Trickle charge lithium-ion battery after it is fully charged

[drkntz]

I am working on a multichemistry battery charger as a feature on a product. The charger is slow, 0.1C or less.

One of the issues I face is that I have to accurately and quickly determine if the battery is present. This would be fine with a normal Li-ion battery, as I could just turn off the regulator and measure the voltage once it's done charging, or measure the current while it is still charging. However, the customer has several types of battery packs, some of which have built-in fast charge regulators and series diodes to act as protection. So, if I turn off the regulator once charge is terminated, I am left with an indeterminate (could be zero!) voltage reading, as I am reading the input of a circuit and not reading the cell directly.

A solution I am working on that works quite well is to give the battery a few mA of charge current through a series diode. Then, I get a nice 0.2V if the diode delivers current to a load ("battery present") or if there is no load, a pull-down resistor ensures the reading is zero.

Is there any issue of pumping a few mA into a 2500mAh Li-ion battery if it has already charge terminated? I read online that the self-discharge is 5% in the first 24 hours, so presumably if I kept my current under 5%/24 hours (5.2mA), I shouldn't overcharge the battery?

The image below is a simplified schematic of the battery charger showing the series diode I use to sense battery presence. (The problematic diode is the one in the battery pack, not the one shown here)

[iMo]

I asked the same here a while back and the answer I got was, afaik - a couple of mA cannot hurt provided the voltage at the cells will not exceed their max voltage (4.20V in your case)..

https://www.eevblog.com/forum/renewable-energy/trickle-charging-of-18650-li-ion-cells/

[wraper]

You could apply small current once in a while, say for a few ms every few seconds. Trickle charging lithium batteries is not safe. There is barely any self discharge, so if battery is left like this for a prolonged time, it's really a safety hazard.

[drkntz]

Thanks, and reading that thread I realized I should not use the term "trickle charge" because the point of this charge current is distinct from a typical NiCad or NiMH trickle charge.  I figure if I terminate at 4.1V, or 4.0 maybe, then set the regulator to maybe 5mA current limit or something like that, I can keep within a pretty safe margin. But to ensure the current is always available, I'd have to set the voltage higher than I would expect the battery voltage to go, maybe 4.5V?

I was also thinking about the pulse current thing, since I have access to a microcontroller. This would lower my average current significantly, if I did a short duty cycle. Say I pulsed 10mA every half second for 50ms, I'd get an average current of 1mA, which I would think is well below the self-discharge rate of the battery, not?

[wraper]

average current of 1mA, which I would think is well below the self-discharge rate of the battery, not?

depends what battery. If it's say 2000mAh lithium cell, its self discharge will be way lower and 1mA is more than enough to overcharge if enough time is given. I personally would not make an average current more than 10uA. There is no reason to apply current for whopping 50ms unless initial detection already happened and you want to verify more reliably.

[tridac]

Have a stack of battery power tools, some nicad and some lithium. Nicad life can suffer if left permanently on the charger, but nicads can die in a few weeks max if not kept topped up. Way I got round that is to have a timeswitch charging board, powering up all the chargers for around 10-15 minutes per day. keeps the batteries topped up, but won't overcharge. The key thing being that modern lithium appliances have very smart chargers, which are voltage as well as current limited. They also detect the charging curve for the cell type. All in single chip now. Companies are ever more cautious about litigation, consumers leave on the charger permanently often, so if the charger is voltage limited, should be pretty safe, even if on all the time. On a 2ah battery, a few ma test current won't affect anything much if pulsed infrequently.  Ymmv, of course. If in any doubt, build a simple test rig and see what happens ?...

[Faringdon]

One of the issues I face is that I have to accurately and quickly determine if the battery is present. This would be fine with a normal Li-ion battery, as I could just turn off the regulator and measure the voltage once it's done charging, or measure the current while it is still charging. However, the customer has several types of battery packs, some of which have built-in fast charge regulators and series diodes to act as protection. So, if I turn off the regulator once charge is terminated, I am left with an indeterminate (could be zero!) voltage reading, as I am reading the input of a circuit and not reading the cell directly.

Thaks, but it seems there are issues with the premise of your question.....so are you saying you cant turn a very light load on and measure a voltage across it?......so what happenes when the battery does  need to supply power?.......surely it must be possible to draw some power out of the battery if its connected?...or be able to just measure its voltage...what circuitry have they put there that stops you being able to measure its voltage.
Youre right to be worried about overcharging it with the modus operandi that you spoke of.....well, not really...because you could charge it a bit to see if its there...then discharge it back down again straight away.

[drkntz]

so are you saying you cant turn a very light load on and measure a voltage across it?......so what happenes when the battery does  need to supply power?.......surely it must be possible to draw some power out of the battery if its connected?...or be able to just measure its voltage...what circuitry have they put there that stops you being able to measure its voltage.

I can't really draw power out of the batteries, because they have two sets of terminals. One set is the input terminals, one set is the output terminals. I can only access the input terminals with my circuit. The designers of the battery packs have put various fast charge controllers and series diodes in their devices, so I can't rely on discharging the batteries to detect presence.

The batteries are like this:

[ input terminals ] -> [ input charge regulator and/or diode ] -> [ Battery cell ] -> [ output discharge regulator ] -> [ output terminals ]

[tridac]

Looks like an impossible task, since you must detect actual battery voltage to build a safe and effective charger. I would talk to the client to perhaps take out the diode from the charger input. Otherwise, turn down the work, because you will get the blame if the slurry hits the fan later. Clients demand all sort of stuff, but it's often a wishlist and open to discussion, such as: why do they mandate a diode on the input ?. Never compromise design principles that could affect safety...

[uer166]

The answer here is simple enough: a fully charged battery that has been at 4.2V for a while is indistinguishable from no battery at all, since no current can flow.

If you can constrain it to only allow partially charged (e.g. 4V per cell), then battery detection becomes trivial enough. In practice a few "battery presence test" mA at really low duty will probably not fully charge the battery, but you still need to limit it to 4.2V per cell and not blindly inject current.

[uer166]

@tridac you do not need to detect actual battery voltage directly to build a safe charger, if you account for lowest possible diode Vf. Plenty of products in the field charge through a diode np problem, and it's a net safety adder to avoid reverse current flow in fault conditions.

[tridac]

Sorry, that is wrong. Lithium and nicad have a voltage / time charge curve that can be used by a smart charger to detect when the battery is fully charged. Avoid overcharging issues and contributes to safety. A diode vf in series will vary with temperature, and you already need  temperature compensation to take account of battery variations with temperature. If the charger is smps, then there are already rectifier diiodes in the output circuit to prevent reverse discharge, so you don't  need yet another diode in the circuit...

[Siwastaja]

I read online that the self-discharge is 5% in the first 24 hours

Don't believe everything you read online. Self-discharge is basically zero in the first 24 hours, and very small even in long run. So putting more and more charge into the cell will start overcharging it. Now the silver lining is that if your purpose is to detect "is there current flow?", then you can make it very short (milliseconds?) and therefore any overcharge completely negligible, as long as you implement some kind of reliable timer so that your short pulses cannot occur repeatedly with intervals so short that total charge put in the cell (in worst case, like the cell sitting in your charger product for years while powered) would become significant.

[drkntz]

Don't believe everything you read online. Self-discharge is basically zero in the first 24 hours, and very small even in long run.

Agree, but where do I find the actual answers? Am I to believe batteryuniversity (where I found this info), am I to believe some guy on a forum, start digging through IEEE journals? Obviously I'll have to do some testing to be sure.

the silver lining is that if your purpose is to detect "is there current flow?", then you can make it very short

If you can constrain it to only allow partially charged (e.g. 4V per cell), then battery detection becomes trivial enough. In practice a few "battery presence test" mA at really low duty will probably not fully charge the battery

I think this is the direction I am heading. It will require some experimentation to ensure the battery never overcharges. Typical use case for these batteries is short bursts of use then back on the charger. So if I charge maybe 4V, then blip the charger a couple mA at a low duty cycle & limit to 4.2V max cell voltage, I'll probably get a tolerable compromise between battery charge and detection.

[drkntz]

Sorry, that is wrong. Lithium and nicad have a voltage / time charge curve that can be used by a smart charger to detect when the battery is fully charged. Avoid overcharging issues and contributes to safety. A diode vf in series will vary with temperature, and you already need  temperature compensation to take account of battery variations with temperature. If the charger is smps, then there are already rectifier diiodes in the output circuit to prevent reverse discharge, so you don't  need yet another diode in the circuit...

Would you argue that undercharging (4.0V) and blipping a few mA with max cell voltage capped at 4.2V is unsafe? I figure I would follow a typical Li-ion charge cycle: precondition -> "fast" charge (0.1C, really not that fast) -> cap at 4.0V (instead of 4.2) -> current terminate at 0.01C -> blip (if necessary) to poll battery presence.

If the charger is smps, then there are already rectifier diiodes in the output circuit to prevent reverse discharge, so you don't  need yet another diode

Yes, the charger is SMPS, but it's a buck converter topology, so the freewheeling diode is more like parallel-ish to the battery cell. On another note, I can't get the customer to change the battery packs or redesign or whatever, because the end user buys the battery pack and the device it drives from one of several tool companies. Turning down the work is not in the cards for me.

EDIT: or are you saying the diode I added to the circuit is a problem? I'm not really worried about that one right now, as I am reading the battery voltage directly and can correct the charger voltage.

[ejeffrey]

Would you argue that undercharging (4.0V) and blipping a few mA with max cell voltage capped at 4.2V is unsafe? I figure I would follow a typical Li-ion charge cycle: precondition -> "fast" charge (0.1C, really not that fast) -> cap at 4.0V (instead of 4.2) -> current terminate at 0.01C -> blip (if necessary) to poll battery presence.

As long as you never exceed 4.2 volts at the battery terminals, it's safe.  Derating your charge termination voltage to equal the worst case diode drop, plus a margin for pulsed presence detection should be fine with two caveats:

1) You will be sacrificing a significant fraction of the battery capacity.
2) If it's possible for someone to go insert a fully charged (4.2 V) battery into the charger, you need to make sure that you are still operating safely. This may mean that your battery detection will not work.

One alternative is to measure the reverse leakage current through the protection diode.  Disconnect the charge circuit and try a pull-down resistor, see how much current flows through it.  Power diodes and especially schottky diodes will have plenty of leakage to measure.  One trap here is that the leakage is exponentially dependent on temperature, make sure it will operate in the lowest temperature someone might use it.

[wraper]

Would you argue that undercharging (4.0V) and blipping a few mA with max cell voltage capped at 4.2V is unsafe? I figure I would follow a typical Li-ion charge cycle: precondition -> "fast" charge (0.1C, really not that fast) -> cap at 4.0V (instead of 4.2) -> current terminate at 0.01C -> blip (if necessary) to poll battery presence.

You won't be able to detect batteries charged above certain level. And with a diode in series it means not detecting unless nearly empty.

[wraper]

I would first check if there is voltage = battery with no diode. If voltage detected, do not apply any current for detection purposes. If no voltage detected, apply a short current pulse to detect batteries with a diode.