Li-ion cells and cold temperatures – Part 2

26F Cold tests 2 Descr

Please click on the image to enlarge it.

After the first cold tests I decided to go a step further and do some tests with even colder temperatures to see the behavior of the cells.

For the Orange tests the cell was padded this way:


Side view with padding on the sides removed:


These tests are interesting for people like me that uses lithium ion cells in cold climates. Especially in my case, I use the cells in an electric bicycle. It’s important to have some sort of thermal insulation on the battery when riding in cold temperatures, the performance improves dramatically and the battery cycle life is preserved.

Li-ion cells and cold temperatures


Please click on the image to enlarge it.

I was curious about the behavior of lithium cells at low temperature. In particular I was expecting capacity loss and voltage sag  compared to the standard test temperatures of 20-25°C.

The tests above were conducted on a single cell, a Samsung ICR18650 26F, discharged at 1C.

The voltage sag and loss in capacity are lower than I expected. I am planning to do some -10°C tests to have even more data to compare.

2013 in review

The stats helper monkeys prepared a 2013 annual report for this blog.

Here’s an excerpt:

The concert hall at the Sydney Opera House holds 2,700 people. This blog was viewed about 12,000 times in 2013. If it were a concert at Sydney Opera House, it would take about 4 sold-out performances for that many people to see it.

Click here to see the complete report.

Mixing Battery Chemistry : Li-ion and Li-po


I have about fifty brand new li-ion cells (18650 format) in my lab and I am tempted to build another battery pack for my bike .

The cells in question are 25 pieces of LG ICR18650C2 ( 2800mAh) and 25 pieces of Panasonic NCR18650B (3400 mAh) . 50 cells are not enough to make a battery composed of 20 cells in series , because of the low discharge capacity ( C rating). To buy more cells would be the easiest solution , but considering shipping , customs and cost of the cells (at least 2-3 dollars per cell with a minimum order of 100 pieces) is not a cheap solution.

Looking on the net, a well-known seller of RC toys and products ( HK) has very affordable prices for lithium- polymer cells. There is also a warehouse in North America and the shipping is not that expensive compared to the li-ion 18650 cells.

So I decided to mix the two different battery chemistry’s , Li-ion and  Lipo to get a battery with adequate capacity ( about 8- 10Ah is what I aspire to ) and discrete C -rating. On average I use between 4A and 10A on my bike on a level surface , with peaks of 35A (rare) when I am in a hurry.

I could not find any feedback on the feasibility and the behavior of the cells, so I decided to do some testing on my own.

The two types of cells have the same charge voltage ( 4.2v , 4.1V if you want to conserve battery life, prolonging the useful life cycles ), but different levels of low voltage , Lipo does not like to be discharged under 3.4v ( under load) , while with the li-ion can reach 2.5v ( under load) .

I have done several tests and the LVC mixing the two cells  (li-ion  and lipo connected in parallel) seems to be 3.0v . This way, at the end of the discharge LiPo cells are still cold woth no signs of “stress” , while the li- ion batteries are hot as usual , a typical behavior of this chemistry.

How  the tests were performed.

Cells used :

– Ah Turnigy 5.0 20C ( Lipo )
– LG ICR18650C2 ( 2800mAh)
– Panasonic NCR18650B (3400 mAh)

The cells were charged all at the same voltage and then connected in parallel . This way I obtained an 11.2Ah nominal battery . Then the battery was subjected to a discharge test using the CBA4 battery analyzer .
The discharge tests were performed at 5A, 10A and 15A continuous , stopping the discharge at 3.0v .

The tests compare two kind of batteries:

  • “Mixed” Li-ion +  Lipo battery which is 11.2Ah nominal.


  • “Pure” Li-ion battery which is 11.4Ah nominal.


Here are the results :

Let us look at the behavior of a battery composed of only 18650 cells with similar capacity (11.4 Ah nominal vs 11.2 Ah  nominal), discharged at the same current with the cutoff at 2.5v :

Now, we directly compare the test results of mixed cell with the “pure” li-ion cell of similar capacity.

As can be seen from the graphs the lipo cell contribute for the most part of the graph, trying to limit the voltage sag. When they are at the end of their capacity (3:45-3.60v on the graph), you can notice a drastic drop in voltage because only cells li-ion are contributing to the load.

Cell-logs as LVC BMS


In the video below you can see my latest creation.

It’s a low voltage cutoff that uses cell-logs.

Cell-logs are battery monitors for lithium batteries. They display the voltage of each single cell in a battery pack. Each cell log can be connected to up to 8 cells. Cell-logs can also send an acoustic alarm if one or more the cells is above or below the preset voltage. This voltage can be programmed by the user.

On top of the acoustic alarm, the cell logs have an alarm port that closes or open (like a switch opening or closing a circuit) if the alarm is triggered.

In my project I exploit the alarm port of the cell-log and use it as a low voltage cutoff.

Because the alarm ports of the cell-logs are not isolated, it’s not possible to connect the alarm ports together if you are using more than one cell-log in your battery pack.

Therefore I designed and produced a board that isolates the alarm output of the cell-log with an optocoupler.

Using my board you can connect as many cell-logs as you like, connecting the output of the optocouplers in parallel.

This video explains how it works:

Using this system has many advantages, some of them are:

  • Can be adapted to any lithium chemistry
  • It’s programmable, the alarm can be triggered selecting several options
  • The single cell is monitored
  • Can be used with any current drain
  • Can be used in large battery packs
  • It’s more flexible than the standard BMS
  • It’s portable, the same system can be disconnected from one battery pack and used on another, even if they have different chemistry and capacity

This system saves your cells and your batteries from being over-discharged, saving them from getting damaged.

Feel free to comment or contact me using the site contact form (on the top right).

New Electric Bicycle. MKIII


Finally I found the time to write about my new electric bicycle!

This is my latest build and my best build so far.

As you can see from the photo, the bike has a small back wheel (20 inches), front suspensions, disk brakes and a central battery box that contains the battery (obviously), the controller and all the other connections (e-bike computer for example).

Here below I describe the process and all the modifications:

I started by buying a bicycle a few months back. I wanted something light and with front forks and disk brakes.


I bought the above bike for 250$, used and in excellent condition.

In the meantime I also got an used motor, a 9C9807. Originally the motor was laced in a 26-inches rim and had the hall sensors burnt.


I decided to use a sensorless controller, so I decided not to replace the hall sensors, but I replaced the original phase wires with 14AWG copper stranded wire as shown below.

Original wires (thin) vs new ones:




I decided to use a cheap sensorless controller, KU123 from BMSbattery. I upgraded the controller with 100V capacitors and IRFB4110 high quality mosfets to support battery voltages up to 100v. I also “beefed-up” the main power traces on the controller board to improve current flow on the traces.

Original controller:


High quality IRFB4110 mosfets:


Controller modified with new capacitors and mosfets. Please note that the capacitor legs do not touch and they are quite apart, it’s just a camera angle “optical illusion”.


Original board traces:


Traces “beefed-up”:


Then I got to work on the battery box. I wanted something better and larger than my previous bike, so I used a 11.5cm wide wooden board  and built it. Photos are pretty much self explanatory.






Side covers:


I attached the battery box to the frame using zip ties. It holds really well.

Lastly I decided to lace the motor into a 20 inches rim,spokes in a radial pattern, which turned out to be a pretty good choice. The bike has very good torque and can handle steep hills without overheating the motor.



I also added an e-bike computer based on arduino to complete the job, so I have real time data on battery usage, speed, range left and so on: