trying to make a LiPo charger/balancer

[dirty_d]

this is what ive got so far for the analog part that limits the charging voltage and current, the analog balancing stuff isn't in this schematic, the voltage and current limits are adjusted by two potentiometers. i tried it out in an electronics simulator and it seems to work, but maybe someone with real experience can look at it and see if everything is ok. one of the outputs of the AVR will go to VEE and be able to start and stop the charging.

EDIT: oops R13 should be 300 ohms.

 

[safe]

Wow, that chip is tempting. I could do all kinds of programming with it. Hmmmmm... the price is just insanely cheap. I'll be interested in how it all turns out.

 

[dirty_d]

get one, theres no risk, you can make a programmer with nothing more than a parallel port connector with a old IDE ribbon cable and a few resistors, and the programming software is free. it would be very very easy to make a brushed motor controller with one and you could do the current limiting or whatever in software, it has a 6 channel analog to digital converter so you could do all kinds of crap like measure temperature and rpms and stuff.

 

[fechter]

Analog part looks great. I don't think I'd use a LM324 though. I thought those had an open collector output or something weird like that (I'd have to check the datasheet). There are newer amps around with better performance. In this application, performance should not be an issue (slow=OK), so it would be fine as long as the output can swing high.

I'll have to admit I'm pretty out of date on what amps to use too. I should dig around on National or Maxim to see what looks good these days.

 

[dirty_d]

fechter, im not really sure what you mean about the op-amp, but ill look into it. this is what i came up with for the balancing analog circuitry, i guess you could hook up as many cells as you want like this.

 

[safe]

This circuit looks very interesting. It appears that you are using a Bipolar Junction:

...and at the same time an optocoupler in the OPPOSITE configuration of the LVC. I love it... this place is filled with so much creativity.

Now could you sort of explain the flow of how this thing balances. I sort of get the flavor but would like the long version using good old fashioned plain human english for dummies. :wink:

 

[dirty_d]

safe, sure, well there is some additional wiring for each cell, a LM324 quad op-amp which is used in a voltage difference arrangement, it takes the voltage from the negative terminal to the positive terminal on each cell and converts that to the same voltage relative to ground. for example, say the 3rd cell has a voltage of 3.95V and all the other cells have 3.9V, relative to ground the positive terminal of the third cell is 11.75V, the op-amp takes the voltage from the negative terminal and positive terminal and outputs the difference between them, which is relative to ground so that the AVR can measure the actual cell voltage.

but anyway heres how it actually works, the avr samples each of the cells voltages one after another, then compares them, it calculates the difference between the lowest voltage cell and the highest voltage cell and if it is over the tolerance(12mV in the code). it will output +5V to the VEE in the schematic above of the highest voltage cell. the resistor is to limit the current to 10mA, this lights the diode which activates the photo-transistor, there isn't much of a gain though, the opto-isolator collector current is only about 21mA, so its fed into a darlington transistor which has a current rating of 800mA and a dc gain of over 10,000, that lets me choose how much current i want to bypass the cell by selecting the right value resistor on the collector of the darlington, at 500mA the voltage drop across the collector to emitter is about 1V, so if the cell voltage is 4.2V then the amount of current bypassing the cell would be (4.2 - 1.0) / 6.5 = 492mA, so the end result is that the highest voltage cell is being charged with 500mA less current than the rest of the cells, and they are able to catch up while the highest one charges slower.

 

[fechter]

I checked the datasheet on the LM324 and it looks great. I was thinking of some other part??(brainfart).

I guess one other consideration is when the first cell gets high enough to trigger the shunt, you may need to drop the current on the charger to make sure the shunt element can keep up with the charging current. The worst case would be the shunt has to handle the entire charging current. This might only occur if a cell goes bad, but I'm not sure. Normally, I would guess the 'balancing current' will be much less than the overall charging current. You might be able to do this by modulating the cutoff signal to the current limiter. Otherwise, if you have another dedicated output available on the processor, you could switch a resistor into the reference for the current limiter to drop the limit to a lower level. If the main charger current is set low enough, you wouldn't need to switch it, but then charging may take too long.

 

[dirty_d]

yea i set it up so it only bypasses 500mA from the cell not all of the current. so during the beginning and before the end of the charge the cell just gets charged slower, but when the charge current is less than 500mA the cell will actually be getting discharged to balance the pack, im not really sure if thats bad or not.

this is the first time ive ever made anything like this and im having some problems, the schematic layout isn't bad, but i cant figure out how to actually lay out the traces. i have a program called multisim for the schematic and utiboard for the pcb layout, it has a "autoroute" feature, but it seems like it can never finish, its gets to about 90% of all the traces complete but then it seems to reach a dead end. i don't think i would be able to lay it all out by hand. is there some other free program that does this stuff?

 

[flip_normal]

Hi dirty_d,

I realy admire your have a go attitude with this, I hope you can get it working. I got one of the ATMega168 chips the other day but took the easier route and opted for the Arduino Diecimila board. My EE skills are aren't up to your level but here's a list of some other free/trial PCB Layout/Design/Schematic Capture/Autorouting Software which might help.

http://www.everythingpcb.com/p14120.htm

 

[safe]

This circuit diagram from Thundersky made me a little. Along the same line of thinking that you have been going on with your "voltage targeting" cell ladder circuit (or whatever you call it) you can just leave the series connections in place as long as the power out loop is disconnected and then you can apply selective charge currents across the lines that divide the cells.

At rest with three cells you might have:

0v
3.2V
6.4V
9.6V

...during charging you raise everything to:

0v
4.25v
8.5v
12.75v

...but when cell two is full you now have:

0v
4.25v (hold cell two to 3.6v)
7.85v
12.1v

...if you do all your calculating in software the wires and circuitry sort of disappear. You end up with just a few wires and lot of logic.