Fantastic or fantasy? Testing LiIon battery fakes

In some Chinese shopping sites you will find LiIon batteries with fantastic capacities. You can find 18650-type batteries from UltraFire with a capacity of almost 10Ah.


That doesn’t sound too good to be true? It is.

Let’s do some fact checking first and look at the UltraFire web site. You will see that the UltraFire 18650 LiIon batteries have a capacity of 2500mAh. Here it is clear that these batteries must be fakes. Can somebody build a battery of the same size with almost 4 times the capacity? No! This just isn’t possible. These batteries are fakes and the capacity isn’t specified correctly!

But do they provide at least the same capacity? Let’s check. The setup for the capacity check is simple. The battery will be fully charged and then discharges with a constant current of 500mAh (this value had been chosen to limit the stress on the batterie, but also make sure that the measurement doesn’t take too long). You should not discharge a LiIon cell below 2.7V, therefore this is the point where the battery is “fully discharged”. Now you only have to measure how long it takes from full charge until the cell voltage is down to 2.7V. Then the capacity is very easy to calculate as time x current.

This measurement is easy for me as my electronic load that is used for power supply tests has a battery mode that not only turns off the load at a given end voltage, but also automatically calculates the capacity.

Measurement alternatives
You can setup a measurement like this even with a simple resistor, but the current won’t be constant. However, for a rough approximation, you can just assume that the cell voltage is constant at 3.7V the whole time. The real capacity will be a bit lower. Just make sure you turn off the current when the cell voltage is at 2.7V. Fully discharging the cell will most likely damage it and make it unusable.


Now comes the funny part. I did not expect that the capacity is even close to the specific value, but have a look at these values of this “9800mAh” battery:

LiIon capacity test

Really? Around 1100mAh? That’s 11% of the advertised capacity. It also means that this cell performs worse than standard NiMh cells. It is hard to understand how you can build a LiIon battery that performs that bad.

If you wonder, why the display shows 3.143V: the battery voltage recovers when turning off the load. However, if you turn on the load again, it will drop again to 2.7V very fast.

I’ve tested two 18650 batteries. The perform both similarly bad.

I’ve also tested two 16340 batteries that are advertised as “2300mAh”. Their real capacity was also around 10-11% of the advertised capacity (230-240mAh).


While there are some cheap offers of LiIon batteries on Chinese shopping sites, don’t even try these things. They’re not worth any money. Also note that this test doens’t say anything about real UltraFire batteries. I will try to find some original batteries and perform tests with them in the future.

MP1584 buck converter module

This small module MP1584 buck-converter module seems to be a good solution to power small circuits from higher voltages. Especially cool with this chip is that it accepts input voltages up to 29V. This makes it a perfect candidate for additional circuits that connect to a KNX bus. But it’s not limited to KNX buses. If you want to build a WiFi interface for your Roomba, you also have to down-regulate the 15V battery voltage to 5V or even 3.3V. You want to something to power a circuit from a car battery? This seems to be a perfect circuit for there use cases.

mp1584With it’s tiny dimensions of 17 x 22mm it’s size is around the size of 2 MicroSD cards (and much smaller than a single SD card).

The output voltage of the module is controlled by a tiny potentiometer. You will need a multimeter to check the output voltage before connecting it to a circuit.

While the plain voltage range data seems to be quite good, how does the circuit behave on different loads? Some Chinese dealers claim that similar modules should handle 3A load. If you just look at the size of the inductor you might already start thinking that this most likely isn’t true. However, DC resistance of the inductor is approximately 10mOhm (I miss my Kelvin probes and therefore could only measure this very roughly).

Let’s start with 29V input voltage – this will bring everything to the limit. Output voltage is fixed to 5V in this experiment. Without any current drawn, it still looks a bit rough:


However, a voltage swing of (Vpp) 98mV is no problem at all. Between 0.1A and 0.9A the circuit behaves quite well. Vpp is around 0.2V (a bit lower on low currents, higher in higher currents)


We reach the end at 1A. Now the regulator doesn’t provide a stable output voltage anymore:

But what about lower input voltages? Let’s go the the other “extreme”: 9V (it will work with even less, but let’s give it a bit headroom).

Almost perfectly clean with no load:

A bit more stable than at 29V for currents between 0.1A and 0.7A:


And at 0.7A we reach the end of the useable current range:


Conclusion: This module works fine for almost every embedded use case: Arduino, ESP8266 WiFi modules and even a Raspberry Pi without any additional USB devices plugged in.

Here are 2 animation that shows the output voltage under different loads:

29V input, 5V output, 0-1A in 0.1A steps:


9V input, %V output, 0-0.7A in 0.1A steps:


What’s inside a cheap Chinese LED light?

LED lights have become cheaper and cheaper in the last years. But they are still a bit more expensive than Halogen lamps. To save money, some people might have a look at Chinese shopping sites. I wanted to know what you get if you pay less than 5$ for an LED lamp.

My delivery arrived like this:


Yes, it fell apart already during shipping. It consists of some PCBs with LEDs soldered together by hand and a cheap plastic cap. You don’t really want to use something like this on mains voltages – right?

However, this gives us the chance to look what’s inside of this thing.


Ok, this is really the simplest circuit you can think of. You see a suppressor capacitor (the red one), a bridge rectifier (the small black part with the 4 pins), a small 4.7uF electrolytic capacitor and a few resistors.

The two 51Ohm resistors are connected in parallel and they limit the current flowing through the LEDs.

Better designs use switching mode power supplies with a fixed current output. However, you won’t get this if you buy the cheapest stuff on the internet.

Especially the fact that this ding falls apart on the slightest touch and every part of the circuit runs at mains voltages makes them extremely dangerous. After the the rectification you can expect almost 300V DC on the board. This is even more dangerous than 230V AC.

Do not use something like this!

433MHz window sensor

On eBay, Alibaba or other web sites, you can find cheap Window sensors that will send a data packet to inform you about the state of your windows. They often costs less than 10$ per unit, which makes it easy to test them.

They look like this:


But what’s inside? Let’s have a look. The part that is mounted on the window itself is basically just a magnet. You might already know what will be on the other side: a reed relay. But what else:

The electronic circuit is based on the HS1527. The nice thing about this chip is the 1uA standby current. Even with a small battery, the battery life will be relatively high. It is powered by a small 12V battery – I haven’t seen this battery type before.

The good thing about the chip is that you can use a lot of cheap 433MHz receivers connected to a Raspberry Pi or Arduino to receive the signals.

However, there is one major downside: This sensor sends only a data packet when you open the window, not when you close it. This is ok if you want to use it to detect break-ins, but it won’t work if you want to detect if a window is open (e.g. if you want to turn off the heating).

Therefore I can’t recommend this sensor type for home automation use.