Primary and rechargeable cells in handheld application

[PeterFW]

Hello!
Currently i am designing two handheld devices, one on the back burner and one active as a platform for various projects.
Consisting of a main PCB with MCU, voltage regulator, display, human interface, storage and RTC wich has a header for a "expansion board" that will hold the circuitry for a specific task.
The PCBs are not routed yet, only the schematic is done.
I thought for the longest time i had everything sorted out, the prototype has been build and right now i am working on the case.
But i forgot about something...

It was designed to work of two AA cells in series and drain them until the power fails, the startup voltage of the regulator is 0.9V and it will work down to 0.3V.
Now i thought about using rechargeable cells and that is the problem, they will not like that.

A simple undervoltage lockout will take care of the over discharge but will prevent squeezing every last drop out of the primary cells.
But below 1V the alkaline cell will have not that much charge left and a NiMh as well, a cut off at about 1.8V to 2V would mean the primary cell gets discharged reasonably low and a NiMh will not be damaged.

Would i be right with that assumption? In that case i would only have to put some sort of very low power device (supervisor, reset ic, comparator) at the enable input of the boost converter and i can call it done.

Additionaly i decidet to use a polyfuse and big TVS diode for a reverse battery protection since it could be possible to put batteries in the unit that have been discharged low enough that the gate-source voltage of a p-channel fet will not be high enough to disable it.
At least that is what the datasheet predicted and my test have shown.

If you ask yourself, why i do not want to use more cells or even a lithium cell...
Ease of use, swapable cells, readily available replacements, size/case restrictions and the "big brother" uses a lithium cell with integrated charger.
So i would like the little brother to be as low cost as possible

Greetings,
Peter

[PeterFW]

Too much text, too stupid question? 

[tszaboo]

It is not stupid.
I think your assumptions are correct. With two NI-xx cells you dont need to worry about balance. Maybe youll even find a device which has UV/OV pushbutton and a reverse protection.

[paulie]

In reality you don't need to worry about over discharge with NIxx chemistry. Lithium, which does experience some reduction in internal resistance with sustained undervoltage, is often confused with NIxx which actually benefit from such treatment. For example surplus NiCd from the space program that I purchased some years back arrived with a shorting bar across the terminals.

[PeterFW]

Hello and thank you too for your replys!

In reality you don't need to worry about over discharge with NIxx chemistry.

Mhm, i just dit a bit more reading, about NiMh and it seems you are right when it comes to single cells. Looks like i my information was a bit wrong.
But i came across another aspekt...

My concern is not about active use, when the device is active it monitors the battery voltage of the two series-aa cells and can go into a permanent power down state when the battery voltage gets too low.
I do not want to include a hardware switch and want to rely only on the power down mode of the main MCU since i want to use it for "long term data aquisition".
Temperature, humitidy, radiation, sound.

In power down the circuit will draw a constant ~250µA due to the regulator iq, RTC and low efficiency at the light load.

If the two cells are charged slightly different the regulator will run them down to 0.3V before it disables it self.
With two series cells one could be 0.3V and the other 0V, but they should self balance them self.
The Wikipedia article (must not be correct, it is wikipedia) states:

omplete discharge can reverse polarity in one or more cells, which can permanently damage them. This situation can occur in the common arrangement of four AA cells in series in a digital camera, where one completely discharges before the others due to small differences in capacity among the cells. When this happens, the good cells start to drive the discharged cell in reverse.

I only run two cells in series, how much of a problem could that be?
Lets say it is running on fumes and i put it in a drawer and forget about it for
a month.
Could that cause a problem?

I could switch the power to the RTC and let the MCU wake up once a day, check the voltage and if it gets below a certain point disable the RTC, and send the MCU into permanent power down from wich it can not wake up without power cycling/hard reset (equals to changeing the batteries).

Or am i too overly cautious and should not worry about this?

[splin]

Hello and thank you too for your replys!

In reality you don't need to worry about over discharge with NIxx chemistry.

Mhm, i just dit a bit more reading, about NiMh and it seems you are right when it comes to single cells. Looks like i my information was a bit wrong.
But i came across another aspekt...

My concern is not about active use, when the device is active it monitors the battery voltage of the two series-aa cells and can go into a permanent power down state when the battery voltage gets too low.
I do not want to include a hardware switch and want to rely only on the power down mode of the main MCU since i want to use it for "long term data aquisition".
Temperature, humitidy, radiation, sound.

In power down the circuit will draw a constant ~250µA due to the regulator iq, RTC and low efficiency at the light load.

Presumably that's at 2.4 to 3V, and will increase as the voltage drops? So 500uA+ if one cell is flat and down at .3V the current will increase to 2.5mA - or even more as the boost convertor efficiency will reduce as the battery voltage drops?

If the two cells are charged slightly different the regulator will run them down to 0.3V before it disables it self.
With two series cells one could be 0.3V and the other 0V, but they should self balance them self.

They won't. Ni-mh cells have a very rapid voltage drop off when nearing depletion so it doesn't take much capacity difference for the weakest cell to drop to 0V and below whilst the other is still above 1V. This reverse charging damages the cell a little bit more causing it to lose even more capacity, ensuring that it discharges earlier and suffers more reverse charging on subsequent cycles. Its reduced capacity also means that it becomes fully charged before the strong cell and can then start to overcharge, causing yet more capacity loss, before the voltage of both cells rises sufficiently for the charger to cut off. With only two cells this shouldn't be a big problem, but the cutoff voltage will change with temperature and condition of the cells.

This is a big problem for high voltage batteries operated at high currents such as power tools, but even in your case I would be concerned.

I only run two cells in series, how much of a problem could that be?
Lets say it is running on fumes and i put it in a drawer and forget about it for
a month.
Could that cause a problem?

Yes. Reverse charging Ni-mH and Ni-cd cells is a killer, but the relatively low reverse current will mitigate the problem - although that could be 2.5mA or more as the reverse charged cell's reverse voltage increases. The issues are the frequency that it occurs, the rate and the length of time the cell is reverse charged; the latter will depend on the amount of imbalance but could be months from your description.

See: http://data.energizer.com/PDFs/nickelmetalhydride_appman.pdf

That describes the reverse charging damage mechanism, which may not be a big problem for you as the 2nd plateau voltage where both electrodes are reversed, might not be reached or the low current might not cause excessive gassing if it is. However:

Consumer usage batteries intended for storage for extended periods of time (past the point where they are fully discharged) should be removed from the device. In particular, many portable electronic devices place a very low-level drain requirement on their batteries even when in the "off" position. These micro-current loads may be sustaining volatile memory, powering sense circuits or even maintaining switch positions. Such loads should be eliminated when storing batteries for protracted periods. When nickel-metal hydride batteries are stored under load, small quantities of electrolyte can ultimately begin to seep around the seals or through the vent. This creep leakage may result in the formation of crystals of potassium carbonate, which detract cosmetically from the appearance of the battery. In extreme cases, creep leakage can result in corrosion of batteries, or the device components.
 

and:

Prolonged Storage under Load
Maintaining a load on a battery past the point of full discharge may cause irreversible changes in the battery chemistry and promote life-limiting phenomena such as creep leakage.

So having even a tiny load on a discharged battery can be damaging even without reverse-charging.

I could switch the power to the RTC and let the MCU wake up once a day, check the voltage and if it gets below a certain point disable the RTC, and send the MCU into permanent power down from wich it can not wake up without power cycling/hard reset (equals to changeing the batteries).

Or am i too overly cautious and should not worry about this?

It depends on how long you want the batteries to last and how reliable it needs to be. If it only occasionally gets over-discharged or the batteries are easy to replace then its probably not worth worrying. If the equipment is somewhere remote and/or hard to get at (ie. not in the drawer!) then perhaps it is a concern. However, you'd ideally need to ensure that the current is reduced to considerably less than 250uA - would disabling the RTC make that much difference? I'd think about using a low leakage mosfet to effectively disconnect the battery when flat. The mosfet drive arrangement would need to ensure it causes minimal drain from the battery in this state.

Another possibility is to operate the cells in parallel - not normally recommended but it might be preferable in your case as it eliminates the reverse charge problem, but may worsen the over charge problem. That won't be serious if they are slow charged (< 1/10 C) or the charge cut off before full charge.

Alternatively you could put Shottky diodes in parallel with the cells to divert most of the reverse charge current, but make sure their leakage, at your equipment's highest operating temperature, is low enough not to cause more problems than they solve.

Even when using Alkaline batteries, my understanding is that they are much more likely to leak when discharged so disconnecting when either cell drops to .8V or so may be a good idea to avoid damage.

Splin

[PeterFW]

Hello and thank you very much for the extensive reply!
You pointed a few things out i forgot about or dit not realise

Presumably that's at 2.4 to 3V, and will increase as the voltage drops? So 500uA+ if one cell is flat and down at .3V the current will increase to 2.5mA - or even more as the boost convertor efficiency will reduce as the battery voltage drops?

Up to 3.3V, everthing else is correct.

See: http://data.energizer.com/PDFs/nickelmetalhydride_appman.pdf

Very good read, thanks!

However, you'd ideally need to ensure that the current is reduced to considerably less than 250uA - would disabling the RTC make that much difference?

The 250µA, those were a mistake, tiny little fubar because i forgot that it will be considerably less if i disable the I2C pullups in power down. My test gear to measure µA is rather coarse, mostly i go by the datasheet for now.
My worst case numbers on the 3.3V line are, a bit less.

MCU = 10µA (power down without watchdog timer)
RTC = 5µA (TCXO disabled)
EEPROM = 5µA
IO Expander = 1µA
Voltage regulator: 30µA

I'd think about using a low leakage mosfet to effectively disconnect the battery when flat. The mosfet drive arrangement would need to ensure it causes minimal drain from the battery in this state.

I thought about that too but those are terribly low voltages to be of anny use in disabling a n-channel FET, to my knowledge.

The boost regulator draws 5µA in shutdown, i get reset ICs with a push-pull output that work in the same region, that would be 10µA load if the undervoltage trips.
Looks like that might be my best option.

Even if i would only use primary cells i should have a undervoltage lockout but i still have to remember about the damn thing in the drawer.
Things are much clearer now, thanks! 

Greetings,
Peter