Momentarily interrupt of all loads attached to the SBMS0

[Peter Flynn]

I have an SBMS0 and a Multiplus running the Two-signal BMS support Assistant.

Frequently but not every time when I reach full charge on the SBMS0 and the charger on the Multiplus is turned off it also momentarily interrupts all loads attached to the SBMS0 and all my devices flash.

Pretty sure the SBMS0 is causing this because the 12 volt devices also flash and they run off the Orien converter and that is only triggered off the SBMS0.

I think it is hitting the OVLK then quickly dropping below the threshold to turn everything back on once the Charging is removed. I think it only occurs during cell balancing, but I have no solid evidence of this.

I have raised the OVLK to 3.59 and lowered the OV to 3.40 it is better but still occurs.

This has become more noticeable since I replaced the AIMS inverter/charger with the Multiplus inverter/charger and is now charging at a much higher rate, I see this behavior much less when charging strictly from solar.

I have 16 CATL 314 AH Cells in a 24 volt 2P8S configuration.     

I have one, cell #3 that was always the first to reach full charge much earlier than the others and is always the cell to stop the charge cycle. Example, when the next highest cell is at 3.76, #3 will be at 3.44, I have balanced all the cells a few times.

In an attempt to cure this issue I swapped Cells so that the first Cell in the Cell #3 pair become the second Cell in #6 pair which was typically the lowest pair and the second Cell in the Cell #6 pair become the first Cell in the #3 pair.

The Cell #6 pair is now exhibiting the same behaver as the Cell #3 pair did.   

A little background here, when I bought cells originally in march 2021, I bought 8 cells from a group buy on DIY solar that went bad, instead of nice new cells, after many months I received 8 poorly packaged bloated cells some with bent posts, two entirely unusable, I sucked it up and bought a second group of cells from Docan power, 10 new CATL 314 cells, adding 8 new cells and replacing the two unusable cells ending up with a 24 volt 2p8s pack.

This worked quite well for a long time, in the last 8 months I have been experiencing the current issue.

It is not a loose or bad cell monitoring cable, I checked them all when I swapped the cells around.

My next thought is to replace the second Cell in the Cell #6 pair, it is one of the original bloated cells and replace one other cell that is also bloated with new ESS grade CATL 314ah that are available from Docan Power.

It has not been convenient to get photographic evidence when it occurs so pictures are few.

Any thoughts or suggestions on if you think this will cure my issue or any other constructive comments.

Peter  

[Dacian Todea (electrodacus)]

Peter,

The default OVLK is 3.75V Not sure if you modified that or you increase the OV threshold from default 3.55V to 3.59V ?

OV should stay at 3.55V as with 3.4V you will not get a full charge (deepening on charge current it may stop charging even at 50% SOC)

I will suggest reverting back to default settings and if you have issues with OVLK that means you have bad contact at one or more of the cell balancing wires or a high resistance contact between cells.

If you have a bad contact resistance between cells you will be able to see when you have the cells under heavy load as the cell with bad contact resistance will show a significant lower voltage than the others.

If that is not an issue it only leaves the cell balancing wires as they will have voltage drop across them when cell balancing is enabled (during charging only).

Worst case with included cell balancing cables is 50mV drop per wire and so absolute worst case is showing a 100mV higher voltage than real cell voltage for a cell between to cells actively balancing and so with default settings OV at 3.55V and OVLK at 3.75V there is a 200mV delta thus the 100mV will not trip the OVLK but with a bad cell balance wire (high resistance) is possible to exceed 200mV and then OVLK will be tripped during cell balancing.

[Trydan Am Ddim]

Peter,

Some feedback on OVLK as I’ve been looking into this for a while. Your results are puzzling…

I’m wondering if you know that OVLK causes the SBMS0 to switch off the loads it controls as well as expected, all the currents. In this situation DFET on does not mean the SBS0 will let you discharge anything under its control. I guess the firmware designers were super cautious and thinking that OVLK should never happen, but if it does, let’s switch everything off and run away to safety!

So OVLK is causing your loads to turn off, but you mentioned that they flicker – implying things come back almost immediately. If I’ve read the ISL94203 data sheet correctly, OVLK and OV have the same rest threshold - the Over Voltage Recover setting which I believe defaults to 3.35V in the firmware. While your OVLK issue might be a “peak” high voltage reading caused during the balancing period with poor balancing wires, your OV flag is also set, meaning that you also exceed the OV threshold during the non-balancing period, (assuming you have the correct settings for balancing on/off times and for the overvoltage delay). This OV is a real voltage at the cells even if you don’t have optimum balance wire connections as the voltage read is done with a tiny current - but as Dacian correctly says we should fix any issues with balancing wires.

If any cell is at the OV threshold, all cells need to go below the OV recovery voltage before the OV flag is removed and charging source will be off until this threshold is reached. Similar if OVLK threshold was only momentarily reached, all cells need to go below OV recovery before the OVLK flag is removed and during this time both charging and loads are off. If there is no load on your battery, except the SBMS0, when OVLK kicks in, and one of your cells really is at least at OV if not higher, I’m very surprised that your cells get down to the OV recovery threshold almost immediately. My situation would take many hours – not a flicker to drain the batteries sufficiently to turn things back on. I’m wondering if your OV recovery value is reasonable?

As far as your cell issue is concerned you could debug your system by temporarily downgrading to a 1p8S system – first using what you consider your “best” 8 cells to initially solve the above “flicker” problem. If all cells behave well for a decent number of cycles, you could rotate in one of your other cells. In that way could you eventually locate any poor performing cells that previously seem OK. 

[Dacian Todea (electrodacus)]

Trydan,

Peter has messed with advanced settings so it is hard to know exactly what happen.

Normally OV will be triggered when any cell exceeds 3.55V (default OV threshold) for more than 6 seconds (again default value). When that happened charging will have stopped so will cell balancing thus the OVLK in that case will have happened if the charger was not controlled by SBMS0 and continued to charge the battery after OV flag was set.

You also mentioned in earlier comments about wanting to change the firmware so Loads are not turned OFF when OVLK is set.  I actually added that in firmware at a latter date when I seen multiple people usually using Victron Inverter chargers that used just the EXT IO3 to control ON/OFF the inverter part but not the EXT IO4 set as type 1 to control the grid charging part of the inverter charger. Thus many overcharged their batteries.

I think is a good decision to leave the OVLK flag disconnect not just the chargers but also the Loads as in some circumstances it can help prevent the battery from overcharging and in any case OVLK should not happen thus turning OFF the loads will be more visible to user and it will be more likely to investigate the reasons for that.

[Trydan Am Ddim]

Dacian, When an OVLK condition occurs, the way in the firmware you force the values of CFETError and DFETError which then cause all IO controls to be off is a very safe way to respond.

My OVLK errors were so infrequent there was no evidence for me to see what happened. Often by the time I noticed the problem, the cells might not even have been above the OV threshold, but of course were above OV recovery.

Once triggered I initially thought the simplest recovery procedure was to switch on my Gainfel inverter and suck power out of the battery to get each cell below the OV recovery threshold, but I had designed things so the SBMS0 had control over the inverter and each time I turned on the inverter the SBMS0 dutifully sent it a pulse that turned it back off since now loads were not allowed! Hence my though process about allowing loads during this error condition.

Power cycling the SBMS0 would clear the OVLK, but I’m always a bit hesitance in disconnecting and reattaching the power / balancing cable to the SBMS0. There is some brief anxiety each time whilst you "hope" the SBMS0 will come back on! Also when it is mounted, access to the cable is not easy.

In modifying the code, I was thinking that the existing control types could keep there their current functionality and perhaps a new “control type 7” could be added that would basically remain on as long as DFET was 1. Users would be warned to use Type 7 at your own risk!

HOWEVER, now that I understand how OVLK can be triggered without OV being triggered I’m no longer concerned that there is some strange intermittent error in my SBMS0; especially since OVLK has not manifested itself for about 3 years.

[Peter Flynn]

After posting I went back and looked, I had done a few typos sorry about that, the OVLK is set to 3.80.

The original issue started when I had OVLK set to default, 3.75 and OV set to 3.55

After reading here on the group I checked all the connections and raised the OVLK to 3.80.

This did not affect the issue at all. I then lowered the OV to 3.45, this made it occur less.

I may have used the incorrect term when I used the term Flash, when all the loads (including the lights) go off it can last 20 or more seconds, by the time I get to the SBMS0 the voltage spike has returned low enough that the lights come back, on occasion I have to power cycle by pulling the ribbon cable to get the lights back on.

I have removed cleaned and tested all the Balance wires.

Since this has been occurring, I thought it might be that the SBMS0 being in a moist hot environment might be the cause, I bought a new SBMS0 and moved the new one into a dry temperature controlled environment replacing all the wiring including the balance and power ribbon cable.

The issue still occurred, the only thing I have done that affected it was when I moved the bloated cell 1 from pair #3 to cell 2 of pair #6, and cell 2 from pair #6 to cell 1 of pair #3, the same behaver happens only now the voltage on  Pair #6 spikes to end the charge first.

Due to this being the only lithium pack I have ever worked on, I do not know if this is a way a LiFeP04 cell can fail?

I do believe it is the Pair #6 cell 2 causing the spike.

Is there a way to take measurements from this cell to determine if it is bad?

I just bought an ST-link programmer and have my original, spare SBMS0 that I intend to load the new firmware on to run the new testing selections, I also want to replace the bad cell incase it decides to fail in a bad way.

When I am off Grid with the Bus, the pack works the way it should, I am pleased with the operation, this issue does not seem to affect the operation, in a typical scenario running off the batteries overnight I only bring the charge down to 70% with heavy usage.

It only occurs at the very end of the charge cycle.

When I get to the SBMS0 quick enough to look at the difference, Cell #6 voltage is significantly higher, I have included a photo from when it was still cell #3 set at 3.55, and you can see my earlier photo of after I set it to 3.45.

 Trydan Am Ddim, I like your suggestion of reducing the pack to 1P8S for testing, although that is a lot of work.

 

 

Peter 

[Peter Flynn]

[Peter Flynn]

On Saturday, July 25, 2026 at 1:39:52 PM UTC-4 Peter Flynn wrote:

[Peter Flynn]

For some reason I can only post one picture at a time

[Dacian Todea (electrodacus)]

Peter,

There is nothing abnormal in your photo of the cell voltages.

In that photo cell 3  (parallel cells are no different from a single cell with higher capacity) is maybe 0.1% closer to fully charged than the other cells but there is nothing unusual or the issue related to your problem.

It can be any of the other cells that spikes above 3.8V and thus turning OFF your loads.

With your current setting of 3.45V OV and 3.8V OVLK there is a 350mV delta and if you still get OVLK it means you have huge extra resistance on at least one of your cell balancing wires.

In properly installed and connected cell balancing wires max voltage drop is 50mV so worst case a cell not balancing surrounded by two cells that have balancing enabled you will have 100mV of voltage drop meaning that one cell in the middle could spike 100mV above the real voltage of the cell and that is not enough to ever trigger the 3.8V OVLK even if you have the limit set as default at 3.55V.

So you have a sense wire with 7x the normal resistance or two wires with over 3.5x the normal resistance.

This issue has nothing to do with your damaged cell.

A photo of the battery could be useful as another thing that could cause this will be having long bus bar between two cells (like maybe you having the pack split in two half and having a long wire connecting this).   All wires connecting the cells in series need to be well below 30cm (1ft)

But if that is not the case (you do not have long bus bars anywhere) then you clearly have one or more high resistance connections on one or more of the 12 cell balancing wires.

[Trydan Am Ddim]

Peter,

It might be wise to focus on one issue at a time. If you reduce your battery back to 1P8S using 8 cells that you “know” are good, then you can completely remove from your mind the thought that the issues you are seeing are due to the cells. Working with ½ the cells also speeds up the test described below as less charge time will be needed. After fixing the OVLK / OV issue you can then move back to the task of working out if you have a bad cell.

In your settings you need to establish what you are trying to do. For example, when you said:
“then lowered the OV to 3.45, this made it occur less”.  I think you interpreted this as a positive result; I’ll attempt to explain why it isn’t. If you are about to use your batteries on a critical voyage and what to minimize problems lowering OV could be useful; however if your current goal is to discover what is wrong, you should maximize the chance of a problem. If the problem occurs more frequently you get more chances of catching it happening and thus finding the cause.

A suggestion on how to debug….
In your 1p8S good cells only setup, put ALL the settings at their recommended values and repeat a test in which you progressively make your system worse; taking a video directly off the SBMS monitoring screen during the test so that you can later look at the voltage peaks you get. What do I mean by making your system worse – gradually lowering the OVLK by e.g. 0.1v - 0.2V and keeping everything else the same. Though you want to capture an OVLK event happening you should start with settings which are reasonable “stable” (e.g. no point in starting with OVLK = OV) but not so stable that it takes days for the OVLK to occur.

You start the experiment with cell reasonably balanced, cell voltages as high as possible but with them all below OV recovery and no loads except the SBMS0; when everything is set start videoing and charging. If the SOC is initially reasonably correct with this and the charge current rate, you will have some idea of how long it takes to fully charge the cells. Leave charging in place and when you return either you will see OV, OVLK, or OV+OVLK. If you only get OV the test run is not useful – you need to discharge the cells so that the OV flag is cleared. You could repeat the setup hoping this time for an OVLK, or you could repeat the experiment with a slightly lower OVLK. Repeat until you get a recording with the OVLK set. You can then examine the recording and find the peak voltage reading that caused the event, see which cells are involved and is the peak voltage reasonable.  The object is to capture an “unreasonable” cell voltage reading. If you get an OVLK but the immediate voltage peaks are within the range that Dacian says they should be, then I believe you have gone too far in your reduction of OVLK. Basically, you need to generate an excessive voltage peak. If you could achieve that with the default value of OVLK that would be ideal, but I don’t know how long “on average” that takes to occur in you system.

I suggest capturing the video data directly from the SBMS0 screen as I believe it updates more frequently than the web feeds. From the video you could isolate key screen shots before and after the OVLK event and report here on the OVLK setting you had.

Of course WAY quicker than doing any of this is to upload the photos that Dacian suggested. However, if you already know all your cell connections are short then you could do the above series of tests.

It should be obvious, but I should mention if you do this test with 1p8s with default settings and no OVLK happens then clearly there is a issue with your 2p8s setup.  

[Peter Flynn]

I do not have a current picture of the pack, I will see if I can get one in the next few days.

This is the most recent picture from a few months ago, just after I replaced the AIMS inverter with the Multiplus and relocated the SBMS0 inside, the SBMS0 in the picture is not active, I have also included part of what I made as a wiring diagram for how most of it works, the picture show the 817 jumpers on but in real life they are removed.

Trydan Am Ddim, Thank you for the trouble shooting suggestions, before I remove half the cells, I should try and capture the issue occurring with the current setup before I make any drastic changes. It just involves sitting there waiting, I’m not good at that LOL. 

[Dacian Todea (electrodacus)]

Thanks for the photo of the battery.   Those solid busbars are terrible in any circumstance but in a high vibration environment like yours they have no chance of working.

You will need flexible busbars else even if you fix the issues they will appear again after a few km of driving.  Cells could also be damaged as the case is made of thin aluminium and that could easily crack with solid bussbars and vibration.

The charge current from an inverter is not a clean and constant as from solar. If you set for example a 50A charge current on the inverter/charger the peak current will be 1.41x 70A = 99A        100 or 120x per second

A photo of my battery showing flexible busbars and flexible cables to connect the cells in series. Also compression so cells while moving due to thermal expansion or normal expansion with the state of charge will not put any pressure on the cell terminals. 

[Dave McCampbell]

Hi Dacian and All,

While we are discussing LFP system improvements here are some thoughts and questions.  

On our cruising sail boat I too started construction of my LFP system with a plywood cell box and solid buss bars.  As time went on and I read more about LFP systems on boats and RVs, I came to realize that there was a better way that would improve my system safety and security.  I changed the compression box to one built out of heavy fiberglass, well secured to the boat, and with a clear polycarbonate top.  I changed the cell connecting bars to thicker ones with flexible construction and a small screw connection in the center for the sense wires.  As I mentioned in an earlier post I also changed the control wires to my various relays and the shunts from solid CAT 6 to small multistrand tinned wire.

Dacian, looking at your big red wire terminations, are the ends of those wires bare, soldered or with a ferrule going into your screw terminals?  I have read elsewhere that it is best to use ferrules for wire ends going into screw terminals without a plate between the screw end and wire.  I am looking at my terminations now and trying to decide if I need to buy ferrules for mine where that is the case.  

What are your thoughts on this?  And thanks again for your review of resistor calculations above.

 Dave

[Dacian Todea (electrodacus)]

The cell balance wire is very thin so I folded the wire in two or three so it fits the slightly larger ferules and then I crimped them on to wires and then had those inserted in those massively oversized lugs that I had around.  There where zero issues for the past few years.

Soldering the wires should be a great option maybe having some strain relief so vibrations do not affect the solder point. Soldering will be great in marine environs where atmosphere is much more corrosive than in my climate controlled and dry house.

On an older GBS battery I had some trough hole block terminals that I took the plastic spacer out and used those individual terminals with the corresponding leads that I soldered directly to the middle of the busbar and then screwed the wires directly in there without eve using ferules. But those where quality terminals have a conductive and elastic lame between the screw and the wire. Also there was no problem for those used many years but again in my ideal conditions without corrosive environment or vibrations   

As you have likely an M3 or similar screw mounted on the middle of the busbar wires soldered on to a ring terminal should work great. Maybe solder the wire closer to the ring so the plastic insulation can be glued to the end of the terminal so that vibration is not directly transmitted to the soldered joint.