Voltage based Charge/Discharge Controller?

[IpigiEpos]

Good afternoon everyone. This is my first post on EEVblog and have recently decided to hone my amateur electronics skills.

A bit of backstory: I have been working on an off grid, solar powered, cross band repeater project as part of my Amateur Radio hobby. This is meant to allow communications with friends during camping and fishing trips out in the Canadian Wilderness, where cellphone service is not available. As such, the system must be autonomous for about 2 weeks using only solar and battery power. Standby power user is around 4.5-5W, but active use of the repeater can bring up power draw to 100W.

Basic setup:
50W - 12v Solar Panel
12V 26AH Sealed Led Acid Battery (Main Battery)
2x 11.6V (9v-12.6v range) 9Ah Li-ion Battery (around 100Wh each, "Backup" Batteries)
DC-DC converter (to normalize the Li-Ion battery output for the solar charge controller)
Solar Charge Controller (cheap chinese stuff, but works "well")

The Issue:
1) Unpredictable use of the repeater and cloudy days may drain main battery and not allow the system to fully charge
2) During very sunny days, solar power is wasted as it has nowhere to go (can easily get 46W output from that 50W panel at peak sunlight)
3) Backup batteries used to keep main battery topped off, but only compounds issue #2 as the system will just draw indiscriminately from solar/Li-Ion batteries

The solution:
1) Timer turns off load during off hours (2300-0500), saving around 30W of power - implemented and works great.
2) Design a "circuit" to only enable backup batteries to charge main battery if main battery is below specified voltage (12.2v, for example)
3) Circuit in #2 to reroute power to a secondary DC-DC converter to recharge Li-Ion Battery when Main Battery is above certain voltage (13.5v, for example)

I was looking at using an Adruino board to do something like that. The alternative would be to use two relays, one NC and one NO, triggered when solar panel has power, but would be less precise (and potentially less efficient) than using some kind of voltage detection.

Any suggestions, comments or alternatives are appreciated.

[penfold]

How do the Li-ion batteries charge the lead-acid? Is the DC-DC converter bi-directional or is there another one? Is the solar charge controller an MPPT type or is it one of the really-cheap ones that just acts like a linear regulator?

I think just for "accuracy" and configurability, an MCU/Arduino solution sounds best generally, it's not like they're too power-hungry or anything.

I don't know enough about repeaters to comment whether 5W is a lot or not. But, would it be possible to design a receiver that could detect an incoming signal and use that trigger a power on, which consumed less than 5W?

[IpigiEpos]

The charge controller is not MPPT, just a really cheap one that I purchased years ago for another small project. Probably not the most efficient, but least expensive for the time being.

The Li-ions are fed into a DC-DC that uses constant current, variable voltage output. Both the solar panel/DC-DC converter just feed whatever voltage the charge controller "wants". The DC-DC converter is not bi-directional and does have protection to prevent power to flow back in the return direction, so another one would be needed. It would most likely need to be a buck converter as the solar panel would feed higher voltage that what the BMS is expecting under most operating conditions.

I'm not too worried about the 5W idle load. This includes the timer, fan controller for heat management and radio itself. I've been using the system so far with some fairly good weather and just a few cloudy days while the Li-ions are disconnected. I've used the repeater as a piggy back and have been using it fairly regularly for the last two weeks. This is how I noticed the panel would actually top out the battery around 1400-1500 in the PM and just keep the lead-acid battery at standby voltage. 

 

 

[cortex_m0]

I'm wondering why use both SLA and Li-Ion batteries. Seems like it would simplify matters to use only one or the other. If the answer is "because they are paid for", that's good enough :-)

As far as your controller, an Arduino would work, but this seems like it should also be easy to implement in hardware. A wide-VCC range comparator (LM3302?), a voltage reference (LM4040?), and a few resistors.

If the charging circuit you are controlling has an enable, the output of the comparator would go there. Otherwise you'd need an additional switching transistor to turn on/off the supply to the charging circuit based on the comparator output.

[penfold]

That's interesting then. Is the 46W measured at the output of the charge controller or from the panel under an ideal load? Would be interesting if you could measure to power on the panel side and the battery side to see how efficient it comes out.

The MPPT advantage isn't absolute, though, if the sunlight is either a bright clear sky OR heavy overcast, it's probably not worth it because it'll either be sat somewhere around 17V or in "give-up/trickle charge" mode, but if there's a continuous variation of cloud cover over the day, there's a stronger argument for it. One way to work all that out is to do some data logging over the course of a trip and work out whether the times that you're running low on charge are due more to lack of available light or poor conversion and being too far from the MPP.

If you were so inclined, an MPPT buck converter can be a relatively interesting little project. With an Arduino measuring the voltage and current at the input and using one of the PWM'd "analogue" outputs at the input to a PWM modulator it can be simple enough to reasonable efficiency without really doing any serious design effort and it's quite tolerant to errors in inductor value.

[IpigiEpos]

There are a few reasons for the mix of batteries, but the first and foremost is indeed that it just happens to be what I had laying around. Other reasons include energy storage capacity (the 26AH is the largest I own), the ability of the SLA to provide high current draw, and the SLA's nominal voltage. I've been investigating sources/costs for replacing with LiFePo4, or even some old Tesla/Leaf batteries as a future upgrade.

WRT nominal voltage, the radio expects 13.5v +/- 10% as operating voltage according to the spec sheet, but testing has shown that it can work off as low as 11.5v , so the bulky SLA has a bit of an advantage in that it doesn't require voltage manipulation for operation. Mixing the two allows me to maximize energy density and weight, based on what I had on hand.

The neat thing about the setup is that I can feed the DC-DC converter anywhere between 3-30v. I could potentially recharge the thing using USB power banks or off a laptop charger, if I wanted to. Some scenarios are not efficient, but flexibility is there. 

The 46W rating from the solar panel was measured at the battery with all loads disconnected, except for the timer module which also acts as on/off/auto switch. I purchased one of those meters from amazon which provides some statistics for runtime, capacity, Watts, Voltage, Amps and "Battery Meter", which is essentially just a voltage range setting. Not sure how accurate it is, but testing with my multimeter seemed to match up.

I'm finding this to be quite an interesting project from a learning perspective, but I may be making things harder on myself with all this mixed battery stuff.

[penfold]

I'm finding this to be quite an interesting project from a learning perspective, but I may be making things harder on myself with all this mixed battery stuff.

I think the mixed chemistry approach is a wise idea, you could go more power-dense with straight Li-ion as you say, but you'd be compromising the peak power handling and margin of error in overcharging that SLA gets you. Depending on the duty cycle of the load, another option could be to use Cyclon batteries (spiral-wrapped SLA) which give huge peak power delivery, so better W/kg but not necessarily capacity Wh/m^2 though I forget the exact figures.

If you have a campfire going, there's always a thermoelectric generator, there are a few that are clearly only designed to go on the top of a gas stove, but I do remember seeing one that looked kinda sensible for an open fire, I just can't for the life of me remember any more detail about it.

[IpigiEpos]

I have though of using thermoelectric for other projects.

In this case however, the unit would be placed as high as possible on a hill, some ways away from the camp. This is essentially the staging area that leads into the camping area where we leave our cars. The antenna has to be placed as high as possible in order to clear any topographical obstructions. I'm even bringing a portable mast to raise clearance from ground by another 10 or so feet.

Thank you for your feedback so far, much appreciated!