PC Battery Cell Measurement
Charging multicell battery packs is simple at slow rates (0.1C = 1 full charge in 10 hours). But at higher rates with unmatched cells (cells having slightly different capacities due to age or manfacture) there is a good chance that one or more cells will be full (and therefore madly gasing) before the entire pack has reached capacity. Good chargers may detect this with thermal or voltage analysis but a more reliable method would be to monitor each cell individually. This would have the added benefit of detecting weak cells before (in the case of Radio Control application) disaster strikes.
The following circuit was intended to be used with a PC Analogue to Digital Interface. I used the Discovery Series K-2805 Parrellel Port Interface (as distributed by Dick Smith electronics) but any unit with 4 or more analogue inputs could be used (or even 4 individual digital voltmeters).
Shown below is the measurement circuit fitted with a typical plug for a 4 cell battery. Optionally the charger could utilise the same plug connection but currents would be limited to less than an amp.
Shown below is the measurement circuit fitted with a dummy plug for measurement of the voltage of the entire pack (where individual cells cannot be accessed. In this internal connection to the battery charger means than no connection is needed between the plug and the pack. The diode is there to protect the circuit from negative voltages that can occur with my home grown charger (described in PC Controlled Voltage and Current Source.htm)
Each of the 4 modules M1 to M4 is based on one of the four op.amps in the LM324.
Since the PC interface I used has 10 AI Inputs, I have two such sets of modules connected through one 25 pin sub-D female connector plug which also includes pins for the current source load voltage so that with a dummy plug as shown, there is no need for a seperate connection to the battery pack. In my circuit shown, the maximum AI is clamped to 5V which could represent 1 or 2 NiCad cells. Through the use of changeable plugs (such as the dummy plug shown in the circuit, I can measure 2 sets of packs (each 4*5V=20V max. each) or 1 pack (8*5V = 40V max. - though Vpos is only 30V so the true max. is 28V). As shown with M#g connected to ground, the module provides unity gain and mearly buffers the cell voltage to bring it in reference with ground. To double the maximum input voltages, a 100K resistor could be fitted in series with the M#+ nand M#- connections.
I've brought M#g to the plug connection so that smaller voltages can be measured on M#+ by fitting a resistor between M#g and ground on the plug (e.g 100K would amplify the M#+ signal by two) but this is not useful for battery cell measurements as the M#- signal remains unamplified.
Even with the dummy plug measuring only the total pack voltage from the charger leads, there is the advantage of having greater Analogue to Digital accuracy. The PC interface I use has a resolution of a 256th of the supply voltage being 5V. If the input voltage from a 12V pack is scales to 5V and divided by 256, the conversion accuracy is 0.05V at best. By having each module only measure a maximum of 5V, the conversion accuracy is increased down to 0.02V.
The battery charger software included in charger includes provision to terminate charging based on the cell with the greatest voltage, the cell which ceases to increase in voltage (delta peak detection ) and the cell for which there seems to be no increase in the rate of increase (looking for the point at which the cell starts to gas). All levels are adjustable for experimental purposes. It is perfectly happy to have more that one cell per voltage tapping, though the chance of spotting that unmatched or weak cell is thus reduced.