off-grid - battery & winter 12V-generator?

On 7/30/26 8:11 AM, Richard Smith wrote:

Over the last 6-7 years the number of brands of self-contained (battery, mains charger, solar controller, mains output inverter) "solar generators" or "solar power stations" has exploded. The battery chemistry is now exclusively LiFePO4 for both safety over Li ion and charge cycle life. The usual arrangement is to have separate power switches for DC output (12 V and USB charging ports) and mains AC output so you can optimize the standby draw, and battery life is usually at least 80% capacity remaining after 3000 full cycles (so one cycle per day is over 8 years). Portable sizes range from 100 watt-hours to about

The most fuel-efficient way to use a gas generator with a solar generator is to only run the generator to charge the solar generator, and to match the gas generator rating to the maximum charging power of the solar generator so you are charging near the maximum rate the solar generator allows and the gas generator is close to it's rating (say withing 20% of the max for each). That way the generator is near full throttle for best fuel efficiency and the noisy run time is minimized. The solar generator DC charging input for 12 volt charging is limited by the current rating of 10 amps on the usual cigarette lighter socket on a car and 12V * 10 A is only 120 W so charging even a smallish 500 whr solar generator will take nearly 5 hours. Most larger solar generators need at least 30 volts DC input for charging to get faster charging and to keep the wire size from getting out of control so the simplest way is to use a standard mains output gas generator and charge using the mains input on the solar generator. Again, to keep wire size reasonable the usual plan is to wire solar panels in series so the current stays small and the voltage goes up so for medium and large solar generators the DC input (you can feed this port from either a battery bank or DC generator or solar panels) range is usually about 30 V minimum and 60-200 V maximum. Many large generators will accept 12 VDC for charging either on the main port or on a separate lower power input but that's mostly for historical reasons and probably never used on any solar generator larger than 500 whr because waiting 2+ days to charge is just not practical :-).

My quick take as someone who started with a 500 whr unit and a small solar panel six years ago to make sure I could keep my CPAP running in case of a power failure and now has a 2400 watt solar array and 3000 whr solar generator that I use to charge my PHEV car and home essentials backup if needed.

Minimizing generator size is critical to efficiency, but 12V car alternators are optimized for cost and size, not efficiency.

I've got a UPS-style inverter/charger setup based on Amazon components for my fridge and communications equipment. My generator is grossly oversized, luck of the second-hand draw. Something like a Honda EU-series or equivalent would be much more appropriate if planning ahead.

This is a fairly deep rabbithole, I'd suggest starting the burrow at alt.energy.homepower to keep it on-topic.

Thanks for reading,

bob prohaska

Somethign you folk from North America will be familiar with.

There is a 12V light, fridge and occasional inverter to mains-voltage system with 12V "leisure" lead-acid battery and photovoltaic (PV) charger and charge-regulator.

For winter when our days are short and sun not bright - would it be an efficient solution to have a small gasoline engine running a 12V generator eg. from a car? Would be efficient to have a small engine which runs full-power for a shortish time - something like an hour? - then stops. Smallest Honda "GX" engine is 100cc. Small engine+12V-generator & battery is much more fuel-efficient than running a generator supplying power on-demand? Tempting thought is - that would happen when it is dark and *cold* - therefore if you coudl recover the waste heat of the generator engine it would be welcomed.

I have no knowledge of this - anyone want to get this going?

Regards, Rich Smith

----------------------------------- Realistically the recommended charging current for a 100A Lead or LiFePO4 battery is 20A which means 5+ hours to fully recharge. The power from the generator is around 300W per battery. You can charge at higher current at some indeterminate risk to battery lifespan.

Inverter generators don't need to run at full speed to produce full voltage like a standard genny, they throttle down and consume less fuel at lower loads. Inverter circuits are the DC equivalent to AC transformers, they can convert one combination of voltage and current to another of slightly less power.

I'm about to test such a setup. My DC-AC power station consists of two 100Ah LiFePO4s powering a Vevor 2500W inverter, chosen for low standby loss if running the fridge overnight. A 30A solar controller, overload and accidental short protection and voltage, current and charge level monitoring added to its cost and complexity. I design and build more electrical than mechanical equipment.

The AC powered charger is a homebrew made from a 50A stick welding transformer, rectifier and capacitor, with an input Variac to adjust the power. The transformer isn't ideal for the task but I had it, hadn't found an alternative transformer and could tweak the circuit to work well enough, and since it's self-limiting to tolerate a stuck electrode it's suited to performing tests such as circuit breaker trip current, up to 70A so far. It can deliver 30A at 28V for 24V batteries and 40A at 14V for 12V, matching the 2x 20A suggested charging current for the batteries.

The test will be what overheats and needs upgrading, likely it's the reverse input protection diodes for which I bought a higher current upgrade. A forward diode in series keeps battery current from surging into the capacitor if connected when power is off, another in reverse to minus shunts the current from a battery connected backwards until the fuse blows. Commercial DC sources may lack this protection. The Variac brush current is a hard limit, replacements are scarce, expensive and tricky to make. I have a thermal imager to find hot spots and a thermocouple meter to more accurately measure them.

Grid power here costs $0.22 per KWH, I figure genny power at $0.60 to $1 depending on gasoline cost and electrical efficiency, which is poor for the welding transformer at higher loads. Nevertheless it can charge effectively on the 900W max of my Honda EU1000i, and at full power on my 2200W HF Predator, the first model. A Kill-A-Watt meter lets me keep the load within the Honda's limited capacity. The Vevor 2500W invertor runs my microwave and starts and runs a window air conditioner without a hiccup. Its no-load loss is around 10-15W, much better then the 50W of my APC1400 UPS which wastes half the battery charge while the fridge has cycled off.

For small AC loads an inexpensive Bestek 300 works well, usually it's all I need for the computers and TV. It's on batteries retired on schedule from hospital mobile laptop stations. My larger batteries were bought new to qualify for the now discontinued 30% tax credit.

The genny must be outdoors away from doors and windows for CO safety, which rules out collecting its heat, but I have free firewood for that. For theft safety it may need a 3 side + lid sound enclosure which I made from fireproof suspended ceiling tiles rimmed with shop made sheet metal U channels. The noise is from the engine, a larger muffler didn't help.

I reworked my clothes dryer vent to be the inlet for the extension cord to the driveway where the Honda genny needs only a small space cleared of snow and likely ice. Behind the house the cord can go through a hatch I made to load in firewood. The larger genny is fueled and in a shed there, but I have to shovel half way around the house to reach it and the snowblowers.

The little Honda is stored indoors without fuel, which is in an EPA sealed container stored in a closed bin. I made a rubber stopper and bulb pressurizer for the fuel tank inlet that lets me prime it for use and then completely empty it through the carb bowl drain before bringing it indoors with the wood stove.

Except for the weatherproof wall entry near equivalents to most of this setup could be purchased and plugged together. I didn't look hard at battery chargers because their experimental use is too restricted. I need the adjustable voltage output and current limiting plus meters of lab supplies.

This or its droogs can operate directly from a 12V or 24V battery.

Managing home power generation can become a nuisance chore that frequently interrupts you, even if the tasks are brief like feeding the stove. ... Speaking of which, I just read the electric meter several hours late and put the solar panels back in the truck bed, the only place they can be concealed in front of the house. In this overcast they produce about 5% of rating.

--------------------- My quick take as someone who started with a 500 whr unit and a small solar panel six years ago to make sure I could keep my CPAP running in case of a power failure and now has a 2400 watt solar array and 3000 whr solar generator that I use to charge my PHEV car and home essentials backup if needed.

Most larger solar generators need at least 30 volts DC input for charging to get faster charging and to keep the wire size from getting out of control so the simplest way is to use a standard mains output gas generator and charge using the mains input on the solar generator.

-------------------------------

Agreed. I've wired industrially for up to 1000 Amps with 4/0 cable, and No.

Anderson Powerpole connectors are common for DC because the connections don't have gender like pin & socket where power flows only one way, batteries both give and receive current, both sides of connectors may be live/hot. All connectors of the same current rating and color coding will mate together and are shrouded when unplugged. The color coding is for separation by voltage but it can be milled out of the housing if not needed.

Any system like this will span the range of design and build it yourself to buying a complete unit off the shelf. To start Richard's education I figured it best to let him see what an off the shelf unit looks like so he can appreciate all of the component subsystems and start getting a feel for the specs so he can decide what he needs. Also, since whatever someone constructs for themselves will inevitably be compared to what is available as a finished product it's useful to keep up with the commercial products. My unit has a 3071 whr 48V LiFePO4 battery, a 3600 watt max 4200 watt peak 120VAC inverter that can start a 12 amp vacuum cleaner and 15 amp table saw at the same time, and an MPPT 25-120 VDC input solar charger and 120VAC input charger both capable of up to 1600 watts charging but which can be turned down in 20% steps in the control menu so you can make the choice of time vs. battery life. It has six

I am reading and thanks, yes this is a journey into the topic. BTW with "car alternator" I was thinking get one from a scrapped car and spin the alternator with the small gasoline engine. Even the smallest "GX" engine at 100cc is looking a little overpowered for this. On the plus - I hear many car alternators have a built-in regulator ? Rich S

More thanks.

The refrigerator is going to be inherently 12V DC. Apparently something like 30A-hr to 50A-hr per day. Hence the small size of the system.

I am reading and thanks, yes this is a journey into the topic. BTW with "car alternator" I was thinking get one from a scrapped car and spin the alternator with the small gasoline engine. Even the smallest "GX" engine at 100cc is looking a little overpowered for this. On the plus - I hear many car alternators have a built-in regulator ? Rich S

--------------------------------- I've considered this too and have an alternator, besides incompatible belt types the stumbling block is that car voltage regulators aren't adjustable and I like to follow maker's numbers for the lead acid bulk, absorption and float stages, plus DC desulfation if needed. For LiFePO4 I use the AGM or similar float voltage and occasionally raise it to take the battery to its cutoff and balancing voltage. Also I update my current-measuring charge level monitor to 100%.

Drok offers a voltage adjustable and current limited supply that's nearly what I want but the voltage apparently doesn't go quite high enough, I'd like 16V to allow an anti-backflow diode.

told me that 13.6V is a good charger voltage, just below cutoff. The battery still charges rapidly to around 90%, then the current tapers down, but you wouldn't know this unless you display the current as well as the voltage.

My power supply works and didn't overheat with the output protection diodes bypassed, new ones rated for 162A tested but not installed yet. The welding transformer's efficiency is dismal, 33% at 44A out. The lost power mostly heats the secondary winding. This is how a "constant current" stick welder transformer operates, by wasting power, and why they have a low ON duty cycle and need 8 minutes out of ten to cool off.

The limit was the input Kill-A-Watt beeping at its 15A limit which is also the Variac brush rating. So I can check off preparing for hurricane season.

A Meanwell 15V 40A power supply or equivalent would work if its output was adjustable enough, or this if a company was paying.

More thanks.

The refrigerator is going to be inherently 12V DC. Apparently something like 30A-hr to 50A-hr per day. Hence the small size of the system.

--------------------

Solar power systems tend to grow until they hit a limit, in my case good places for panels.

We've discussed some of the tradeoffs between build and buy. LiFePO4 seems to be the agreed choice for batteries, partly from hope that they will deliver on their promised long life; AGMs sometimes didn't. Their disadvantage is the full charge cutoff where they disconnect and may allow full panel voltage to enter the inverter. Unlike lead they have no place to put excess charge current, lead batteries electrolyze and vent the water with it. Charging to 13.6V solves that but slows final charging rate and may allow the cells to drift off balance. You become a part time power plant engineer tending the batteries.

MPPT solar controllers are usually preferred for somewhat higher efficiency, especially in cooler climates. When the panels get hot their advantage over PWM decreases because the panel voltage drops. The most I saw was 10% higher current. MPPT wants sole control of the panel array while multiple PWM controllers can share it, an advantage if your loads need to be in separate areas of the house. They can allow a discharged battery to draw current from a charged one on another controller which current meters and individual DC-rated circuit breakers can detect and stop, if you catch it.

MPPT permits the higher voltage and lower current of panels in series, which reduces wire cost. I don't know how well it handles the cutoff disconnection, that may vary between units and takes lab equipment to detect. Even with PWM I've seen 100V kickback spikes when the controller shuts off the current. For that reason I have analog voltage and current meters on the panel downleads. The digital meters on the battery side are not subjected to the same static spikes or slowly rising voltage that may interfere with their power-on reset or calibration.

With inverters the loss from idle current can be an issue. Some get around it by detecting no load conditions and shutting down until they sense demand. This is good when the only load is a refrigerator overnight. The Vevor I bought doesn't sense loading but it idles at only around 10W. For DIY current sensing relays aren't expensive.

Some inverters may include a UPS function without the limitations of computer units value-engineered (cheapened) to run only long enough to shut down. I have an APC1400 server UPS that can operate indefinitely on a large external battery at somewhat reduced load. It's good for nights when a predicted storm may cause power loss but you don't want to expend the limited battery capacity needlessly in extended cloudy weather that prevents solar recharging. The APC's weakness is high inverter idle power which approaches what the compact refrigerator takes. It accepts external serial port commands but not to turn on the inverter when on battery, that takes a special button pushing sequence which they emphasize is Not Normal.

In the US the electrical code requirements for solar are rather demanding and expensive, partly to protect firemen spraying water on the roof.

I made solar panels free standing by adding a folding V leg of 1/2" electrical conduit to the back, 2 legs from each 10' length. I can rotate or move the panels to follow the sun.

Just bear in mind solar tends to be less efficient than you might think. I ran that gamut some years back setting up burglar alarm systems models here there was no power, and in un-powered warehouses. I found that if I didn't plan for 4 times the estimated max demand I was getting constant low battery communication from the cellular communicators. This is in SW Arizona where arguably it is among the best locations in the world for solar power.

Just bear in mind solar tends to be less efficient than you might think. I ran that gamut some years back setting up burglar alarm systems models here there was no power, and in un-powered warehouses. I found that if I didn't plan for 4 times the estimated max demand I was getting constant low battery communication from the cellular communicators. This is in SW Arizona where arguably it is among the best locations in the world for solar power. Bob La Londe CNC Molds N Stuff

---------------------------------

Yes, I might get 50% of rated output at the meter panel due to shade and suboptimal fixed angles and the resistance of 10 AWG (5.26mm^2) wire. They only test around 100% of label wattage at the panel connectors on a very clear and cool day with a variable resistive load adjusted to find the peak, which varies in voltage between identical panels. Barely noticeable haze or cirrus can drop them to 50%, overcast to 10% and rain below 5%. Fixed mount panels lose power off-angle, you may get only 2 hours near max power and 5 total per day, unless you can move them. I am not in a good location for solar power unless I move panels several times a day.

I have never done this, nor have I ever seen it first hand, but I heard described a very simple tracking system once. Panels are mounted on a swivel axis. A box with a slit is placed on either side of the panel. A small photovoltaic cell is used to trigger a circuit which moves the panel with an actuator depending on condition. Sound complicated and fiddly, but the idea seems pretty simple. Elevation is adjust every so often through out the year. I don't recall hearing how it was rest to the sunrise horizon at the end of travel. Maybe a timer. No signal from either trigger for so many minutes and it trips a circuit that powers the panels back to a limit switch using something like a

Given today's abundance of advanced flexible processors it might be easier to program a small computer system with a 365 day timer, to just move the panel/s by dead reckoning every few minutes, and then swing back to the sunrise horizon at sunset. Not at sunrise obviously because if all backup power is consumed overnight there would be no energy to restart the system when the sun rises.

A friend did that using LEDs as photo sensors. They actually work in reverse, turning light to voltage instead of voltage to light - triggering an arduino

I have never done this, nor have I ever seen it first hand, but I heard described a very simple tracking system once. Panels are mounted on a swivel axis. A box with a slit is placed on either side of the panel. A small photovoltaic cell is used to trigger a circuit which moves the panel with an actuator depending on condition. Sound complicated and fiddly, but the idea seems pretty simple. Elevation is adjust every so often through out the year. I don't recall hearing how it was rest to the sunrise horizon at the end of travel. Maybe a timer. No signal from either trigger for so many minutes and it trips a circuit that powers the panels back to a limit switch using something like a

Given today's abundance of advanced flexible processors it might be easier to program a small computer system with a 365 day timer, to just move the panel/s by dead reckoning every few minutes, and then swing back to the sunrise horizon at sunset. Not at sunrise obviously because if all backup power is consumed overnight there would be no energy to restart the system when the sun rises. Bob La Londe

------------------------------------ I started to make the 200W panel pole mount track by machining a stainless steel thrust ball bearing to fit its water pipe supports and buying a small DC geared motor. The sun sensors can be LEDs on either side of a baffle, they also output current when illuminated, though not much.

I played with an Arduino a little, didn't have a good use for its limited capabilities, then went back to programming on an old laptop running DOS and QBasic which has a nicer interpret or compile IDE, the structured programming syntax of Pascal minus its restrictions and all the legacy computer hardware free to use (unlike in Windows), KBD, LCD and HDD for files plus COM and LPT whose registers are addressable in the I/O space. LPT is the digital interface to custom external hardware, I've wired it to a DAC chip and comparator as a successive approximation voltmeter. A mouse can be added through a Call Absolute to a pointer to INT33 code stored as a String.

Ideally the supporting pole would be aimed at the North Star, a polar instead of the present equatorial mount. The panel already pivots vertically with a turnbuckle to adjust for mid-day sun elevation, which I actually measured with a sextant for practice. The structural complexity and strength to withstand wind and snow in a location with morning and afternoon tree shade makes that approach highly uneconomical for the small power return. In sunny weather I can often just leave the panels aimed South to fully recharge batteries.

The output doesn't vary that much between summer and winter except around Christmas when pine trees block the low sun, I think because winter's clearer sky and colder panels increase their output. The highest is in spring and fall, summer heat and haze definitely reduce power.

I've found that solar experimenting may be valuable for my education but it is barely worth the cost and effort. Before it I was consuming 4 KWH from the grid per day, now I average 2 except when running A/C. That's $0.45 to $0.60 saved per day depending on New England's varying weather and power rates. The only significant change is the T60 freezer added due to Covid supply disruptions, and it ran on solar from the start.

Battery backup saves food if power fails during the night or you intentionally disconnect from the grid before thunderstorms, and allows a generator to be sized for average instead of peak motor starting current draw. The solar current maintains lead acid batteries. A simple minimal system could be a Jackery type power bank and matched portable solar array to top it off outdoors for occasional emergency or off-grid recreational use. A clothesline for drying laundry and shrinking your refrigerator size and power use to what you really need may save more.

Lead batteries can cost as much in depreciation as they save from free electricity. A 100Ah battery that costs $100 and lasts for 500 equivalent full discharges, which is very optimistic, costs about $0.20 per KWH. Flea market AGMs that had been replaced on schedule may be much cheaper but their remaining lifespan and thus system reliability varies considerably, not good for an emergency backup. Their fading capacity may not be evident in daily cycling. I do get to feel good about repurposing them, for whatever that's worth. I hope LiFePO4 performs as advertised, at least they reduced my income tax bill by 30% of their price. I met the requirement to charge them only with solar in exchange for taking the credit by charging the second hand batteries with grid or generator power.

The very last actual computer program I ever wrote was a Quick Basic (4.5) executable running under PCDOS 7.0. Due to a partial Y2K bug and dead cmos battery I called it in my autoexec to bring up a menu to set time and date upon boot up. I used that computer until the end of 2016 for remote programming some older lines of alarm panels. In 2017 I gave it to the company that purchased my alarm accounts.

The very last actual computer program I ever wrote was a Quick Basic (4.5) executable running under PCDOS 7.0. Due to a partial Y2K bug and dead cmos battery I called it in my autoexec to bring up a menu to set time and date upon boot up. I used that computer until the end of 2016 for remote programming some older lines of alarm panels. In 2017 I gave it to the company that purchased my alarm accounts. Bob La Londe CNC Molds N Stuff

---------------------------------------- You can access Visual Basic through MS apps, Write and Excel, though they don't allow the direct I/O port addressing I needed to control external hardware, which was lab development and customer application boards for new ICs.

The very last actual computer program I ever wrote was a Quick Basic (4.5) executable running under PCDOS 7.0. Due to a partial Y2K bug and dead cmos battery I called it in my autoexec to bring up a menu to set time and date upon boot up. I used that computer until the end of 2016 for remote programming some older lines of alarm panels. In 2017 I gave it to the company that purchased my alarm accounts. Bob La Londe CNC Molds N Stuff

-------------------------------- I used QB for a program that communicates with my APC1400 UPS, on a non-standard serial port cable. Since it talks by COM instead of LPT the .exe can run under Windows, monitoring load power and remaining battery capacity. One port pin has battery 24VDC on it which may be intended as an input for production line testing without installing batteries.

Another QB program reads the data port of PC-connected multimeters. Some transmit ASCII, others the LCD bar patterns of the numbers which is simple to decode with a Select Case structure. There are programs available to read them and record the data but none I've found can send an output command triggered by a reading, a control loop.

The trickiest one I which wrote when I was on dial-up parsed the modem log for the amount of data transmitted on a connection it hadn't seen before and kept track of the monthly total. It was an exercise in the analysis and handling of complex, variable conditions more than programming.