4

I have a design that I'm working on that is to be outdoors year round, with a 1 watt solar panel trickle charging a 6 Ah li-ion battery during the day.

I expect there to be about 3 to 4 months of subzero temperatures during the winter given the local climate.

I understand that charging lithium polymer batteries when they are at sub-zero temperatures is a bad idea, since it will cause lithium plating, thereby compromising the battery's future performance.

So, when the battery is too cold to charge (below freezing), my charge controller (MCP73871) automatically stops charging the battery.

In such a case, I would like to be able to take the solar energy that isn't being used and send it to some nichrome wire i'll have wrapped around the battery. I'm thinking that some sort of configuration like this would work... any thoughts?

schematic

simulate this circuit – Schematic created using CircuitLab

macdonaldtomw
  • 1,420
  • 17
  • 32
  • 2
    I fouond that anything special I want to do with battery charging and conditions justifies for me a µc because it adds flexibility and I know that in a few months I get a new idea what to add to such a thing – PlasmaHH Mar 20 '15 at 20:26
  • 2
    As PlasmaHH mentioned, you probably want to control this with a MCU. You probably have one, and if you have a spare ADC-capable pin there, go ahead and monitor temperature that way. One thing I would do is to add some hysteresis around the "freezing" point. The circuit as it is may have the tendency to oscillate around it by quickly turning the heating on and off. – anrieff Mar 20 '15 at 20:33
  • I was tempted to use my MCU to control the mosfet, however 99% of the time the MCU will be "asleep". The only way to wake the MCU is through a digital interrupt, or have it wake up and poll conditions every so often. This means that I would still need some sort of external ADC in order to fire an interrupt to wake up the mcu, unless I wanted the MCU to kick on every so often and poll the heating circuit. I'm weary of leaving the heating circuit on while the MCU is asleep, and I'm reticent to use unecessary power on the MCU, when it is scarce to begin with. – macdonaldtomw Mar 20 '15 at 20:42
  • 1
    Think about it: waking the MCU for a few µs each second is not going to waste any significant power. And you can always increase the interval if power is *really* that scarce. – anrieff Mar 20 '15 at 21:02
  • 1
    Thermistor needs a pullup - use a high R thermistor. | Pot should be large for small I / Gate pulldown can be MUCH larger and not needed at all with "ordinary" opamp. || Best of all - Look at using LiFePO4 battery! | Thermal use of sun can be 3+ times more effective if you can open/close a thermal path at minimal energy use. – Russell McMahon Mar 20 '15 at 21:50
  • 1
    Also adding lots of insulation and putting every circuit inside might be enough to keep the thing warm enough. – PlasmaHH Mar 20 '15 at 23:07
  • @Russel McMahon: I don't think my thermistor needs a pullup resistor in this case. It is being biased with a 50 uA constant current source (from the li-ion battery charging IC). – macdonaldtomw Mar 22 '15 at 13:43
  • @Russel McMahon: I don't think you can charge LiFePO4 batteries below freezing either... not sure what advantage you think they would have, other than better discharge performance at low temps. – macdonaldtomw Mar 22 '15 at 13:44
  • 1
    [**Related and of some relevance**](http://electronics.stackexchange.com/questions/97798/how-to-efficiently-heat-a-lithium-ion-battery-so-that-it-can-be-charged-in-sub-z?rq=1) || LiFePO4 stated minimum charging temperature varies with manufacturer. Some spec 0C as minimum while others say down to -20C. – Russell McMahon Mar 23 '15 at 01:51
  • 1
    This is a useful reference - [**One Laptop per Child battery page**](http://wiki.laptop.org/go/Laptop_Batteries) - OLPC are competent and carefully technically. Here they say LiFePO4 charging ranges are 0 to +60 for Goldpeak and -10 to + 60 for BYD. These two companies are amongst the top few Chinese battery makers. [I was told that Donald Trump owned 30% of BYD at one stage but, true of not, they are a large and competent battery maker]. – Russell McMahon Mar 23 '15 at 02:45
  • 4+ years on: 1> Did you do it? If so, how did it work out? 2. My new answer here may be of interest. https://electronics.stackexchange.com/a/467545/3288 – Russell McMahon Nov 16 '19 at 12:29
  • I didn't end up doing the project after all sadly. I'm still curious to see if it would work though so if anyone tries it out let me know! – macdonaldtomw Nov 19 '19 at 06:45

2 Answers2

3

Not a bad idea. I disagree with the other comments about using the MCU to control this. Why waste the quiescent current? Periodic brief wake ups do have a real effect over time. Just add some hysteresis to the comparator so that it's not oscillating. You could consider having the output of the comparator connected to an interrupt pin on the MCU if needed.

Also, it's probably just a drawing mistake, but I noticed that your FET is in backwards. You should flip it around so that the source goes to ground.

edit: typo

mblem22
  • 196
  • 5
  • 1
    He is running from solar power, a bird pooping on the solar cells has a much bigger effect... – PlasmaHH Mar 20 '15 at 21:44
  • 1
    I'm not saying that MCU wake ups are a huge current draw, just that a proper comparator circuit will draw less current on average. I work in the automotive electronics industry and quiescent current requirements are very strict, so we put a lot of effort into things like this. In general, it's better to have very low quiescent draw discrete circuits than to have to wake up your micro periodically. Of course it depends on the time between wake ups and a lot of other variables. Just my two cents. – mblem22 Mar 20 '15 at 21:50
  • 1
    All other effects are dominating in current draw. In the winter he heats up a wire, the total power saved would probably prolong the heating period in the order of microseconds. In the summer there is more than enough sunlight to compensate. He probably already has a µc. Efficiency is not always an absolute minimum in current draw, it can be reduction of parts, failure points and design time too. – PlasmaHH Mar 20 '15 at 22:02
  • 1
    Just because it's warm doesn't mean the sun is shining. There is still value in lowering quiescent draw. I agree that there's also value in shrinking the BOM. It's all about trade odds and what's best for the application. – mblem22 Mar 20 '15 at 22:18
  • 1
    Let me put it in another way: If the power savings by a comparator make the difference between the thing working 24/7 and eventually failing in the summer, then he should check each hour for bird poop. – PlasmaHH Mar 20 '15 at 22:30
  • Except that you can design this circuit to account for quiescent current, but you can't design it to account for bird poop. Fact is, we both have proposed methods of accomplishing the goal here and they both have advantages and disadvantages. OP can decide on their own which method best suits their needs. :) – mblem22 Mar 20 '15 at 22:35
1

Thanks to those who weighed in.

For me, ultimately, it comes down to saving a couple of pins on the MCU.

I've totally maxed out the usage of my I/O pins, though I do still have an ADC and a DAC pin I could use, but I'd rather save them.

In order to achieve hysteresis, I'm planning on using a double inverting op amp configuration, with the op amp serving the thermistor configured as a Schmitt trigger. The outputs of the two opamps feed into a dual input NOR gate. The output of the NOR gate drives the base of the N-channel MOSFET. Here's my plan:

schematic

simulate this circuit – Schematic created using CircuitLab

This should create the following truth table:

enter image description here

macdonaldtomw
  • 1,420
  • 17
  • 32