Lithium Balancer / Charge Guard
PIC-based balancer circuit
Balancing lithiums is an essential requirement but has some complexities and risk. So let me start by recommending that you buy a charger with an integrated balancer. Let me also recommend that you measure cells rather that always balance them. Balancing may be needed but balancers have been known to damage packs. That said, this page describes some of the issues involved in balancing lithium cells. It also presents a cheap and cheerful circuit that works well. I also have a more sophisticated PIC-based circuit.
It is important that you understand the characteristics and risks of lithiums which I describe on my main lithium page. Consider A123 cells if you want a more forgiving chemistry. Most lithium cells are not a very robust so it is common for the voltages of individual cells to drift. This results in 'low' cells going below voltage in use and 'high' cells being overcharged. Both damage cells. Both means you throw them away! YOU HAVE TO KEEP LITHIUMS IN BALANCE!
Two types of balancing circuits
Charge Guards - A charger cannot normally see the voltage of individual cells. So it's easy for high cells to be overcharged which damages them. The 'charge guard' approach requires a circuit for each cell to bleed off its share of the charge to prevent it from being over-charged. They are therefore most effective if calibrated for your charger. Many are set to 4.3v/cell which makes them largely ineffective!
Balancers - I use this term to describe the device which simply levels the voltages of all the cells. It does this at its own rate independant of (and likely to be lower than) the charge current. Inefficient or low current designs (as most are) may take longer to balance the cells than to charge them. So on cells that are significantly out of balance it may help balancing before charging. Balancers tend to be cheaper than charge guards and usually require no calibration.
Please note that all circuits draw current even when their balancing or 'charge guarding' duties are done. So all designs will flatten and potentially kill your pack if left connected for long periods (eg: overnight). So please remember to remove your balancers when the job is done.
This rest of this page covers a simple cheap balancer which works very well on 2 or 3 cells. It's good for general use but is a bit inefficient so if you want a better one, higher currents or for more cells than consider my PIC-based circuit.
The circuit on this page was started by Sebastian Zettl and I think first appeared in this thread. Richard's Szym added components to allow it to balance at higher currents. I have merged a couple of his suggestions and made some minor changes based on suggestion by James Hopper and our experiences with the design. I have 'our' design as a veroboard layout, in Eagle for those than want that format, and an image file for PCB creation (best option if you can etch). The layout of the PCB version is broadly the same as the veroboard one, but much simpler to construct with only one link on top.
The circuit uses a resister-based voltage divider to provide a reference in exact proportion to the number of cells. These are the three 10k 1% resisters and are key to accurate results. Perfectionists would buy extra resisters to match them and improve balancing accuracy slightly. With this approach it makes no difference whether the pack is empty or full and therefore requires no calibration. The circuit is powered by the pack whether connected to a charger or not so can be used while charging or on its own.
An op amp (the LM358) compares the above voltage reference to what it sees from each cell junction. It sources current from the top of the pack when the junction is low, and sink current to ground when high. There are scenarios when low cells may be charged slightly, but in general it sets up discharge loops which eventually reach equilibrium but in a slightly inefficient way. So cells tend to get discharged more than they need to but despite this it does a good job with 2 and 3 cells. I don't believe this approach has a natural equilibrium with 4 or more cells.
The role of the second op amp (L272) is purely to increase the current capacity from the LM358's 50mA to a theoretical 700mA or more. In reality this is throttled by some high wattage 47ohm resisters to around 250mA. Even at this current they get hot enough to melt the heat-shrink covering so I now add an aluminium plate above them to act as a heat sink. The L272 also gets hot when working to capacity.
Both chips actually comprise two op amps so one of each is suitable for a 3 cell pack. Experience has shown that club mates sometimes connect their balancer to the packs the wrong way round. This reverses the polarity and tends to fry the L272 so if you use IC sockets, replacement will be easier.
By default, this circuit is designed for three cells with its three resisters. To balance just two cells requires you to bypass the middle resister with 'JP1'. Both 2 and 3 cell packs must be connected to the + and - of the balancer. A 2 cell pack has just one middle cell junction which can be connected to either of the two middle inputs. Please note that if you plug a 3 cell pack onto the balancer with the jumper bridging out R2, the balancer will utterly drain one cell (so be VERY careful!). No jumper = 3 cells.
In a 3 cell pack there are two cell junctions. The LEDs indicate whether there is any balancing activity. Two LEDs are required for each junction because LEDs are undirectional devices and the balancing current could be flowing in either direction. At any point in time either two, one or no LEDs will therefore be illuminated (never three or four). The LEDs are only visible in daylight when the balancing current exceeds 60 or 70mA so leave the circuit connected for 30-90minutes after they are no longer visible. Do not leave balancers connected for long periods of time as they can flatten and kill small packs. Larger capacity packs naturally take longer than smaller packs. Measure voltages with a voltmeter if unsure (they must stay within their limits).
The components for the circuit shown cost under £5 in the UK. I have indicated Farnell part numbers because they had everything I needed. Please note that the plug is part of the circuit and any differential resistance will result in incorrect balancing. It is therefore essential that the plugs make good electrical contact and can handle balancing currents. Poor connections can destroy a pack.
Drop me an email if you get stuck.
TESTING THE BALANCER
The easiest way to test the balancer is to measure each cell in a pack first. Then connect the balancer for a while making sure it does not get hot quickly which could indicate an error. Then check that the cell's voltages have changed correctly. Over time the pack should balance to within 1 or 2 hundreths's of a volt. This approach can work but does not test all scenarios.
A better way of checking the balancer is to feed it with known voltages and make sure the LEDs come on as expected and the balancer tries to correct an imbalance situation. To test a 3 cell balancer properly requires 3 variable power supplies. Even if you have these it can be very difficult to get the voltages exactly equal. So a simpler, cheaper and better way is to set up a string of identical resistors. With the correct number (2 per 'cell') you can divide any input voltage with great precision.
I have used 27ohm resistors fed with about 4v per resistor pair. I use pairs so that I can halve the resistance representing each 'cell'. So to test a 3 cell balancer I use 6 resistors and supply it with 12v. The current through these is 74mA (12v / 162ohms) so the resistors get a little warm. 'Ordinary' resistors should be fine but they should be rated at at least 0.5W. Higher wattage and slightly higher resistance resistors would work as well. It is important for the pairs of resistors to be identical. I buy 1% resistors from Farnell who only sell 50 at a time. So you may need to buy a dozen and use a DVM as I do to select those that are closest in value. Mix and match so that each pair is the same.
The 6 resistors in series are used to divide the 12v supply into three identical 4v feeds to the balancer (see circuit below). All LEDs should start off being off (test 1). If one resistor is shorted out / bypassed with a temporary jumper (ie: the resistance representing that 'cell' is halved), you will now only have 5 resistors in series and the one tap will have a lower voltage. This is an imbalance situation that the balancer should try to correct (tests 2, 4 and 6). If you then short out one resistor in each of two pairs, those two will be low relative to the third thus forcing the third to be high (tests 3, 5 and 7).
The suggested circuit diagram is shown below. I also list the 7 tests mentioned above together with a truth table of what the LEDs should do. Please note this depends on which way round your LEDs are connected; your results may be the inverse but should have the same pattern. The last photo shows how I have extended my tester to handle 7 cells, and I have added coloured leads to make it easier to short resistors out.
Top of page
Back to main Lithium page
Click on the links below for other pages.
Art of the Possible ¦
Absence of Matter ¦