SBMS cell balancing explained.

[Dacian Todea (electrodacus)]

I often get questions about cell balancing and people thinking their battery is imbalanced when it is not in reality.

So here is an example I got today from my 8s2p EVE 280K battery I installed about a year ago.

I saw the battery is close to fully charged and so I took a few photos to be able to explain what happens around the end of charge.

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

Photos are in chronological order so this first photo was taken 3:35:04pm when cell balancing was enabled and ON on cells 1 and 6 as those are the highest voltage cells.

The cells look like they are the lowest voltage because of voltage drop on sense wires due to cell balancing current.

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

This next photo was taken at 3:35:06PM so just 2 seconds after the first photo as cell balancing is set to be ON for 3 seconds and off also for 3 seconds.

Now this time since cell balancing is OFF the real cell voltages can be seen.

Notice the 20mV cell delta few minutes before charge ends about 6 minutes from this point so at average of about 40A charge current in to a 520Ah real battery capacity it means about 0.77% SOC increase.

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

Another two photos taken at 3:39:09 PM and 3:39:11 PM so again 2 seconds apart showing the cell balancing ON and then cell balancing OFF

Now more cells are balanced 1,4,6,7,8 and this is a better photo to explain how cell voltage drop on sense wires affects the reading while cell balancing is ON.

For example looking just at cells 3,4,5,6 the real voltage is

3) 3.410V

4) 3.435V <

5) 3.414V

6) 3.456V  <

Then when cell balancing current is turned ON you get this voltages

3) 3.451V      (+41mV)

4) 3.356V  <  (-79mV)

5) 3.496V      (+82mV)

6) 3.414V   < (-42mV)

This means voltage drop on each sense wire is around 40mV but each cell is measured with two sense wires so when current flows trough both sense wires you see 80mV and when is trough only one of the sense wires you see about 40mV

Depending on the current direction trough sense wires the 40mV can be added or subtracted to the real cell voltage.

Cell 3 is not cell balanced but one of the neighbor cell (cell 4) is balanced and they share a cell sense wire thus cell3 will read with about 40mV more than real cell voltage.

Cell 4 is balanced and none of the neighbors are thus it reads with 80mV less than real voltage

Cell 5  is not balanced but both his neighbors are so it reads with about 80mV more than real voltage.

Cell 6 is balanced but also has a neighbor that is balancing so it reads with 40mV less than real.

Cell 7 is balanced but since both his neighbors are also balancing everything cancels out and it reads the real cell voltage.

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

Now a photo at 3:40:14 PM of the real cell voltages just before the charging stops

Then below photo taken at 3:41:13 PM just seconds after the charging stopped.

Below at 3:41:13 PM about one minute after the charging stopped with about a 45A load

And finally at 3:43:02 around 2 minutes after charging stopped.

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

In conclusion the battery imbalance is below 0.5% and that is after about a year of use when it started with higher imbalance than this.

Meaning SBMS0 is able to keep the 540Ah battery in balance with normal daily use (mostly electric cooking everyday and in winter some electric heaters).

[jedics]

So when I see up to 3.76v charging of a cell before charging stops that is still ok?

[Dacian Todea (electrodacus)]

What SBMS model and what settings ?

The peak I have seen with Over-voltage set at 3.55V and over voltage delay of 6 seconds was just slightly over 3.6V but for me it was cell 1 so just 40mV max drop during cell balancing thus max will have been 3.55V + 0.04V = 3.59V but since there is also that 6 second delay and about 40A charging witch is not much I can have up to around 100A in full sun and with no load.

If it was some cell in the middle I will have had 80mV drop so 3.55V + 0.08V = 3.63V + a bit more due to 6 second delay (I should have set that to 4 seconds as my cell balance ON and OFF are just 3 seconds).

With a bit longer cell balancing wires than standard included cable and a long over-voltage delay I can see you being able to get around 3.7V but with 3.76V you will trigger the safety over voltage lock limit that is normally set at 3.75V.

So I will say check the settings and based on that I can give you advice of what you need to modify not to trigger the over voltage lock.

[Chris Sailor]

Hi Dacian,

you can see that the SBMS is in the limits with a 540AH regarding its balancing. You have a clear runner cell no1 and i am quite sure 1 year ago is was cell No1 too.

Also No8 is your 2nd runner. You also stated it was a bigger imbalance a year ago. Over a year SBMS should have catched the runner, did but only partly.

Andy from Offgrid Garage did some tests and on a 280AH EVE from 3,45 to cut off without absorption is around 1AH, so 540A but about 3,41V will be around 2-2,5AH deviation.

Looking at 153MV directly after switch off in a 540AH bank the diffence will be around 2-2.5AH. SBMS has only 150mA, does it have this per cell? How much devation must be to get the full 150mA and how much for 75mH?

So now you take a double sized bank means the delta is 4-5AH but the cell balancing is still only 150mA means the balance cannot get that sorted anymore as any devation right away doubles and the raise cannot be compensated by the balancer anymore.

with 45A load after 2 min you can already see that they reach the flat area where differences are much smaller, curious how that would look without a load means let the cells sit and see how big devations are without the drop voltage from charge to discharge. additional the LifePo4 resistance is not constant, it changes with SOC in the 85 till 95% area is lowest depending on the cell manufactuerer and the actual cell model.

So the lower you go the less deviation and obviouly less balance current. Also due to flat area and no constant resistance it can also happen that cells that are actually the lowest in the upper knee are higher in voltage but lower in SOC in the 85% till 95% SOC area and gets less charge current from the balancer which is contraproductive. Thats the reason you should only balance in the upper knee.

so what i expect happen is that the balancer screws the balancing in that area up (but very little as the devations and therefor balancing current is not big) and then fixes that afterwards in the upper (and lower) area when devations get bigger with higher balancing current. Till this 500AH range that works still ok but if you double capacity it doesn't anymore.

[Dacian Todea (electrodacus)]

I have not manually balanced the cells about a year ago when I installed them as they where not that far off to require that.

Yes the imbalance between cells is around 2Ah. This is first full charge after about 4 days (it was cloudy and I was discharging to use all the available energy around 10 to 15kWh/day)

If I was waiting for more cycles yesterday the delta could have been reduced from 2Ah and in a few days with multiple full charge the delta can be reduce below 1Ah but then after multiple days with no full charge it will increase again to similar 2Ah.

This cells are not matched in capacity or internal impedance and they are factory reject as probably all EVE cells that can be purchased by individuals.

Whit or without a load the cells will still get to same voltage levels after a full charge it will just take a bit longer without load but the delta is about the same 2Ah. To be sure of the exact delta I can stop the charge when full have no loads and then manually charge all cells to 3.55V CC and CV and measure the exact energy I put in all cells to get them to same level.

The delta in capacity between cells can be as high as 5Ah (I did not tested the individual cells) and cells are below 260Ah each in the 3.55V down to 2.8V range that I tested them still well below the 280Ah claimed thus it is clear why they failed quality control and got on the black market.

The date this cells where build are all the same low capacity (I so some other people test cells from same manufacturing date with similar low capacity results). But that is not a problem the price was still fair for what I got.

So unless you want to do constant voltage charging you can not bring this cells below 2Ah or at best 1Ah delta but they will also not exceed this 2 or 2.5Ah imbalance over next 10 to 15 years.

In spring when I did the capacity test the cells 1,4 and 8 where the lowest voltage at discharge so this cells based on yesterday full charge have the lowest capacity and likely the highest internal impedance.

The difference in internal impedance is what creates the imbalance both during charge and during discharge and that is what the cell balancing needs to compensate and has no problem doing so for this 500Ah+ capacity battery.

At 1000Ah there will still be no problem assuming similar delta between cells so the same max 0.5% imbalance.

Cell balancing can not fix / repair the cells. The cells will always have differences between them and while you could do a slow constant voltage charge to perfectly top balance them that perfect balance will be there until you start to discharge them and they will have an increase imbalance again.

That time spent at the top to try and force current in to cells to balance them will only be detrimental to cell's life and the 0.5% possible increase in available battery capacity will not be useful as in this sort of applications you never fully discharge the battery (not even close usually I never drop below 40%).

Even if you did fully discharge the pack you will not have gained anything since as I mentioned this cells that get to full charge first are the lowest capacity cells so getting the higher capacity cells top balanced will not help as pack will be disconnected when this lower capacity cells get to 2.8V so having 2 or even 3 or 4 Ah extra energy in the other cells will not help as it can not be used.

If the cell balancing was turned OFF on the SBMS0 the cell imbalance will have grown to maybe even more than 10 or 20% imbalance (after a year) making the available usable capacity just 80% of battery capacity.

So the cell balancing works as good as it is possible based on this cells and if it was insufficient the cell balancing will have increased since battery was installed.

Pack is only as good as the weakest cell.

If cell imbalance grows above 1 or 2% then you start to have more than 10mV delta even when SOC is below 99% thus cell balancing can work for more time to correct that.

1000Ah (4 of this types of cells in parallel) will not be a problem. Much above 1000 to 1200Ah is just not useful for an SBMS0 due to how large of a PV array you will need.

For this about 500Ah 24V battery (about 13kWh usable capacity) I have a 9kW PV array in dual PV array mode. To take advantage of a 1000Ah+ I will need an 18kW PV array and that is getting quite large.

For where more than that is needed is much better to have separate systems so multiple SBMS0 and batteries (fully independent) for redundancy.

I have the A123 system battery that is made with 8s10p pouch cells and that is kept closer balanced around 0.2% max imbalance but that is because of better more equal cells and the fact that there are 10 cells in parallel selected at random makes them even more equal in characteristics.

Will be very curious to test the EVE battery capacity in March a year after the first capacity test. Up to this point I used about 2000kWh from the pack that is about equivalent with 150 full DOD cycles (of course the only full cycles are made when I do a capacity test other than that battery is almost never below 40% SOC).  This is nothing compare to the 6000 cycles at 80% DOD in spec but the calendar aging will play the main role and that was not provided in spec.  I do not expect them to be as good as A123 Systems battery that degrades at about 1% per year but if if 2% or less it will still be very good.

[jedics]

sbms120. All setting default except for delay which I changed to 1 second after seeing such a high number, I don't remember what it was before. Maybe I am mistaken of the exact OV number as it was a few months ago since it is still set to 3.75 OV.

[Dacian Todea (electrodacus)]

You mean the Overvoltage Lock is set to 3.75V the Overvoltage should be set at 3.55V by default.

In any case this limits can be set to whatever you are comfortable with but with the delay set at 1 second and likely the cell balancing OFF more than one second no cell will exceed 3.55V in this case maybe 3.56V or whatever it can get to in that one second delay. I mean real voltage as you can read higher voltages on SBMS120 during cell balancing but those will include the voltage drop and can be as high as 80mV maybe even 100mV depending on how long are those cell sense wires and what resistance they have.

[Chris Sailor]

Hi Dacian,

in boats we think differently, you cannot have enough capacity, limit is space or money. Same with solar which is extremly limited, that's why we buy the highest efficency panels per sqm we can get. A typical cat has 8sqm sapce on the solar arch and mabye another 4 sqm on the roof which is partly shaded by the boom.

at the moment i have 3x365W bifacials delivering in best case 1300W, average is about 900W. 2 will go on the roof and 3x590W bifacial will go on the arch. that will bring me to 2300W total, more doesn't fit. So for the big banks we have we are always underpaneled.

that means that there are times especially side seasons i only get to full charge once a month when i go into marina for restocking and cleaning the boat for 2 days a month. then shorepower  charges it fully or i run the mobile 3500W gen.

My cells are matched and grade A+ from my buddy that builds powerbanks, so they should have minimal deltas.

i disagree with 1000AH is no problem, i see on mine it is. Also the 5A balancer has 1.3A balance current at 180mV devation and even that has a hard time to get it sorted with 1088AH, 1% is 10AH and balancing just makes sense in the knee so the time spent their in normal operation is small.

As I explained balancing in flat area is unbalancing the cells. in my pack cell 7 is highest and 8 lowest but at 3,30V the 7 is lower then 8 because thats the area on Lishen where you have lowest internal resistance 7 is lower becaus it has a higher SOC but 8 has a higher voltage because its exactly in the low resistance area while 7 due to higher SOC has a higher internal resistance. That will fool the electrodacus balancer and its balancing cell 8 while it actually should balance cell 7. and because here is still summer i am always in the upper area 70 till 90 normally and that creates the imblance. A week ago the bank was at 50MV devation with the active balancer connected, then i switched active balancer off so only SBMS balances and today cell8 is 195mV lower then 7, means SBMS brought it out of balance. no have use 130A for some minutes for nespresso machine was max it got.

I will know use the shorepower charger and put absoprtion and float to 14,2V, connect the active balancer and let the bank be in the higher area till the active ballancer could sort it.  then i will change balancing SBMS from 3,4V only and add the active balancer to only work above 98% SOC when charging and see what happens. Thats how on all other BMS the balancing works, only in the knee.

[Dacian Todea (electrodacus)]

In one month of no full charging with acceptable quality cells you can not get an imbalance higher than 0.5% without using any cell balancing.

That was the result I got after running an old GBS battery for 4 months with cell balancing disabled.

In the flat area the cells will not need balancing unless charged at very high rate in witch case the weakest cells will have higher voltage due to larger internal resistance so they will be balanced.

3.3V can be almost any SOC 

Currently my A123 system battery is at 46% charging with 18A so about 0.1C and all cells are around 3.330V about 4 to 5mV delta.

The EVE battery is at 55% and discharging with about 10A and all cells are around 3.308V with a 8mV delta.

The SBMS will only attempt to balance if there is a charge current and delta is above 10mA

As it is cell balancing is not enabled on any of my two batteries that are around 50% at the moment  and of course the EVE is discharging now so no matter the delta cell balancing will not be active.

When you say the bank was 50mV delta and 195mV delta you are referring to fully charge or very close to that just before charging is disabled or immediately within seconds after full charge ?

 Because if so that does not mean there is any significant imbalance and the reason I made this post.

You seen 150 to 170mV peak delta on my EVE pack that was perfectly balanced with no more than 0.5% imbalance.

[Dacian Todea (electrodacus)]

Small update:

The A123 System battery is now at 80% all cells around 3.370V and 7 to 8mV cell delta with a charge current of 33A that is about 0.18C for this battery.

The EVE battery is at 66% now and currently charging with 39A (cloudy just about 1A from each panel) and cell voltage around 3.338V and cell delta fluctuates around 10 to 11mV. So from time to time there is one to 5 cells that are being balanced so maybe for this battery since is not as good quality as the A123 System I can increase the cell balancing delta to 15 or even 20mV so there is no cell balancing this early.

But even as it is with default 10mV delta the battery has remained at just 0.5% imbalance over the past year so battery is being kept in balance as 0.5% imbalance is not significant in any way and is mainly due to difference in spec between cells as they are less uniform compared to the A123 Systems.

[Chris R8]

Update on my side:

Done what I said and put absorption and float at 14.2V with shorecharger and active balancer and let it cycle between 3.4V recharge enable and 3.55v for one full day. Bank came down to 40mV devation direct after cut off charge at 3,53V, highest 3,56V cell 7 and cell 8 at 3,52V. So the active balancer done its job and got them top balanced. Active balancer permanently off now.

Then I discharged a while with 80A till 98% and connected MPPT which will now absorb till 14.2V and get the rest of top balance sorted. Voltage straight after charge enable is 3,35V and SBMS still standard 3,2V balancing exactly the wrong cells 2 and 8 which are the lowest in the knee above 3,42V till cutoff but higher between 3,34V and 3,42V. This proves my assumption that actually the SBMS balancing in the flat area causing the imbalance. 

changed it now to 3,42V start balancing SBMS, strange thing the balancer only engaged later at around above 3.50V.

Whats causing this?

[Barry Timm]

You're mistaken, in your assertion that the SBMS0 balancing routine is incorrect. Hundreds of users of the SBMS0 have ZERO balancing issues. My 7.8 kwh battery comprised of 16 x 150ah cells in an 8S2P configuration retained remarkable balance over my last 3 years, seldom ever seeing over 50 mV at certain charge times and almost always well within the 11 mV range during normal usage phase, using all default charge and balance parameters. 

[Chris R8]

Hi Barry,

It's in general by theory wrong but works well especially in offgrid use with big arrrays of solar, salso bigger banks. But till around 500Ah in my experience for boating. As you can read boating is significantly different from setup, mostly huge banks significantly underpaneled. Getting daily full, nope often a month not full...means balancing happening mainly in the flat area.

 I installed over 20 SBMS now. And all work fine as typically 2p4s or 2p8S with Max 600AH (they have 300-800W more doesn't fit) except 2 installations, mine with 4p4S 1088AH and another with 5p4S with 1516AH. I had before 840AH 3p4S and there it was working partly OK with low C use but high C use the balancer couldn't bring them back to balance, it's just 150mA balancing current and that has limits. 40mV deviation with 1000AH in upper knee is 1AH, 200mV 8-12Ah.... 150mA is too small here.

I installed an active 5A balancer, which is actually too small in theory too but that got all sorted balancing only in the upper knee during charging and discharging.

Search for offgrid garage, Andy did a series of videos about balancing with actual live testing and demonstrated what happens if you permanently balance.

[Dacian Todea (electrodacus)]

Cells can not be lower under some limits and then higher above others unless they have dramatically different temperature or if charge or  discharge current is very different at that level thus the effect is from the cell internal resistance + the cell connection resistance.

Also you do not need absorption setting on any charge source or leave them at default 14.4V typical so that it will never get to that point.

You can not have the cells balanced discharge 2% of battery SOC charge back that 2% and then cells are imbalanced. It takes tens or hundreds of 40 to 50% DOD cycles to even get a 0.5% imbalance.

Your pack seems perfectly balanced so just disconnect the external cell balancer. Use the battery for at least 6 months to a year. Then do a full charge with the SBMS0 and from there take a lab power supply and charge each cell individually to 3.6V and measure the energy needed to get there for each cell then the delta in energy will be the imbalance. If after a year of full time use that cell imbalance delta is higher than 1% it means the SBMS0 can not keep the battery balanced. 

If it is below 1% as in my case around 0.5% imbalance after about a year of daily use then cell balancing is sufficient to keep that battery in balance. The cells can not be maintained at 0% delta as that will require cell balancing after the charge is done by using small absorption current and cell balancing and that will be way more detrimental to battery than maintaining the 0.5% imbalance.

Using absorption for any type of Lithium battery degrades the battery (with some type of chemistry very significantly) so should not be used.

If you had 10% imbalance after 6 months or 1 year then that will be considered significant and will mean that cell balancing is not working or is working incorrectly (unlikely as the cell balancing logic is done in hardware not software and that was designed by Intersil) .

Voltage above 3.4V has no meaning or relation with SOC and there will always be a cell to be first at 3.55V and charging be terminated.

With constant current charging and no absorption it will be impossible to have just a 40mV delta immediately after end of charge as you obtained with absorption for an entire day.

If voltage is set at 3.55V the cell that gets there can be even above that say 3.6V while the lowest cell is guaranteed to be around 3.4V so as much as 200mV delta but that will not mean much in therms of SOC delta between the cells that will typically be around 0.5% as shown on my EVE battery.

If due to large variation in charge currents or high charge rates you think some cells at some point are balanced while they should not be wasting some energy then increase the cell balancing delta to 15 or 20mV to prevent that from happening.