I'm in early stages of planning a low power IoT project, trying to figure out a method to preserve lithium battery health despite weather below -20C. From what I've read it's extremely bad to charge below 0C.
There will be solar power and I'm looking initially at using the DFR0535 to manage charging and regulation.
One idea I had was to simply disconnect the solar input via relay in the event my MCU senses low battery temperature. But that's wasting an opportunity to use the solar energy to power heat tracing and remove the temperature issue so that charging can proceed.
So what I think I'm looking for is some control logic to both put limits on the direction of current flow to the battery as well as enable /disable heat tracing.
Here's some rules I imagine I'll need:
Heat tracing should only be ON when temperature is less than 0deg.
Heat tracing should only draw power from solar source (never from battery)
Battery should only charge when temperature is greater than 0deg.
Battery should be allowed to discharge into the load independent of temperature (ignoring battery ESR effects).
In absence of sufficient power to run the MCU, charging/heating logic should maintain last state.
In the event of a dead battery the load (MCU +LTE module) should be able to operate off the solar supply alone (if there's some sun).
Does this make sense or am I going in a weird direction with it?
I'm imaging having an MCU-controlled bistable relay switch the battery's connection to the charge controller between two paths: one with a diode oriented to block charge current, and one without.
Another bistable relay would switch the heater connection to the solar panel terminals.
In theory I think this satisfies all the above except rule #2, unless a diode is added to prevent the heat tracing drawing current from the battery via the DFR0535.
My load is approximately 150mW and I'll be looking for the above control system to not add significantly to that (other than the case where the heater is running, where inefficiency is good)
Initially thinking of a 15W solar panel and 20Ah 3.7V LiPo to give you a sense of the currents and voltages invoved.
If the project box can be heated directly by the sun, that would be more efficient than indirectly heating it via solar electricity.
Glazing and project box insulation will require careful thought, in any case.
Lead acid batteries are much more tolerant of low temperature operation (generally rated to -20 C for both charging and discharging, whereas Li batteries crap out at 0 C).
If the project box can be heated directly by the sun, that would be more efficient than indirectly heating it via solar electricity.
Agreed, although it may be beneficial/necessary to have both. A benefit of solar electricity heating is it can be delivered direct to the battery cells where it's needed. I don't need the chassis or other components to be warm.
I'm also not yet sure if overheating during summer will be a concern. If it is I may want the project box shaded by the panel (won't be an option to move it seasonally).
Glazing and project box insulation will require careful thought, in any case.
Wait, by "glazing" are you suggesting the box be constructed of glass to allow light energy into it?
Lead acid batteries are much more tolerant of low temperature operation (generally rated to -20 C for both charging and discharging, whereas Li batteries crap out at 0 C).
Good to know about LA. I didn't mention this but size is also a constraint, and from what I gather Li generally has much better energy density which was my reason for pursuing it.
Can you give a detailed description of what your system is doing?
WIth battery-powered systems my first idea is to use a microcontroller that has a sleep-mode with drastically reduced current-consuption.
Example an ESP32 in deepsleep-mode goes down to 20-50 µA.
Though the WiFi-components of an ESP32 drives up the current to 100 mA if it is full active
So if you don't need WiFi something else than an ESP32 is a better choice.
Could your system work with beeing 90% of the time in a sleep-mode and only 10% active?
0.15W / 3.3C = 45 mA
If it has to be active all the time did you think about using something like a super low-power microcontroller like MSP430?
No, that doesn't follow at all - the PV panel area is likely to be larger (much larger)
than the box, able to collect much more power. Exposing a box to the sun means exposing it to the wind too...
Perhaps all you need is a change-over relay, switching the solar power between
charger and heating system.
Note that any relay for DC needs to be suitable rated, anything above 28VDC is likely to be a specialist device.
A 24/7 weather station.
At one time I did the whole sleep for 5 minutes, wake up the ESP32, do the thing, go back to sleep but with an over abundance of power available, I now just let it run 24/7.
I am using 30W of solar cells fed into a PWM Charge Controller that is connected to a 12V 16Ah LiFePo4 battery.
Can you give a detailed description of what your system is doing?
I'm the OP and my device is a sound level meter - constantly reading 48,000 samples per second over I2S, applying a frequency weighting filter, calculating a running average of the result. I can't sleep the MCU but I do sleep the LTE module and wake only every 15 minutes to upload a calculated average.
My code runs on the ESP32 downclocked to 80Mhz but downclocked any further isn't stable.
I would like to explore reducing the MCU consumption perhaps using another MCU, a low power FPGA or some kind of ASIC but that's another concern - Independent of how low I can get the MCU draw I will have cold weather and the need to solar charge.
A concern here is that with a dead battery scenario where the MCU has shut off: the system will be stuck in heating mode - it will thaw the battery but not switch back to charging mode because the MCU is not powered, so it doesn't sense the temperature and toggle the relay. Perhaps there could be a secondary MCU devoted to managing the heater which is powered directly from the solar terminals (with regulator between) or something? Otherwise I feel rule #6 from the OP is important here.
That's good to know, I hadn't gotten to looking for parts yet. The panel I was considering has max V_open_circuit of 30V so I guess that's not a good choice. I'm looking for panels with lower voltages then... Or maybe I can count on the fact the terminals will never be open circuit - always connected to either a heating or charging circuit so the real voltage will be lower?
Are there any solid state switching solutions like MOSFETs but that are non-volatile (retain switch state with power loss)? I don't expect so but I'm constantly surprised by technology.
A glazed solar absorber is one of many possible approaches. It doesn't have to be the electronics box. Solar panels capture only 15 - 20% of solar energy.
