Around two weeks ago, a reader contacted me with an offer of a Qualcomm Quick-Charge 3.0 capable USB car charger that they had obtained for cheap but was not certain about using. In fact, they weren’t even certain if it was quick-charge capable as it had claimed. After confirming their willingness to send it all the way to me, blocking out some time in the evenings to run some tests and analyse the data, I seem to have arrived at an answer.
The Item
The charger looks like most modern car chargers, featuring a curved design and a shiny black plastic finish. The contacts protrude out of the case subtly, but are shaped such as to be enclosed.
The only identification is seen on the underside where it carries the Qualcomm Quick Charge 3.0 logo and a model number of LZ-681. It claims to accept an input of 12-32V, making it suitable for both car and trucks. The output claims to be capable of 6A at 5V, 2A at 9V and 1.6A at 12V. It is also marked with FCC and CE markings, which I am skeptical about. It seems to be a misuse of the Quick Charge logo, as this model is not listed in the official list of approved devices.
The tip positive connection is surrounded by a gold coloured ring, but this is merely plastic. The unit does not appear to have any replaceable parts (e.g. fuses) from the outside.
The user-facing end has two orange coloured USB ports, with the top port marked for Quick Charge and the bottom port marked for 3.1A (i.e. non-quick charge). Normally, regular speed ports are coloured black or yellow but I suppose this is merely convention.
Teardown
In order to test this effectively, tearing the unit apart was a prerequisite as I don’t have a car nor do I have any cigarette lighter sockets to hand.
The front fascia is secured by plastic pegs to the body and is made of partly translucent blue plastic. From there, I couldn’t find an easy way to extract the PCB despite pushing all the bits in …
… so I got out the hot knife and just cut through the casing to cut near the mid-line of the case.
The board is internally marked TE-348, dated 1st April 2020. The design appears to have two separate converters – the normal port using U2/L1 near the input capacitor E1. The switching converter chip is marked SD2006BH, although I could not identify the make. There is a footprint for C4, likely a ceramic capacitor, omitted to save costs. The output is instead filtered through E2.
The quick-charge port uses U2/U3 and the large toroidal inductor L2. U2 is unmarked, however, U3 is marked KTG6601Q. This seems to be a clone of the Fitipower FP6601Q Quick Charge controller IC which is on the approved list, but as it is a clone, there is no guarantee that it performs as well as the original. This merely provides an input to the switching converter U2, which remains anonymous.
A number of ceramic capacitor positions were unpopulated – C1, C2 and C5 which suggests that the ripple performance may be a bit impaired.
The underside does not have much in the way of connections, but mostly solid copper planes being used as a heatsink with stitching vias to the other side. Depending on the efficiency of the converters, this may not be enough to ensure stable operation at high temperatures and high loads. Notably absent in this board is any fusing whatsoever – in case of catastrophic failure, the board may take out the fuse in the car instead.
The low cost extends to the electrolytic capacitors, all of which are “FMZ” branded. This is not a known reputable brand. It’s interesting to see that the input features only a capacitor for filtering, which is not likely to do much for higher frequency noise components from the vehicular power supply.
The output capacitor for the Quick Charge port is rated 220uF at 16V, as the output can reach 12V. The regular port uses 220uF at 10V instead, with an output fixed at 5V. The input capacitor (which was slightly “crushed” during disassembly) is rated 100uF at 35V, so it can handle the “rated” 32V input.
Testing
Output voltage versus current and efficiency tests were completed concurrently. Power was supplied from a Rohde & Schwarz HMP4040 supplying various voltages simulating a 12V and 24V automotive system at various levels of battery charge (12V, 13.8V, 14.5V and double for 24V). The outputs were loaded, one at a time, using a B&K Precision Model 8600 DC Electronic Load with a home-made USB jig. All connections from power supply and to load were four-wire kelvin connections to ensure power loss in the cables were compensated. Testing was automated using Python 3 and pyvisa, with an Arduino Leonardo and QC3Control used to test the Quick Charge capabilities.
Efficiency calculations are based on total consumed power and total delivered power on the port being tested. The quiescent power for the other unused port has not been compensated for, thus the efficiency figures reflect the efficiency of using just one port at a time.
Ripple and noise was measured using a Rohde & Schwarz RTM3004 oscilloscope and RT-ZP10 10Mohm passive probes. Signals were measured using a 20MHz bandwidth filter on the input at a range of currents with the output at 5V.
Note that as testing is performed “on the bench” using high quality power supplies, the observed performance is considered best case – in a true automotive environment, there may be copious amounts of noise on the input power which can interact with the converters causing additional stress and increase the voltage noise on the outputs. Another consideration is that the PCB was taken out of its enclosure, thus received better ventilation than it would have received inside its original casing.
Quiescent Current/Power
While plugged in, the unit does not start generating output until about 6.2V. From there, quiescent current is about 32-36mA with a quiescent power of about 0.4W (12V) to 0.8W (24V). This is not terrible, but if you’re using it in a car which does not switch the lighter socket, you’d want to remove the unit to prevent the additional drain on the battery.
Voltage versus Current – Port 1 at 5V
The first port showed very consistent behaviour for the most part, with the delivered current reaching above the 3A claimed, up to 3.2A before the unit drops the voltage significantly and then “cycles” on and off.
The developed voltage at light loads slightly exceeds what is permissible for USB, reaching about 5.27V. This is unlikely to cause immediate damage but is definitely out of specifications. It shows an interesting “dip” in output voltage before rising again, slightly.
Voltage versus Current – Port 1 at 9V
At 9V output, the unit far exceeded the rated current of 2A, instead delivering 2.7A and even reaching 3A for lower input voltages. I did not decide to push it further than that.
The voltage standard for Quick Charge is not as strictly defined, but the output is definitely about 0.2V above the requested voltage with the same characteristic curve.
Voltage versus Current – Port 1 at 12V
Running at 12V output, the unit again delivered much more current than the rated 1.6A, reaching up to 2.7A and even reaching 3A depending on input voltage.
With an input of 12V, the output is unregulated as the buck converter cannot boost the voltage to meet the requirements. As a result, a car with a low battery may not deliver the full 12V to a connected QC 2.0/3.0 device. At 13.8V, the output was regulated, again about 0.2V above the requested voltage.
Voltage versus Current – Port 2 at 5V
Testing of the second port showed that the unit is incapable of sustaining an output of 3.1A as it claimed on the front, instead delivering about 2.2A at most for 27.6V and 2.68A for 12V. The bigger issue is that when powered at 29V, the converter failed entirely and supplied 10V transiently to the USB port before ceasing to function. From there on, the unit could not deliver any useful current and was not able to maintain 5V and suffered permanent damage. As a result, it seems the second port on this unit is not suitable for use in trucks especially when their batteries are approaching full.
A look at the converter voltage output behaviour shows that for higher voltages, especially at light loads, the USB output voltage exceeds the maximum 5.25V expected from USB. This also happens for lower input voltages at heavy loads. While perhaps not high enough to damage a USB connected device immediately, it is out of specifications. At lower input voltages, there is still a “bumpy” appearance – this may be due to a change in switching modes. The output voltage increases as current increases, which does help overcome some of the resistive losses in the cable, but may also confuse some devices which try to detect the power supply’s capabilities by looking for a voltage drop.
Efficiency – Port 1 at 5V, 9V and 12V
Conversion efficiency for the first port was generally quite good. As expected, efficiencies were better for smaller differences in input and output voltages, but peak efficiencies of about 90% were achieved for 5V which was quite impressive considering the losses in quiescent current for the other port was not taken into account.
Efficiency – Port 2 at 5V
The second port was less spectacular, achieving about 85% peak efficiency, which is not bad if we consider the contribution of the other port’s quiescent current is included. However, this does suggest that perhaps heating could have played a role in the failure of this port.
QC 3.0 Continuous Mode Voltage vs Step
Testing with a load of 500mA, each voltage step was around the 200mV as required, although sometimes the step was a bit smaller at 180.5mV. The lowest voltage reached in continuous mode is about 3.85V.
Ripple and Noise
Ripple and noise could only be tested for Port 1 as Port 2 failed during earlier testing. The observed ripple voltages were as follows:
- 500mA – 144.41mV peak-to-peak, 28.682mV RMS
- 1A – 156.11mV peak-to-peak, 29.908mV RMS
- 2A – 191.07mV peak-to-peak, 32.127mV RMS
- 3A – 231.81mV peak-to-peak, 37.75mV RMS
The output showed some very short duration spikes which could cause radio interference but seems likely because of the omission of fast-acting ceramic capacitors. The ripple voltages were above the 50mV that the ATX standard for PCs usually requires and above the 120mV that many “original” OEM wall-plug chargers are capable of, but not by much. This may be not ideal, but when combined with the slightly high output voltage, it seems quite probable that it could stress some USB devices beyond their intended limits.
Conclusion
The LZ-681 QC 3.0 + 5V/3.1A Dual-Port USB Car Charger is a cheap product and its performance seems to show it. While featuring the Qualcomm Quick Charge 3.0 logo, the unit is not officially approved. Internally, its design features a minimum of input filtering, limited cooling, no fusing at all and uses a clone port controller IC for quick charge functionality. While the unit is capable of providing QC 3.0 functionality, both USB ports featured voltages which strayed slightly beyond the upper limit of 5.25V permitted by the USB standard at times. Further to this, the second port failed under load while being tested at 29V, making it unsuitable for automotive truck usage. The ripple and noise tests showed sharp switching transitions which increased the measured figures, a consequence of the cost-saving measures of omitting SMD ceramic capacitors.
While the unit may seem to function just fine, especially for 12V car usage, the output of the tested unit is not strictly within the expected limits and could cause stress on the connected devices which could lead to heating, unexpected operation or their premature failure. The results are expected to be even worse when used in an automotive environment due to the “dirty” power coming in and the enclosed nature of the device operating in a (possibly) hot ambient environment. As usual, you get what you pay for.