Review, Teardown & Test: Portable USB 18650 Li-Ion Rechargeable Fans (x3)

It’s summer in Australia and this one is another scorcher! Our summers do get somewhat hot, but not usually this hot. Data from my own weather station in my backyard shows that the temperature over the period from Christmas Eve to now has been unexpectedly warm.

The yellow line is at 30°C, the brown at 35°C and the red at 40°C. We’ve reached or exceeded 40°C for four days in a row, with ten days in a row reaching 35°C or more. Of course, the accuracy of my sensor and the way I’ve placed my weather station counts – but my records agree with the Bureau of Meterology to the 2°C level. No matter how you look at it, it’s been hot and uncomfortable.

That made me think of ways to escape the heat. Fanning yourself is one way which can improve comfort. I can still remember as a kid, owning a fan operated by two AA batteries and a motor attached to a hub with two flexible blades that relied on centrifugal force to deploy. Those units were very primitive and the current drawn by the motors could get the batteries “burning hot”. Fond memories indeed.

But move into the present day and now there are inexpensive USB rechargeable Li-Ion 18650-based portable fans that you can carry around. I’ve seen them rise in popularity recently, especially amongst tourists, so I thought I’d get myself a few just to see if they were any good. They weren’t too expensive either – about AU$4.50 each including postage from China, although you’d have to supply your own 18650 cell.

Variant #1 – JPG PG-FS (Blue)

As practically all of the units in this review share the same sort of construction, I’ll detail this first variant in the greatest detail and explain the differences in the other variants instead. I’ve decided that the blue unit will be my first victim.

The front of the unit is occupied mostly by a fan grille with curved fins, a badge label in the middle with a snow-flake design and a control panel featuring four LED positions and two buttons. This is covered by a label with a legend – noting that the label doesn’t quite fit the recess in the body correctly. If aligned to fit the recess, the buttons don’t line up properly – [sarcasm] that’s how you know you’re getting the best quality. [/sarcasm] Another neat thing is that it is a rechargeabie fan according to that label. There is a slot at the top to allow for a strap or lanyard (not included).

The rear is mostly dominated by an inlet grille featuring straight edges this time. It houses a battery door and positions for the four screws that hold the unit together. The label claims it to have an output voltage of 5-9V, power of 4.5W.

The battery door opens to reveal a location for an 18650 cell to be installed. In my case, my listing did not include any such battery, so I used a spare Panasonic 3400mAh cell, noting that the unit can accommodate “flat-top” cells, e.g. those harvested from old laptop batteries.

The grille extends all the way around the unit, with a white 5mm LED on the right side and a microUSB socket on the left side for charging.

The base of the unit is flat, with no feet. Unfortunately, this particular unit is not very well made, suffering some vibration in operation due to an off-centre fan blade, so the unit has a tendency to walk across the table on its own.

The unit is easily opened by undoing the four screws. This reveals the relatively simple insides.

The unit basically runs from this single-sided single-layer paper-type PCB with an unmarked 16-pin SMD microcontroller doing most of the work. The unit obviously has some switching-conversion going on to step up the voltage to the motor, but aside from that, it looks like all functions including charging and protection are handled by this one microcontroller. While this is a cost-reduction measure, any bug in the software will likely compromise the safety of the unit – one of my units managed to draw my Li-Ion cell down to 0.6V, so I can’t say I trust the software entirely.

I have dubbed this unit the JPG PG-FS based on the logo and text markings found on the silkscreen on the PCB.

The fan-blade can be pulled right off, revealing a regular motor in a metal shell. The shape and size strongly suggest this is a spindle motor for a CD/DVD player, which may be in excess supply given the fact that optical media is starting to become obsolete.

Variant #2 – SINAO-FAN005 (Purple)

Because of my curiosity, I decided to take apart all the units. This revealed three of the four units were internally identical, but the purple unit was a little different. The front did have F95B on the label.

Taking it apart, there is a subtle difference to the plastic mould with the buttons having a different profile on the rear, making them not quite perfectly interchangeable.

The PCB is still single-sided and paper-type, but with a much more sophisticated-looking design. This one is dubbed the SINAO-FAN005 based on the markings and instead uses an unlabelled 4-pin microcontroller. This one appears to have an old-fashioned discrete Li-Ion charger IC, which may be slightly safer, but still does the same trick of stepping-up the voltage to run the motor.

The motor is practically the same one – but in this case, it was so well attached to the blades that removing the blades ended up removing the motor itself from the rear casing. The motor was not marked at all, but externally, the unit claims the same output. This one was much better balanced, so didn’t vibrate like the blue one did.

Variant #3 – Everplus EP102-V3.1 (Black)

The third variant is not pictured above as this variant was purchased by my father locally at a higher price, inclusive of the 18650 cell. This one looks visually very similar, sans spelling mistake, but the lanyard slot is a rounded diamond shape instead.

This unit is branded HADATA and instead claims to be 5W power. There is an additional warning for the battery to observe polarity. This unit also differs slightly in having two rubber-feet strips to stop it from “walking” across the table.

Unlike my units, this one features a fairly nice fibreglass PCB with tin plating. It is still single-sided, using an unlabelled 16-pin microcontroller to do all the work. The PCB is labelled Everplus EP102-V3.1 and is dated 29th April 2016. One positive difference is that this fan obviously uses a brushless DC motor similar to that of computer fans, which should be more efficient and smoother in operation. It’s already clear that not all these similar-looking fans are made the same.

Performance Testing

Here at the Gough Institute for Portable Rechargeable Fan Research*, I have utilised a plethora of test equipment to characterise just how these fans perform electrically. To do this review, I used my Rohde & Schwarz HMP4040 Programmable Power Supply and Keithley Model 2110 5.5-digit Digital Multimeter to power and measure the performance of the circuitry.

* Not a real research institute, even though I am a fan of research …

In order to obtain the results, various modifications were performed to the circuit board using 0.5mm^2 (20AWG) wire to supply power and measure voltages/currents while the unit lay secured to the bench using a few blobs of Blu-Tack.

Operation Modes

Before we get into the actual test results, it’s worth mentioning that despite all units having four LED indicator holes and two buttons, their mode of operation is slightly different and this may be able to help distinguish units.

The JPG PG-FS unit cycles through low-medium-high-off when pressing the green button. The LED is activated by holding down the green button and can run concurrently with the fan. The red button needs to be held down to turn off all functions on the fan. When short pressing the red button, the LEDs indicate the battery remaining capacity, although very inaccurately. On low battery, the operation mode indicator blinks before powering down.

The SINAO-FAN005 cycles through low-medium-high-LED-off with the green button and does not allow for simultaneous use of the LED and fan functionality. The red button turns off the fan immediately, with no battery testing capability. The unit shuts down when power becomes too low.

Finally, the EP102-V3.1 cycles from high-medium-low-off on clicking of the green button. To activate the LED, the green button needs to be double-clicked, whereas the red button stops the unit immediately with no battery testing ability. On low battery condition, the unit shifts down in speeds before powering off entirely.

Input Power vs Input Voltage

Two of the fans claimed to be 4.5W, with the final one claiming to be 5W. Is this really the case? The easiest way to determine whether this was even plausible would be to measure the power input – the power output cannot exceed the input.

For the JPG PG-FS, in high output mode, the maximum power draw was measured about 2.8W. As a result this is definitely not a 4.5W fan. The power output is relatively stable with voltage, which is a good feature.

Unfortunately, the fan posses very poor characteristics when it comes to voltage tolerance – high mode operates only down to 3.5V meaning a relatively restricted run-time of only about half the battery’s capacity. Medium is better, with low cutting out at about 3V, protecting the battery from overdischarge.

The SINAO-FAN005 has a maximum power output on the high mode of about 4.1W. This is another fan which is not 4.5W, but does get close. But this is only achieved when the battery is full – this unit has terrible regulation with a power output that is very much dependent on battery voltage. As a result, its average power output is realistically not any better than the JPG PG-FS above.

The unit basically operates all modes down to about 2.7-2.9V, meaning all modes can utilise the full battery capacity without too severely discharging the cell.

The EP102-V3.1 seems to have a power consumption that varies due to its relatively coarse regulation abilities. In the high mode, it was seen consuming about 5.8W at its peak, so there is a possibility this could be a 5W fan, but it would have to be very efficient to do that.

During this trial, we saw the fan operated in the high output mode right down to a shut-down, without stepping down into a lower speed mode. This seems to be slightly anomalous, as a later test for efficiency seemed to show that it did step down to low as expected before shutting off. I wonder if this may be a software flaw or an interaction with the lab style power supply which may have much lower internal resistance compared to a Li-ion cell.

The fan does step down modes at certain voltages, meaning that higher modes do not operate when the battery is partially depleted, with the unit cutting out at about 2.7V allowing for the safe utilisation of the full capacity of the cell.

Output Voltage vs Input Voltage

The second big question I wanted answered was whether the unit did output 5-9V as claimed on the label on the rear.

For the JPG PG-FS, the output voltages were well regulated throughout the operating range, with low putting out about 4.5V, medium putting out 5.3V and high putting about 6.3V. The low output was below 5V and the high output was quite short of 9V, so the label is being very liberal about its ranges.

On the other hand, the SINAO-FAN005 is very liberal with its output voltage regulation. In fact, I hazard to guess that there is no output voltage regulation and it merely may be relying on fixed duty cycle operation instead. The output, as a result, ranges quite significantly – low is about 2.95V to 4.75V, medium is about 4.1V to 6.35V and high is  about 5.1V to 7.95V. The voltages fall very much below 5V for most of the operating window and the high voltage does not reach 9V either. That’s a disappointment.

The Everplus EP102-V3.1 seems to have some limitations in its duty cycle regulation steps, but it does seem to have feedback regulation of sorts. The voltages for low ranges from about 3.4V to 5.15V, medium from about 6.2 to 7.2V and high from about 7.8V to 9.15V. This one does meet the voltage output expectations on the high side, but the low side does dip below that claimed. So far, it seems the Everplus may well be the “real deal”, with the others emulating it … or maybe they’re all just emulating another product entirely.

LED Current vs Input Voltage

I thought it was also good to investigate how the LED is being driven. From what I can tell, it’s nothing special – all three show current-voltage relationships which suggest they are basically resistor current limited, controlled by a transistor and the microcontroller. The LED current of the JPG PG-FS is higher than the others, reaching 23.844mA, with the EP102-V3.1 reaching 16.199mA and finally the SINAO-FAN005 reaching just 9.2mA simulating a fully charged cell.

On all units, they cut-out at about 2.7-2.8V which should avoid over-discharging the lithium ion cell. Of note is that the Everplus seemed to increase its current consumption below 2.35V – I wonder if this is because the microcontroller used has breached its minimum voltage at that level – whereas the JPG PG-FS appears to show this behaviour at about 1.95V. In all cases, it seems that it’s not a good idea to store these units fully depleted as the quiescent current does not seem to be that low, especially for the JPG unit. This is further investigated later in this review.

Power Conversion Efficiency

It’s all fine to understand how much power is going into the unit from the battery, but how much of that power is making its way to the motor? In order to test this, I had the motor output measured in dual-measure mode on the Keithley Model 2110 – the current measurement introduces a small burden resistance, so results will never be quite 100% accurate, but they should still be within a percent.

While testing of the JP PG-FS on low, it seems that the unit powered off unexpectedly during the test and I did not catch it. As a result, we don’t have too many values for the efficiency, but we can already estimate average efficiencies of 81%, 77% and 70% for low, medium and high settings respectively. This is not a particularly great result.

The SINAO-FAN005 fares much better with fairly stable efficiencies 85%, 83% and 80% respectively. The downside is that the voltage regulation is very loose.

Finally, the EP102-V3.1 shows quite a strong variance of efficiencies, depending on battery voltage – but still averaging 85%, 83% and 78% respectively.

Of note that this only focuses on the electrical conversion efficiency of the circuit and does not account for the further losses in the motor (for example). So the actual true work-done by the fan will be even less than the powers implied by the efficiencies given above.

Charging Current Profile

Another item worth checking is the charging current profile to see if it respects the Li-Ion battery (ensuring safety) and whether it charges quickly and efficiently.

The JPG PG-FS displayed something along a typical charge curve with a few minor bumps. The unit was observed trickle-charging the cell at a lower current before stepping up to a bulk charge. The technique was a basic “linear” charge, dropping excess voltage as heat, regulating the constant-current phase fairly well at about 650mA. While this isn’t the fastest charge possible, it is still similar to what many discrete charging ICs offer. The current tapers off and the charge terminates at a current of 80mA at a voltage of 4.229443V, slightly high but considered within tolerance.

The SINAO-FAN005 was a big disappointment. While it seemed to offer a discrete charging IC, its performance was very unexpected to say the least. Charge current was averaging about 350mA, with a performance that seems potentially thermally constrained due to the paper PCB. As a result, long charge times were the result, with an initially average 525mA charge current falling to just 250mA initially, before rising as the charge progresses . The charge terminated at about 70mA with a termination voltage of 4.131765, slightly undercharged but safe.

The EP102-V3.1 was tested charging its supplied 18650 cell, which doesn’t appear to be very high in capacity. Regardless, it was observed that the EP102 uses the same coil as a buck converter during charging, more efficiently charging the battery at a higher rate than the graph might indicate (as it measures USB current input to the fan). The current did seem to vary in some discrete steps, with a slight blip downward at one stage for unknown reasons. The downside was the termination voltage was a little high for my liking at 4.286202V which could pose a lifetime-shortening risk to the battery and reduce safety slightly. The termination current appears high, but this may have been artificial as a result of the measuring of a “spiky” switching load coupled with losses in the converter (the inductor was quite hot). The behaviour could be a side effect of the low apparent capacity of the supplied cell as well.

Motor Characteristics

As I didn’t know what the ratings of the spindle-motor like motors were, I decided to run a sweep test to check what the motor’s voltage, current and power characteristics were from 0-15V (which appeared the absolute limit before possible melt-down).

It seems the motor is probably happiest about 0-7V for a “long” life – at 15V, we can see signs that it was moments from potential burn-out with the sharp increase in power draw. If the units did provide 9V to the motor, the power consumption still would not have reached 5W – so the labelling of power is very loose to say the least.

Summary & Other Spot-Checks

In the end, the data can be summarised in the table above. It’s clear that the EP102-V3.1 leads in the majority of categories, whereas the JPG PG-FS mostly lags behind. Results not covered in the sections prior are the quiescent current results, showing the EP102-V3.1 having a very healthy 18.3uA quiescent current, with the SINAO-FAN005 using 148.1uA and the JPG PG-FS consuming a whopping 462.7uA. High quiescent draws will waste battery power and can lead to over-discharge of the battery especially if stored for long periods in a discharged state.

Note that the output power is the power into the motor – the inefficiency of the motors will further reduce the true work done and the actual output power.


Hot summers can be relatively uncomfortable. One way people have traditionally made themselves more comfortable in such conditions is to fan themselves. These relatively inexpensive USB-rechargeable portable fans are gaining in popularity, often boasting a relatively “generic” design that is almost indistinguishable at a glance. However, as this testing shows, what you get can be completely different even when it comes to operation.

Of the four fans I purchased, there were two variations based on inspection of the internals – the JPG-PG-FS and the SINAO-FAN005. This differed from a more expensive fan purchased locally by my father – the EP102-V3.1. While they share similar external appearances and rating labels, the EP102-V3.1 was ahead of the rest boasting a brushless DC motor, higher output powers approaching the label ratings, low quiescent current draw and switch-mode charging of the cell. The main downside appears to be a slight over-charging of the Li-Ion cell. It pretty much wiped the floor compared to the others.

The JPG PG-FS comprised three of the four units I purchased and was unfortunately the most disappointing design, falling short of the label rating by about half in terms of output power, boasting the lowest efficiency of the bunch, cutting out at a higher voltage (i.e. not offering higher airflow modes despite ample remaining cell capacity), average charging rates and high quiescent current draw. The only thing it was best at was having a higher output on low mode and higher output on the LED, although it too slightly overcharged the cell. While this was the only unit to have a battery check and low battery indication, the feature was hampered by its inaccuracy.

The SINAO-FAN005 was a mixed bag. While its design appeared more sophisticated. it instead charged slowly with a rate more akin to a Bluetooth speaker and terminated slightly early. The charging current profile was inconsistent, the LED drive current was the lowest in the batch and the output of the fan varied significantly as a function of the battery voltage indicating poor regulation. The average power output was, for the most part, equal to the JPG PG-FS, although the maximum output was greater. The SINAO-FAN005 offered quite competitive electrical efficiency and use of the cell capacity, with a reasonable quiescent current draw.

Both the SINAO-FAN005 and JPG PG-FS units appear to repurpose CD/DVD spindle motors in the design, possibly due to a market surplus now that optical media is starting to become obsolete. These motors are slightly less smooth compared to the brushless motor used in the EP102-V3.1.

Unfortunately, as with such “generic” products, it’s unlikely that you will know exactly what you are getting when you buy them without being able to inspect very closely or take the unit apart. Unfortunately, it seems both the JPG PG-FS and SINAO-FAN005 have some severe compromises, likely due to the cost-cutting measures needed to bring them down to the AU$4.50 price point (without battery). While the EP102-V3.1 was more expensive (cost not known), it did come with a battery and has the superior performance based on a more sophisticated design.

So while it seems the EP102-V3.1 may be the ‘real deal’ with the others merely being lesser imitations, I wonder if all three may well be copies of another product entirely.

About lui_gough

I'm a bit of a nut for electronics, computing, photography, radio, satellite and other technical hobbies. Click for more about me!
This entry was posted in Electronics and tagged , , , , , , , , , . Bookmark the permalink.

Error: Comment is Missing!