16 hours ago, BritishRacingGreen said:

Hi @jumper  there is no way the bms can limit the current,  but it can instruct an  inverter to limit  the charge current, via suitable bms comms 

That's my understanding too. But as I showed in my opening post, if the BMS displays and communicates inaccurate values the inverter would be misled and act accordingly false. Therefor I rather trust on the battery voltage.

29 minutes ago, Beat said:

That's my understanding too. But as I showed in my opening post, if the BMS displays and communicates inaccurate values the inverter would be misled and act accordingly false. Therefor I rather trust on the battery voltage.

Yes I get that, truth be told I like setting my own values, with the reasoning I know what i am doing (hope so ☺️).  

6 minutes ago, BritishRacingGreen said:

Yes I get that, truth be told I like setting my own values, with the reasoning I know what i am doing (hope so ☺️).  

I do the same, no BMS coms with my inverters, I set my own parameters and so far everything running smoothly.

However, my battery or should I wrather say the Seplos BMS cycle count still puzzles me, if ones drain 20% of SOC, 80% still available, the moment it gets a bit of charge it counts one cycle. 

Interesting to see how members use the voltage when there is comms available. Some of us are forced to use USR settings.

3 minutes ago, Scorp007 said:

Interesting to see how members use the voltage when there is comms available. Some of us are forced to use USR settings.

Yes,  lets take livoltek lithium as an example, i like the limits that the pack impose on the inverter, except for the 58V charge voltage limit. Now this manufacturer knows their battery inside out, and i am still a year old apprentise. But i dont like that setting, too hi for me. Ideal would have been a user interface for bms settings where you can impose a conservative limit on the default limits. Sadly thats not an option. 

Furthermore if bms comms breaks down then the inverter is left hanging with very liberal values that was previously issued to it. I dont like it. 

Having said all of this, the basic promise of the bms comms function  is sound, and will only get better.  I think pylontech is one of few that has battle hardened bms comms, and cotresponding inverter drivers are pretty matured and stable. 

47 minutes ago, Antonio de Sa said:

I do the same, no BMS coms with my inverters, I set my own parameters and so far everything running smoothly.

However, my battery or should I wrather say the Seplos BMS cycle count still puzzles me, if ones drain 20% of SOC, 80% still available, the moment it gets a bit of charge it counts one cycle. 

The rules of engagement regarding cycle count has been dark territory for me, thanks for shedding light. Well at least if i project my own cyclecount i will still get good number of years out of my battery, fortunately my pv yield is very strong in relation to my load demands at the moment, meaning battery augmentation to pv power is relatively low. Except of course clody days, but thats more exception than the rule. 

6 hours ago, Antonio de Sa said:

However, my battery or should I wrather say the Seplos BMS cycle count still puzzles me, if ones drain 20% of SOC, 80% still available, the moment it gets a bit of charge it counts one cycle.

I found the setting for this on the seplos bms (Cycle cumulative capacity) and is set incorrectly by default to count a cycle at 20% instead of 80%. I've changed my setting to 90% as my battery is sold as 90% DOD and now it counts a cycle every 3 days.

I would like to give the manufacturers the benefit of the doubt and say it's an old lead acid setting and they are unaware of the issue and aren't trying to scam anyone out of a warranty, but nevertheless there will be a heck of a lot of batteries out there counting cycles 4 times faster than they should.

6 hours ago, BritishRacingGreen said:

I think pylontech is one of few that has battle hardened bms comms, and cotresponding inverter drivers are pretty matured and stable.

Interestingly, the seplos bms can imitate a pylontech and a few other protocols. They provide a pulldown to change the protocol of the CAN port in their bms software depending on the inverter you run.

24 minutes ago, jumper said:

I found the setting for this on the seplos bms (Cycle cumulative capacity) and is set incorrectly by default to count a cycle at 20% instead of 80%. I've changed my setting to 90% as my battery is sold as 90% DOD and now it counts a cycle every 3 days.

I would like to give the manufacturers the benefit of the doubt and say it's an old lead acid setting and they are unaware of the issue and aren't trying to scam anyone out of a warranty, but nevertheless there will be a heck of a lot of batteries out there counting cycles 4 times faster than they should.

The problem with changing anything on the BMS like the way it counts will void your warranty on many of them , i remember reading that in the Hubble AM2 warranty.  

1 hour ago, Nexuss said:

The problem with changing anything on the BMS like the way it counts will void your warranty on many of them , i remember reading that in the Hubble AM2 warranty.  

For sure, if warranty is an issue I would suggest gathering heaps of documentation of the count being faulty and send that to Hubble or whoever the manufacturer is.

I've saved all the evidence I need if warranty ever becomes an issue and will deal with it if/when it arises. If worst comes to worst I've sacrificed my warranty in order to share evidence with the community that there is a simple issue with a setting on the bms and not the fairytale replies I've been reading on this forum and I'm ok with that.

Today one of my BMS almost sent me in panic mode. Sunny day, the packs got fully charged. However while my first 3 packs showed 100% SoC at 52.5V, the 4th and newest one remained at 80%. Its cell voltages showed weird things, several cells went into balancing mode while the pack current remained low. I was afraid it had taken damage in a recent total discharge during unexpected 4 hour blackout. However suddenly the BMS jumped into normal display, 100% SoC, no more balancing. What a relief! It looks like the battery itself is behaving correctly while the BMS was getting weird.

I'm getting goosebumps thinking such a BMS would communicate with the inverter for the charge management. I feel safer managing the charge by the inverter based on the battery voltage that never lies.

Interesting thread forumites and it seems majority of installers are not people conversant with coding or interested in verifying how batteries record cycle counts as most refer end-user back to manufacturer who may be unaware of issues due to focus on pushing products out. Which means in time a lot of people will be prematurely out of warranty, with no recourse as installers and sellers may be ignorant or not interested.🤦‍♂️

Now most manufacturers will INSIST that the inverter and battery should have comms e.g. Sunysnk which puts most end-users up the creak. Now am wondering if the BSL battery warranty would be void if there's not comms with an inverter... probably it would. Food for thought.

Edited November 30, 20232 yr by Moffat

Another strange BMS behavior: This morning SoC of all packs went low. The 1st 3 packs show less than 29% SoC with a "Low SoC" warning. The 4th and newest one reports 10% SoC (witch can't be right) but no warning. Since this one pulled significantly less charging current I presume that its real SoC was quit higher than the SoC of the others.

11 hours ago, Moffat said:

Now most manufacturers will INSIST that the inverter and battery should have comms e.g. Sunysnk which puts most end-users up the creak.

I would simply not buy batteries with such conditions. LEOCH, sold by Averge do not even mention comms in their instructions.

1 hour ago, Beat said:

Another strange BMS behavior: This morning SoC of all packs went low. The 1st 3 packs show less than 29% SoC with a "Low SoC" warning. The 4th and newest one reports 10% SoC (witch can't be right) but no warning. Since this one pulled significantly less charging current I presume that its real SoC was quit higher than the SoC of the others.

I would simply not buy batteries with such conditions. LEOCH, sold by Averge do not even mention comms in their instructions.

Interesting that you get way different charge rates. Normally the newer battery would charge faster and if there is a difference between them also discharge faster. 

Something weird going on. BMS still seems to be the weak link but comms should keep them closer than what you are seeing. 

1 hour ago, Scorp007 said:

Normally the newer battery would charge faster and if there is a difference between them also discharge faster. 

As a matter of fact it does, despite being at the end of the row wiring. I concluded that its internal resistance must be significantly lower than the older ones, probably due to improved manufacturing.

7 hours ago, Beat said:

As a matter of fact it does, despite being at the end of the row wiring. I concluded that its internal resistance must be significantly lower than the older ones, probably due to improved manufacturing.

Might be worth investing in a decent test meter to actually also check voltages individually and across the packs. Worrying bit is if you have a sudden high induction continuous load, wonder how this would affect the system.

18 hours ago, Beat said:

As a matter of fact it does, despite being at the end of the row wiring. I concluded that its internal resistance must be significantly lower than the older ones, probably due to improved manufacturing.

I have 3 x 5.2kwh LFP batteries, wired up in parallel and using my own charge settings (so no comms between battery pack and inverter) but batteries are connected to each other with interbattery comms. 

One battery was purchased 7 months after the other two. In this period the manufacturer (BAK) changed their cell design and manufacturing process (unknown to me when I purchased the 3rd battery). While the batteries look exactly the same, the internals are not. 

The impact of this is rather interesting. The older batteries reach 100% SOC at 55.8v while the newer battery reaches 100% SOC at 55.1V. Voltages measured at the battery poles which agrees to the charge voltage displayed on the inverter screen at the same time. Voltages provided by BMS is approximately 0.1V lower across all 3 at 100% SOC.

These batteries use 4 different criteria to determine when the battery is at 100% SOC, but the one most frequently resulting in the BMS determining that the batter has reached 100% SOC is when any individual cell reach 3.65V (these are LFP cells).

They also do not discharge and charge at the same rate (Amps flowing to/from older batteries are always higher compared to same paramater on never battery). BAK confirmed that this is due to change in cell design and manufacturing which the results in internal resistance differences. 

The never battery always gets to the 100% SOC before the older two (normally the older two would be at 90% SOC at the ame time). At this stage the BMS on the never battery disconnects the cells from the charge circuit, normally it would balance the cells and reach its resting voltage of about 53.2V an hour or two later. The two older batteries would continue to charge irrespective of what the never battery is doing. The sequence of events is lower the same (get to 100% when one cell reaches 3.65V, BMS disconnects the carging circuit, cells balance and resting voltage is about 53.4V on the older two).

As a result of the different resting voltages and the divergent internal resistance the two older batteries also start discharging earlier compared to the never battery (older batteries will provide full load and new battery zero load until the battery voltage are align).

The older two has done +- 1300 cycles and the never one +- 1100. One cycle is counted for each 4200W discharged (approximately 80%). Cycle count therefore indirectly ignores charge cycles. 

Changing the battery sequence in the pack or the sequence of the wireing has no impact on the above behavior.

 

11 hours ago, Moffat said:

Might be worth investing in a decent test meter to actually also check voltages individually and across the packs. Worrying bit is if you have a sudden high induction continuous load, wonder how this would affect the system.

I did! Recently invested in a DC clamp multimeter. That allowed me to verify the current readings on the packs. It confirmed the current readings of the 4th packs BMS while its SoC readings are havoc. What individual voltages of the packs are concerned, as long as they are interconnected with 25mm² no differences are measurable with common multimeters.

I do not understand your concern about induction load. In DC circuits there is no such thing. On AC it's an inverter matter.

11 minutes ago, I84RiS said:

get to 100% when one cell reaches 3.65V,

Thank you for your post. It largely confirms my observations. 100% SoC readings concerned: I observed that the packs BMS report 100% SoC when the voltage has not yet reached bulk or floating level and charge continues. I suspect that the BMS calculate SoC with the in and out Ah metering but neglect the internal losses.

Further more I observed that reaching floating charge level the packs equalize themselves. No intervention needed.

Yet another happening that shows weird BMS behavior.

Yesterday load shedding at 20h after a rainy fogy day. Batteries running low so the inverters shut down. The inverters setting is: low V shut down at 42V. Note that BMS supposed to shut down at 40.5V according to specs. However in the BMS parameter settings I find UV protect at 37.5V.

During shutdown I could still read the BMSs (battery powered laptop). The 3 first BMS read 42.5V at around 10% SoC. All showed UV alarm. The last pack with significantly lower Ri read 0% SoC at 47.5V. A flagrant contradiction! Obviously it had disconnected from the battery bus. It also signaled UV protect. At what V did it disconnect to recover to 47.5V ?

When power came back within a few seconds the system powered up in line mode and started to recharge the batteries from grid (all automatic restarts enabled). However the 4th pack remained disconnected and all the provided current from the inverters went to the 3 active packs. After those packs reached around 47.5V the 4th pack's BMS reconnected and joined the others with recharging.

Every thing is now back to normal and I did not have to intervene. However the event documents the weird behavior of the BMS in pack 4. So I basically ignore its SoC readings and rely on the pack voltage.

PS: I wonder what device inside the BMS capable of over 125A current (built in circuit breaker rating) is it that disconnects the pack?

 

9 hours ago, Beat said:

PS: I wonder what device inside the BMS capable of over 125A current (built in circuit breaker rating) is it that disconnects the pack?

 

Mosfets or IGBTs but more likely Mosfets. The true definition of "dynamite comes in small packages". Usually a few connected in parallel to share the total current being conducted. Pretty amazing the amount current they can handle for their physical size.

A BMS disregards voltage as a source of direct correlation to SoC as a percentage. Main reason is that the Lithium % to voltage chart is not linear. It's steep at the bottom of the voltage SoC charge close to 0% & gets steep from 92% odd to 100%. The majority of the operations is between 10% & 90% & the individual cell voltages range between 2.9v to 3.4v. So the voltage is just not a good source of what the exact capacity is. 

Most BMSs employ a shunt or current measuring hall effect sensor. Basically let's take a Victron smart shunt as a very good basic but very effective tool. The smart shunt doesn't care what battery you connect it to. All it does is counts Amps (Coulomb Counter). You setup your initial capacity ah & the smart shunt will count the amps flowing into during charge & amps flowing out during discharge. A very crude tool but makes for the most reliable way to get SoC. 

Now at some point there needs to be a starting point. So some BMS when you first turn them on, need a point of start reference & the manufacturer usually makes some initial assumptions. Like 2.7V is 0% 2.9v is 10% & so on. Because you need the shunt to start somewhere. Once a full charge up to voltage absorb & float the shunt will auto correct to 100% thereafter the calibration can start. Then a full discharge & there is now the first cycle where the Amp counting from a basic shunt will be most reliable. 

Now just imagine the complete setup where a pack of 4 batteries exists. Imagine the poor internal BMS shunts trying to figure out what the hell is going on. So some basics are to never pair batteries on different SoC levels. Rather get one to 100% or voltage full cut off. Then the other battery as well separately, then once this is done then combine them. Not to mention wire lengths etc. So there are so many variables that can upset the apple cart that it's not surprising that one BMS can be a tad forgiven for being lost in SoC time & space. We live in an imperfect world & an even more imperfect SA. 

11 hours ago, Steve87 said:

All it does is counts Amps (Coulomb Counter).

Thank you for your explications. As I presumed in an earlier post the BMS only meters Ah in and out but neglects battery internal losses. That's why it reaches 100% SoC before reaching bulk or floating voltage, the difference being the internal losses. That goes for the BMS that work somewhat normal, but my 4th BMS looks like going havoc.

In my reported event I wonder if the BMS disconnected when falsely reaching 0% SoC. There is no way the packs terminal voltage could be any different from the other packs. By the way, its current readings seam to be correct as I verified with a clamp Ammeter.

13 hours ago, TenMilSocket said:

Mosfets or IGBTs but more likely Mosfets.

How much are the junction voltage losses?

12 hours ago, Steve87 said:

Basically let's take a Victron smart shunt as a very good basic but very effective tool. The smart shunt doesn't care what battery you connect it to. All it does is counts Amps (Coulomb Counter). You setup your initial capacity ah & the smart shunt will count the amps flowing into during charge & amps flowing out during discharge. A very crude tool but makes for the most reliable way to get SoC. 

Yes i agree very effective tool even when using 2 x 12v batteries in series it will take mid-point deviation into account.