Battery and Charger Strategy - any ideas?

[rkl@veecad.com]

Our product is an automotive battery tester which recharges its
internal 6V SLA battery from the same clips which perform the battery
test. Owners tend to clip the tester on a spare battery when driving
between breakdown service calls, and it takes a "drink". Internally,
we have temp-compensated voltage charging, so the SLA is charged
quickly and prolonged charging only floats the battery nicely. If the
SLA runs flat, you can get testing again after a few minutes clipped
to a battery.

Now we have to replace the SLA with Li-Poly or NiMH or other non-lead
technology, but it gets tricky.

NiMH batteries tend to be used in cycle use, where you fast charge up
to a voltage inflexion or delta temperature. Trouble is, we can't
initialise a fast charge every time the tester connects to a battery
because 50 fast charge cycles a day will destroy the battery!

GP Batteries rate NiMH cells for 0.1C trickle charge over long times,
so one option is to charge at 0.1C whenever the clips are connected.
Unfortunately, this means a 16 hour charge time from flat, whereas the
SLA got 90% charged in an hour or so.

Li-Ion or Li-Poly also tend to be used in cycle mode. They do voltage
charge, like the SLA. But they don't like to be at full voltage
indefinitely - you float for 3 hours after charge, then stop. Trouble
is, when the clips are connected, you don't know if you are continuing
a previous long charge, or if the battery is ready for the 3 hours.
In addition, keeping a Li-Poly at full voltage tends to reduce its
capacity, and mechanics tend to keep the tester on charge. I wonder
if we can reduce charge voltage slightly and put up with 85% of
capacity, to lessen the capacity deterioration effect when constantly
under charge.

I can think of clever microprocessor schemes, such as coulomb
counting, but we'll need field trials and the project is "interesting"
- ie. will have unexpected, time consuming and costly lessons to teach
us. Also spark plug pulses will reset the processor at times. The
coulomb counting IC typical circuits seem to show horrendous
quantities of parts.

I'm not keen on Li-Ion or Li-Poly, because of the brutal handling by
mechanics. Testers reach 60C sitting under the windscreen, get driven
over, dropped into the fan, zapped by spark plugs, connected to 240V
mains, and we can't tolerate a single battery combustion incident.

Please feel free to ask questions, make suggestions, or solve the
problem!

thanks,
Roger

 

[Eeyore]

Our product is an automotive battery tester which recharges its
internal 6V SLA battery from the same clips which perform the battery
test. Owners tend to clip the tester on a spare battery when driving
between breakdown service calls, and it takes a "drink". Internally,
we have temp-compensated voltage charging, so the SLA is charged
quickly and prolonged charging only floats the battery nicely. If the
SLA runs flat, you can get testing again after a few minutes clipped
to a battery.

Now we have to replace the SLA with Li-Poly or NiMH or other non-lead
technology, but it gets tricky.

NiMH batteries tend to be used in cycle use, where you fast charge up
to a voltage inflexion or delta temperature. Trouble is, we can't
initialise a fast charge every time the tester connects to a battery
because 50 fast charge cycles a day will destroy the battery!

Why not incorporate a microcontroller which looks at the battery's charging
'history' ? It also ought to be possible to get a reasonably accurate figure for
its state of charge. Combine these data and you can establish the best charging
regime at any point in time.

Graham

 

[Eeyore]

us. Also spark plug pulses will reset the processor at times.

Not if it's properly designed.

Graham

 

[MooseFET]

On Apr 16, 9:37 pm, r...@veecad.com wrote:
[...]

Now we have to replace the SLA with Li-Poly or NiMH or other non-lead
technology, but it gets tricky.

NiMH batteries tend to be used in cycle use, where you fast charge up
to a voltage inflexion or delta temperature. Trouble is, we can't
initialise a fast charge every time the tester connects to a battery
because 50 fast charge cycles a day will destroy the battery!

I assume that the device has a micro in it etc. that remembers things
and uses the battery as a power source to hold that memory up.

I don't see any problem with having the micro also remembering the
time of the last charging of the battery and other stuff like that.
You can gate the charger onto the fast charge mode only if the
estimated state of the battery is nearly discharged.

GP Batteries rate NiMH cells for 0.1C trickle charge over long times,
so one option is to charge at 0.1C whenever the clips are connected.
Unfortunately, this means a 16 hour charge time from flat, whereas the
SLA got 90% charged in an hour or so.

You can charge at a very low rate almost all of the time. I don't
think having two levels of charging makes the circuit much more
complex.

I'm thinking of a circuit like this:

Two bits from the micro run to level shifters that make 0 to +12V gate
drive signals of them.

Two N - MOSFETs with source resistors that will bias on a couple NPNs
if the current tries to go too high. Note: I assuming the positive end
is connected directly and the neg. is the one switched. Turn stuff
around if you can't do this.

The NPNs pull down the gate drive of the MOSFETs to regulate current.

When the external battery is disconnected, the body diodes of the
MOSFETs conduct allowing the battery to take over.

An ADC channel in the micro circuit will monitor the battery voltage
so the micro can decide.

 

[john jardine]

[...]

Now we have to replace the SLA with Li-Poly or NiMH or other non-lead
technology, but it gets tricky.

[...]


What's the "have to" business?. Is it specific to your industry, or are we
all going to be buggered by new regulations?.
Personally I wouldn't choose to specify anything other than sealed lead
acid. It's Victorian technology but a damned sight more
reliable/predictable/cheaper than the newer fluff.

 

[Terran Melconian]

NiMH batteries tend to be used in cycle use, where you fast charge up
to a voltage inflexion or delta temperature. Trouble is, we can't
initialise a fast charge every time the tester connects to a battery
because 50 fast charge cycles a day will destroy the battery!

I would not say this is always the case. Qutie a lot of work regarding
NiCd behavior was done for space missions, for example, and those
applications did not generally involve full cycles or fast charges.

I am aware of three types of charge that NiCd and NiMH can commonly
accept:

* Fast charge, 1C or more, with appropriate dV/dt or dT/dt termination

* C/10 charge, for several hours after it's full but not forever

* C/300 charge forever

You can get a fairly good impression of the state of these cells from
the voltage, perhaps temperature compensated - certainly better than you
can get from a lead acid using just the voltage. It seems like the best
approach for you is to set a charge cutoff below which you will perform
a fast charge and above which you will perform a slow charge.

I would recommend NiCd, which is more robust in most ways than NiMH, and
is probably the most robust and least expensive technology remaining
once you eliminate lead acid for whatever reason is forcing you to do
that.

If you can accept less than a full charge, you can charge to a
(temperature-compensated) voltage level; you can't get to 100% using
this technique like you can with lead-acid, though, because the voltage
starts going back down for reasons my chemistry is too weak to properly
understand.

Forget about lithium; the safety issues will double your engineering
work in exchange in an application where none of the advantages matter
much. If you have to go with them, look at A123's cells.

 

[rkl@veecad.com]

Why not incorporate a microcontroller which looks at the battery's charging
'history' ? It also ought to be possible to get a reasonably accurate figure for
its state of charge. Combine these data and you can establish the best charging
regime at any point in time.

Graham

That's what I called "coulomb counting". Laptop battery monitors do
this. Its the development time, parts count that put me off. Its
still an option, perhaps in simplified form.

 

[rkl@veecad.com]

I would not say this is always the case. Qutie a lot of work regarding
NiCd behavior was done for space missions, for example, and those
applications did not generally involve full cycles or fast charges.

I am aware of three types of charge that NiCd and NiMH can commonly
accept:

* Fast charge, 1C or more, with appropriate dV/dt or dT/dt termination

* C/10 charge, for several hours after it's full but not forever

* C/300 charge forever

NiMH are more like NiCd than before, so much of your NiCd ideas now
apply to NiMH. GP Batteries NiHM specs permit C/10 for 12 months, C/
20 forever. The NiCd space application angle is a good idea - I look
up info.

You can get a fairly good impression of the state of these cells from
the voltage, perhaps temperature compensated - certainly better than you
can get from a lead acid using just the voltage. It seems like the best
approach for you is to set a charge cutoff below which you will perform
a fast charge and above which you will perform a slow charge.

I would recommend NiCd, which is more robust in most ways than NiMH, and
is probably the most robust and least expensive technology remaining
once you eliminate lead acid for whatever reason is forcing you to do
that.

Lead and Cadmium are out, this is for EU.

If you can accept less than a full charge, you can charge to a
(temperature-compensated) voltage level; you can't get to 100% using
this technique like you can with lead-acid, though, because the voltage
starts going back down for reasons my chemistry is too weak to properly
understand.

Thanks. I think something similar could work with NiMH. Switches to
C/10 when voltage comes up above a set point, below that current
increases to C/3 etc. Temperature compensated.

Forget about lithium; the safety issues will double your engineering
work in exchange in an application where none of the advantages matter
much. If you have to go with them, look at A123's cells.

Your advice much appreciated.

 

[Rich Grise]

That's what I called "coulomb counting". Laptop battery monitors do
this. Its the development time, parts count that put me off. Its
still an option, perhaps in simplified form.

Don't car batteries have some kind of exemption? Couldn't you get a
similar exemption because you're in the car battery maintenance business?

As far as a replacement - I'd go with Ni-Cd - you've got good cold
cranking amps, and they're easy to keep charged. I don't trust any o'
them newfangled batt'ry tecknowledgies, y'know! ;-)

Or have they banned cadmium too?

I guess we'll see the asymptote when they finally ban carbon itself. >:->

Good Luck!
Rich

 

[rkl@veecad.com]

[...]

Now we have to replace the SLA with Li-Poly or NiMH or other non-lead
technology, but it gets tricky.

[...]

What's the "have to" business?. Is it specific to your industry, or are we
all going to be buggered by new regulations?.
Personally I wouldn't choose to specify anything other than sealed lead
acid. It's Victorian technology but a damned sight more
reliable/predictable/cheaper than the newer fluff.

RoHS lead and cadmium are out. We had a good run with SLA.

 

[rkl@veecad.com]

Not if it's properly designed.

Wrap the spark plug leads around almost any appliance and it will
crash or break. With thousands of units in the field, processor
crashes are my assumption.Yes, a watchdog is required. Yes, you can
shield stuff at a price.

The problem is that I replace a switchmode chip and a few parts with a
fairly sophisticated solution. Time and money. Plus politics.

 

[Eeyore]

That's what I called "coulomb counting". Laptop battery monitors do
this. Its the development time, parts count that put me off. Its
still an option, perhaps in simplified form.

It's not actually very difficult. Microcontrolelrs typically have A-D converters on
board now. Add a serial EEprom for data backup and you're away.

Graham

 

[Eeyore]

Wrap the spark plug leads around almost any appliance and it will
crash or break. With thousands of units in the field, processor
crashes are my assumption.Yes, a watchdog is required.

A watchdog is simply good practice !

Yes, you can shield stuff at a price.

Shielding after the event is the wrong way to do it. You make the equipment
tolerant of the environment *by design*.

The problem is that I replace a switchmode chip and a few parts with a
fairly sophisticated solution.

I don't follow you.

Graham

 

[Eeyore]

RoHS lead and cadmium are out.

Not in batteries they're not. Whatever gave you that idea ?

Graham

 

[Eeyore]

Lead and Cadmium are out, this is for EU.

You apparently misunderstand the relevant EU legislation.

Graham

 

[rkl@veecad.com]

Not in batteries they're not. Whatever gave you that idea ?

My terminology was wrong. RoHS does not apply to batteries, rather the
EU Battery Directive:

http://eur-lex.europa.eu/LexUriServ/site/en/oj/2006/l_266/l_26620060926en00010014.pdf

Our batteries are sealed, user replaceable, hardly "industrial", which
means cadmium is out as I read the above doc. In any case, we are not
going to leave lead to go to cadmium!

However, you may be right about lead - maybe we have a few more years
left with lead - my boss and EU distributor are calling the shots on
this, so I'll check the story with them.

Roger

 

[rkl@veecad.com]

On Apr 18, 1:20 pm, Eeyore <rabbitsfriendsandrelati...@hotmail.com>
wrote:

This is get it right first time. We have found that a new battery
solution requires 2 years of field use before the real story comes in,
and it can mean 100% product replacement. Temperature, battery life
under wierd charging patterns, battery explosions, user expectations.
If I can leave out a processor, I will. The main processor is not
available for the job.

This is a cost controlled PCB in a plastic case.

I'm removing risk and cost factors before I start. I not going to be
clever. This isn't a $70K project, its a$2K project. That is why I
appreciate all the insights I can get from you people - and the
challenges to my assumptions.

Roger

 

[rkl@veecad.com]

You apparently misunderstand the relevant EU legislation.

I don't claim to understand it!

Our EU distributor is pushing this. The new EU Battery Directive
provides for equipment manufacturers to "participate" in the cost of
battery recycling, as each member state sees fit. Over the whole of
the EU, this looks like a wide open exposure to the distributor - he
gets nervous.

 

[Terran Melconian]

clever. This isn't a $70K project, its a$2K project. That is why I

For $2k you can hardly afford to make a few prototypes, much less pay
for any engineering time. I respect your willingness to try, but I
would consider recognizing up front that a project with your stated
reliability requirements is simply not going to happen on a budget like
that, and not waste your time trying until adequate resources are
available. Of course, if the company politics dictate that the only way
you will ever get the resources is to start and then go overbudget...

Your best bet is probably to find a firm which has already developed the
technology and license it from them.

I saw elsewhere that you are primarily concerned with
recycling/disposal; my other comments regarding cadmium mostly apply to
NiMH as well.

 

[ehsjr]

Our product is an automotive battery tester which recharges its
internal 6V SLA battery from the same clips which perform the battery
test. Owners tend to clip the tester on a spare battery when driving
between breakdown service calls, and it takes a "drink". Internally,
we have temp-compensated voltage charging, so the SLA is charged
quickly and prolonged charging only floats the battery nicely. If the
SLA runs flat, you can get testing again after a few minutes clipped
to a battery.

Now we have to replace the SLA with Li-Poly or NiMH or other non-lead
technology, but it gets tricky.

NiMH batteries tend to be used in cycle use, where you fast charge up
to a voltage inflexion or delta temperature. Trouble is, we can't
initialise a fast charge every time the tester connects to a battery
because 50 fast charge cycles a day will destroy the battery!

GP Batteries rate NiMH cells for 0.1C trickle charge over long times,
so one option is to charge at 0.1C whenever the clips are connected.
Unfortunately, this means a 16 hour charge time from flat, whereas the
SLA got 90% charged in an hour or so.

Li-Ion or Li-Poly also tend to be used in cycle mode. They do voltage
charge, like the SLA. But they don't like to be at full voltage
indefinitely - you float for 3 hours after charge, then stop. Trouble
is, when the clips are connected, you don't know if you are continuing
a previous long charge, or if the battery is ready for the 3 hours.
In addition, keeping a Li-Poly at full voltage tends to reduce its
capacity, and mechanics tend to keep the tester on charge. I wonder
if we can reduce charge voltage slightly and put up with 85% of
capacity, to lessen the capacity deterioration effect when constantly
under charge.

I can think of clever microprocessor schemes, such as coulomb
counting, but we'll need field trials and the project is "interesting"
- ie. will have unexpected, time consuming and costly lessons to teach
us. Also spark plug pulses will reset the processor at times. The
coulomb counting IC typical circuits seem to show horrendous
quantities of parts.

I'm not keen on Li-Ion or Li-Poly, because of the brutal handling by
mechanics. Testers reach 60C sitting under the windscreen, get driven
over, dropped into the fan, zapped by spark plugs, connected to 240V
mains, and we can't tolerate a single battery combustion incident.

Please feel free to ask questions, make suggestions, or solve the
problem!

thanks,
Roger

Time for re-design that does not require any nternal battery.
Run your circuit from the battery under test. Charge up a
supercap from that battery if it needs memory between events.
You'll probably need to re-design for low current. If an
automotive battery can't provide enough power to drive your
circuit, it's shot anyway - no need to test further.

Ed