Experience of Lipo (Lithium polymer) batteries required.

[Blueteeth]

Hi,

I've been using Lipo batteries in a few jobs, specifically these ones:

https://www.sparkfun.com/products/8483

2000mAh, 4.2V (max voltage). Along with the MCP73832, the charging circuit used is taken directly from the microchip datasheet:

https://www.electro-tech-online.com/custompdfs/2012/11/21984e.pdf

I chose the 4.2V option, '2AC'.

Now, this is the same chip used in the sparkfun charger, albeit with an open-drain status output (doesn't change functionality). However, in my design the battery is permanently connected to the circuit, which can drain a constant 9mA, or 30mA if there is a fault and it doesn't sleep.

Often the battery's built-in protection circuit kicks in, due to low battery voltage. This can happen if it is not charged for a while, the constant drain lets the battery drop to <~2.9V so the internal FET disconnects it. Of course the only way to reset this condition is to start charging it, and this is a bonus because it means the battery is protected.

Now, the problem I have is, in two cases, I will leave the battery charging over night with the charging current set to 300mA (this has been measured). Twice after several hours the battery has disconnected itself, and once put back on charge the battery voltage was 2.9V. The charger of course now indicates its fully charged (or not charging) because it connot detect a battery present. Has anyone experienced anything similar?

I do not understand why the battery would drain so significantly *whilst* on charge. The charging power supply is more than enough to provide the current the circuit draws, and whilst it takes a while (2000mAh battery, charging at 300mA) this current was set to reduce the heat dissipation of the charging chip.

Unfortunately, the battery has to sit on top of the charging board, which, taking in 5V, and providing 3-4.2V to the battery @ 300mA warms a bit. (0.6W heat dissipation max) and this warms the battery. Only to maybe 30C.

The only thing I can think of is.... the charging chip determines the 'end of charge' state when the current drawn from the battery (with circuit attached!) drops below 7.5% of the set charging current., In my case thats 0.075 * 300 = 22.5mA. With the circuit drawing 9mA, that means that the battery stops charging when it draws less than 13-14mA. It appears this is too low, but I don't see how such little ammount of current, trickling into the battery, can cause the internal circuit to shut down, AND make the battery voltage drop to <3v.

If anyone has experience of 'dodgy batteries' or this particular charging chip I would appreciate any quirks/problems that have been encountered. I am hoping is it simply a dodgy battery, but if it turns out its my design, a complete overhaul would be required.

A very worried, BT

 

[cowboybob]

Sure sounds like a battery issue... Can't otherwise imagine (lacking a schematic of your actual circuit) how a

...battery would drain so significantly *whilst* on charge.

And battery voltage levels are very suspect.

 

[Blueteeth]

They really are.

Also, whilst on charge 'in' the system the battery voltage quickly rises to ~3.8V, no surprises there as these tend to do that - but - there it stays. I have yet to catch the battery voltage going over 4V, which means the charger remains in constant current mode. It could well be that the battery's internal circuits are kicking in because this forced current is causing the battery to heat up. But then I still don't get why it would then drop to the absolute minimum of 2.9V - which really is a drained lipo voltage.

Not sure if I can post much of the schem since I sold the design, sorry :/ But if this persists through today, I'll knock up a schem just of the charging. The circuit its powering is tried and tested, and whilst of course there could be shorts, or faults with it after heavy use in a factory, I have ruled that out by testing it separately. So in the test schem I'll just indicate a current sink for the circuits draw.

I'll get back to you aftre another 5 hour test, this time with an ammeter directly in series with the battery (looking at current in and out), as well as a voltage meter across it, and an ammeter on the charging input (to rule out a duff charger chip).

Bah, debugging.

 

[RCinFLA]

Are you sure the battery has disconnected itself? Did you try to load it.

Other then something wrong with battery, if you are using the SOT version of the charge regulator chip it probably cannot dissipate 0.6 watts of heat sufficiently causing the die to overheat and shutdown.

 

[Blueteeth]

Are you sure the battery has disconnected itself? Did you try to load it.

Other then something wrong with battery, if you are using the SOT version of the charge regulator chip it probably cannot dissipate 0.6 watts of heat sufficiently causing the die to overheat and shutdown.

Yeah I'm sure, the whole jig died, then disconnected the battery and measured the voltage directly across it - nothing. Only when I put it on charge with an external charger (same circuit as in system, just without any load connected to the battery) did it reconnect, with its voltage at the bare minimum.

Yep, using the SOT version as my boss prefers to avoid using things that many seem to be scared of soldering (I'm find with DFN, but if I'm away and they need replacing, not many at work are comfortable with it). And yes, the upshot is, even with significant copper (but few vias..) it really can't dissipate the heat. I thought this would happen, but the datasheet says it goes into 'thermal regulation mode' (it shutsdown, waits for the temp to drop 10C below max, then starts again), but I prefer not to rely on the chips internal temp cut-off to do the job.

All this means I will have to lower the charging current even further. For a 2Ah battery, it'll take an age to charge. Currently looking at simple switching solutions, but its a pity as these little chips cover everything from pre-charge, to charge termination with very little in the way of external components. Other manufacturers solutions seem to want you to set all of that externally. I usppose I could still go for linear charging with a 5V supply, but choose one of microchips chargers that requires an external pass-element - that would mean I could use adequate heatsinking on the board and up the charging current to 600mA+.

I don't wish to use the board as a place to rant.... but... with battery charger chips it *seems* like one either goes for simple integrated solutions - that have very limited charging capability, or much more complex multi-chip systems - which bring with them their own issues (switching noise, higher parts count, size, cost etc..). I've yet to see switching charger that doesn't require a lot of external components, bit like this microchip charger with a built-in buck converter. The previous times I@ve had to make chargers for NiMH, I've designed things from scratch, given the increased complexity of lipo's, and the safety concern, I went with simpler solutions this time.

 

[Blueteeth]

Ok,just sorted it.

Turns out the ceramic cap on the output, between charger chip and battery had ~30ohm resistance >.< I should have checked any shorts/resistance across the battery terminals first - lesson learned. So, just to wrap it up - that resistance put a constant drain on the battery, roughly 120mA. With a full battery that would have taken (again roughly) 15 hours to completely discharge it, and yet it took less than half that in my tests. And that is because.... with a low resistance across the battery, the charger provided a constant 300mA, but with that dodgy cap taking 120 of it, the battery was only charging at 180mA - so it was barely charged after the test, then got discharged by that cap.

Long story short - multimeters are our friends. And even though the ceramic cap (4.7u X5R) was rated for 10V, it simply failed. If there weren't size constraints I would have replaced it with a bog standard electrolytic which seem to be far more reliable.

Still, nice to see some real-time debugging on the forum, thanks for the replies guys, I needed a sanity check.