Green Power

On another forum I also posted about this and got the suggestion:
"52.5V is way too high for a float voltage.
Recommend you set:
Bulk: 51.75V
Float: 51.0V"
A post from Coulomb in January suggested to set Bulk at 52.5V and float at 51.8V although these were generic recommendations not specific to my current issue.
Here is the discussion I had with Pylontech (earlier, and not in reference to this specific issue):
PYLONTECH: "please set the following parameters that are on the battery side.........:
Bulk charge voltage 52.5V.
Float charge voltage 52.5V."
ME:"If float charging voltage of 52.5V is OK, then I guess a lower float voltage 51.0V or 52.0V might be OK for charging to less than 100%. But never >52.5V. Do you agree?"

PYLONTECH: "That's correct! Because sometimes the voltage sensors on the inverter terminals are not very accurate, the full battery voltage is around 53.2V(smaller than 54V). So 52.5V is a charge voltage value that has been limited in a safe range! You do not need to worry about this. "
Would be open to recommendations from others on float and bulk settings. e.g. @Coulomb@Beat@Donand others too.

 

Yes a electrician could do an insulation resistance test on the solar system using a normal Megger tester and provide you with an solution. This method will take time as individual solar panels needs to be tested isolated from the solar wires and mc4 connectors. Should all the panels pass the IR(>2MΩ, test, then all the wiring needs to be tested. 
A solar expert with a string maintenance tester could pin point your problem within a short period of time with 1 test and all components intact. My company have invested in this tester very very expensive but worth while in the long run. Maybe you can enquire of such a solar expert with a string maintenance tester. Below picture of the tester and connection diagram as how it pin points the fault.


 

Yes, that makes perfect sense. There will be leakage to earth somewhere in the panel wirings or one or more panels. This somehow pushes up the bus voltage, and the inverters stop with a fault code (often F08) to protect themselves.
The fact that the fault doesn't show during the day suggests to me that it's a condensation issue. You might be able to find the problem yourself merely by inspection, but there are lethal voltages up there. The problem is that these are not straight forward faults; it might take a clever solar expert to find it.
Definitely sort out what's on the roof first, because the next inverter will likely have the same problem.

The battery BMS board (tarjeta de control in Spanish) has been replaced and the battery returned to me and the system is now working again with two batteries
Since the error previously occurred occasionally in the past with gaps of days or weeks in between we won't know for sure if this was the correct solution to stop the error until some time has passed - maybe a month or two.
This also confirms that the guarantee's vaidity has been accepted in spite of using the battery with an inverter with no communication. I was not charged for the replacement.
 

At first glance I would say yes. It looks like the 2 terminals on each polarity are on the same copper bar. However I don't know how the main switches on the packs are wired up. Do they cut the terminals from the battery directly or via a relay? If they do, then there would be no charge nor discharge.
On my Averge packs there are 125A circuit breakers that cut the terminals from the batteries.

Sorry for the late reply; life.
I don't know much about Pylontechs, but yes, I think that's OK. The BMS will just not turn on the MOSFETs inside, so that the battery cells will be isolated (not connected to the inverter).
Of course, leaving it off for extended periods of time, even an hour or so, will cause it be be at a different SoC than the other one. It will be good to reconnect when the SoCs happen to match (within a percent or so) when you reconnect. Otherwise, you'll have issues with balancing. Maybe the BMSs can sort this out; I suspect not, but could easily be wrong.

Surely with the 1 battery faulty it does not matter if the faulty one goes to a lower SOC.
I think the wiring will still work OK with the faulty one is switched off.
The BMS in faulty unit should disconnect the battery when reaching the lower level.

Piper, did you ever have the battery beep/alarm and all lights flashing?
I once discussed this with Pylontech by email as I mentioned in the thread: I will paste below the email exchange.
PYLONTECH: "please set the following parameters that are on the battery side.........:
Bulk charge voltage 52.5V.
Float charge voltage 52.5V."
ME:"If float charging voltage of 52.5V is OK, then I guess a lower float voltage 51.0V or 52.0V might be OK for charging to less than 100%. But never >52.5V. Do you agree?"

PYLONTECH: "That's correct! Because sometimes the voltage sensors on the inverter terminals are not very accurate, the full battery voltage is around 53.2V(smaller than 54V). So 52.5V is a charge voltage value that has been limited in a safe range! You do not need to worry about this.

Here's my story with the 53,2 v setting.
 

Yes, exactly. So you have main firmware version 71.80, probably an Axpert MKS II by the looks of it. (There is also an Axpert King firmware with the same number, but the display looks totally different with the hexagons).
If I get time, I'll check that specific firmware; it's in my collection.

Sounds like you need the KettkeKomp™ feature of patched firmware. But it's not available for the Axpert MKS II.
None that I'm aware of. I sometimes run like that, though usually with 2 V difference.

@Green Power
Always glad to hear after all the chat from the time your lead acids were no longer up to it that all is working fine.
We are in a loadshedding 2 days from yesterday at 17h up to Wednesday morning for now if more generators don't pack up.
Enjoy your system going forward.

I actually have setting 12 at 47V. However if load shedding looms I set it to 48V and 13 to 50V.
There is no way to automate that setting 13 change in my knowledge. However theoretically it could be done with software. But as a matter of fact I leave it most of the time at 49V. So fare I had no negative experience with it.
I highly suggest setting 16 to CSO. With my suggested settings for 12 and 13 it works very well and will never charge from grid as long as PV power is present. Except of course if you empty the battery during the day with heavy load that cannot be covered with the PV power.
My understanding is that LFP batteries are less affected by deep discharge than lead batteries. The data sheet of my batteries guarantees 3500 100% discharge cycles. That would last almost 10 years with one 100% cycle per day. But with above settings it never reaches 0% SOC. Design live is 20 years. Therefor I do not worry about battery live saving. As a matter of fact I lately bought a Canon picture camera. Its Lithium battery came willingly discharged. That means to me that those batteries do not suffer from being left discharged.

I have also found my Axpert clone does not always switch at the set voltage but is does not bother me a lot.
Some settings you might have to see what works for your situation.
As was said the load of 900W can drop tth value rapidly to below your setting. One must measure it and not just assume a certain load should not drop the volts by what you would like it to be.

The difference between settings 12 and 29 has been well explained, so I don't have to repeat it. However you must consider this: BMS and inverters have each their own independent measuring systems. Those have naturally their inherent measuring inaccuracy, same as multimeters have. The range of their inaccuracy may be 0.5 to 1V. The display of more than 3 digits therefor is nonsense if already the third digit is doubtful. For you that means that setting of 29 just 0.1V above the value stated for your BMS cutout is most likely not gonna work even if your logic is correct. I suggest to take a margin of at least 1V.
Another concern is that battery voltage is also strongly dependent on battery current due to battery inherent resistance. In particular the discharge current lowers that voltage, the so called kettle effect. That means that under heavy load the battery voltage will reach the limits of settings 12 and 29 earlier than with little load.

Taking all what is said it seems 46V is what you should aim for. We know your inverter is know for not always measuring true values. Thus min as in inverter shut down set at 46V gives some margin to the absolute min of 44.5V.
The reason why you get a range of values provided for 48V lithium is Pylontech use 15 cells while most others use 16 cells. This then give you about 3V lower on the discharged side.
The 46V is then about in line with the value @Coulomb
mentioned. I have found that my Axpert would at times display the same Watt value and VA for inductive loads which is wrong. Just accept it.

Agree!  The datasheet of my LFP batteries says that the BMS cuts off at 40.5V (0% SOC). So I set the inverter low voltage cut off to 42V. "Back to utility bypass" at 47V, however 48V if load shedding looms. It worked well so far. It cut out once at night when load shedding came as a surprise and the geyser heating remained on. It happened 10 minutes before power came back.
LFP batteries seem to be less vulnerable to deep discharge than lead batteries.

There is an undocumented interaction between the battery low cutoff voltage setting and the warning going off. For unpatched firmware like yours, it will be 2.0 V. This is acceptable for lead-acid batteries, but makes little sense for LFP with their relatively flat voltage versus SoC curve. Hence "just not correct".
Since you have a BMS to save your battery anyway, you could also just reduce the low DC cutoff voltage setting until it goes off only at a sensible voltage. Then you can keep the beeper on in case of any actual warnings.

I registered mine just to get an extended warranty period, from 7 to 10 years if I remember well.
 

Thanks everyone. I checked the battery university article that shows 75%-25% beating 100%-50% in longevity and it has a source that leads us to this study:
https://www.researchgate.net/publication/303890624_Modeling_of_Lithium-Ion_Battery_Degradation_for_Cell_Life_Assessment
It states: "Section V presents a set of model parameters derived from LMO battery degradation test data". Within section 5 it states "Table I shows a set of model coefficients tuned using LMO battery degradation test data from the same manufacturer."
So it looks  like the data showing 75%-25% wins is for the LMO chemistry (lithium manganese oxide) rather than LFP.
 
I believe I may have been confused by the Western EV forums where they recommend to charge to 80% daily rather than 100% but I think many Western EVs are NMC rather than LFP.
I just managed to find this: "Here’s what the revised Model 3 owner’s manual for the US says in this respect.“If your vehicle is equipped with an LFP Battery, Tesla recommends that you keep your charge limit set to 100%, even for daily use, and that you also fully charge to 100% at least once per week.” Also "Another recommendation covers the instance where a Model 3 has been parked for longer than a week, in which case “Tesla recommends driving as you normally would and charge to 100% at your earliest convenience.” Source: https://insideevs.com/news/557527/tesla-model3-lfp-charging-recommendations/
So OK, I probably will allow the battery to charge to 100%. Seems like LFP is an ideal choice for home solar.
 
EDIT February 16th: I have now bought the batteries and preparing to install. Discussion continues here: https://powerforum.co.za/topic/11579-correct-inverter-settings/
 
EDIT May 16th: The battery already has a fault and they are going to send me a new BMS board. At the moment, it looks like they will respect the guarantee.
EDIT June 14th: They put in a new board and nothing was charged, so guarantee has been shown to be valid in this case.
Discussion continues here: https://powerforum.co.za/topic/12071-pylontech-high-voltage-alarm/
I was also able to register the battery online a while back and stated which inverter I was using, and got back a registration confirmation that seems to imply the warranty is valid.
 
 
 

The way these BMSs are designed, you should definitely be charging to 100% everyday. Although the cell life is higher when cycling 75-25%, that would already be built into the battery by over-provisioning. The SoH algorithm of the BMS likely also relies on the fact that the batteries get charged to 100%.

No. Balancing does not take place overnight. It can take up to 14 days to balance all the cells.
If the OEM states that you should achieve 4000 - 6000 cycles out of a battery, it is assumed you fully charge the battery to 100% SOC every day and it allows the cells to balance.
Why would anyone charge the battery between 25-75% every day and think that is good for a battery? I would love to see the response of the OEM when you return the battery for a warranty claim and show them your charts that you only charged the battery between 25-75% daily and only fully charged it to 100% one day every two weeks for balancing purposes. Don't expect a pat on the back for doing the right thing. 
It is your battery, you paid for it, you do what you think is best. I personally would not recommend it. 

Unfortunately, you can't reliably use voltage as a measure of SOC.
You could use say a Victron BMV to count coulombs, and give an accurate measure of SOC, but without charging to 100%, if will gradually drift off and become very inaccurate.

These specs are for EV batteries which use mostly NMC chemistry . LFP cells are not the same and can be cycled 100% to 10%SOC (90% DOD) daily without negatively affecting lifespan. 

No, the amps only determine how long it takes to charge your battery to reach your bulk charge voltage and then float charge voltage. The lower the amps, the longer it will take to charge. If you want to limit the SOC, you need to reduce the bulk charge voltage and float voltage, not the charging amps. The amps is the rate at which you charge the battery. You can charge a battery to 100% SOC at 2A, it might just take a few days.