I am making a circuit to discharge an 18V Li-ion battery, with multiple cells, in order to condition it. Each 18V cell has a capacity of 100Wh when fully charged. I would like to control the condition of each cell individually with PWM (to control the current flow), in order to regulate the discharge based on battery and heat sink temperature. A common power resistor is diode or'ed between the 4 switches (Only 1 switch is on at a time).

My question is this, what type of FET or switch circuit (U1) could I control with a 3.3V MCU PWM and also ensure the lowest amount of leakage from the battery when not in discharge mode? Protection against over current, 0 leakage, and lowest footprint are a bonus. It needs to be able to handle currents of up to 1A.


simulate this circuit – Schematic created using CircuitLab

  • \$\begingroup\$ What do you mean "condition"? What you are doing does not seem to be at all necessary for LiIon cells. It reduces lifetime with no apparent gain. If you explain IN DETAIL what you want to achieve and NOT how you think you can achieve it we MAY have a better answer. Or not :-). \$\endgroup\$ – Russell McMahon Jan 11 '15 at 4:25
  • \$\begingroup\$ Conditioning is used when there is a battery gauge regulating the Li cell. After several charge cycles, the gauge becomes out of sync with the true capacity of the battery cell and needs to be completely discharged. This is not a function that is performed often. \$\endgroup\$ – topgear Jan 11 '15 at 5:48
  • \$\begingroup\$ Charging each battery (you say cell) separately to full capacity for equalisation can bring each full capacity if done well. Each battery (not cell) is presumably 5 x LiIon cells. If they are out of balance how do you get individual "CELLS" to true zero point? What cutoff voltage do you use? If one true cell is low in capacity you will drive it very very low if you simply deal with the whole pack. There are numerous balancing chargers designed to equalise LiIon batteries. | Why do you think you can deal with 5 cell batteries all at once? \$\endgroup\$ – Russell McMahon Jan 11 '15 at 7:01
  • \$\begingroup\$ I understand re getting out of sync BUT discharging to a semi-arbitrary point does not guarantee that batteries are in sync. But charging each CELL (not battery) to eg 4.2V and then fleeting it charge at 4.2V until I chg drops to say 5% of mAh rating (in this case about Wh/V/20 = 100/18/20 =~~~ 225 mA) will ensure ALL cells are as charged as they are going to be.If you then discharge so that the worst cell is always safe you get best results. I believe :-). \$\endgroup\$ – Russell McMahon Jan 11 '15 at 23:15

First, I hope you're aware that controlling discharge current via PWM is intrinsically more stressful on the battery than using a DC current level. The reason is that the battery chemistry is intrinsically non-linear, being less efficient as current increases. Even assuming no difficulties along that line, a battery pulling 2 amps with a 50% duty cycle will dissipate twice the internal power as one pulling 1 amp continuously, since doubling the current for half the time give 4 times the power for half the time, or twice the average power.

Second, I'd really recommend using about a 20 ohm resistor, rather than a 10 ohm unit. The reason is that if you somehow have a failure which turns a cell on continuously, the current will be about your desired maximum of 1 amp. This will, in effect, give you overcurrent protection naturally.

Third, the simplest high-side switch for your application would be a P-type MOSFET, and you cannot drive it directly. Try


simulate this circuit – Schematic created using CircuitLab

It also has the advantage that an open input will produce no load current.

Frankly, small footprint should not be a consideration, since you will need a fairly large heat sink to dissipate 20 watts.

Idss, the MOSFET leakage current in the off state, will depend on the MOSFET you choose, but it can easily be less than 10 uA for room temperature. I suspect that your cell voltage measurement will draw more than that. Note that a 1 Mohm meter will put 18 uA just reading the cell voltage.

  • \$\begingroup\$ Thanks for your answer! I was thinking about using a p type FET but didn't know if there was a better solution (such as an ic) that existed. Just for clarity, the heat sink with power resistor is located away from this circuit. I like the idea of limiting the current by sizing up the resistor, I suppose that instead of PWM, it could be constant on, and when the heat sink exceeds the spec threshold, it could be turned off until it cools back to a safe temperature. \$\endgroup\$ – topgear Jan 11 '15 at 6:02
  • \$\begingroup\$ Why not just size the heat sink properly? \$\endgroup\$ – WhatRoughBeast Jan 11 '15 at 12:52
  • \$\begingroup\$ As an alternative to the P-FET route, do you think something such as the Fairchild FPF2701 would work as well? It seems that It could offer me a nice current protection option, as well as a built in FET with Charge Pump. \$\endgroup\$ – topgear Jan 13 '15 at 0:19
  • \$\begingroup\$ I don't see why it wouldn't work. I'd suggest, though, that you look very carefully at the PCB requirements. Making a board to play with the IC looks like it would be a bother - not to mention expensive. \$\endgroup\$ – WhatRoughBeast Jan 13 '15 at 4:10

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.