Charging Lithium-Ion Batteries

Shortly after releasing my article on hacking SNES and NES controllers for the Wii I received several emails from an individual going by the alias of "nightjumper". Nightjumper provided me with a wealth of valuable information regarding the safety issues (or rather, the lack thereof) with regards to charging iPod batteries by connecting them to a constant voltage source. My personal tendencies towards overt recklessness stem partly from experimentation and partly from my experience flying RC planes, where we charge lithium batteries the same way and make sure we don't leave them on charge for too long!

With Nightjumper's kind permission I am including his emails and circuit diagrams for safe recharging on this page, I highly recommend anyone with any semblance of self-preservation to read what he has to say. The charger circuit he provides is very easy to make and, if connected to a mono socket, will work fine without requiring changes to the transmitters.

- Mark


Only one bit of constructive criticism -

PLEASE use a Li-ion charger chip to charge your battery (they're way too expensive to destroy / reduce the lifetime by incorrectly charging them) AND lithium batteries have some serious issues with what the manufacturers call 'catastrophic disassembly' (ie. they burn or explode if they're overcharged).

I've been researching (and designing) charging circuitry for lithium batteries for 2years, and the battery manufacturers all say to be VERY careful charging the batteries.  They recommend a 1% reference tolerance on max charging voltage. 

Most of the 'protection circuits' I've found physically attached to lithium batteries (especially the ones that go into iPods) merely protects against under-voltage and over-current - they turnsoff the output when the voltage gets too low or the output current gets too high.

Because the 'protection circuit' doesn't protect against over-charging, the charging circuitry has to take care of that.

I've built (and used quite successfully) lithium chargers with the following chips (FREE SAMPLES from MAXIM - should be able to get freebies in Australia):

MAX1811 - 100ma/500ma charging current - very easy to build - comes in an easy-to-use SOIC package and only needs two SMT tantalum capacitors (very cheap - from junkbox).

MAX1501 - 1.4A charging current - VERY tricky to build by hand-soldering, since it's in a .8mm-spacing TQFN package with a thermal pad on the bottom and thermal-management on a package this small is very difficult on a home-built circuit board, since it's dissipating 1.5watts maximum in a 5mm square package.

haven't built (but I've got samples) a circuit for the MAX1555 (100ma/280ma charging current) - comes in a SOT23-5 package, needs two X7R (not cheap) ceramic capacitors (might work with higher-value tantalum caps).  Since the package is so small, thermal management might be an issue (but the MAX1555 reduces charging current by 17mA/`C if the chip gets too hot).

The circuit board for the MAX1811 is non-critical and I built mine with an xacto knife  and a magnifying glass.  Since the board is only 0.45x0.35",it 'almost' fits into a USB plug with two T1-LEDs and a two-wire cable going to a Futaba-J male plug. 

I've standardized all my lithium-batteries with 2-pin 0.025" square wire-wrap posts to mate with the Futaba-J high-current connectors (they can be found on eBay or at a hobby store - they use them for radio-controlled servos in model planes and cars).

Should be possible to use an xacto knife to build the board for the MAX1555, but haven't built one yet 'cause the MAX1811 works so good. 

The 'best' protection circuit board I found was inside a 1400mAH replacement battery for the Dell X50 PDA -- high current output (about 2Amps) AND a pad to turn the output on or off.  Inside the plastic case was two 700mAH lithium batteries wired in parallel.  I'm using that protection / control circuit board in a 5volt supply  (1A ouput) with a pair of 18650 batteries and a PIC 12F683 reading a push-button switch and a TL431 reference (to read battery voltage).

Only change I made to the circuit for the MAX1811 was to add a green LED to show power was connected - like this (might only work in fixed-width font)

+5 --------\/\/\/\--------|>| ----------|------------|>| ---------- GND
              resistor      RED LED         |                GRN LED
                             CHG- pin8    ----|

Placed side-by-side, the GRN LED 'overrides' the RED LED since both are on in low-charge and maintenance mode.  This works 'cause the human eye is much more sensitive to green light and the green LED has a higher light output than the red (6,000 mcd green -vs- 3,000 mcd red with the T1 LEDs I used).

My MAX1811 circuit board (remember the copper was cut with an xacto knife and the circuit was hand-soldered, so the actual board doesn't look anywhere near this nice) looks like this:


I've attached the PDF files for the MAX1811 and MAX1555 to 'help'. (Editor Note: The latest version of these can be found by doing a search on the Maxim web site).

The white wire [on the iPod battery] is a 10K negative-temp-coefficient thermistor used to measure battery temperature while charging -- remember I said the batteries are a little-bit tricky to charge... manufacturers recommend you stop charging if battery temp exceeds 45`C and not even start charging if the battery temperature is less than 0`C.  If you've got a dedicated charging circuit that 'properly' charges it, battery temperature should never get that high (and you probably don't want to try attaching a charger if the temp is less than 0`C).

I've attached a couple of PDF files that'll get you started (and hopefully, scare you a little about Li-ion batteries).  Just remember lithium batteries have very high energy densities -- up to 700WHr/kg - that's 60% the energy density of TNT (1300WHr/kg).  Even though lithium batteries don't 'disassemble' with a 5000 feet-per-second shock wave.  Since ALL batteries use the same chemistry, the manufacturer's datasheets are similar enough so that differences are slight.

There's a bunch of videos on the internets that shows what happens when a lithium battery undergoes 'catastrophic disassembly'.  Last year, a UPS plane (a 757 in Memphis, if I remember right) was destroyed on the ground after a shipment of lithium batteries decided to 'ignite'.  Lots of stories on the internets also about cell-phone batteries melting holes in people's pants (and legs).

Other thing to remember is that lithium battery capacity decreases by about 20% per year, so a 3-year-old battery only has 40% of the initial capacity.

If you've been involved in electronics for a while, you know that ALL electronic devices work because they are filled with 'magic smoke' at the factory.  When the 'magic smoke' comes out of the device (whatever it happens to be), it stops working. 

A girl at the local RadioShack didn't believe me about the 'magic smoke' until she tried to power a DVD-player that worked on 12vDC with a 12vAC adapter.  The magic smoke came out of the unit, and it stopped working.

And the initial power-on test is always called a 'smoke test'.

My most memorable experience with lithium batteries involved an 8-pack (2-parallel, 4-series) 4400mAHr notebook battery that wound up with a 24-gauge wire shorted between the positive and negative terminals (before the protection circuitry).  The wire and the insulation vanished -- totally vaporized.  I don't even want to think how many amps were going through that wire.

Back when I was a bench tech, we had a ready supply of 2.2uF 10v tantalum caps.  We'd hook them up with a power cord with alligator clips, turn off somebody else's power strip, and plug it in.  When they'd turn on the strip, the tantalum cap would detonate -- sounded like a .22ga rifle.  After a couple of experiences with white-hot pellets flying around the room (they hurt when they hit), we always put a roll of scotch tape around the 'assembly' so the pellet would fly straight up (and cool before it hit us).  Not as impressive as low-voltage electrolytic caps across 120vac (otherwise known as 'confetti generators'), but the noise made up for the lack of confetti.

Back in the old days, liquid-cell NiCds (like car batteries) could output 100C at 0`Celcius ('C' is the amp-hour rating, so 100C is 100x the amp-hour rating - in Amps).  A 10-amp-hour battery could put out 1000Amps - the military used them for starting jet engines, 'cause they worked so good below freezing. 

Some of the specialized LiPo batteries being sold for model airplanes are 'rated' for 15C continuous -- one of these 5AHr batteries can output 75Amps continuous.

Standard ratings on 'normal' lithium batteries are 2-3C, so you could get 1-2Amps out of your 700mAH battery (for 15-20 minutes - - if the protection circuit would allow it - they probably don't, but you can most likely get at least 700mA out of it).  But a short-circuit directly on the battery is only limited by the internal resistance of the battery and the poly-switch protection (hopefully) built into the battery.

The protection circuit is 'supposed to' turn off the output if there's a short-circuit - you might look at the datasheet for the Si8241 (PDF attached) - one of the (OEM) iPod batteries I took apart had a 5-pin SOT23 control chip that 'looks' like the Seiko chip (the other has a 6-pin SOT23 chip that 'looks' like the Chinese chip).  According to Seiko, the chip protects against over-charging. 

If your battery was made in China, it probably uses the (Chinese) SC451xx control chip (PDF also attached) - it also says it protects against overcharging.  Don't worry too much about these two datasheets -- there's a LOT in information in there, very little of it pertinent to what you're doing.

I guess you 'pays your money and you takes your chances'.  I've gotten scared enough about properly charging lithium batteries (and failure mechanisms that generate fire, smoke, and fumes) that I'm ONLY using dedicated chips on mine.


Both the MAX1555 and the MAX1811 Lithium-ion chargers are just TOO easy to work with.  The MAX1555 charges at 280mA, and the MAX1811 charges at 480mA (no thermal limiting evident on either part, and within the tolerance of the chip and my measuring setup).


A couple days ago, I built the MAX1555 280mA lithium charger.  First circuit board was cut with an xacto knife, second board was done by drawing the pattern (using the actual chips and a 5x magnifier) onto bare copper with a 'Sharpie(r) fine point permanent marker" and an xacto knife to 'clean up' what I didn't want protected (human eyes and human hands just were not made for the tolerances of small SMT parts - and these are BIG SMT parts - some of the new ones are REAL SMALL).

The circuit board for the MAX1555 is only 8mm square (final revision), so I cut a 1/2" wide strip of blank board and used Scotch tape to mask what I didn't want etched.  Before I started drawing, I put a plastic container full of Ferric Chloride into a pan of hot tap water (warm etchant is much more active, and etches much faster). 

The 'Sharpie(r)" is just an office-supply-store marker and isn't made for protecting copper from Ferric Chloride, so I want the etching to go as quick as possible (also helps to have 1oz copper - etches in half the time of 2oz copper). 

Blank copper board came from (real cheap - although I got one board that 'felt' like the copper was real thick in one small area, or I might have been sloppy cleaning the board - it took forever to etch that one small area, rest of the board etched quickly, so it was taped over to keep it from over-etching).  Just remember the board is so cheap because it's either factory surplus, or factory seconds.

It took less than 15 minutes to draw and clean up each layout, and another minute or so to etch the board by holding one end of the board and constantly stirring the etchant with the board I was etching.  Everything I didn't want etched (most of the board) was protected with Scotch tape.

After about a minute, I rinse and gently dry the board and check it (under the 5x magnifier) to make sure the pattern I want to keep stays on the board (and to touch up any traces where the marker isn't holding up to the etchant), and to scrape off any 'stray' marker that didn't get completely cleaned off the first time.  At this point, I usually just re-draw the entire layout to make sure everything stays protected.

First pass with the marker doesn't have to be perfect - just 'real good' - 'cause it's checked at least once during etching and any errors get (hopefully) corrected.

After a couple of times (this isn't my first hand-made surface-mount board), you can 'best-guess' how far apart to draw pads for 0805 and 1206-sized parts, and then double-check to make sure the pads are big enough and properly spaced by putting the actual parts on the drawn pads.


Capacitors on the MAX1555 board are 1uF 1206-sized ceramic caps 'salvaged' from some junk SMT boards I had.  Since ceramic SMT capacitors aren't marked, I categorized the values by building a 555-timer-based astable timer with the cap 'gently' soldered between the test pads.  Counting the number of flashes in 30 seconds tells me the value of the capacitor, using a spreadsheet.  If I did this more, I'd probably build one of the PIC-based capacitance meters you can find on the internets.  At the time, I wasn't PIC-knowledgeable, so that wasn't an option (or just buy the parts - I used what I had, tolerances aren't critical).

LEDs on both boards were laid out for 0805 parts, but I made the pads big enough to solder T-1 LEDs (leads cut VERY SHORT - about 0.075".  I still can't find high-brightness SMT LEDs - 80mcd is about the brightest I can find from DigiKey (for 12-cents each).

After the board is etched, it's cut to size with a pair of scissors (another nice thing about 0.020" circuit board), leaving the 'extra' length that was used to hold it when it was used to stir the etchant.  That way, you can tape it to your workbench to keep it in place.  (Did you ever try soldering on an 8mm-square circuit board while trying to hold a multi-lead chip with 0.020" leads on a 0.030" pad AND keeping the part oriented properly ??  It's a lot easier to tape the board down to remove one source of movement.  After you've got one lead on each part soldered, you can remove the tape, and tape it down again before soldering capacitors and resistors.)

Swab the board with solder flux (I don't even wipe off the marker - the flux lifts it and it's wiped off with a paper towel after the copper is tinned) and 'flash-tin' it with a solder-wetted soldering iron tip (there's so little copper to be tinned, it hardly takes ANY solder - like almost-none).   First lead to be soldered on a chip is the one with the most solder left on it when it was tinned (that way, the chip gets closer to the board).  If all the pins have the same amount of solder, I just solder a corner pin to 'anchor' the chip.

Almost all new SMT parts are ROHS-compliant, and they solder 'differently' from what I'm used to.  I'm using .015" and .020" 2%-Silver solder (available at Radio Shack in the States), and a LOT of flux (also at Radio Shack) to solder them.  All soldering is done under the 5x magnifier.

After the board is populated and tested, the final cut is made with diagonal cutters to get it to the right size (and final trimmed with an xacto knife and a sheet of sandpaper).  I've had components break when I used scissors to make the final cut (which is NOW always on the shortest dimension - learned my lesson the hard way).

Since I had another MAX1811 that wasn't being used, and I needed 'another' 500mA charger to go into the NiMh charger that had some empty space (1/2" x 3/4" x 1/4" - lotta room), I revised THAT layout and got that board down to 10mm x 9mm (earlier design was 9mm x 12mm, and the LEDs were on the side -- new layout has the LEDs on the top). 

The 'funny-shaped' part on the right of the boards mates with a female Futaba-J connector (mine were made from a single-row SIP connector modified to fit over a 1206 capacitor).  All my Li-ion connections are compatible with a Futaba-J connector (0.1" spacing and fits 0.025" pins perfectly).  The 'key' on the top polarizes the connector, and using a 3-pin connector eliminates reverse-polarity problems -- you CAN'T hook it up backwards.

The MAX1811 board uses two 4.7uF 1206 tantalum capacitors (also salvaged from junk SMT boards - at least tantalum caps are marked, even though you might have to find the 'code' on the internets - mine were marked AS6 - that translates to 4.7uF, 10v). 

Since the T1 LEDs are so bright, I used a 2K4 ohm 0805 current-limiting resistor on each board, and the LEDs are bright enough to see across a well-lit room.


The full layout (top surface in red, bottom surface in green, and part outline in yellow) for both boards looks like this:

And just the top (component) side of the boards looks like this:

Since input and output are made to capacitor terminals, I made the pads big enough to solder small-gauge wire (I'm using 22- or 24-gauge wire) alongside the capacitors.

Back to the main page...