Balancing issues and circuits
Charging lithiums can be an extremely simple exercise. All they require is a supply with voltage and current limits. Many adjustable power supplies can do this. Simple supplies are cheap and easy to make. Lipo chargers made from power supplies work in the same way as very expensive chargers as the graphs below show. Their added advantage is that they are normally simpler to use, usually having just current to adjust.
This page is about building your own lithium battery chargers. Three projects are presented although the last one (LM2577) is largely obsolete now. I assume that you understand the charging requirements and risks involved (particularly fire) which I describe on my main LITHIUM page. The information and designs are free for personal use but may not be published elsewhere or exploited without my permission. I share my experiences to help you but accept no responsibility for its accuracy or anything you do with this information.
L200 DESIGN:- Variable current up to 2A
I've always wanted a fully adjustable current DIY lipo charger (and recently higher output). The L200 datasheet has an LM741 op amp design to achieve this. It works well but has a very non-linear action (worse with a log pot). So with James Hopper's help I have tweaked this design to produce near linear current adjustment with a standard linear pot. In my version, current is variable from about 100mA - 2A. The circuit is slightly more complex but is nicer to use. Here it is in EAGLE and image formats below.
R3 is a current sensing resistor which feeds the LM741 with a voltage that varies according to current flow. This gets amplified by the op amp which tells the L200 to change current. I have used a 100mm length of 0.5mm insulated copper wire for R3 (= ~0.010 ohms). The precise value is not critical. However, it interacts with the gain of the op amp and affects the minimum charge current and overall current range. Although I have tested with a value half this size (0.005 ohms), R3 should be larger rather than smaller than the 0.010 shown (eg: between 0.010 and 0.100 ohm). A larger R3 will give a smaller starting current.
The gain of the op amp is set by the ratio created by R2 (470ohm), R6 (10k pot) and R7 (250k trimmer). It also depends on the value of R3 above. If you want a lower start current it is best to increase the value of R3 as suggested above. However, reducing R2 (say to 220ohms) will have a similar effect.
L200's are internally limited to about 2A and this circuit makes an obvious correction if you go over this (you are likely to see current suddenly drop a bit as you rotate the pot over the 2A level). R7 is used to fine-tune the op amp's gain and I recommend you adjust it so that max current can only get to just under 2A. If you can't find a 250k trim pot then use a 100k trimmer with 150k resistor in series between R6 and R7 (eg: as shown in the first photo above). R7 does not require great precision so a single turn trimmer is fine.
The output voltage is set by the combination of R4 (1k) and R5 (5k trimmer). These values are suitable for 3 lipos (12.6v). A multi-turn trimmer is not essential but it does make it much easier to set the voltage accurately. A 10k trimmer is also OK.
The L200/LM741 design allows some leakage from cells connected while the circuit is not powered. This leakage is <20mA so is not very significant. Nevertheless, D1 prevents this loss, but more importantly protects the circuit from reverse polarity and any disasters that this may cause. It is therefore optional but I think strongly recommended. D1 would normally be a Schottky diode. It needs to be rated in excess of the circuit's output current and voltage, and it is desirable to have a low a forward voltage drop and high reverse current capability. I'm using SMT versions which although small are not hard to use. Mine are rated at 3.5A, 30v, 0.35v and 535A. Not shown, but it would also be normal practice to protect the input from reverse polarity if it is being connected directly to a supply that can be reversed.
The L200 needs an input voltage that is at least 2v higher than the output to be able to yield the full rated output. You also need sufficient volts to cover the forward voltage drop of D1. With a 0.35V D1 drop, the input to the L200 needs to be 2.35V higher than the desired output (eg: 14.95v in for 12.6v out, perfect for a 15v supply...).
UNIVERSAL CAR LAPTOP ADAPTORS - Fixed current chargers up to ~8A
In addition to being an excellent base for boosting 12v to 15v for the above L200 circuit, these adaptors can be used for charging lipo and A123 cells direct. They can only boost voltage so are only suitable (on their own) for 4-6 lipos or 4-7 A123's (more cells should be possible but require more changes).
Commercial chargers in the UK cost around £1 per watt output (eg: the 50W chargers cost between £40 and £60, 100W usually over £100 and 250W like the Schulze 636+ up to £300). So an 80W charger for £8 and 120W for £20 seems a good deal to me if you can accept single-purpose use. Here are the 80 and 120W adaptors:
Here you can see photos of my 7 cell A123 charger based on the 120W adaptor. It's set to 25.2v so I can use it for 6 Lipos as well. To use these adaptors as standalone chargers, you need to consider the mods described below.
1. Voltage output setting
A multi-turn trim pot is needed to replace the switch to adjust voltage accurately. 1k trimmer is perfect for 6 cells and slightly larger (2k) for 4 or 5 cells. The first photo below shows where I glued my trimmer (on an 80W version). In the second photo, the green arrow shows where I cut one trace on the board, and the red and blue arrows indicated where I connected the leads to the pot. The charger needs a load to regulate accurately so set the output voltage with say a 2k resistor as a load.
2. Current output setting
The adaptors control how much current they produce. It is set by a current sense resistor labelled R1. It is a short length of resistance wire about 0.030 ohms in value. The 120W adaptor has 2 in parallel. Shorten the wire if you want more current, and replace with a higher resistance if you want less. I have had success with a 4W3P rotary switch to select different sense values when I connected all 3 poles in parallel to reduce switch resistance as illustrated below. The yellow wires are ordinary single strand 'bell' hookup wire to create different sense values.
3. Current limiting
The adaptor is only designed to provide a output voltage greater than the input. However, that output will be held low by a flat pack until it accepts a charge and its voltage starts to rise. It is possible when charging four cells for the input voltage to be higher than the output. In this situation, the device will not be boosting so will provide a direct pass-through from the source to the load. This effectively connects your pack direct to the 12v source and will allow excessive current through until the output becomes higher than the input and the adaptor has to generate the voltage and currents set in 1 and 2 above. For instance, my 80w adaptor charger allows up to 20A through during the early stages of charging four A123 cells. This can be limited with a transistor or P-channel MOSFET as described on this external page. So to repeat, this is only likely to be a problem when charging 4 cells (and remember these adaptors can never be used on its own with fewer than 4).
4. Input reverse polarity protection
It would be wise to protect the input from reverse polarity. Various option are described on this external page. I've used Option (c) on my 120W charger except I put a 100k resistor between the Gate and 12v supply. I've also used a relay with 2 small diodes and many sites on the web show how to do this.
5. Output reverse polarity protection
You can rely on polarised connectors, but adding a schottky diode between the adator's output and the cells would be more foolproof (similar to option (a) in the above link but on the output side). However, this introduces another component carrying a high current so reduces efficiency.
A more elegant solution is to modify the adaptor. The FET nearest the output on both the 80 and 120W adaptors rectifies the output (Q6/pink arrow in the 120W example below). It would be more normal in a SMPS such as this for this component to be a schottky diode and the FET can be replaced with one. A dual schottky (eg: 2x10A) would slot right into the existing holes. The middle hole already goes to the output which is the normal configuration for dual schottky's. But the two outer connections need to be common. This requires (i) removing R8 under the board or cutting the trace to R8 (red arrow) and (ii) soldering a big blob where indicated by the darker blue arrow (so that the two outer leads of the schottky become common). This mod also improved efficiency on my 120W charger by 10% (I've not actually tried it on the 80W yet). Optional mods are to remove Q3 (pale blue arrow) and diode ZD2 (yellow arrow). Thanks to James Hopper for this diagnostic.
6. Low voltage cutoff
The 3843A chip used in my adaptors has a low voltage cutoff (LVC) but it is set to under 9v and does not latch so is too low for 12v batteries. The output of the 3843A is disabled when 'pin 1' is pulled low so I have a Pic-based circuit to do this when the input falls below a pre-set level (eg: 10v). With this mod it also becomes essential to have a diode in the output path (5 above). Fierce bad things happen if you don't! (the lipo can activate the output FET Q6 and discharge itself back into the 12v battery on the input side). Pin 1 is near the white arrow in the photo above.
Cooling at rated output is a bit marginal. Either add larger heat sinks or use a fan. Email me if you want the above Pic-based temperature controller and LVC circuit.
My adaptors have 25v output capacitors so are on the limit with 6 lipos / 7 A123 cells (25.2v). Using them close to or above their operating voltage will shorten their life. If replaced you need to select caps designed for switch-mode supplies.
I have an older LM2577 switch-mode design. It works well but only yields about 1.8A output at 15.6v from 12v. It costs more than the laptop adapter, has less power and may not work (if you build it wrong). It usually 'sings' at higher currents which the laptop one does not. No point really in building this now.
Also shown is simpler L200 circuit. It uses a rotary switch to change a fixed current sensing resistor to vary current in steps from about 300-1000mA. The photo on the right shows my original 4-way 'milking station' that I have been using for some years with the LM2577. This simpler version of the L200 is perfectly adequate where variable current is not needed.
Top of page
Click on the links below for other pages.
Art of the Possible ¦
Absence of Matter ¦