Overdischarge protection for a li-ion powered circuit

[milestag]

Why not use the onboard AtoD convertors? 10-bit precision should be more than is needed.

Are you using a voltage regulator to power the PIC? If it is relatively precise (1% ish) then you should be fine to use that as your reference voltage. Just use a voltage divider to feed the AtoD input.

Otherwise you may need an external voltage reference. 3-pin voltage refs are cheap and plentiful. I'd guess you need around a 2.5V reference.

Set your voltage divider so it doesn't exceed 2.5V when the battery is fully charged. Then just do the math to determine A/D value at 3.0V.

Here's some code for 18F2525 so may need to change register names, etc. But the concept is the same. Note, I only use 8 bits to keep it simple. It's accurate enough for my app:

battery_monitor: '
TRISA.0 = 1 'make AN0 input
ADCON1 = 14 'AN0 = analog in
ADCON0.1 = 1 'start AD
wait_ad:
pause 1
IF ADCON0.1 = 1 then wait_ad
batt_v = ADRESH
ADCON1 = 15 'return pins to digital
TRISA.0 = 0 'output
IF batt_v < 150 then batt_warn

[ErnieM]

I just did some testing for a similar circuit to see how hard we had to drive a flashing LED.

I used a green led, gave it a trickle of 10 mA, and surprisingly enough there was little change in intensity if I drove it any longer then 20 milliseconds. The human eye is a peak detector, and the short burst is enough to let you see it, even from across the room; I could get about 30 feet away and it was still plainly visible in room ambient light.

Milestag has a good idea, although I wouldn’t bother with a regulator between the battery and the PIC, it’s just a waste of power (which batteries don’t have much of). Drive it direct, and use a low current low voltage reference to reference the A2D. Divide down the battery voltage to put it in the range of the lower reference voltage and just measure.

[peu]

Good ideas, I found this ldo voltage reference: http://www.national.com/pf/LM/LM4121.html, it may do the trick.

One more question, do I need to use 2 pins? or could I compare this voltage reference to any internal value?

Thanks!

[milestag]

It looks like you would need 2 pins. GP0 is the A/D input and GP1 is the Vref.

But you can sometimes do multiple functions on a single pin.

In my circuit I have one pin performing 2 functions alternately as Digital output (driving an LCD) and Analog input (read battery voltage).

[Ron Marcus]

If you have a free on board A to D, it should be pretty easy. Get a silicon diode, say a 1N4148. These have a nominal voltage drop of .7 volts. The cathode goes to ground, and the anode is connected through a high value resistor, say 100K to B+. Connect the A/D input to the junction, and measure the voltage relative to the power supply voltage. As the battery drains, the number will rise, because the .7 volts becomes a larger portion of b+. Take samples at the target voltages, and tell the processor what to do at the preset trip points. You can even use an output pin from the processor as b+, to turn the circuit on or off for even more current savings. You may have to put a small bypass cap across the junction to ground as diodes can get noisy in forward bias. I'd try it without and see how it works first.

Ron

[Acetronics2]

It is not written in the specs ...

but the low to high voltage level change occurs for a quite precise voltage !!! ( TTL inputs ... of course).

a 1.5 v value, i.e.

so, a simple voltage divider is generally enough for these measurements

( a scope closer look with the RCTime function pin could enlight that ...)

Alain