Battery Temperature Measurements for Charging

[MrAl]

Hello,

I was looking (again) at using the change in temperature
that accompanies a cell being fully charged up to trigger
the end of charge cycle for a NiMH charger. The current
would be high enough (probably 1 amp) to induce such
a change easily detectable by today's very inexpensive
thermistors.
The reason for this is simplicity -- the whole charger
comes out very simple and each cell could be charged
individually without adding too much to the circuit.

The whole idea however, depends entirely on the mechanical
thermal interface between cell and thermistor. The
thermistor has to be held tight against the cell body
or else the charger would never know when to turn off
and thus destroy the cell.

So my question is, has anyone ever had to mount a
thermistor against a cell? Apps that use battery
packs dont count, because what im talking here is
the (say AA size) cell has to be free to go into
any flashlight or other device, so the thermistor(s)
can not be epoxied to the cell(s).

Any ideas would be great!

Take care,
Al

 

[georges80]

I built a thermistor based charger a bunch of yrs ago before the more sophisticated charger chips with delta V etc detect came on the market. I used 2 thermistors, one as the reference for ambient and the other on the battery pack. That allowed for a delta measurement and turn off at a predetermined temp difference. Of course this is pretty crude since you really want to cutoff at some absolute value as well.

But, overall it is better to detect the voltage curve as the primary cutoff since it will allow stopping of charge before the battery temperature rises excessively. If you look at fast charge curves you'll see that by the time the temperature hits a high level the pressure inside the battery has already reached levels close to pressure release. It takes time for the core heat of the battery to work its way out to surface. Voltage monitoring gives you a head start on determining charge completion.

And as you mention, the big issue is how to reliably and temporarily make thermal connection with the battery body. My thermistor based charger was for some home made battery packs and so I could heatshrink a thermistor to one of the cells.

Consider if it was easy & reliable to use a thermistor per cell, then commercial cheap fast chargers would use that scheme rather than the more complicated delta V system...

With todays cheap and powerful uC it would be safer & more reliable to implement a multichannel charger that was delta V based. Even thermistor based you would require separate current limited 'channels' for the individual cells. So, the major complexitity difference would be in the firmware. With a uC that has multiple A/D channels it would probably be simpler from the hardware perspective to implement the delta V and also provide maintenance charge, battery conditioning, bad cell detection etc...

Or you can just spend $30 - $50 and buy a commercial charger that does all that already /ubbthreads/images/graemlins/grin.gif

george.

 

[MrAl]

Hello there george,

Well, i've reviewed the pro's and con's of thermistor
cell temperature measurement and was still considering
giving it a shot, but i actually had a second motive for
bringing the thermistor mechanical problem into focus...
That is, if i decide later to go with some form of
delta v measurement i'd like to back it up with a thermal
profile as well rather then a 'timer', or perhaps all three.
Because of this, i'll still need some form of mechanical
mechanism to hold the thermistor firmly to the cell.
Some unusual things pop up:
With the 'x' direction defined as along the axis of the
cell and the 'y' direction the height of a cell lying
on it's side,
any firm pressure in the 'y' direction tends to reduce
the effect of pressure in the 'x' direction holding
the connector pieces firmly to the battery tabs.
I guess force x = 10*y might take care of it though,
but was wondering if anyone else had any ideas on
this type of setup in general.

So far all i've come up with is to mount the thermistor
under the cell, then clamp the cell down, then turn
over (so the thermistor is on top).

I'd like to be able to present the final project for
other's to build also, so it cant be too hard to do,
and it has to use parts that are very VERY common
(no exotic ic's).

My first choice was the PIC, but then i'd have to
program a whole bunch and sell 'em, and that would
mean some people overseas would have a hard time
getting them due to shipping fees to some areas
of the world. I figured if it was all linear with
very common parts, many people could build and
completely understand it too. Of course with the
purely thermistor based unit they would have to
acknowledge the fact that the thermistor HAS to
have direct contact with the cell at all times.

I've also thought about mounting the thermistor to
the negative contact piece so the only way the therm
could become disconnected from the cell body is if
the terminal itself disconnects, which would stop
all current flow anyway. Only catch is i havent studied
the 3d thermal profile function from internal to negative
plate...(is it better or worse than from the side).
I guess it cant be 'too' bad can it (?)

Any more ideas, opinions, or other appreciated...

Take care,
Al

 

[georges80]

I just tried a magnet on a few nimhs I have - it sticks. So, maybe a small powerful magnet to just hold the thermistor to each battery - pot the thermistor in thermal epoxy to the side of a battery to 'take the shape'.

Just a thought...

george.

 

[MrAl]

Hello again george,

Oh yes, good idea...i'll have to try that one.
I was thinking about the same with the epoxy :-)
having two strong magnets (one on each side of the therm)
in with the potting.
The only thing i dont like so far is that the thermistors
will probably have to dangle at the end of a two lead
wire. Since i'll probably be using NTC devices and the
failure mode for the wire will be open circuit, if one
becomes disconnected it will charge forever. Maybe
i better switch to PTC devices instead.

Take care and thanks,
Al

 

[jamaica]

Mr. Al,
FYI: The Energizer CH30MN 30-minute NiMH charger has 4 bays and one
thermistor per cell. Each bay's thermistor has one leg soldered
to the back of the negative battery contact. Contact pressure
presumably ensures adequate thermal contact with the battery
(though these thin-gauge contacts don't look especially beefy).

Jamaica

 

[MrAl]

Hello there jamaica,

Oh ok thanks, that means it's been done with some
success already so it should work at least to some
degree.

I may even look at early temperature rise detection
when the cell just starts to rise in temperature.
I've compared temperature rise to the negative delta
method and found that with the negative delta
method (which seems to be preferred) it does allow
some temperature rise before turn off.
I'll combine the temp rise with georges idea of
comparing to ambient with another thermistor.
This should produce a charger anyone can build and get
parts for.

I would think the peak voltage detection would be best,
but perhaps reliability issues come into play trying
to detect extremely low changes in terminal voltage
(-3mv).

At least we know for sure it's possible now, so
thanks much!

Take care,
Al

 

[tonyb]

I have a fast charger I designed many years ago (20+yrs.)using two thermistors one sensing ambient and one sensing cell temp 5deg Centigrade rise terminates charge. It works quite well and I have never had a problem with it.
Digi-Key sells thermal cut off switches for NiMh and Nicads just for that application.

 

[MrAl]

Hi tony,

Oh that's cool.
So you know some part numbers for those parts at DigiKey?

Take care,
Al

 

[NewBie]

Have you thought about utilizing non-contact thermal sensors for measuring the battery temperature? Think long band IR.

Prolly more expensive though.

 

[MrAl]

Hello again Newbie,

That's a very good idea Newbie...

I would probably do it
if i could find two sensors cheap enough.
I'd even put up with the cost for my own charger as
long as the price wasnt outrageous.

I did a little more testing a little while ago, and
found that with the AA type NiMH cells i now have
the temperature rises quite a bit before a negative
delta of even 10mv (that's 0.010 volts) shows up.
It seems the short distance the heat has to travel to
reach the surface of the side of the cell means it
wont be too much hotter inside. With the sub C size
i think that's a different story.

I should be able to adapt a circuit to almost any sensor
output. If you spot any let me know.

It's thoughts like yours that eventually lead to
something that works really great. I think it
would make an exceptional charger from what i've seen
so far with the close correlation between case temperature
and minus delta V.

I even noticed that if you are willing to check them
about once every 5 mins or so, you can tell when they're
'done' <chuckle> by feeling for a temperature rise
with your fingers.

The only problem i had with the testing was that every
time i went to 'feel' them to see if they were getting
warm yet i noticed that even the slightest movement
jarring the batteries and the contact resistance changed
enough to cause sometimes a change in measured terminal
voltage of 50 millivolts. That's more then enough to
upset a reading when trying to detect a minus delta of
only 10 millivolts. I guess next time i'll have to
try more contact pressure with the battery holder.

Interesting thing that is surfacing here is that if you
are measuring temperature, you dont really need a
constant current source, which could mean a really
low down cheapest possible NiMH charger :-)
Heck, a wall wart and resistor and check every 5 mins
after say 1 hour by feeling the case of the cell(s).

Take care,
Al

 

[NewBie]

Al,

Have you thought about utilizing a dual battery contact, much like using a Kelvin connection, one which supplies the current, the other to read the voltage, on each end? That way you could get away from some of the errors the connection makes.

Also...

Okay, for starters, look here Al

These are mid-wave, and may work fine for what you need, I forget the best wavelengths at 25C, I think it was 10 µm, but I think the emissions went way up to 2 µm, it has been four years since I worked in that field. The PbSe works from 1.3 µm to 5.2 µm, which should work for you.

Hamamatsu mid-band IR

These run on down to 6.2 µm:
InSb and InAs photovoltaic detectors

 

[tonyb]

Al,

Just looked up the NiMh page in digikey catalog and found out that it is a thermostat from Texas Instruments for cell pack applications. Sorry I don't have a number for you, thats all they referenced to in the catalog. The NiMh Technical Handbook Part# P001-ND may prove useful. I know I have seen controller IC's from various manufactures that address the requiremnts for charging Nicad, NiMh and Li-ion batts. Maxim has a special purpose IC for batt. charging apps.

tony.

 

[MrAl]

Hello again,

tonyb:
Thanks tony, i'll be looking into some of those chips in
addition to what im doing with the thermal sensing too.

Newbie:
Oh yes, very nice! Now one last question immediately
comes to mind... Any idea what the price for those
guys are and where to get say two or three of them?
This could work out very very well if they arent
extraordinarily high priced.
Also, i couldnt seem to find any data on the sensors,
like what do they output and how much?
Thanks very much for finding those parts. I hope it
turns out i can find some of them and use them. If
not, i guess i'll have to go back to ordinary thermistors.
Heck, at 50 cents each it does make a heck of a lot of
sense i guess :-)

Take care,
Al

 

[NewBie]

Convieniently, Hamamatsu samples if you call. They are not the cheapest.

 

[MrAl]

Hello again,

Oh ok. Well for now i'll probably just use the cheap
thermistors, then later update the whole design with
the better temperature measurement devices.
I really want something anyone can build up first anyway,
so it wont hurt to use the regular parts first.

I forgot to reply to your Kelvin sensing idea before.
What i was going to say was that i considered using small
contact probes (about the size of a needle) but the
way i was doing it at first was i had two cells in
series in the same battery holder, and they were mounted
'inline' so that one contact touched the other in the
middle of the 'pack'. This meant there was very little
room to squeeze a probe (even a small one) between the
batteries (1/16 inch or so) so i didnt want to fool
with it, but i can always modify that so that the
cells each have their own holder. I think my contact
spring wasnt strong enough either, but i'll fix that
too. I only need to make the measurements during the
tests anyway, but it's interesting how much difference
the contact pressure can make when you're running 1 amp
through. If i was measuring for minus delta using an
automatic test setup and the cells were jarred just a little
or perhaps vibrated by someone slamming a door maybe,
the darn thing would have probably detected much more
then 10mv change either plus or minus. Minus of course
would have triggered end of charge :-) and thus they
wouldnt have been charged properly for that cycle.
What this tells me is that contact pressure has to be
great enough to avoid these kinds of errors in any
battery charger that doesnt use separate contacts
for measurements -- very good to know i think.

When i did my Sub C tests the batt's had welded tabs
so there was no problem with this, but with cells that
rely on proper contact it's really gotta be there! :-)

Take care and thanks for looking up the parts,
Al

 

[NewBie]

Anytime Al.