Converting toy power source from batteries to LiIon cells

  Most electronic toys for young use non-rechargeable batteries to run. For some of them it's ok, because power consumption is low enough, but for the others it's terrible choice made by manufacturers.
One example of such a power hungry toy is this Winnie the Pooh.

  It contains one motor for swinging, one electromagnet for mouth movement, and quite powerful speaker. All of this is supposed to run from three AA batteries. First batteries lasted for three days, second cheap set lasted for two.
  Since 3 fully charged AA give little above 4.5V. I originally wanted to use small external Sony-Ericcson power bank laying in my drawer. Unfortunately this power bank has quite a logic built in, and expects steady load under 500mA. It worked until teddy started to move its mouth. This caused overload condition, and cut power off. That was confirmed with external power supply - it was drawing around 1A in peaks.
  It was clear that higher power output is necessary. I looked in my parts drawers, and I found old non-functional HP laptop battery. Battery had 9 cells in 3 banks, and by the look, and measuring voltage I confirmed that 3 of them were dead (reverse polarity, heavy corrosion). Other 6 had to be tested, voltage on each of them was around 1V (they were in this drawer for at least 5 years).
  Fully charged LiIon cell has >4V on it's terminals, and gradually discharges to around 3.3V. I verified with power supply that teddy will sing happily in this voltage range.
  LiIon charger was sourced from old Palm Tungsten E2 with dead mainboard. These mainboards (actually most boards in Palm devices) are nicely laid out in blocks.

• LiIon charger with 1A/500mA current selection (decoded as TI bq24023)
• 3.3V switching regulator (decoded as TI tps62020)
• Processor block with memories, and local voltage regulator 1.2V (probably TI lm3674)
• Bluetooth, IrDa, Audio, and small LDO regulator, possibly for SD card slot (tps73633)

  You can simply cut board part that you need. I needed battery charger block only, but I decided to keep 3.3V regulator block as well.

  First it's good to remove all connectors. Battery will be soldered directly to the board, and power supply will be extended to other connector to the side of the toy. Hot air gun is a tool of choice for removing large connectors. Wrapping board in an aluminum foil protects other components from accidental desoldering.

  Now it's time to test cells extracted from laptop battery. Chip bq24023 used on Palm board supports preconditioning, fast charge with current up to 1A, and charge status indication. Complete setup contains:

• 5V power supply
• LEDs connected to status indicator pins
• LiIon 18650 cell

  Status leds inform about the end of charge, and fault condition.
These tests took a while, but I confirmed that all 6 cells were good. Three of them (with lower initial voltage) sometimes ended charge with fault that was probably caused by a charge timer expiration. This charger chip is optimized for palm batteries that are slightly smaller capacity as I remember. Also these cells were deeply discharged, and very old. Eventually these also fully charged.

  Next step was to actually test teddy with these batteries. I started with one cell, and it worked fine - all motors were running, and electronics wasn't resetting itself. I added one cell in parallel, just to be sure that it will last long enough between recharges. Now was a time to separate charging part of palm main board from the rest. There are two main benefits:

1. We're ending with smaller board that is easier to fit. This is also a reason for removing bulky connectors.
2. We're physically cutting components that could potentially drain battery.

  This was more like an experiment in this case, because there was plenty space in base part. Palm boards have very neat power rails design. Power supplies are separated with 0 ohm resistors from load. You can just desolder these to cut power to the rest of a board.

Anyway, here is charger, and 3.3V switching power supply cut.

• Battery pack, and LiIon charger board
• External 5V power supply connection
• Battery output
• Status LED

  Status LED is connected to test pad on the other side of the board. Status pins were definitely routed to main processor, but I didn't found any vias or traces on the top. They must have been routed in one of inner layers. Brown connection goes to cathode, and yellow goes from 5V input to anode. LED is powered only when external power supply is connected. Second status line was mainly showing that pre-charge cycle is in progress, so I didn't bother connecting LED.
  I decided to keep the power supply, because 3.3V regulated power may be handy one day. I just wanted to disable this chip to prevent current drain. It had disable pin connected to other 6 pin IC that appeared to be something like window comparator. I didn't find any datasheet for this one, but on one side there was string of 3 resistors connected to the ground, and other side to battery positive terminal. This suggested that it's some under voltage detection circuit. Upper and lower resistor values were in mega ohm range, so I just shorted upper one to trip under voltage condition, and effectively disabled 3.3V switching regulator.

• Comparator chip
• Resistors
• Switching regulator disable line

  Last part included securing battery and charger board (you don't want flames in child toy when batteries short to something accidentally ;) with electrical tape, and attaching everything to base plate. I previously tried hot glue method, but it didn't last long. These toys are often thrown around, and hot glue is just not sticky enough. This time I went for Power-Tape. It has very strong glue, and is practically indestructible.

  For power input I used RCA connector - just because I had few laying around, and they can be easily secured to plastic case. I just needed to drill a hole, put connector through, secure with lock screw, and solder cables. It's also rugged enough to withstand fall on the floor. I press-fit LED through 3mm hole drilled.

• Charging status LED
• RCA power supply connector

  Finished  product connected to standard 500mA phone charger using modified USB cable with RCA jack at the end.

UPDATE (2014-12): After several months of not being used LiIon cells died. Charge cycle was starting, and falling back into error mode after few seconds. It happened that cells completely discharged to 0V. This battery pack has no undervoltage protection, as LiIon should have - only charge circuit.
Lesson learned: keep accumulators charged if not in use. Two cells were replaced with second pair from the same battery pack that way lying in the drawer, and everything was back to normal.