GGoodrum said:
Thanks, Bob. This does look like it would be easier to solder on a proto board.
I'm still confused about how this would work, and what the rest of the circuit would look like, but I'll keep studying. :?
-- Gary
i'll try a non-engineer's explanation, or at least as close as i can come with engineering being my "first language" and anything else being a translation that inherently loses something:
the mcp111 is open drain output, meaning that the output will pull down a line to which it is attached when the supply voltage drops below the limit.
the ebrake line is typically pulled up to 15v or less with a resistor of 10k or so, meaning that a driver must sink a few milliamps to pull that voltage down close to 0v.
you could put one device on each cell, with the outputs connected together, and the common output line would be pulled down if any of the sensors detected a voltage below their limit. they have devices with a number of preset trigger points, and my experience with the a123 m1 cells is that they have <20 milliohms series impedance,so at 30A they would drop about .6v, and the standard 2.32v part would then cut off at a 2.9v cell voltage which my testing has shown is a good point for the m1 cells.
this way you could put as many cells in series as you like, put an mc111 on each, tie the outputs of the mc111s together and to the ebrake input, and if any cell falls below the trigger voltage the brake would be asserted.
if you wanted to just use one device to tell when the voltage of a serial string dropped below a preset limit, you would need a resistive divider to let the device see the appropriate fraction of the total voltage.
this quickly becomes a problem, because it is ok if each cell of 10 drops .1v but not ok if one cell drops 1v. it quickly becomes apparent that the voltage of each cell must be sensed.
the first device i mentioned has the input voltage separated from the sensed voltage, so you would just need to power it from 1 or 2 cells and then use a divider to set the trigger point. the device uses less than a microamp so unbalancing the pack is not an issue. the second device has common supply and sensing input, but once again the current drain is so low it is not an issue.
suppose you put a 2.32v mc111 on each cell, and tied the outputs together. the device draws much less current than the self discharge rate of the cell, so it will not measurably affect the power you can get out of the cell.
any time the voltage under load of a single cell drops below 2.32v which would be 30a load and a cell voltage of 2.9v, the output open drain fet would turn on and pull down the common output, which could be connected then to the ebrake line. you could use the 2.62v devices or the 2.90v or 2.32v devices, depending on your criteria.
if you wanted to use one sensing device to measure a higher voltage, you could use the first device i suggested, and power the device from 1 or 2 cells, then divide down the total pack voltage to where your cutoff voltage would trigger the device.
because the mc111 is only a 3 terminal part it is a bit more difficult to divide down the voltage, but still possible. suppose you had a 10 cell pack and wanted to cut off at 29v. you could put a 100k pot across the pack, and adjust the wiper to supply about 1/10 of the total voltage to the mc111. since the supply current of the mc111 is so low, the current flowing through the pot would be plenty to keep it running.
when the voltage from ground to wiper dropped below 2.9v the mc111 would turn on, pulling down the output. the output current needed to pull down an ebrake signal would then become dominant in the voltage divider, so this would be a problem unless you buffered the output with a fet or op amp. using a 10k pot might be a compromise that would work, but it would suffer from inaccuracy with temperature and voltage change.
a way to avoid this problem would be to supply power to the mc111 through a regulator, with an output voltage less than the trigger voltage. you would probably need to buffer this through a diode to keep it from driving the sense voltage. this may be a bit over the top, but if you used a 3.3v regulator and a .7v drop diode and connected the diode cathode to the power input of the mc111 along with the divided down pack voltage, the regulator could supply the power to drive down the output after the device triggered on the input going below 2.9v, without too much error from the supply voltage moving.
a regulated voltage fed to one input of a comparator and the sense voltage to the other is the basic technique, but these modern devices have simplified the task considerably, also preventing false triggers from brief drops in voltage.
this way when the mc111 turns on and sucks down the voltage divider the regulator could supply the mc111 power to pull down the output. the alternative would be to use a voltage divider that supplied enough current, say using a 1k pot on a 10 cell pack, but that would present a 33 ma load and shorten the pack output a bit unless switched off when not in use.
the mc111 and other solutions like it provide the combination of a reference source, and a comparator with some hysteresis to prevent false tgriggering.
using a voltage reference, voltage divider, and op amp can provide the same functions, but not in a convenient small package like this.
the need to reset a microprocessor in a laptop when the battery dies has spawned a number of devices like this which will serve this purpose. i have used the mc111 in a bms prototype myself, and it works very well.
the higher voltage versions of the part could also be used to cut off charge when a cell rises above the desired trigger voltage, but none of the voltages provided standard are quite as well suited as the standard parts for low voltage cutoff.
using one of the mc111 parts for low cutoff and one for high, with a bit of tweaking, can provide the basis for a simple reliable accurate bms for the a123 cells. my experience has shown that unless you can ignore the loss of a few cells capacity it is necessary to monitor each cell.