charging LiFePo4 with solar cells

[www.recumbents.com]

I'm thinking about charging a 24Vx20ah LiFePo4 pack through it's BMS with 25V of solar cells (3A max). I know I should use a diode to make sure that the solar cells only charge when they are at a higher voltage than the battery. I'm assuming that the BMS will prevent overcharge, right?

Thanks,

-Warren.

 

[amberwolf]

Depends on how the BMS works. If it has a relay or FET that cuts off input voltage to the battery once it is full, then yes. If not, then it probably has no way to protect it and you'd need to add something.

 

[amberwolf]

Also, remember that 24V of Lithium is generally not actually 24V even. LFP should be about 3.3V/cell nominal, but as high as 3.7V "hot off the charger", depending on the chemistry/version.

Assuming 7 cells in the pack, the actual nominal voltage is going to be 23.1, and would more likely be 8 cells for 26.4V nominal.

If 8 cells, it's going to need a charging voltage of up to 29.6V, and 25V of solar cells won't be enough.

Also, if you're just directly using bare solar panels attached to it, you will only get a high enough voltage to actually charge it at all when you have direct full-strength sunlight falling on it. If only partially lit for whatever reason, you may not have enough voltage to do what you want.

So a better solar charger will take a higher voltage of panels, and use a regulator system on that, even if it is as simple as a wasteful linear regulator, but better a slightly more complex PWM / SMPS output setup. The PWM setup would just switch the input voltage on and off at a fast rate, thru an inductor and then across filter capacitors, to give a final output voltage that is equal to your charge voltage. Just have a feedback in there that monitors the output voltage and keeps changing the PWM duty cycle to maintain the constant charge voltage.

 

[mettleramiel]

I have a solar array at home and I think that you might find that you have the opposite problem that Amberwolf is suggesting. Most panels are made to compensate for not getting direct sunlight, so they put out much more than the 24v they are rated for. This changes dramaticaly by manufacturer, usually a 12v panel have a working voltage of anywhere from 16v-22v, so a 24v panel can easily have over 40v come out of it! Don't know how your BMS would handle that.

While it's rather wasteful, myself I would just use the solar with a charge controller to charge a battery and then use an inverter with your regular charger for your LiFePO4 pack. If you had a rather large battery bank, this would also allow you to have extra power saved up for when there's no sun.

 

[amberwolf]

Ah; I wasn't aware of that particular problem. Most of the cells I've played with, including a couple of 20/30-year-old 1sq ft cells, put out only about the rated voltage when at max insolation. I am guessing that the difference might be that panels are rated differently than the cells they're composed of, due to the different market for them and the different needs for data about them.

It is actually better if the panels he gets work the way yours do, because the voltage can be chopped up (bucked) by SMPS/PWM effects easier than boosting it, while still getting higher currents out of them into the batteries.

 

[dak664]

PV panels act mostly as constant current sources, and the specs usually give a nominal voltage at 25C ambient. The voltage changes something like 0.35 per cent per degree C, so you have to choose a panel that will have enough voltage on the hottest day and then allow for a 20-40% percent overvoltage on the coldest day. A linear regulator would have to dissipate the excess internally, so for a panel of any size this pretty much requires PWM or a buck converter. The buck converter has the advantage of converting excess voltage into additional charging current.

 

[jkeny]

There is a specify chip for charging LiFePO4 from solar power from LT - LT3652 - pdf attached

 

[www.recumbents.com]

That looks like a nice chip, but it appears to be designed for 12V batteries. It seems perfect otherwise. Do they have one for 24V?

Looks like no easy answers here, but I have not yet built the solar array so I have some time to contemplate the options.

Thanks,

-Warren.

 

[tailwind]

You might want to check out the products from http://www.genasun.com.

For example the GVB24-6 will use a '12v' panel as input and output 24v.
A bit pricey though.

It may be cheaper to have two panels and two 12v charge controllers.
Each panel and controller would charge a 12v half of the 24v pack.

I purchased a GV-4-Li-14.2V from them a couple of months ago. The spec sheet has a panel Voc from 0 to 27V and maximum output current of 4A.

It has MPPT and will charge the pack up to 14.25v and then the current will drop to almost zero. This works out to 3.56v per cell on average for a 4s pack, but it does not do cell level balancing.

I use it in combination with a 5w panel to charge up a small 2p4s pack of 'under spec' A123 M1 cells.

Greg

 

[bigmoose]

Those chips are good to get peak power out of the arrays, but there is a simpler way.

A shunt regulator on the bus. As said above, an array appears as a constant current source. Design your shunt regulator to bypass, shunt, or waste the excess power of the array once the battery reaches it's cutoff charge voltage. If you BMS has cell shunt currents that are greater than the array can produce current wise in full sunlight with say a 50% safety factor, you are likely OK with a suitably sized array and a BMS.

 

[jjackstone]

Warren,
There are a number of inexpensive 24V charge controllers on the market.
Here are a couple links.

http://www.virtualvillage.com/6a-12v-24v-solar-panel-charge-regulator-controller-007808-026.html?utm_source=googlebase&utm_medium=shcomp

http://www.ledtronics.com/products/ProductsDetails.aspx?WP=C774K512

Can't vouch for their quality but for $15 to $30 they might be worth a try.

JJ

 

[www.recumbents.com]

Very cool that's exactly what I was looking for.

Thanks,

-Warren.

 

[myzter]

I was looking into this aswell, if your PV array is 12V

look for a Constant Voltage Regulator
24 V battery @ 6 amps max from a 12 V, 10-160 Watt nominal Solar Panel
Input Voltage (V) 0 - 24 V (12 V nominal Solar Panel)
Output Voltage (V) 28v / +/- 2%

there are also 6v step up to 28v modules aswell

but if your working on the 0.5v cell level with solar, & 13.1v ultra-capacitor booster pack .. like myself
wiring 6 solar cells in series per 2.7v capacitor (schottky dioded)

 

[Opi]

I'm new to LifePO4 batteries and the systems surrounding them, so if I could extend this thread into 12v solar charging as well, I'd like to get some insights into using a standard solar charge controller as the method of charging a LifePO4 battery pack or see what a BMS by itself would do.

For information on what a solar charge controller can do for a 12v lead acid battery, here's what you get:
A "typical" solar charge controller can usually be adjusted to work with wet or flooded lead acid batteries or SLA/sealed cells. The difference is the voltage during the initial absorption stage and whether or not an equalization charge is also performed after so many cycles or days.

The charge voltage for a lead acid flooded cell is 14.4v and for a sealed cell is 14.1v so if I understand things correctly, a "typical" solar charge controller would never fully charge a 12v, 4 cell LifePO4 pack which needs to get up to 14.6v for a complete charge. There are some programmable charge controllers such as the Morningstar Sunsaver MPPT which can be adjusted to handle different voltage charge points but for the sub $100 controllers you are stuck with fixed charge voltages built into the controller. Note however, if the solar charge controller has an equalization mode which can't be turned off, you could overcharge a LifePO4 battery from the higher 14.9 voltage used during that stage. Note: A 12v solar charge controller is only .2v from the correct max voltage of a LifePO4 cell but a 24v solar charge controller would charge at 28.8v max. or .4v less than than what a 24v LifePO4 battery would need to completely charge. By not having a fully charged battery, how much capacity is lost by that .2v on a 12v system and .4v on a 24v?

One question deals with the PWM type charging method that many solar charge controllers designed for lead acid batteries use. I've heard or read that LifePO4 batteries prefer a smooth constant current vs a pulsed current. Is this correct, and if so, what effect would a pulsed current (PWM) have on a LifePO4 cell? A pulsed current helps a lead acid battery but does it help or hurt a LifePO4?

As mentioned in earlier posts, 12v nominal solar panels have an open circuit voltage between 18v-22v but once you put a load on them the voltage drops down between 15v-18v depending on the panel and amount of sun shining. A post earlier mentioned that some BMS's use a shunt to turn off (burn off) the power to the battery pack protecting it from over charging. How common is this type of BMS or is the overcharging aspect generally handled by the battery charger in concert with the BMS? I deal with smaller solar panel arrays from just 30 watt to 200 watts with current produced < 1 amp to about 15 amps.

Would some battery management systems allow me to connect a higher voltage power source such as a 15v-18v solar panel directly to the BMS to charge the battery but then shut off the solar panels when the battery voltage hit 14.6v? How minimalistic can we get with charging these LifePO4 batteries?

With a lead acid battery I can use the voltage to determine how "full" that battery is. However, the voltage curve of a LifePO4 battery is pretty flat compared to a lead acid, so a voltage measurement alone doesn't appear to work as it does with a lead acid cell. What is the best method of determining the remaining capacity in a battery as it is being discharged?

 

[commanda]

I have this setup and working as a 12 volt system. 8 x Thundersky 90 AHr cells in 2P4s for 12 volt 180 AHr. Every cell has it's own shunt regulator ala Goodrum/Fechter. There is a constant current source and a solid state relay (in parallel) between the negative lead of the battery and the solar panel, which is 120 watt. Any shunt low switches off the solid state relay, so charging continues on the constant current source (about 400mA). Been running for about a year now. Similar setup to what I did for my EVT scooter. Search Commanda's BMS for more details.

 

[Mike B]

Commanda, I am looking at ways to direct connect PV to a smallish (two kwh or so usable (3 kwh then if LiFePO)) 12 V system to power some loads. I have been wondering if the G/F shunt regulator would work on a 4 serial LiFePO arrangement, and so seeing you have one running is fortuitous.

Could you put up some more info on how the cells are working in a solar charge regime? Have you metered the watts in/watts out for example? Do you keep them mostly charged, or discharged, etc? Do you run loads at the same time as charging? Are your loads 12V, or are you inverting to 230V? Are the cells showing any signs of deformation? Thanks for any info.

I think it inevitable that we will see PV charge controllers specifically set up for charging LiFePO stacks, with all the cell balancing/monitoring built in, but I am impatient to get something going.

GM has a paper out on a direct connect for vehicle charging. The full paper is behind a paywall but here is the title and part of the abstract:

"Solar photovoltaic charging of lithium-ion batteries"--General Motors Research and Development Center, Mail Code 480-106-269, 30500 Mound Road, Warren, MI 48090-9055, USA
The solar energy to battery charge conversion efficiency reached 14.5%, including a PV system efficiency of nearly 15%, and a battery charging efficiency of approximately 100%. This high system efficiency was achieved by directly charging the battery from the PV system with no intervening electronics

 

[myzter]

Mike B,

It is inevitable that we will see PV charge controllers specifically set up for charging LiFePO stacks,

an alternative might be to use a cell balancer..

 

[Mike B]

"an alternative might be to use a cell balancer"

OK. Pardon my cluelessnes, but will the balancer work when the battery is only 70% full, as might obtain at the end of each solar charging day, many days in a row?

Also, the PV charge controllers are built to last many years in service, so there might be an advantage in having a cell monitoring system built by a company with that kind of experience (and reputation to protect). If the LiFePOs can actually extend the cycle life out into a calendar life of 10 years or more, the BOS should match.

 

[commanda]

Commanda, I am looking at ways to direct connect PV to a smallish (two kwh or so usable (3 kwh then if LiFePO)) 12 V system to power some loads. I have been wondering if the G/F shunt regulator would work on a 4 serial LiFePO arrangement, and so seeing you have one running is fortuitous.

Could you put up some more info on how the cells are working in a solar charge regime? Have you metered the watts in/watts out for example? Do you keep them mostly charged, or discharged, etc? Do you run loads at the same time as charging? Are your loads 12V, or are you inverting to 230V? Are the cells showing any signs of deformation? Thanks for any info.

I think it inevitable that we will see PV charge controllers specifically set up for charging LiFePO stacks, with all the cell balancing/monitoring built in, but I am impatient to get something going.

GM has a paper out on a direct connect for vehicle charging. The full paper is behind a paywall but here is the title and part of the abstract:

"Solar photovoltaic charging of lithium-ion batteries"--General Motors Research and Development Center, Mail Code 480-106-269, 30500 Mound Road, Warren, MI 48090-9055, USA
The solar energy to battery charge conversion efficiency reached 14.5%, including a PV system efficiency of nearly 15%, and a battery charging efficiency of approximately 100%. This high system efficiency was achieved by directly charging the battery from the PV system with no intervening electronics

Mike,
I'm not using my system in anger yet. I've only got the one panel (100-120 watt) connected (I've got 350-400 watts total), and it's not in an ideal location. Summer is pretty good, but right now in winter it only gets about 2 hours sun, at best. I've currently got a little trickle charger (about 1 amp) keeping it topped off. It powers my lcd tv in the bedroom, which consumes 4 amps, and runs for 4-8 hours daily. I've also got an accumulating amp-Hr meter, based on a pic chip and an lcd (2 x 20) display. I've got 2 current shunts, charge & discharge, and count AHr' s in vs AHr's out, and calculate the current % capacity remaining. It resets to 100% when the total cell voltage exceeds a certain value.

There is some photos and some discussion over here:
http://fieldlines.com/board/index.php/topic,130173.0.html

Let me know if you want detailed schematics, they're on the computer at work, but not hard to pdf them.

 

[Mike B]

Amanda. tks for the quick response.

Read the thread over at Field Lines. I am much the wiser. As for schematics, you have already clearly explained what you have built, and left enough schematics elsewhere, thanks for the thoroughness.

My main interest is in how well the LiFePOs are doing in the PV charged RE setting. They do great under the EV regime -- discharge followed by full recharge -- but there is surprising little info out there I can find on how well this chemistry does in a random walk situation between fully charged and 60% DOD. Probably an NREL paper out soon.

Do you have any figures for round trip charge discharge efficiency?

I am curious as to why you have a trickle charger on them also. Tks again, hopefully when there is more sun you can start thrashing these things a little.

A little background on what I am doing wouldn't hurt. Have a 5kw grid tied system here in Pennsylvania and there is an annual surplus of about 2,000 kwh. The surplus and more will get used up when the cars are running on electricity. So I am looking at how to take some loads off grid with a smallish battery based RE system. I am familiar with AGM based systems, having installed a large one a few years back (5 kw Sanyos, two Sunny Islands 5048s, and 25 kwh of AGM). It was expensive, and, I think, decidely old tech. The SI 5048s are nice though--can't wait to seem them set up for LiFePo.

I think a small (2 or 3 kwh) but powerful LiFePo bank, a 2 kw inverter, and about 1 kw of new panels would be neat. Put in some 12V circuits for a few things (lights, fans, solar thermal pumps, this computer) and have the inverter for the bigger appliances. Where it gets interesting is how to couple the PV to the batteries, and whether LiFePos are actually cheaper in the long run then the AGMs. I have this hope the LiFePOs will not care if their SOC wanders around between 100% and 40%. This would be ideal if your managing a few loads and intermittent RE.

Mike

 

[commanda]

Amanda. tks for the quick response.

Read the thread over at Field Lines. I am much the wiser. As for schematics, you have already clearly explained what you have built, and left enough schematics elsewhere, thanks for the thoroughness.

My main interest is in how well the LiFePOs are doing in the PV charged RE setting. They do great under the EV regime -- discharge followed by full recharge -- but there is surprising little info out there I can find on how well this chemistry does in a random walk situation between fully charged and 60% DOD. Probably an NREL paper out soon.

Do you have any figures for round trip charge discharge efficiency?

I am curious as to why you have a trickle charger on them also. Tks again, hopefully when there is more sun you can start thrashing these things a little.

A little background on what I am doing wouldn't hurt. Have a 5kw grid tied system here in Pennsylvania and there is an annual surplus of about 2,000 kwh. The surplus and more will get used up when the cars are running on electricity. So I am looking at how to take some loads off grid with a smallish battery based RE system. I am familiar with AGM based systems, having installed a large one a few years back (5 kw Sanyos, two Sunny Islands 5048s, and 25 kwh of AGM). It was expensive, and, I think, decidely old tech. The SI 5048s are nice though--can't wait to seem them set up for LiFePo.

I think a small (2 or 3 kwh) but powerful LiFePo bank, a 2 kw inverter, and about 1 kw of new panels would be neat. Put in some 12V circuits for a few things (lights, fans, solar thermal pumps, this computer) and have the inverter for the bigger appliances. Where it gets interesting is how to couple the PV to the batteries, and whether LiFePos are actually cheaper in the long run then the AGMs. I have this hope the LiFePOs will not care if their SOC wanders around between 100% and 40%. This would be ideal if your managing a few loads and intermittent RE.

Mike

Mike,
The trickle charger is only connected in winter, when my temporary mount solar panel doesn't get enough sun because of shading from the neighbours trees. In summer it is switched off. Last summer it went a few times, up to 7 days, where it didn't reach full charge, with no problems. Other than drift in the AHr counter due to a zero offset in one shunt amplifier giving about 0.3 amps when it is in fact zero.

I have seen a figure of 95% or better for charge efficiency. I actually put this function in my meter, but my implementation is obviously seriously flawed, because it currently reports 55% efficiency.

Other than charge control when full, the solar panels connect directly to the batteries. PV panels approximate a constant current source. You can drag the output voltage down to zero, and they will still produce current. In fact, this is how we check the short-circuit current with an ammeter.
Once you get any shunt low, you need to limit the charge current to whatever your shunts can handle.

I KW of solar will make 70 - 80 amps at 12 volts. Not trivial to handle. With increasing system size, the arguments for going to 24 volts become more compelling.

I think a good rule of thumb for sizing solar systems, is 100% charge on a good sunny day. Assuming you have weather that typically gives a significant number of sunny days. Say you can expect 6 hours of good sun, then a 1 Kw solar panel would charge 6 KwHrs of batteries in one day. Conversely, a 3 KwHr battery would equate to 500 watts of panels. Just something for you to think about in sizing the system. There are online calculators out there to help with these sizing decisions.

Because of the flat discharge curve of LiFePo4 vs lead-acid, I think an accumulating AHr meter is a must-have. You can't just measure the voltage and get an idea of SOC.

Amanda

 

[Mike B]

Amanda,

In PA we see about 3 kwh each day from each 1 kw of panel during Nov-Mar. Not that simple of course, with day to day variations. One idea on what to do with a nearly full battery and SUN would be to move some AC loads off grid with relays (listed for this purpose, like IOTA's). For example, take the fridge and freezer off grid temporarily along with a thermostat override on the appliances to force them on. Or you could start an air source DHW heat pump like the GE GeoSpring. The relay would have a minimum ON timer to prevent short cycling. The NEC here requires redundant overcharge protection on RE battery systems, so a second dump load would be required, preferably not through an inverter.

PV has gotten so cheap, $2.60/watt, having some extra watts in the summer isn't a big concern, especially if the overall system is small with respect to the overall house load and one can switch other loads in.

I was thinking a 12 volt system having just 4s of batteries is (4 times ?) less likely to have mismatch problems than a similar number of cells in 8s. By staying under a full charge most of the time the need for giant 80 amp shunts would be avoided. Once a week maybe, do a balancing charge overnight at some reasonable amps.

I think we will see in the future is a charge controller that was built to work with a 4s or 8s LiFePO . How would that work?

Next thing I should do is spend a few bucks and build this up at 1/10 scale with some A123s, 100 watts of PV, etc and meter the whole thing. I have never found actual numbers for round trip efficiency for A123s, watts in versus watts out, so getting those numbers will be worth it.

Mike

 

[texaspyro]

OK. Pardon my cluelessnes, but will the balancer work when the battery is only 70% full, as might obtain at the end of each solar charging day, many days in a row?

A standard resistive shunt balancer would have problems in that situation. You would need some sort of active balancer. I have been playing with a switched capacitor balancer that runs whenever the pack is in use or being charged. It wastes almost no energy as heat and will balance all cells to the same voltage level. It works by switching capacitors across adjacent cells at a fairly high frequency. Charge is transfered from the higher cells to the lower ones.

It only balances a few milliamps of current per 10mV of cell imbalance, but like the slow but steady tortoise, it gets there. Since it runs at all time the pack is in use, it tends to keep the cells well balanced. It is totally independent of the cell technology, passive in its operation, and relies on no precision components or calibration.

 

[Mike B]

Texas,

Well, I was wondering how to find out if there was an active balancer that looks at V to V.

Can it scale up? And with reference to Amanda's comment on flat V vs SOC in the middle, does this affect how they are working?

Mike

(edit) Bigger capacitors for scale up?

 

[myzter]

... round trip efficiency for A123s

I would like to see the PV efficiency for A123 aswell, I will probably do a proper test soon and log the details and post them here...

I have so far got 14 cells mounted on my bike, all mounted in front of the bike.. in an arrow shape that should also drop the wind resistance a fair bit...
- the total weight of the 5 panels is approx. only 5 lbs.. it is the weight of the mounting hardware Im dealing with right now.....
next I will mount an additional 14 cells over the back, and maybe a couple along the side of the bike...

PV has gotten so cheap, $2.60/watt,

I picked approx. 30 - 6"x6" - 4 watt cells @ $1.79/watt (0.5-.6v @ 8A)

Here is a shot of the panel frames without PV's installed or wired...
I positioned the frames as close I could so you can see the general arrangement