Parasitic charge current into lithium coin cell

[Guest]

Hi,

Our product uses a CR2032 lithium coin cell as a CMOS battery for an
embedded PC. The cell is OR'ed with a 3.3V supply which becomes
available when the product is powered on, the intention was to extend
the cell life. I think desktop PCs do a similar thing using the standby
power.

However the diodes we are using have a reverse leakage of a few
microamps at room temp, rising to a few tens of microamps when hot.
This means the cell could see a 'trickle-charge' current of say 30uA
under worst-case conditions. Is this likely to damage the cell?

The reverse leakage current also means that the cell is discharging into
the standby power circuit when the product is powered down. I have
found some alternative diodes with a spec'd worst case reverse leakage
of 200 nA, but I am trying to get to the bottom of a series of early
cell failures we have had - cells completely flat in a matter of weeks.

The extra leakage current would certainly reduce the life, but not by
this much? The CMOS input pin on the PC draws 5 uA, for a 225 mAh cell
this should give about 5 years' life, or about 9 months if it had to
supply the extra 30uA leakage continuously as well.

Any suggestions welcome.

 

[[email protected]]

Hi,

Our product uses a CR2032 lithium coin cell as a CMOS battery for an
embedded PC.  The cell is OR'ed with a 3.3V supply which becomes
available when the product is powered on, the intention was to extend
the cell life.  I think desktop PCs do a similar thing using the standby
power.

However the diodes we are using have a reverse leakage of a few
microamps at room temp, rising to a few tens of microamps when hot.  
This means the cell could see a 'trickle-charge' current of say 30uA
under worst-case conditions.  Is this likely to damage the cell?

The reverse leakage current also means that the cell is discharging into
the standby power circuit when the product is powered down.  I have
found some alternative diodes with a spec'd worst case reverse leakage
of 200 nA, but I am trying to get to the bottom of a series of early
cell failures we have had - cells completely flat in a matter of weeks.

The extra leakage current would certainly reduce the life, but not by
this much?  The CMOS input pin on the PC draws 5 uA, for a 225 mAh cell
this should give about 5 years' life, or about 9 months if it had to
supply the extra 30uA leakage continuously as well.

Any suggestions welcome.

Why not just use a better diode? You'd get 5 years' battery
life, and not have to worry about the effect of backcharging
the Li cell.

If you need low forward drop, use a FET.

HTH,
James Arthur

 

[[email protected]]

Suggestion:  see "Troubleshooting Analog Circuits" by Bob Pease, page
66.

Pease suggests using a transistor's collector-base junction, instead
of a discrete diode, if you want low leakage... avoid transistors
using a gold-doped architecture as these are leakier.  "You can easily
find such 'diodes' that have less than 1 pA leakage even at 7V."

Such CE diodes are quite slow (compared to fast switching diodes) but
that would be irrelevant in your application.

--
Dave Platt <[email protected]>                                   AE6EO
Friends of Jade Warrior home page:  http://www.radagast.org/jade-warrior
  I do _not_ wish to receive unsolicited commercial email, and I will
     boycott any company which has the gall to send me such ads!

Yep, that's a good diode, if he can stand its forward drop.
The description sounds like he's using schottky diodes to
avoid the voltage loss.

James Arthur

 

[Nemo]

In general ANY reverse current into a coin cell is bad 'cos they lack
the vents other cell types have. (Example: if you look at say an AA
cell, you'll see some gaps under the pip at one end - or on some cells,
some grooves carved into one end, these are points of deliberate
weakness so the cell ruptures when internal pressure is just say 2-3
atmospheres so any bang is small and controlled).

In the UK the law ued to be that you must have two current limiting
components (eg a diode and a resistor) between any Li coin cell and a
voltage source to stop them exploding in a spectacular manner. Wish I
could find a link to this because the Health & Safety Executive used to
have a very explicit pdf on this with graphic descriptions of real life
examples of unintended charging which they had investigated. Does
violent explosion meet your criterion of damaging the cell? Even 1uA
into a CR2032 would really worry me. I *did* find one battery
manufacturer's app note saying "don't charge more than 3% of the
battery's capacity in its entire life":

http://www.73.com/specs/th693.pdf

I'd suggest as a start, use low leakage diodes (I'd start at
www.centralsemi.com, they have low leakage ranges with leakages of e.g.
one nanoamp at 100V). These are normal diodes though, I'm not sure what
their low Vf Schottkies are like.

There *are* some rechargeable Li coin cell chemistries but, when I
looked at these a few years ago, they had very limited cycle life and a
memory effect.

How much design freedom do you have? We could go on about "you should
use such and such a chemistry cell with a capacity of at least X" but if
the circuit's already laid out, and the moulding won't permit a larger
cell, there's no point discussing the optimal blue sky solution. Maybe
you should just accept that the cell needs occasional changing (is it in
a holder, or a tabbed soldered one?)

Nemo

[email protected] enquired

 

[[email protected]]

[email protected] enquired






the vents other cell types have. (Example: if you look at say an AA
cell, you'll see some gaps under the pip at one end - or on some cells,
some grooves carved into one end, these are points of deliberate
weakness so the cell ruptures when internal pressure is just say 2-3
atmospheres so any bang is small and controlled).

In the UK the law ued to be that you must have two current limiting
components (eg a diode and a resistor) between any Li coin cell and a
voltage source to stop them exploding in a spectacular manner. Wish I
could find a link to this because the Health & Safety Executive used to
have a very explicit pdf on this with graphic descriptions of real life
examples of unintended charging which they had investigated. Does
violent explosion meet your criterion of damaging the cell? Even 1uA
into a CR2032 would really worry me. I *did* find one battery
manufacturer's app note saying "don't charge more than 3% of the
battery's capacity in its entire life":

http://www.73.com/specs/th693.pdf

I'd suggest as a start, use low leakage diodes (I'd start atwww.centralsemi.com, they have low leakage ranges with leakages of e.g.
one nanoamp at 100V). These are normal diodes though, I'm not sure what
their low Vf Schottkies are like.

There *are* some rechargeable Li coin cell chemistries but, when I
looked at these a few years ago, they had very limited cycle life and a
memory effect.

How much design freedom do you have? We could go on about "you should
use such and such a chemistry cell with a capacity of at least X" but if
the circuit's already laid out, and the moulding won't permit a larger
cell, there's no point discussing the optimal blue sky solution. Maybe
you should just accept that the cell needs occasional changing (is it in
a holder, or a tabbed soldered one?)

Nemo

(Nemo, for your info and flame-proofing: SED's
preference is to quote the post you're responding
to, then respond, i.e. post your reply at the
bottom. That keeps things in chronological order.
Top-posting is disdained.)

I was going to suggest this:

Q1
PNP
+3.3v >--o----. .----o-----o----> Vbackup
| e \ / c | |
| --- [R4] |
| |b | |
| o------' |
[R1] | | +
| [R3] --- B1
| | -
| o----. |
| | | |
| |/ | |
o----| Q2 '---.|--'
| |>. ||<-.
[R2] | ||--+ Q3
| | |
=== === ===
GND GND GND


When +3.3v supply exceeds threshold set by
R1-R2, Q2 turns on. Q1 saturates, supplying
+3.3v to the load, and Q3 is cutoff,
disconnecting the battery B1.

When the +3.3v supply is too low, Q2 cuts
off, Q1 cuts off, Q3 turns on, connecting B1.

There are no direct reverse leakage paths. In
battery mode: cutoff Q2 blocks c-b conduction
through Q1, and Q3's gate takes no current.
With +3v is applied, Q3 disconnects B1.

Have I missed any leakage paths?

Cheers,
James Arthur

 

[Guest]

[email protected]>, [email protected]
says...

Why not just use a better diode? You'd get 5 years' battery
life, and not have to worry about the effect of backcharging
the Li cell.

If you need low forward drop, use a FET.

HTH,
James Arthur

Sure, that's the plan. I have tested OnSemi's MMDL301 as a replacement
and the measured leakage current is << 1 uA.

My other question was, is 30uA of parasitic charge current lilkely to
kill the cell in a matter of weeks, or do I need to look for some other
fault?

I found some partial answers in Panasonic's Lithium Battery Handbook

http://products.panasonic-
industrial.com/datasheets/en/Panasonic_Lithium_Handbook_Part1.pdf

where it states "the total charge applied over the life of the cell
should be less than 3% of the capacity".
(225mAh x 0.03)/(3 x 7 x 24) gives about 13 uA so it's certainly
possible this is the cause.

 

[Guest]

I was going to suggest this:

Q1
PNP
+3.3v >--o----. .----o-----o----> Vbackup
| e \ / c | |
| --- [R4] |
| |b | |
| o------' |
[R1] | | +
| [R3] --- B1
| | -
| o----. |
| | | |
| |/ | |
o----| Q2 '---.|--'
| |>. ||<-.
[R2] | ||--+ Q3
| | |
=== === ===
GND GND GND


When +3.3v supply exceeds threshold set by
R1-R2, Q2 turns on. Q1 saturates, supplying
+3.3v to the load, and Q3 is cutoff,
disconnecting the battery B1.

When the +3.3v supply is too low, Q2 cuts
off, Q1 cuts off, Q3 turns on, connecting B1.

There are no direct reverse leakage paths. In
battery mode: cutoff Q2 blocks c-b conduction
through Q1, and Q3's gate takes no current.
With +3v is applied, Q3 disconnects B1.

Have I missed any leakage paths?

Cheers,
James Arthur

Neat - I like it!
I'd still be concerned about leakage through Q3 body diode and Q1 E-C.
I know low leakage small-signal FETs and bipolars are available - any
recommendations? Most manufacturers only quote worst-case specs at
125C, all lower temperatures just have 'typical' figures. Our product
never gets above 50C internally.

As I mentioned in another post, I've found a very-low leakage Schottky
which looks a suitable replacement, but the circuit above looks like a
good alternative to another battery/mains diode-OR we have in the
product, which has to pass several amps and wastes a fair amount of
power. This battery pack is 2.2 Ah and is rechargeable, so leakage
current specs aren't so tight. I'll have a play with it in SPICE...

Thanks

R.

 

[Guest]

In general ANY reverse current into a coin cell is bad 'cos they lack
the vents other cell types have. (Example: if you look at say an AA
cell, you'll see some gaps under the pip at one end - or on some cells,
some grooves carved into one end, these are points of deliberate
weakness so the cell ruptures when internal pressure is just say 2-3
atmospheres so any bang is small and controlled).

In the UK the law ued to be that you must have two current limiting
components (eg a diode and a resistor) between any Li coin cell and a
voltage source to stop them exploding in a spectacular manner. Wish I
could find a link to this because the Health & Safety Executive used to
have a very explicit pdf on this with graphic descriptions of real life
examples of unintended charging which they had investigated. Does
violent explosion meet your criterion of damaging the cell? Even 1uA
into a CR2032 would really worry me. I *did* find one battery
manufacturer's app note saying "don't charge more than 3% of the
battery's capacity in its entire life":

http://www.73.com/specs/th693.pdf

Yes the Panasonic app note I found says more or less the same thing. It
doesn't present it as a safety risk though, more that the cell won't
meet its capacity specs.

I'd suggest as a start, use low leakage diodes (I'd start at
www.centralsemi.com, they have low leakage ranges with leakages of e.g.
one nanoamp at 100V). These are normal diodes though, I'm not sure what
their low Vf Schottkies are like.

OK, thanks, I haven't looked at this company before. They seem to do a
few different parts in the package I need.

There *are* some rechargeable Li coin cell chemistries but, when I
looked at these a few years ago, they had very limited cycle life and a
memory effect.

How much design freedom do you have? We could go on about "you should
use such and such a chemistry cell with a capacity of at least X" but if
the circuit's already laid out, and the moulding won't permit a larger
cell, there's no point discussing the optimal blue sky solution. Maybe
you should just accept that the cell needs occasional changing (is it in
a holder, or a tabbed soldered one?)

The board layout is all done, the cell is in a holder but it's a real
PITA for our service guys to get at, so we want as long a life as we can
get.

Thanks

R.

 

[[email protected]]

I was going to suggest this:

               Q1
               PNP
+3.3v >--o----.   .----o-----o----> Vbackup
         |   e \ / c   |     |
         |     ---    [R4]   |
         |      |b     |     |
         |      o------'     |
        [R1]    |            | +
         |     [R3]         --- B1
         |      |            -
         |      o----.       |
         |      |    |       |
         |    |/     |       |
         o----|  Q2  '---.|--'
         |    |>.        ||<-.
        [R2]    |        ||--+ Q3
         |      |            |
        ===    ===          ===
        GND    GND          GND

When +3.3v supply exceeds threshold set by
R1-R2, Q2 turns on. Q1 saturates, supplying
+3.3v to the load, and Q3 is cutoff,
disconnecting the battery B1.

When the +3.3v supply is too low, Q2 cuts
off, Q1 cuts off, Q3 turns on, connecting B1.

There are no direct reverse leakage paths. In
battery mode: cutoff Q2 blocks c-b conduction
through Q1, and Q3's gate takes no current.
With +3v is applied, Q3 disconnects B1.

Have I missed any leakage paths?

Cheers,
James Arthur

Neat - I like it!
I'd still be concerned about leakage through Q3 body diode and Q1 E-C.  
I know low leakage small-signal FETs and bipolars are available - any
recommendations?  Most manufacturers only quote worst-case specs at
125C, all lower temperatures just have 'typical' figures.  Our product
never gets above 50C internally.

As I mentioned in another post, I've found a very-low leakage Schottky
which looks a suitable replacement, but the circuit above looks like a
good alternative to another battery/mains diode-OR we have in the
product, which has to pass several amps and wastes a fair amount of
power.  This battery pack is 2.2 Ah and is rechargeable, so leakage
current specs aren't so tight.  I'll have a play with it in SPICE...

Thanks

R.

You wouldn't need special transistors--standard junk
units would easily suffice.

When the battery is ON Q1 is held fully OFF by R4, so
leakage will be nanoamps. Likewise, Q3's body diode's
leakage will be tiny. A Fairchild BSS138, for example,
is guaranteed to leak less than 100nA at Vds=30v.

But, if you can get good enough schottkies then that's
obviously easier and smaller.

And yes, I think your CR2032s failed from backcharging.
They don't like that.

Cheers,
James Arthur

 

[Tim Williams]

Neat - I like it!

I'd still be concerned about leakage through Q3 body diode and Q1 E-C.  
I know low leakage small-signal FETs and bipolars are available - any
recommendations?

Around room temperature, I've used regular old 2N4403s down to 100nA
or so, and BS170/2N7000 is good for <1pA gate leakage and about as
much reverse-biased. It's possible that there was some balancing
effect in my measurement, but still, that's not too bad eh?

The circuit I was playing with was:
http://webpages.charter.net/dawill/tmoranwms/Elec_Sample4.gif

I got less than a volt drift on the 1nF cap over a minute, so that's
1V/60s * 1nF = 17pA or so. Might've been even less than that.

And yes, I made this circuit when I discovered I had no JFETs. It has
an awful capture transient due to being biased off in the hold state.
Still turns out fairly fast though.

Speaking of JFETs, some really dinky JFET is referenced in AoE for
miniscule leakage purposes. Probably too small for these purposes.

Tim

 

[[email protected]]

[email protected] says...






Sure, that's the plan.  I have tested OnSemi's MMDL301 as a replacement
and the measured leakage current is << 1 uA.

My other question was, is 30uA of parasitic charge current lilkely to
kill the cell in a matter of weeks, or do I need to look for some other
fault?  

I found some partial answers in Panasonic's Lithium Battery Handbook

http://products.panasonic-
industrial.com/datasheets/en/Panasonic_Lithium_Handbook_Part1.pdf

where it states "the total charge applied over the life of the cell
should be less than 3% of the capacity".  
(225mAh x 0.03)/(3 x 7 x 24) gives about 13 uA so it's certainly
possible this is the cause.

I don't quite follow your calculation.

225mAhr x 3% =~6.8mAhr max. permissible charging.

At 10uA leakage, 6.8mAhr would be exceeded in
6.8mAhr / 10uA = 680 hours, or 28 days.

Likewise, 1uA leakage would exceed spec in 280 days,
or about 9 months.

Another thing to watch out for is dirt. Dirty boards
can easily leak microamps and kill backup cells quickly.

Cheers,
James Arthur