Preferred LiFePO4 USB charger/regulator IC

I'm building some hand-held gear with STM32 and other stuff on a 3.2V 
supply, with consumption up to a watt or two, and want to use a single 
LiFePO4 cell with USB charging, and preferably just run directly off 
Vbatt, with no additional regulator.

The Linear LTC4098 looks nice, and Maxim has the MCP73833.

Any experience using these, or alternatives to suggest?
Hand-solderable would be a plus, maybe through a big via.

Clifford Heath.

On Sunday, November 10, 2013 10:25:13 PM UTC-8, Clifford Heath wrote:
> I'm building some hand-held gear with STM32 and other stuff on a 3.2V 
> supply, with consumption up to a watt or two, and want to use a single 
> LiFePO4 cell with USB charging, and preferably just run directly off 
> Vbatt, with no additional regulator.

1 WattWhat?

We are using 800mAHr on 3.3V, so around 2.7 WattHour.

People are leaving essential info on posting, which is really draining my mind reading power, especially over the net.

Hi Clifford, Edward,

On 11/11/2013 8:45 AM, edward.ming.lee@gmail.com wrote:
> On Sunday, November 10, 2013 10:25:13 PM UTC-8, Clifford Heath wrote:
>> I'm building some hand-held gear with STM32 and other stuff on a 3.2V
>> supply, with consumption up to a watt or two, and want to use a single
>> LiFePO4 cell with USB charging, and preferably just run directly off
>> Vbatt, with no additional regulator.

So, you just want to "regulate" what goes *into* the battery
(i.e., nothing between the battery and load -- just between
battery and USB)?

Are you sure the design will operate with battery at "full charge"
(Vbatt(max))?  I.e., just because you only charge it to 3.2V doesn't
mean you will never see anything higher on Vbatt (3.6V).  And, do you
care about how it behaves when the bottom falls out?  Anything that
disconnects the load so the cell doesn't run into the ground?

> 1 WattWhat?

He's stating the size of his load.  He hasn't expressed any preference
for how long the battery needs to last --so, no claims as to how long
it will take to *charge*, etc.  I.e., your (below) battery mmay give him
an hour or two of runtime.  It may give *you* TEN hours, depending on
*your* load!

> We are using 800mAHr on 3.3V, so around 2.7 WattHour.
>
> People are leaving essential info on posting, which is really draining my mind reading power, especially over the net.

> > We are using 800mAHr on 3.3V, so around 2.7 WattHour.
> 
> > People are leaving essential info on posting, which is really draining my mind reading power, especially over the net.

> He's stating the size of his load.  He hasn't expressed any preference
> for how long the battery needs to last --so, no claims as to how long
> it will take to *charge*, etc.  I.e., your (below) battery mmay give him
> an hour or two of runtime.  It may give *you* TEN hours, depending on
> *your* load!
> 

Hence, my joke about reading his mind over the net.  We need to know the size and type of battery involved, as well as load.  For example, you can pull half a watt out of a 3V 200mAHr coin cell at 5mA.  But at 50mA (typical 32 bitter), it would not last more than one-tenth of a watt.  So, it's better to spec the battery several time more than your load.

We looked at all kinds of batteries, including LiFePO4, but end up going back to NiMH.  We can get 800mAHr (3 x AAA) for $2 and 400mAHr (3 x AAAA) for $1.5.  Nothing Li* for less than $10.

On 11/11/2013 11:47 AM, edward.ming.lee@gmail.com wrote:
>>> We are using 800mAHr on 3.3V, so around 2.7 WattHour.
>>
>>> People are leaving essential info on posting, which is really draining my mind reading power, especially over the net.
>
>> He's stating the size of his load.  He hasn't expressed any preference
>> for how long the battery needs to last --so, no claims as to how long
>> it will take to *charge*, etc.  I.e., your (below) battery mmay give him
>> an hour or two of runtime.  It may give *you* TEN hours, depending on
>> *your* load!
>
> Hence, my joke about reading his mind over the net.  We need to know the size
> and type of battery involved, as well as load.  For example, you can pull half
> a watt out of a 3V 200mAHr coin cell at 5mA.

No.  You can pull 0.5W-Hr out of said coin cell.  But, you're only
pulling it at a rate of 15mW (5mA * 3V).

> But at 50mA (typical 32 bitter), it would not last more than one-tenth of a watt.
> So, it's better to spec the battery several time more than your load.

His point is that his instantaneous current draw from a 3.2V supply is 
~350-600mA -- regardless of the "capacity" (W-Hr) of that "battery".
He wants a charging controller that will be able to eek that out of
the USB connection -- dumping any "extra" power (not used by his load)
into the battery.

I.e., USB port gives you 2.5W (500mA @ 5V) and he wants to (optionally)
use ~1-2 of that in his load and the rest into the battery.  Depending
on the state of his battery (AND his load, at the time), the amount of
time to bring the battery back to full charge will vary.

[And, will obviously be longer if he has a larger "battery"]

[At least, that's how *I* read his post!  :> ]

On 12/11/13 06:09, Don Y wrote:
> On 11/11/2013 11:47 AM, edward.ming.lee@gmail.com wrote:
>>>> We are using 800mAHr on 3.3V, so around 2.7 WattHour.
>>>> People are leaving essential info on posting, which is really
>>>> draining my mind reading power, especially over the net.

Don has it right.

I gave exactly the information that you required. The drain is as 
stated. If I want longer run-time I'll use a bigger cell.

I don't care about the charge time, as long as it shuts itself off 
safely. The device is mainly used mobile, when it's not on charge.

Don, Thanks for asking about the maximum voltage. I expect that the
circuitry will survive 3.6 volts happily, and have to check all the data 
but possibly even 4.2V. So I don't think that's an issue.

The CPU can monitor the battery and shut down to minimal draw on low 
voltage limit.

Now, can we get back to the question? What devices do you prefer for 
this, or can you warn me of any other difficulties I might run into with 
the ones I mentioned?

Clifford Heath

On Monday, November 11, 2013 12:21:34 PM UTC-8, Clifford Heath wrote:
> On 12/11/13 06:09, Don Y wrote:
> 
> > On 11/11/2013 11:47 AM, edward.ming.lee@gmail.com wrote:
> >>>> We are using 800mAHr on 3.3V, so around 2.7 WattHour.
> >>>> People are leaving essential info on posting, which is really
> >>>> draining my mind reading power, especially over the net.
> 
> Don has it right.
> 
> I gave exactly the information that you required. The drain is as 
> stated. If I want longer run-time I'll use a bigger cell.
> 
> I don't care about the charge time, as long as it shuts itself off 
> safely. The device is mainly used mobile, when it's not on charge.

Do you run it while charging?  Our device use around 100mA active, so we design a charger that shut it down and draw around 10mA + charge current.  Sometimes they might leave it charge for days.  One good thing about NiMH is that you can just leave it at 4.2V for a long time without much damage.  However, LiFEPO4 charge at 3.6V.  No sure about overcharging issues.

> 
> Don, Thanks for asking about the maximum voltage. I expect that the
> circuitry will survive 3.6 volts happily, and have to check all the data 
> but possibly even 4.2V. So I don't think that's an issue.

Double check on that.  My PIC32 (and many other ICs) top out at 4.0V.  That is likely a factor of current fab process.  So, i would not be surprise to see many chip max out at 4.0V.

> The CPU can monitor the battery and shut down to minimal draw on low 
> voltage limit.
 
> Now, can we get back to the question? What devices do you prefer for 
> this, or can you warn me of any other difficulties I might run into with 
> the ones I mentioned?

We are still evaluating different approaches.  We are currently using two ways: 1. Charge at constant 4.2V and LDO to 3.3V
2. Charge at 4.2V and relay to 3.6 (and slowly drop to 3.0V).

1. is difficult for LiFePO4 with charging at 3.6V.  The battery quickly drop below critical level, unless we get an expensive LDO.

Hi Clifford,

On 11/11/2013 1:21 PM, Clifford Heath wrote:

> I don't care about the charge time, as long as it shuts itself off
> safely. The device is mainly used mobile, when it's not on charge.

Remember that the characteristics of the port into which
your device gets plugged will also have a role in charging time!

I.e., the amount of "power" that you have available to your device
(which must be used by charging + operating -- if the user opts
to leave the device "on") varies considerably.

If a low power port is feeding your device, you'll have ~100mA
to play with -- instead of ~500mA.

There are tolerances on the supply voltage available *from* the
port.  Plus, potential losses in any interconnecting cable by
which THE USER decides to tether you to the port (I think 5m
is the legal limit?).  So, you'll see IR drops in that cable
as well.

E.g., you could have, nominally, 2.5W available (5V@500mA) or
even as little as ~400mW! (low Vusb and low power port).

Don't underestimate the difference (factor of 5 or 6) in terms of
charge times!  *And*, the impact having the device "powered on"
while being charged (i.e., user's may complain that the device
*never* charges -- if you are consuming 2W and they are on a
low power port, etc.).

> Don, Thanks for asking about the maximum voltage. I expect that the
> circuitry will survive 3.6 volts happily, and have to check all the data
> but possibly even 4.2V. So I don't think that's an issue.

Look at the battery spec.  Even if *you* can't charge it to an
"unacceptable potential", a "new" battery may be "pre-charged"
to a level that causes you grief!

[I'm not going to worry about my misuse of the term "battery" here  :> ]

> The CPU can monitor the battery and shut down to minimal draw on low
> voltage limit.

You just want to make sure a device sitting on a shelf for "some
period of time" doesn't KILL the battery (assuming replacement
batteries are expensive, annoying or "unsupported".

> Now, can we get back to the question? What devices do you prefer for
> this, or can you warn me of any other difficulties I might run into with
> the ones I mentioned?

I've been kicking the can down the road, so far, regarding my
needs (wait for all the software to be done before designing the
hardware).  So, can't point to a "preferred device".

And, not sure how detailed warnings you need...  :<

E.g., hot plugging/unplugging always has design consequences.
If you are looking for hobbyist quantities, some "SMT" devices
can be carefully soldered "by hand".  Others (e.g., those with
thermal connections under the device -- common for power
handling devices esp in small packages! -- which complicate
"hand assembly".

If you can afford it (space), make it easy to break the foil
to the battery so you can insert a current probe, etc.  You
can usually work-around this but remembering that you are
*likely* doing to want to do it can save you an OhShit moment
later.

 From a consumer's point of view, any *holder* you can provide
into which a COTS *tabless* cell can be inserted will get
lots of first-born children named "Clifford" in your honor!  :>
OTOH, with a single cell, maybe they won't be *as* ecstatic
(as they would with a multi-cell BATTERY)  :<

Consider whther or not you want to also allow a "wall wart" to
power (or charge!) your device.  E.g., a GPS tends to see much
more use tethered to a power source than a USB source *or*
completely portable.  There are consequences inthe power handling
there, as well.

AND, clearly/unambiguously indicate connector polarity and
*dimensions*, if possible!  The morons who decided for the
typical barrel connector legend:

+ ---- . )------ -

should be skinned alive!  The ")" representing the sleeve of
the barrel often appears as a CLOSED CIRCLE -- because the entire
symbol (including the + and - "raining wheels") has to be
reproduced at such a small scale!  A smarter approach would have been
a + or - INSIDE a circle (indicating the polarity of the center
conductor wrt the outer sleeve).

As a result of these nitwits, we'll be stuck with this stupidity
for another generation!  Sheesh!!

I'll try to move to a machine with a PDF reader so I can see what,
specifically, the devices you've mentioned are/do.

On 11/12/2013 2:14 PM, Don Y wrote:
> Hi Clifford,
>
> On 11/11/2013 1:21 PM, Clifford Heath wrote:
>
>> I don't care about the charge time, as long as it shuts itself off
>> safely. The device is mainly used mobile, when it's not on charge.
>
> Remember that the characteristics of the port into which
> your device gets plugged will also have a role in charging time!
>
> I.e., the amount of "power" that you have available to your device
> (which must be used by charging + operating -- if the user opts
> to leave the device "on") varies considerably.
>
> If a low power port is feeding your device, you'll have ~100mA
> to play with -- instead of ~500mA.
>
> There are tolerances on the supply voltage available *from* the
> port.  Plus, potential losses in any interconnecting cable by
> which THE USER decides to tether you to the port (I think 5m
> is the legal limit?).  So, you'll see IR drops in that cable
> as well.
>
> E.g., you could have, nominally, 2.5W available (5V@500mA) or
> even as little as ~400mW! (low Vusb and low power port).
>
> Don't underestimate the difference (factor of 5 or 6) in terms of
> charge times!  *And*, the impact having the device "powered on"
> while being charged (i.e., user's may complain that the device
> *never* charges -- if you are consuming 2W and they are on a
> low power port, etc.).
>
>> Don, Thanks for asking about the maximum voltage. I expect that the
>> circuitry will survive 3.6 volts happily, and have to check all the data
>> but possibly even 4.2V. So I don't think that's an issue.
>
> Look at the battery spec.  Even if *you* can't charge it to an
> "unacceptable potential", a "new" battery may be "pre-charged"
> to a level that causes you grief!
>
> [I'm not going to worry about my misuse of the term "battery" here  :> ]
>
>> The CPU can monitor the battery and shut down to minimal draw on low
>> voltage limit.
>
> You just want to make sure a device sitting on a shelf for "some
> period of time" doesn't KILL the battery (assuming replacement
> batteries are expensive, annoying or "unsupported".
>
>> Now, can we get back to the question? What devices do you prefer for
>> this, or can you warn me of any other difficulties I might run into with
>> the ones I mentioned?
>
> I've been kicking the can down the road, so far, regarding my
> needs (wait for all the software to be done before designing the
> hardware).  So, can't point to a "preferred device".
>
> And, not sure how detailed warnings you need...  :<
>
> E.g., hot plugging/unplugging always has design consequences.
> If you are looking for hobbyist quantities, some "SMT" devices
> can be carefully soldered "by hand".  Others (e.g., those with
> thermal connections under the device -- common for power
> handling devices esp in small packages! -- which complicate
> "hand assembly".
>
> If you can afford it (space), make it easy to break the foil
> to the battery so you can insert a current probe, etc.  You
> can usually work-around this but remembering that you are
> *likely* doing to want to do it can save you an OhShit moment
> later.
>
>  From a consumer's point of view, any *holder* you can provide
> into which a COTS *tabless* cell can be inserted will get
> lots of first-born children named "Clifford" in your honor!  :>
> OTOH, with a single cell, maybe they won't be *as* ecstatic
> (as they would with a multi-cell BATTERY)  :<
>
> Consider whther or not you want to also allow a "wall wart" to
> power (or charge!) your device.  E.g., a GPS tends to see much
> more use tethered to a power source than a USB source *or*
> completely portable.  There are consequences inthe power handling
> there, as well.
>
> AND, clearly/unambiguously indicate connector polarity and
> *dimensions*, if possible!  The morons who decided for the
> typical barrel connector legend:
>
> + ---- . )------ -
>
> should be skinned alive!  The ")" representing the sleeve of
> the barrel often appears as a CLOSED CIRCLE -- because the entire
> symbol (including the + and - "raining wheels") has to be
> reproduced at such a small scale!  A smarter approach would have been
> a + or - INSIDE a circle (indicating the polarity of the center
> conductor wrt the outer sleeve).
>
> As a result of these nitwits, we'll be stuck with this stupidity
> for another generation!  Sheesh!!
>
> I'll try to move to a machine with a PDF reader so I can see what,
> specifically, the devices you've mentioned are/do.

What he said...plus.
Is this a REAL USB device that can negotiate power levels
with the host?  Or are you using the host as a dumb power supply?
Is this a production device that is required to meet, and be tested to
international safety standards?  There's lots of stuff in those standards
that many people would consider redundant, irrelevant or not
applicable to your
particular situation.  Those people would be the future unemployed.

On Monday, November 11, 2013 10:21:34 PM UTC+2, Clifford Heath wrote:
> ...
> Now, can we get back to the question? What devices do you prefer for 
> this, or can you warn me of any other difficulties I might run into with 
> the ones I mentioned?

I have been doing the charger "myself" on NiMH, NiCd earlier and
intend to do so for Li-whatever when I need to.
"Myself" meaning that I have written the charge/maintenance code,
run it on the smallest MCU in the system (well last time it was
the only one, an MCF52211).
Takes feedback for voltage at the charger input, voltage at
the battery, current flowing into the device from the charger
and current flowing into the system (i.e. from the battery).
Precision is considered as needed in the particular case, charge
current is PWM regulated (stepdown convertor).
So far has served me well - since I do not have large quantities
to build I want to be able to trim things without soldering etc.

Don's max. voltage reminder may be less easy to shrug away than
it seems though. I remember some years ago I had a tiny MP3 player
which worked for a while until on a very hot day it died (while
not being in use). I opened it, the MCU clock was not starting.
My main suspect was overvoltage kill (I did not go much further
to verify that though). It had a Li-ion battery which powered
it unregulated.

Dimiter

------------------------------------------------------
Dimiter Popoff, TGI             http://www.tgi-sci.com
------------------------------------------------------
http://www.flickr.com/photos/didi_tgi/sets/72157600228621276/