2 runaway cells on 2 banks

[Ucatchmydrift]

You can only effectively balance at the knee. Ignore any cell voltage below 3.5V (or certainly 3.4V).

Bleed off the two runner cells to 3.4V then if you can, charge at a charge current slightly higher than the bleed current at 3.4V and charge until a 3rd cell hits 3.4 while continuing to bleed the two runners.

Now you can stop bleeding and charge the battery until the 3 high cells get up to 3.6 or at least 3.5.

Once you’ve got the highest 3 cells roughly in balance, you can continue the process to get those 3 balanced with the next highest cell, etc…

Once you’ve got the highest 3 or 4 cells in balance, probably time to discharge the battery until a first cell hits the lower knee at 3.0 or at least 3.1V.

If that cell is one of the cells you balanced at the top, that cell is limiting the working capacity of the entire battery and continuing to improve the quality of the top balance will only allow you to increase boost voltage closer to 28.8V but will not gain you any additional capacity.

If another cell hits the lower knee first, boost it’s voltage using a 3.6V power supply until it matches the voltage of the next lowest cell then stop boosting and discharge again until those cells hit the lower knee again).

Now boost both of those cells until they match the voltage of the 3rd-lowest cell.

Repeat the discharge and if none of those 3 bottom-balanced cells hitting the lower knee were one of your 3 top-balanced cells, go ahead and boost all 3 to match the 4th-lowest cell.

Continue to bottom balance to the 5th lowest or even 6th-lowest cell until you have bottom-balanced all lowest cells to the lowest of your 3 top-balanced cells.

You can now charge the full battery back to the top where the first cell to hit the upper knee that was part of the bottom-balanced set is the reference cell determines overall battery capacity and should no longer be touched (neither boosted nor drained).

Boost all other cells to match the voltage of that reference cell.

Getting all the cells balanced to 3.5V should be relatively easy. Getting balanced to 3.6 or even 3.65V just requires a lot of time and attention.

Wherever you end up, that determines your boost voltage and monitoring how quickly the battery strays from that well-balanced starting point will tell you how often you need to go through this painful exercise.

A 5A active balancer that you just attach on whatever schedule you think makes sense is much, much easier…

Thanks for the detailed explanation! I had to read that a few times lol.
That is definitely a lot of hastle, and dont really want to be doing that often, or ever tbh. So yes, if a 5a balancer is enough to deal with it, i may well invest in a couple of them. Ill see how much current is needed to stop the two cells from spiking if i can, by draining it off with a load, hopefully 5a or under will be enough, and assuming the cells are not damaged also.

 

[Ucatchmydrift]

Have you calculated what crypto-mining is paying you on average $/kW for excess solar power?

Its only on on sunny days, and would take forever to pay for its self these days as its a gpu miner (not asic bitcoin) and now etherium is proof of stake, and not proof of work now, profit is about £0.30- £0.40 a day, vs £2-£9 a day a year ago or so. Obviously if the price of crypto goes up in the future, (currently in a bear market, now heading out potentially..), then that £0.30 a day could be worth a fair bit more (he doesnt sell it) in the future. For that £0.30 a day, it uses 7.2kw in 24hrs, so a fair bit. My son built it a couple of years ago and has paid for itself, so anything now is just profit for him.

 

[DIYrich]

If you can, hold 27.1, and let it balance. Every .02v of decrease in cell voltage difference should let you increase the charge voltage by 0.1v.

 

[Ucatchmydrift]

If you can, hold 27.1, and let it balance. Every .02v of decrease in cell voltage difference should let you increase the charge voltage by 0.1v.

Thats good to know thank you ?. Typically the weather looks dull for the next week, so may have to wait a while untill it brightens up. (cant be using grid power to charge them! ?)

 

[fafrd]

Its only on on sunny days, and would take forever to pay for its self these days as its a gpu miner (not asic bitcoin) and now etherium is proof of stake, and not proof of work now, profit is about £0.30- £0.40 a day, vs £2-£9 a day a year ago or so. Obviously if the price of crypto goes up in the future, (currently in a bear market, now heading out potentially..), then that £0.30 a day could be worth a fair bit more (he doesnt sell it) in the future. For that £0.30 a day, it uses 7.2kw in 24hrs, so a fair bit. My son built it a couple of years ago and has paid for itself, so anything now is just profit for him.
View attachment 154580

0.30-0.40 for 7.2kWh is 0.042-0.056 per day. Better than nothing but less than the 0.08 wholesale rate California pays for annual overgeneration…

 

[fafrd]

Thanks for the detailed explanation! I had to read that a few times lol.
That is definitely a lot of hastle, and dont really want to be doing that often, or ever tbh. So yes, if a 5a balancer is enough to deal with it, i may well invest in a couple of them. Ill see how much current is needed to stop the two cells from spiking if i can, by draining it off with a load, hopefully 5a or under will be enough, and assuming the cells are not damaged also.

It’s the charge current. If you charge at a maximum of 5A and have a 5A active balancer, you can prevent any single cell from spiking. But even with a 6A charge current, it’s just a matter of time.

As DIYrich suggested, if you can enter CV mode at 27.1V without the runners spiking to 3.65V, that will reduce charge current to the level below what can be compensated for by a 5A active balancer.

2 x 3.65V = 7.3V, so 27.1V boost voltage entering into CV means 19.8V for the other 6 cells, or 3.3V per cell.

The active balancer delivers the full 5A balance current for mismatch voltage of 100mV or higher, so you’ll get 2.5A into each runner cell when runners are anything over 3.4V when the other 6 cells are at 3.3V.

As long as charge current drops below 2.5A on CV mode (68W) you should be able to eventually balance two runners (or 5A / 136W for only a single runner).

If you end up deciding to get a 5A active balancer, you’ll probably want to speed things up by also bleeding the 2 runners using your 5 Ohm power resistors.

3.4V / 5Ohms = 0.68A, so only +27% over what you’d get with the 5A active balancer alone, but it will help.

Once you have a sense of how much charge you are talking about balancing, the power resistors also have the advantage that they can be used in the flat spot around 3.3V.

When your 2 runners are at 3.305V and your other cells are at 3.295V, the Active balancer will only be balancing those two cells with 250mA each, while your 5A power resistors will be draining 661mA (+164%).

24 hours-worth of bleed current is 15.84Ah or 5.66% of total cell capacity, so you can use the power resistors to bleed the two runners down by ~2-5% overnight even after the battery has been discharged into the flats where the active balancer will be ineffective.

Once you’ve shifted the two runners into laggars, you know what you are dealing with and can focus on bringing them back up to where the other cells are more slowly (boosting 1% at a time using a power supply and power resistors or just letting the active balancerbeing them up into balance over enough full charge cycles.

Once the runners have been pulled under the other cells, you’ll be able to increase boost voltage to 27.2V.

27.2V gets your runners up above 3.65V today, meaning the other cells are at 3.317V.

Once the runners have been pulled under those other cells, the battery will be under 26.53V when the non-runner-cells are at 3.317V.

Now you can successfully charge the battery to 27.2V because this will mean all cells are around 3.4V.

 

[Ucatchmydrift]

It’s the charge current. If you charge at a maximum of 5A and have a 5A active balancer, you can prevent any single cell from spiking. But even with a 6A charge current, it’s just a matter of time.

As DIYrich suggested, if you can enter CV mode at 27.1V without the runners spiking to 3.65V, that will reduce charge current to the level below what can be compensated for by a 5A active balancer.

2 x 3.65V = 7.3V, so 27.1V boost voltage entering into CV means 19.8V for the other 6 cells, or 3.3V per cell.

The active balancer delivers the full 5A balance current for mismatch voltage of 100mV or higher, so you’ll get 2.5A into each runner cell when runners are anything over 3.4V when the other 6 cells are at 3.3V.

As long as charge current drops below 2.5A on CV mode (68W) you should be able to eventually balance two runners (or 5A / 136W for only a single runner).

If you end up deciding to get a 5A active balancer, you’ll probably want to speed things up by also bleeding the 2 runners using your 5 Ohm power resistors.

3.4V / 5Ohms = 0.68A, so only +27% over what you’d get with the 5A active balancer alone, but it will help.

Once you have a sense of how much charge you are talking about balancing, the power resistors also have the advantage that they can be used in the flat spot around 3.3V.

When your 2 runners are at 3.305V and your other cells are at 3.295V, the Active balancer will only be balancing those two cells with 250mA each, while your 5A power resistors will be draining 661mA (+164%).

24 hours-worth of bleed current is 15.84Ah or 5.66% of total cell capacity, so you can use the power resistors to bleed the two runners down by ~2-5% overnight even after the battery has been discharged into the flats where the active balancer will be ineffective.

Once you’ve shifted the two runners into laggars, you know what you are dealing with and can focus on bringing them back up to where the other cells are more slowly (boosting 1% at a time using a power supply and power resistors or just letting the active balancerbeing them up into balance over enough full charge cycles.

Once the runners have been pulled under the other cells, you’ll be able to increase boost voltage to 27.2V.

27.2V gets your runners up above 3.65V today, meaning the other cells are at 3.317V.

Once the runners have been pulled under those other cells, the battery will be under 26.53V when the non-runner-cells are at 3.317V.

Now you can successfully charge the battery to 27.2V because this will mean all cells are around 3.4V.

Sorry for the delayed reply, had some family issues to deal with the last few days.

But,, thanks for taking the time to write such an in depth reply.. I think from what you have said that i will be able to get them back in balance, or close as i can.
I expect ill be reading your posts a few more times as i try, just waiting for some sunny weather now.

I probably soud a bit thick saying this, but, once i have them top balanced again, whats the necessity to use a 5a balancer over a 2a balancer if the banks rarely get above 3.4v per cell? As presumably the balancer wont be working below 3.4v, and if one cell spikes due to them becoming unbalanced again, then the charge voltage from the inverter will still be the same, therefore putting too much current in for the 5a or 2a balancer anyway, resulting in me going through the whole process again? (unless of corse i make sure they get charged to a high soc regurly)

 

[Ucatchmydrift]

0.30-0.40 for 7.2kWh is 0.042-0.056 per day. Better than nothing but less than the 0.08 wholesale rate California pays for annual overgeneration…

Yes, same over here, id get a fair bit more from selling it back to the grid, but, i dont have a grid tie inverter, so just use up the spare power with the mining rig instead.

 

[fafrd]

Sorry for the delayed reply, had some family issues to deal with the last few days.

But,, thanks for taking the time to write such an in depth reply.. I think from what you have said that i will be able to get them back in balance, or close as i can.
I expect ill be reading your posts a few more times as i try, just waiting for some sunny weather now.

I probably soud a bit thick saying this, but, once i have them top balanced again, whats the necessity to use a 5a balancer over a 2a balancer if the banks rarely get above 3.4v per cell?

Your 2A balancer only works above 3.4V per cell. Once you get the pack balanced, you should be able to enter CV above 27.2V and so, as long as you can keep it there for long enough often enough, that may be all you need.

But if you are depleting your battery nightly, a bottom-balance may be better (which is what I have). And for a bottom-balanced battery, you will need an active balancer with no voltage threshold (whether 5A or 3A).

I know my battery is empty every single night for at least 6-8 hours, so anytime I notice a cell dropping, I can just attach the balancer after the battery is depleted (and remove it at sunrise when charging starts).

I’ve got the stuff to rig up an ‘automatic’ co trip that activates the balancer whenever the battery is below 25.6 or 24.8V, but it’s only needed once or twice per year so far, so not worth the effort.

If you are suffering through many data where there may not be enough solar energy to charge the battery past 27.2V for at least an hour or three, bottom-balanced may be the better option for your use-case.

As presumably the balancer wont be working below 3.4v, and if one cell spikes due to them becoming unbalanced again, then the charge voltage from the inverter will still be the same, therefore putting too much current in for the 5a or 2a balancer anyway, resulting in me going through the whole process again? (unless of corse i make sure they get charged to a high soc regurly)

Exactly. If being depleted happens more regularly and for more hours per 24-hour cycle than being charged past 27.2V, bottom balance with a non-threshold Active balancer may be better…

The lower knee is also much less abrupt and hence less dangerous than the upper knee (‘sinking ship’ rather that ‘rocker taking-off’ )

 

[Ucatchmydrift]

Your 2A balancer only works above 3.4V per cell. Once you get the pack balanced, you should be able to enter CV above 27.2V and so, as long as you can keep it there for long enough often enough, that may be all you need.

But if you are depleting your battery nightly, a bottom-balance may be better (which is what I have). And for a bottom-balanced battery, you will need an active balancer with no voltage threshold (whether 5A or 3A).

I know my battery is empty every single night for at least 6-8 hours, so anytime I notice a cell dropping, I can just attach the balancer after the battery is depleted (and remove it at sunrise when charging starts).

I’ve got the stuff to rig up an ‘automatic’ co trip that activates the balancer whenever the battery is below 25.6 or 24.8V, but it’s only needed once or twice per year so far, so not worth the effort.

If you are suffering through many data where there may not be enough solar energy to charge the battery past 27.2V for at least an hour or three, bottom-balanced may be the better option for your use-case.

Exactly. If being depleted happens more regularly and for more hours per 24-hour cycle than being charged past 27.2V, bottom balance with a non-threshold Active balancer may be better…

The lower knee is also much less abrupt and hence less dangerous than the upper knee (‘sinking ship’ rather that ‘rocker taking-off’ )

Thanks again ?
My situation is: in the winter i have a smart plug wired up to control grid supply to a hybrid inverter which will charge the batteries via cheap electricity from 12:00- 07:00am. (i just look at the soc before bed and set the smart plug app to come on for a couple of hours for example). I have 14.4kw storage, and use about 8.5kw in the winter, so dont need to charge them to 100% overnight.
In the summer from end of march (this is my forst year and only been running since mid december) onwards, i didnt need to use any at all from the grid, but ive used up all the power from the solar by firing up the mining rig on sunny days, therefore never getting them to 100% soc in the summer either. They always stay in the 20-80% range (which you hear so many people say this is the best range for lifepo4 for longevity (although im thinking 10-99% is more accurate) but they never mention thet if you keep them in this range, they are likely to go out of balance! Causing a bunch of headaches!)

From now on though ill do my best to allow them to charge to a high soc about once a week and see how they go..

Ive just fitted the two 2ohm 100w resistors to the 2 cells. Supposed to be a mainly sunny day tomorrow, so should be able to get the balancer going too, and with any luck the resistors may have sucked enough power out of the two cells by then to see which ones i need to fit the resistors to next. (if any - ha yeah right!)

(also, im pretty sure the balancers on the jk bmses that i have will allow a bottom balance, as you can adjust the starting balance voltage to whatever you want- not that they ever get that low, but be interesting to test and see)

(im guessing its normal for the resisters to get too hot to touch? - being resistors im guessing thats normal, converting the current to heat ?)

 

[fafrd]

Thanks again ?
My situation is: in the winter i have a smart plug wired up to control grid supply to a hybrid inverter which will charge the batteries via cheap electricity from 12:00- 07:00am. (i just look at the soc before bed and set the smart plug app to come on for a couple of hours for example). I have 14.4kw storage, and use about 8.5kw in the winter, so dont need to charge them to 100% overnight.
In the summer from end of march (this is my forst year and only been running since mid december) onwards, i didnt need to use any at all from the grid, but ive used up all the power from the solar by firing up the mining rig on sunny days, therefore never getting them to 100% soc in the summer either.

So it sounds like you are draining to ~20% pretty much every day (like me).

They always stay in the 20-80% range (which you hear so many people say this is the best range for lifepo4 for longevity (although im thinking 10-99% is more accurate) but they never mention thet if you keep them in this range, they are likely to go out of balance! Causing a bunch of headaches!)

Most hybrid owners fully charge their batteries daily and rarely discharge to 20% (backup power). For that use-case, top-balanced is better.

But for consistent discharge to 20% and only occasional charge to 80%, bottom-balanced works better.

From now on though ill do my best to allow them to charge to a high soc about once a week and see how they go..

Give it a try and see how it goes.

Ive just fitted the two 2ohm 100w resistors to the 2 cells. Supposed to be a mainly sunny day tomorrow, so should be able to get the balancer going too, and with any luck the resistors may have sucked enough power out of the two cells by then to see which ones i need to fit the resistors to next. (if any - ha yeah right!)

Will be interested to hear how you progress…

(also, im pretty sure the balancers on the jk bmses that i have will allow a bottom balance, as you can adjust the starting balance voltage to whatever you want- not that they ever get that low, but be interesting to test and see)

It’s not just a matter of reducing the ‘start balancing’ voltage threshold - you need to reverse the polarity. From ‘start balancing above’ to ‘start balancing below’.

The active balancers include a remote ‘on’ signal so you can either control them with a timer or with a cheap battery voltage monitor.

I bought a battery monitor board but only having to connect the balancer 1-2 nights every 6-9 months makes it hard to justify the complexity.

(im guessing its normal for the resisters to get too hot to touch? - being resistors im guessing thats normal, converting the current to heat ?)

my power resistor is 2.5V so only 64% the power of yours (80% the current) and it gets too hot to touch for more than a second or two.

If you attack your power resistor to a heatsink it will help and you can always add a fan if needed.

With your experience with your crypto rig, I’m guessing you know how to handle heat, but I’d probably suggest you go to the trouble to look up the max temperature your power resistor is rated for and then use an IR thermometer to assure you are staying safely under that.

Because I could touch for longer than I can touch 100C, I’m pretty sure my resistor never got over that even after 8 hours of continuous use.

I’ve been running bottom-balanced for 3 years now. 7.5kWh per day in summer down to 2.5kWh in winter in a 14.4kWh 24V LiFePO4 like yours.

My overnight load is enough to drain that overnight and I’ve been draining to 3.2V / 25.6V but push further down to 3.125V / 25.0V when I decide to balance over a night or two.

It works like a charm and I wouldn’t be changing a thing except I’m planning to eventually add an AIO off-grid inverter, another 1.5kW of panels, and reconfigure my 8S2P battery to 16S1P (48V).

I’ll have more power than I can consume daily with that set-up so I’ll probably move to top-balance for the 48V battery…

 

[Ucatchmydrift]

So it sounds like you are draining to ~20% pretty much every day (like me).

Most hybrid owners fully charge their batteries daily and rarely discharge to 20% (backup power). For that use-case, top-balanced is better.

But for consistent discharge to 20% and only occasional charge to 80%, bottom-balanced works better.

Give it a try and see how it goes.

Will be interested to hear how you progress…

It’s not just a matter of reducing the ‘start balancing’ voltage threshold - you need to reverse the polarity. From ‘start balancing above’ to ‘start balancing below’.

The active balancers include a remote ‘on’ signal so you can either control them with a timer or with a cheap battery voltage monitor.

I bought a battery monitor board but only having to connect the balancer 1-2 nights every 6-9 months makes it hard to justify the complexity.

my power resistor is 2.5V so only 64% the power of yours (80% the current) and it gets too hot to touch for more than a second or two.

If you attack your power resistor to a heatsink it will help and you can always add a fan if needed.

With your experience with your crypto rig, I’m guessing you know how to handle heat, but I’d probably suggest you go to the trouble to look up the max temperature your power resistor is rated for and then use an IR thermometer to assure you are staying safely under that.

Because I could touch for longer than I can touch 100C, I’m pretty sure my resistor never got over that even after 8 hours of continuous use.

I’ve been running bottom-balanced for 3 years now. 7.5kWh per day in summer down to 2.5kWh in winter in a 14.4kWh 24V LiFePO4 like yours.

My overnight load is enough to drain that overnight and I’ve been draining to 3.2V / 25.6V but push further down to 3.125V / 25.0V when I decide to balance over a night or two.

It works like a charm and I wouldn’t be changing a thing except I’m planning to eventually add an AIO off-grid inverter, another 1.5kW of panels, and reconfigure my 8S2P battery to 16S1P (48V).

I’ll have more power than I can consume daily with that set-up so I’ll probably move to top-balance for the 48V battery…

Yes i see what you mean by getting the balancer to balance below a certain voltage, not above, for the bottom balance..
The resistors are rated up to 375 deg Celsius, so pretty sure theyll be ok!
I left the resistors on from about 2pm until about 10am the next day, looking at the voltages on the bmses, i think that may have been a bit too long, ill have to get the power supply out when sunny and im able to get them balanced..
Still waiting for some sun, cant get them above about 80% soc atm due to the weather. Maybe some decent sun at the weekend. If so i can keep an eye on things and do whatever is necassary with the resistors and power supply.

Looking at the bmses today, cells seem to be a bit more out of balance than normal for some reason since i used the resistors (obviously not talking about the first cells), with a low load like atm, usually very close cell difference. But anyway, need to get them balanced and start from there.

 

[740GLE]

At 3.2~v that’s really tough to determine if the cells are out of wack. On my pack with these low voltages my delta hardly higher than .001-2v

Bring the voltages higher, 3.35 or more, and then watch your delta grow for your problem child.

 

[Ucatchmydrift]

At 3.2~v that’s really tough to determine if the cells are out of wack. On my pack with these low voltages my delta hardly higher than .001-2v

Bring the voltages higher, 3.35 or more, and then watch your delta grow for your problem child.

Yep.
Pulling just background house load today shows them back how they normally are, and is more clear the difference between cell 1 on each bank, and what i ment by maybe draining a bit too much.. But like you say, ill wait to see at a higher soc...

 

[Lt.Dan]

I have balancing atm set to start when one cell hits 3.35v+

I haven't read the entire thread so forgive me if this has been mentioned or if you have changed this, but 3.35v is not high enough in my opinion. The knee starts at 3.4v, and I personally prefer to start balancing after 3.45v.

When balancing as low as 3.35v, you might be discharging cells that are actually lower state of charge than the rest, effectively making your balance problem worse.

I setup for start balancing at 3.45v, and balance all the time, even if not charging. With this, my measly 200ma resistive balancing is more than adequate to keep cells within 20mV at the far upper knee.

 

[Ucatchmydrift]

One other annoying thing that seems to have developed lately is one bank drains faster than the other, if the shunts are to be believed. Was at 10 percent difference the other week before getting them charged to a high soc. Plus, they got to 100% according to the shunts, but still had plenty more to get them full. Dont trust them, dont trust the inverter soc either. And the bmses soc percentage is even worse if not charged to a high soc regurly.
I know its difficult to get a soc with lifepo4, but still a bit annoying. A rough guide only it seems.

 

[Ucatchmydrift]

I haven't read the entire thread so forgive me if this has been mentioned or if you have changed this, but 3.35v is not high enough in my opinion. The knee starts at 3.4v, and I personally prefer to start balancing after 3.45v.

When balancing as low as 3.35v, you might be discharging cells that are actually lower state of charge than the rest, effectively making your balance problem worse.

I setup for start balancing at 3.45v, and balance all the time, even if not charging. With this, my measly 200ma resistive balancing is more than adequate to keep cells within 20mV at the far upper knee.

I changed to 3.35v just temporarily as a test as the cells are spiking long before reaching 3.4v meaning the balancer would never even kick in in the first place..

 

[Lt.Dan]

I changed to 3.35v just temporarily as a test as the cells are spiking long before reaching 3.4v meaning the balancer would never even kick in in the first place..

The balancer should kick in as soon as 1 cell reaches that set point.

 

[Ucatchmydrift]

The balancer should kick in as soon as 1 cell reaches that set point.

Yes sorry, im easily confused lol. Had the inverter set to push in too much power for the balancers to cope as one of the cells on each bank was spiking alot sooner than the rest. But yes, it was set to 3.4v before i started messing with things without thinking clearly what i was doing, hence the 3.35v, but was just tying to shift some of the power from the two spiking cells sooner. That was the idea originally anyway. Is now back to 3.4v.

 

[Lt.Dan]

Yes sorry, im easily confused lol. Had the inverter set to push in too much power for the balancers to cope as one of the cells on each bank was spiking alot sooner than the rest. But yes, it was set to 3.4v before i started messing with things without thinking clearly what i was doing, hence the 3.35v, but was just tying to shift some of the power from the two spiking cells sooner. That was the idea originally anyway. Is now back to 3.4v.

Gotcha, no worries. Did you get it figured out?