filtering would be appreciated.
Standby Power Consumption
Using the IEC 62301 Standby Power compliance testing feature of PWRVIEW (which incidentally, doesn’t work under the latest version of Windows 10 and NI-VISA), the standby power was measured to be 0.178W, which is compliant with the requirement for multi-voltage adapters of VI efficiency mark to be <=0.3W.
No review is complete without a teardown, so I decided to take the chargers apart. It was interesting to see that this particular charger differed quite a bit from this other technical review of the BW-S6 EU/US versions which had the Qualcomm logo on them. In fact, now that I look closely at them, it doesn’t seem that those ones had the expected charge controller chip in them so perhaps the one reviewed was a different version or a counterfeit product?
The unit has a screw-less construction, so opening the case entailed “squeezing” the sides until the cover “popped” off after cracking all of the adhesive around it. We can see that the charger is a “tight” fit inside, with copious amounts of white silicone used to secure the PCB in place. Already, we can see a number of welcome design features including the heatsink, primary side suppression capacitors and filter choke, input fuse and output solid polymer electrolytic capacitors. At the least, the design looks to be safe, although the inclusion of the blue LED inside an opaque black case seems to be a bit of an oddity that would only increase the standby power consumption.
With some difficulty, it was possible to cut through most of the silicone to extract the board from the case without damaging it. The green PCB appears to be light in colour with green soldermask and black silkscreen text. The design has isolation from primary to secondary with a wide area of PCB which is “drilled” through. Two ICs breach the isolation – the lower one is an optoisolator, with the upper one being a Power Integrations INN2215K InnoSwitch-CP Off-Line CV/CC Flyback Switcher IC with Integrated 650V MOSFET, Synchronous Rectification, Feedback and Constant-Power Profile for USB-PD and QC 3.0. It is also noted that this chip does not appear in the Qualcomm Quick Charge Device List and thus is not an approved solution, even though it works. There is a bit of concern, as this chip is both primary and secondary side, I wonder if a surge could perhaps compromise the insulation through the chip itself rather than the transformer. I didn’t destroy the unit, so I can’t comment on the transformer insulation either. There appears to be copious amounts of solder on some areas to help sink heat and reduce resistance, but the markings are a bit hard to read.
A look from the top side seems to show that this dual port charger is basically two chargers crammed into one case using two separate transformers. This is a good thing, as it means that the outputs are not “sharing” one beefy (or sometimes, underpowered) supply rail. However, the output filtering is somewhat disappointing, seemingly lacking any significant inductors or ceramic capacitors, relying on the large polymer electrolytics to do the bulk of the work. This may explain the somewhat spiky and high numbers for peak-to-peak ripple voltage.
There is a question as to what the heatsink is actually cooling, or whether it’s intended to redistribute the heat inside to prevent hot-spot formation. There seems to be one transistor attached to it in the top left, but there is also thermal paste along the top of the transformers and this plastic-package diode/transistor that is entirely not attached. A little strange in my opinion.
The biggest downside, in my opinion, are the electrolytic capacitors on the primary side. All of them appear to come from a relatively unknown company, Acon, so whether they will last in the long run especially being so “tightly” packed together and subjected to heating is an unknown.
It’s a shame that some Quick Charge capable phones don’t come with Quick Charge capable chargers and the variety of chargers available from larger brand names generally have EU/US plugs only. The BlitzWolf BW-S6 is an exception, offering for a modest price, a dual-port 30W QC 3.0 charger with Australian plug which features a compact size, two independent outputs, decent circuit design/heatsinking and high efficiency (~85%).
While the charger does operate correctly as a QC 3.0 charger, the charger itself and the IC it uses are both not approved by Qualcomm. Likewise, while the charger has an Australian plug, it doesn’t have the appropriate regulatory markings. Despite this, it does produce safe voltages and exceeds the claimed output current/power capabilities (perhaps a bit much).
On the downside, the ripple voltage is a little high for my liking, but perhaps not unsafe for devices as it is mostly very short switching transients. It seems to reflect the limited secondary-side filtering that is in the design. The primary side capacitors are also of dubious quality, although the unit is properly fused with input suppression, so at least it should not result in catastrophe. The LED inside the opaque case is also a strange design choice.
But more than that, I finally got around to building the necessary equipment that allows me to begin properly testing QC-capable chargers, which was in itself, a worthwhile journey. Perhaps I should put in a little more effort to fully automate the testing …