Is Li-ion now viable for off-grid?

[max2k]

That chart is a very different charge cycle than what anyone suggesting top balance here would use, and some of the risks you are hitting on come from those differences. In addition to the relatively high rate of charge (@pnjunction recommends no more than 0.33C here), there is zero absorb time in that chart. In a top balanced system, with absorb set at a pack voltage equal to or below the cell voltage at which bypassing would begin, the current that those boards need to pass will be much more limited than your chart suggests.

I'm having some trouble following your string of objections to @karrak's experience. If your goal is to understand a grid tie peak shaving application, that is really different than off-grid, and it might be worth starting a new thread so that the different design and operating guidelines for the two applications don't get mixed in a confusing way.

I guess, you're right it's a different application altogether.

[sensij]

Here's another graph where cells look as 'balanced' as on your graph but since the charge current continued to flow you can see it took stronger cells 8 min 20 s to 'catch up' to the same voltage as weaker ones (part between Battery Full and 2h30min point). At charging current 0.4C it corresponds to imbalance of 0.4 x 500 / 3600 = 5.5%. This graph also illustrates the problem bypass board is supposed to solve- during that 8 min 20 s interval bypass boards on the weaker cells are supposed to bypass full charging current of 40A in this case. The graph is for a battery made from Winston 100Ah cells. This also shows the precision demands on voltmeter used- the tick marks on the right scale are in 2.5 mV increments.

That chart is a very different charge cycle than what anyone suggesting top balance here would use, and some of the risks you are hitting on come from those differences. In addition to the relatively high rate of charge (@pnjunction recommends no more than 0.33C here), there is zero absorb time in that chart. In a top balanced system, with absorb set at a pack voltage equal to or below the cell voltage at which bypassing would begin, the current that those boards need to pass will be much more limited than your chart suggests.

I'm having some trouble following your string of objections to @karrak's experience. If your goal is to understand a grid tie peak shaving application, that is really different than off-grid, and it might be worth starting a new thread so that the different design and operating guidelines for the two applications don't get mixed in a confusing way.

[karrak]

The graph shows up now, thank you. I don't agree though that cells are within 1% SOC. LiFePO4 have very 'flat' curves and voltage in the flat part of the curve cannot be used as reliable indicator of SOC. It looks like the Absorb voltage on that system is set too low when none of the cells yet fully charged and reached the 'knee' in their curves. As a result during Absorb stage bypass boards won't even work leaving cells as unbalanced as they are. As I said this doesn't manifest until that day the full capacity is required. Here's another graph where cells look as 'balanced' as on your graph but since the charge current continued to flow you can see it took stronger cells 8 min 20 s to 'catch up' to the same voltage as weaker ones (part between Battery Full and 2h30min point). At charging current 0.4C it corresponds to imbalance of 0.4 x 500 / 3600 = 5.5%. This graph also illustrates the problem bypass board is supposed to solve- during that 8 min 20 s interval bypass boards on the weaker cells are supposed to bypass full charging current of 40A in this case. The graph is for a battery made from Winston 100Ah cells. This also shows the precision demands on voltmeter used- the tick marks on the right scale are in 2.5 mV increments.

I don't think you can use the voltage of the individual cells to calculate the difference in SOC between the cells unless the charge/discharge rate is low or preferably zero because of the difference in impedance between the cells, especially at an SOC close to 100%.

Balance boards do not have to bypass the full current, they only have to generate a differential in the current so that over time the cell voltages will converge. They do have to keep up with any differential between the rate of SOC charge between the individual cells. If they don't the cell voltages will diverge.


The knees, especially at the top end of charge are very dependent on the charge rates. After a lengthy absorb time the charge current is close to zero where the slope of the V-SOC curve is large and small changes in SOC mean a large change in cell voltage.

ChargeDischargeCurves.jpg
Simon

Off grid 24V system, 6x190W Solar Panels, 32x90ah Winston LiFeYPO4 batteries installed April 2013
BMS - Homemade Battery logger github.com/simat/BatteryMonitor
Latronics 4kW Inverter, homemade MPPT controller

[karrak]

your battery cells are suspiciously well matched. From reading around (mostly Texas Instruments power management site) it sounds like simple temperature difference between cells will put them out of balance in few % range even if they were ideally matched at the beginning. Your cells must be in well ventilated area so they are all more or less at the same temperature.

What do you mean "suspiciously well matched"? There would be some ambient temperature differential as some of the cells are closer to a west facing tin wall but I doubt it is enough to get excited about. There would be very little temperature differential due to self heating because of low charge and discharge rates. I don't think my battery is particularly remarkable. My experience matches many others. The amount I have had to adjust the balance only shows the change in balance between the cells over the year. It says little about the total difference in SOC between the cells.

I think I'm having problems accepting your optimism and extrapolations due to different use pattern: if we're talking about true off grid system with 5 days autonomy then such system doesn't go below 80% SOC on daily basis. This use pattern is quite forgiving for imbalance problem until the day comes when that 5 day design capacity must be fully used. At that point such system better have protection from over- discharge on individual cell level. Since this rarely happens ppl can go for years completely oblivious to the problem.

Here are the stats for my battery usage which I posted in an earlier post again
Average Energy Cycled Through Battery Per day: ~1.9kWh (last 19 months)
Average Energy Use Per Day: ~2.9kWh (last 19 months)
Max Charge Current: 47.8A (~0.13C)
Max Discharge Current: 247.7A (~0.69C)
Lowest SOC%: ~8% (April 2017, from cell voltages calculated about ~5% left in battery)
Average SOC%: ~77%
Overall Battery Energy Efficiency: ~96%

My system is regularly below 80%SOC, just spent over 20 days between ~75%-40%SOC because of cloud cover.

I agree about having individual cell monitoring to prevent over-discharge. From the individual cell voltages (3.018, 2.937, 3.019, 3.032, 3.046, 3.068, 3.096, 3.056) at the 8%SOC I calculated the difference is SOC of my cells to be ~4%

Simon

Off grid 24V system, 6x190W Solar Panels, 32x90ah Winston LiFeYPO4 batteries installed April 2013
BMS - Homemade Battery logger github.com/simat/BatteryMonitor
Latronics 4kW Inverter, homemade MPPT controller

---

[max2k]

Ti published interesting article overviewing various balancing techniques. It is for different chemistry Li-ion but I think approaches are the same and chips can be programmed to LiFePO4 voltages. In particular the passive one which uses difference in required charge (not voltage) between cells is interesting. In all cases the balancer manages not a single cell but a group up to 10-16 which relieves them from knowing exact charge curve as all they need to do is to balance cells relatively to each other. As a result most of them try to balance cells during entire charge phase. They can even spread balancing over multiple charge cycles accumulating balancing efforts over time. This is the case where smaller currents could be acceptable: http://www.ti.com/lit/an/slyt322/slyt322.pdf

[max2k]

Thank you for looking into this, I was able to generate the same error by clearing the cache on my computer. I downloaded the file to your server again and it seems to have fixed the problem. ...

The graph shows up now, thank you. I don't agree though that cells are within 1% SOC. LiFePO4 have very 'flat' curves and voltage in the flat part of the curve cannot be used as reliable indicator of SOC. It looks like the Absorb voltage on that system is set too low when none of the cells yet fully charged and reached the 'knee' in their curves. As a result during Absorb stage bypass boards won't even work leaving cells as unbalanced as they are. As I said this doesn't manifest until that day the full capacity is required. Here's another graph where cells look as 'balanced' as on your graph but since the charge current continued to flow you can see it took stronger cells 8 min 20 s to 'catch up' to the same voltage as weaker ones (part between Battery Full and 2h30min point). At charging current 0.4C it corresponds to imbalance of 0.4 x 500 / 3600 = 5.5%. This graph also illustrates the problem bypass board is supposed to solve- during that 8 min 20 s interval bypass boards on the weaker cells are supposed to bypass full charging current of 40A in this case. The graph is for a battery made from Winston 100Ah cells. This also shows the precision demands on voltmeter used- the tick marks on the right scale are in 2.5 mV increments.

Winston_100Ah_charge.jpg

[Mike90250]

Yes, the multi color graph in the original post is showing now. the quoted versions will still be blank, but we get the idea now. Thanks

[karrak]

Yes, I get a "broken icon", no graph. When I hover the mouse over it, the file name pops up. Was it a file that was uploaded, or pasted into the post, or hosted on another server ?

Thank you for looking into this, I was able to generate the same error by clearing the cache on my computer. I downloaded the file to your server again and it seems to have fixed the problem. Has it fixed it on your computer

Thanks
Simon

Off grid 24V system, 6x190W Solar Panels, 32x90ah Winston LiFeYPO4 batteries installed April 2013
BMS - Homemade Battery logger github.com/simat/BatteryMonitor
Latronics 4kW Inverter, homemade MPPT controller

[Mike90250]

Yes, I get a "broken icon", no graph. When I hover the mouse over it, the file name pops up. Was it a file that was uploaded, or pasted into the post, or hosted on another server ?

badGraph_K.png

[max2k]

The battery goes out of balance slowly.I had a look at the spreadsheet I have been keeping on how much balance charge I have had to subtract from the individual cells. The worst cell had ~0.85Ah subtracted over the year. If balanced on a daily basis this equates to ~0.0023Ah per day or ~.008Wh/day. If the balance occurs over six minutes this equates to 80mW (0.080W) of power.
...

your battery cells are suspiciously well matched. From reading around (mostly Texas Instruments power management site) it sounds like simple temperature difference between cells will put them out of balance in few % range even if they were ideally matched at the beginning. Your cells must be in well ventilated area so they are all more or less at the same temperature.

I think I'm having problems accepting your optimism and extrapolations due to different use pattern: if we're talking about true off grid system with 5 days autonomy then such system doesn't go below 80% SOC on daily basis. This use pattern is quite forgiving for imbalance problem until the day comes when that 5 day design capacity must be fully used. At that point such system better have protection from over- discharge on individual cell level. Since this rarely happens ppl can go for years completely oblivious to the problem.

What I'm after is more like EV use pattern where battery capacity is fully utilized daily and all these issues are much more pronounced. I think only automatic monitoring / balancing solution can work there.

[karrak]

total energy might be small but the power dissipated may not be so- depends how fast that energy gets released. If you have say 4 cell battery and 1 cell is weaker its bypass board would need to conduct practically all current over short time of several minutes while the rest 3 cells catching up. It doesn't make it easier for the board this needs to happen only once a year.

The battery goes out of balance slowly.I had a look at the spreadsheet I have been keeping on how much balance charge I have had to subtract from the individual cells. The worst cell had ~0.85Ah subtracted over the year. If balanced on a daily basis this equates to ~0.0023Ah per day or ~.008Wh/day. If the balance occurs over six minutes this equates to 80mW (0.080W) of power.

I can but I don't want to. I'm on academic quest for a feasible home size system which would let me deal with TOU utility tricks. This is pure theoretical exercise for me at the moment but who knows with this crew (CA gvmt). From this perspective I'd like to have my battery to require min attention from me and BMS to be intelligent enough to take care of small annoying things and bother me only when absolutely necessary.

Sounds like you need an automated BMS with individual cell monitoring which will generate an alarm if anything untoward happens and automatic balancing to keep the battery in balance. There are several of these on the market.

can you please elaborate on the 'passive safety' meaning in this context?

I mean that as the battery is in balance all the cell voltages should rise at the same rate so the overall battery charge voltage set by the charge controller should safely end the charge. By active safety I would mean that the BMS would notify the charge controller to stop charging if any individual cells go out of their safe operating range.

After looking at what is involved active balancing looks more and more appealing, especially after they developed specialized chips just for this purpose: http://cds.linear.com/docs/en/lt-jou...300-1-Drew.pdf

at least it looks closer to what I have in mind. May be that's what that 12V drop in battery has inside? This chip would allow to overcome problems I was worried about and it doesn't require switches in line with high current path, it just piggy backs the cells.

IMO the extra complexity of the active balancing is just not worth it in low stress applications like off-grid. Passive balancers are all that is needed. Even Tesla and A123 and I am sure many other manufacturers only use passive balancing in some if not all their batteries.

Active balancers need to communicate with the cells next to them, passive balancers work totally independently. Having a switch in the main line is only there as an extra safety measure. If the battery is balanced properly to start with and there are no faults and the user doesn't do anything stupid they are unnecessary. Maybe this is why Battle Born put them in their battery.

Simon

Off grid 24V system, 6x190W Solar Panels, 32x90ah Winston LiFeYPO4 batteries installed April 2013
BMS - Homemade Battery logger github.com/simat/BatteryMonitor
Latronics 4kW Inverter, homemade MPPT controller

[karrak]

the graph didn't come through, please repost.

I am not having any trouble viewing the graph, anyone else having any problems?

I based my statement on prior SK data he shared about those boards. Sounded like he had enough instances of them doing it. Besides, making something precise to <10mv and temperature stable requires certain efforts so I tend to trust him on this one. What is your statistical sample to state otherwise?

My sample is all the people who have off grid systems who post on the Alternate Energy forum in Australia, on this forum, on the Wind & Sun forum and the Cruisersforum of yachties. I tried doing a Google search on "vampire board lithium" and "cell balancer failure lithium" expecting an avalanche of horror stories and all I got is posts on this forum by Sunking. Maybe you can come up with something.

because I don't understand how it would work . It never worked well for conventional FLA now we put multiple disconnected mini- BMS and somehow it would start working? All those mini BMS have is the input voltage on battery posts and if one ended up with weaker cells (less capacity) how would it know other batteries exist in parallel so instead of bypassing current it is supposed to disconnect its cells? Bypassing in this situation would either fry the bypassing board or cells would need to pick up the slack as neighboring batteries might still be busy taking main charge. I also don't see what mechanism would ensure even charge current distribution between batteries connected in parallel. I guess it all depends on definition of the word 'work' and I use it in a sense that 2 100Ah batteries connected in parallel should look and feel like singe 200Ah battery. I doubt this is possible as something would mismatch and there's no central brain to help.

If you put two Battle Born batteries in parallel with the correct cabling so they share the current evenly they will work just fine. Each balancer on each cell in the different batteries will do their job independently without any knowledge of what is happening with the other cells. In the unlikely event any of the cells in either of the batteries goes outside their safe operating zone that battery's BMS will disconnect that battery from the other one and cut the battery down to a 100Ah battery. Under normal operating conditions the two 100Ah batteries will behave like one 200Ah battery.

Simon

Off grid 24V system, 6x190W Solar Panels, 32x90ah Winston LiFeYPO4 batteries installed April 2013
BMS - Homemade Battery logger github.com/simat/BatteryMonitor
Latronics 4kW Inverter, homemade MPPT controller

[max2k]

Unless there is a fault the cells go out of balance very slowly over a period of months or years. The amount of energy that needs to be dissipated to balance an LFP battery is negligible. My 10kWh battery has not needed any balancing for over a year. The last year that I did have to adjust the balance needed about 10Wh to be dissipated over the period of a year to keep it balanced.

total energy might be small but the power dissipated may not be so- depends how fast that energy gets released. If you have say 4 cell battery and 1 cell is weaker its bypass board would need to conduct practically all current over short time of several minutes while the rest 3 cells catching up. It doesn't make it easier for the board this needs to happen only once a year.

You do not need bypass boards to keep a battery top balanced. To do it manually ...

I can but I don't want to. I'm on academic quest for a feasible home size system which would let me deal with TOU utility tricks. This is pure theoretical exercise for me at the moment but who knows with this crew (CA gvmt). From this perspective I'd like to have my battery to require min attention from me and BMS to be intelligent enough to take care of small annoying things and bother me only when absolutely necessary.

As you say bottom balancing is more "load friendly" which is ideal for large loads that can drain your battery very quickly like EVs. With EVs charged with a standard CC/CV charger you have absolute control over the charging process but not so much control when you are going to run out of power while driving around. In this case the passive safety offered by bottom balancing as long as the balance doesn't drift is useful. With off-grid power systems charged by solar you don't have so much control of the charging process and have more control over the loads. All off-grid system owners I know of try to keep their batteries as full as possible because they don't know when they are going to have cloudy weather. This means that the battery is nearly fully charged quite regularly where the passive safety of top balancing can be useful.

I actually think that the main advantage of top balancing versus bottom balancing is not the passive safety that it offers at full charge but the fact that it is easy to do and you can see if the battery is still in balance every time that the battery is fully charged.

can you please elaborate on the 'passive safety' meaning in this context?

After looking at what is involved active balancing looks more and more appealing, especially after they developed specialized chips just for this purpose: http://cds.linear.com/docs/en/lt-jou...300-1-Drew.pdf

at least it looks closer to what I have in mind. May be that's what that 12V drop in battery has inside? This chip would allow to overcome problems I was worried about and it doesn't require switches in line with high current path, it just piggy backs the cells.

[karrak]

- in the perfectly top balanced battery (cells top balanced manually before being assembled to a battery) the passive bypass board will be subject to much lesser currents at the end of charge. OTOH regardless of its topology it has no choice but to dissipate that cell voltage x bypass current power in form of heat, there's no easy way to make them 'efficient'. The charging energy has to go somewhere so it either can go into cell or heat up the environment, there's no way around this. One could potentially come up with some way to use this energy but that would further complicate already complex system.

I'm not sure how well top balanced battery remains top balanced over time as all unbalances would end up either hitting weaker cells or those bypass boards during charge process.

Unless there is a fault the cells go out of balance very slowly over a period of months or years. The amount of energy that needs to be dissipated to balance an LFP battery is negligible. My 10kWh battery has not needed any balancing for over a year. The last year that I did have to adjust the balance needed about 10Wh to be dissipated over the period of a year to keep it balanced.

Another solution SK was suggesting here recently is bottom balance the cells and then monitor weakest cell during charge to switch to CV on time leaving rest undercharged. In general, a series of cells Ah capacity is limited by weakest cell anyway so you either have to catch this moment during discharge leaving stronger cells with some residual charge (top balance) or during charge leaving stronger cells undercharged (bottom balance). It seems to me the latter is safer, does not require bypass boards and more 'load friendly'.

You do not need bypass boards to keep a battery top balanced. To do it manually you monitor the cell voltages at regular intervals and when they have diverged by say 0.100V at the end of the charge cycle you place a temporary load across the cells with the highest voltages to draw off some charge from those cells. If you don't have the technical ability to do this, can't be bothered or don't trust that you will be conscientious enough to do this for over the life of the battery of over ten years then you use automatic cell balancing.

As you say bottom balancing is more "load friendly" which is ideal for large loads that can drain your battery very quickly like EVs. With EVs charged with a standard CC/CV charger you have absolute control over the charging process but not so much control when you are going to run out of power while driving around. In this case the passive safety offered by bottom balancing as long as the balance doesn't drift is useful. With off-grid power systems charged by solar you don't have so much control of the charging process and have more control over the loads. All off-grid system owners I know of try to keep their batteries as full as possible because they don't know when they are going to have cloudy weather. This means that the battery is nearly fully charged quite regularly where the passive safety of top balancing can be useful.

I actually think that the main advantage of top balancing versus bottom balancing is not the passive safety that it offers at full charge but the fact that it is easy to do and you can see if the battery is still in balance every time that the battery is fully charged.

With my top balanced 4yo LFP battery I am fairly sure that my inverter's 24V LVD would protect my battery. I wouldn't rely on it though...

Simon

Off grid 24V system, 6x190W Solar Panels, 32x90ah Winston LiFeYPO4 batteries installed April 2013
BMS - Homemade Battery logger github.com/simat/BatteryMonitor
Latronics 4kW Inverter, homemade MPPT controller

[max2k]

The following graph shows the balance of my friends LFP 4yo battery at the end of charge. This battery is only slightly out of balance (less than 1%). As you can see by the time the voltages have diverged by 0.100V the charge current is down to nearly zero. If this system had balancing boards that were set to balance at 3.50V it wouldn't be balancing until around the 60 minute mark when the charge current is nearly zero.

the graph didn't come through, please repost.

I haven't seen any evidence to support your assertion. I only know of one case where a balancing board failed and dragged down a cell. This fault was detected by individual cell monitoring before any damage was done.

I based my statement on prior SK data he shared about those boards. Sounded like he had enough instances of them doing it. Besides, making something precise to <10mv and temperature stable requires certain efforts so I tend to trust him on this one. What is your statistical sample to state otherwise?

The Battle Born manual says they can be connected in parallel, why don't you think the batteries can be connected in parallel?

because I don't understand how it would work . It never worked well for conventional FLA now we put multiple disconnected mini- BMS and somehow it would start working? All those mini BMS have is the input voltage on battery posts and if one ended up with weaker cells (less capacity) how would it know other batteries exist in parallel so instead of bypassing current it is supposed to disconnect its cells? Bypassing in this situation would either fry the bypassing board or cells would need to pick up the slack as neighboring batteries might still be busy taking main charge. I also don't see what mechanism would ensure even charge current distribution between batteries connected in parallel. I guess it all depends on definition of the word 'work' and I use it in a sense that 2 100Ah batteries connected in parallel should look and feel like singe 200Ah battery. I doubt this is possible as something would mismatch and there's no central brain to help.

The BMS within the Battle Born battery will disconnect the battery from the outside world if any cell goes outside its safe operating voltage range. It will also disconnect the battery if you try to draw too much current, overdischarge it, or overcharge it. The low resistance switch can easily be made with suitable switching FETs.

outside world will not appreciate that. Can you provide part # for that low resistance FET? They're typically in 1mOhm range and that would be for low voltage ones limiting # of batteries in series so for the current 100A it would need to dissipate 10 W in heat which would require external heatsink. Not so easily IMO. Of course one can put 10 of those FETs in parallel as they have positive Ri coefficient reducing this to 1W but I rather have nothing in the path of that 100A current.