304Ah LifePO Update - Year 2

[geem]

What's fully charged though? And how much difference does "almost" make to lifespan? Offgrid garage has loads of test data,

View attachment 174260

I go up to 13.7v then to 13.35v float with no absorption & from todays data & offgrid garage it's certainly not 100% but well up there. Will be interesting to see how repeatable this is & if pushing some charge into one cell at a time can balance them a bit closer without turning everything off. >

View attachment 174261



From forums I suspect a lot of people get freaked out with active balancing & keep the voltage high with timed absorption to let that kick in every day where it may be be better to keep the voltage lower as much as possible most of the time, just balance now & then. My cells were balanced before installing over a year ago. Bottom line on boats I reckon is nobody really knows. Far too early.

The info I quoted was from scholarly articles so about as good as it gets, but there are so many variables. We have our 100% SOC set to 28v. We don't often get to 28v. When we do, the balancer has very little to do. All 8 cells are typically 4mV apart. The last time we were at 100% was a week ago. It's typically every week or two.
I just helped a friend build 3x280Ah 12v batteries. We bought grade A EVE cells from the USA. JK BMS on each with active balancers. We didn't bother top balancing. The first charge got all three batteries balanced with a 30 minute absorption time. 1 week later and the batteries are all sat with no more than 5mV deviation.
$105US per cell for grade A is pretty good

 

[GHA]

The info I quoted was from scholarly articles so about as good as it gets, but there are so many variables. We have our 100% SOC set to 28v. We don't often get to 28v. When we do, the balancer has very little to do. All 8 cells are typically 4mV apart. The last time we were at 100% was a week ago. It's typically every week or two.
I just helped a friend build 3x280Ah 12v batteries. We bought grade A EVE cells from the USA. JK BMS on each with active balancers. We didn't bother top balancing. The first charge got all three batteries balanced with a 30 minute absorption time. 1 week later and the batteries are all sat with no more than 5mV deviation.
$105US per cell for grade A is pretty good

What did the published papers take as 100% charged though? Datasheet for mine says "0.5 C constant current charge to 3.65 V for cell, then switch to constant voltage charge until charge current declines to ≤0.05C" That´'s sitting at 14.6v til the current drops. Not many boats gonna do that.
Point is data from tests in laboratory conditions doesn't say much about batts on a boat.. We just don't know, from Rod collins it looks like they'll go for a very long time..
From staring at the accurate data for so many hours I also don't think it's actually possible to nail out of balance cells down to single figure mV. Just measuring cell voltage that accurately is near impossible at such resolution outside a laboratory as ohms law is enough to skew the voltages as current varies just with the tiny resistance between the terminal & the battery. The real universe isn't as bang on exact as we like to think it is.
Not that it matters. Close enough is fine.

 

[geem]

What did the published papers take as 100% charged though? Datasheet for mine says "0.5 C constant current charge to 3.65 V for cell, then switch to constant voltage charge until charge current declines to ≤0.05C" That´'s sitting at 14.6v til the current drops. Not many boats gonna do that.
Point is data from tests in laboratory conditions doesn't say much about batts on a boat.. We just don't know, from Rod collins it looks like they'll go for a very long time..
From staring at the accurate data for so many hours I also don't think it's actually possible to nail out of balance cells down to single figure mV. Just measuring cell voltage that accurately is near impossible at such resolution outside a laboratory as ohms law is enough to skew the voltages as current varies just with the tiny resistance between the terminal & the battery. The real universe isn't as bang on exact as we like to think it is.
Not that it matters. Close enough is fine.

I don't think true 100% is that important. The difference in charge as soon as you climb the knuckle part of the curve is small. Stress to the battery occurs here. Leaving cells in this zone will stress them

 

[GHA]

I don't think true 100% is that important. The difference in charge as soon as you climb the knuckle part of the curve is small. Stress to the battery occurs here. Leaving cells in this zone will stress them

Agree, the offgrid garage data shows this. Thus the thought above that so many people desperate to stay up there to balance every day may actually do more harm than good.. Wonder what Rod Collins charge settings were. 100% capacity after 10 years use is worth a look.

 

[gregcope]

I balanced it all on assembly, two years ago.

Since then I have not balanced it.

I might do a capacity test at end of the season to see what capacity I have lost. Might then rebalance then…

A wider thought is how frequent do you need to rebalance, especially if you see no evidence of unbalanced cells? Ie why do it frequently?

 

[geem]

I balanced it all on assembly, two years ago.

Since then I have not balanced it.

I might do a capacity test at end of the season to see what capacity I have lost. Might then rebalance then…

A wider thought is how frequent do you need to rebalance, especially if you see no evidence of unbalanced cells? Ie why do it frequently?

The only reason to balance is if there is a risk the BMS would should down the battery, as you approach full charge, due to a single cell going above the threshold voltage limit. If you have the maximum chaege voltage set low, say 13.6v on a battery that is seeing low to zero load, then the risk of a high voltage event, HVE is extremely low. If you are using large loads and working the battery through a good range of SOC regularly, then there is a higher risk of a HVE. This then makes regular balancing a good idea. How often they need balancing may depend on the quality of the cells.
We do push our batteries fairly hard but being grade A cells, we see very little voltage different across the cells. Even so, we balance every couple of weeks. I did go a month without balancing and the cells were at 8mV difference once I hit 3.5v per cell. Really very good

 

[vas]

never bother going up to 3.5V per cell
balancing starts at 3.45V, DIYBMS has passive balancing and I've configured the system to drop to just under 2A so that even after a 5month on the 40-60%SoC took it two goes to bring the first cell (always trailling) up to par with the others. Throughout use delta is usually between 4-8mV (grade B EVE cells same as gregs)
Even with a 2kW watermaker discharge is way under 0.5C, charging even less so with engine running. So not stressed in that respect

 

[GHA]

A wider thought is how frequent do you need to rebalance, especially if you see no evidence of unbalanced cells? Ie why do it frequently?

Personally if you are looking a plotted cell data then your call really. I really think we are fooling ourselves if we think we can measure out of balance cells down to single figure mV's. The cells themselves aren't sure Getting excited cos the difference is down to 0.005v difference doesn't tell you much, it will be different later, 2 cells today went from 30mV different to exactly the same after going to float, and you can track back to where the current was the same & cell voltage can be very different. Cells 3 & 4 >





Though it does seem plenty consistent enough to be useful, it's change that's interesting even if the exact figures drift. . This guy thinks the same >>
Manual cell balancing

Looking for drift I reckon is more important than just balancing willy nilly, can flag up if some more tiny resistance has built up in the connectors as well, just with the smartshunt that spotted a terminal needing a slight tighten just with the voltage drop being more than it was the week before under heavy loads with a compressor running. Would never have been spotted otherwise, quarter turn made all the difference
Manual balancing next, just to see what happens as much as anything

 

[gregcope]

@GHA remind me how you do that cell level logging? You just have written that up somewhere?

 

[geem]

never bother going up to 3.5V per cell
balancing starts at 3.45V, DIYBMS has passive balancing and I've configured the system to drop to just under 2A so that even after a 5month on the 40-60%SoC took it two goes to bring the first cell (always trailling) up to par with the others. Throughout use delta is usually between 4-8mV (grade B EVE cells same as gregs)
Even with a 2kW watermaker discharge is way under 0.5C, charging even less so with engine running. So not stressed in that respect

I also start balancing at 3.45v/cell. How long is your absorbtion period? I just dropped my absorbtion time to 15mins. Since that is plenty with the active balancers

 

[geem]

Looking at the curve for lifepo4 for battery voltage with a 0.5C charge current, you dont actually hit the knuckle at 3.5v/cell.
Our typical charge current is very much smaller than 0.5C.
The maximum I can charge at with all 3 battery chargers running is 0.3C. This never happens anywhere near the top of the charge curve. Most charging is done at less than 20A at 24v so rarely more than 0.05C on our 385Ah bank at 24v.
Stressing of the cells starts to occur above 3.5v/cell. A long duration above 3.5v on a regular period would have an effect but to be honest, it's pretty low impact in terms of calender aging as its not happening often

 

[GHA]

@GHA remind me how you do that cell level logging? You just have written that up somewhere?

Ended up getting a bit complex, basically 4 op amps measure the voltage of each cell based on something like this --



Then a little ADS1115 board which is a very accurate 4 channel voltage sensor measure the voltage out of each opamp then tells an ESP8266 this interesting info, the the esp talks to a raspberry pi over wifi & sends the data there, and the Pi does all the work after that. Signalk in the Pi does the heavy lifting, no programming required apart from the ESP. That was a load of hours on google & youtube figuring out how opamps work Luckily there's online simulators to blow stuff up virtually these days

So handy being able to cut & paste images straight into a post now.

One day it might make it to jlcpcb to make little boards up with everything surface mounted for prob less than a tenner each. Which would be cool!

 

[GHA]

Looking at the curve for lifepo4 for battery voltage with a 0.5C charge current, you dont actually hit the knuckle at 3.5v/cell.
Our typical charge current is very much smaller than 0.5C.

That's a very fat line
Mine hit the knee at a fraction under 13.6v, day after day. 3.4v per cell. Max about 15A solar into 300Ah.
Never stays there for long though.

 

[geem]

That's a very fat line
Mine hit the knee at a fraction under 13.6v, day after day. 3.4v per cell. Max about 15A solar into 300Ah.
Never stays there for long though.

View attachment 174327

I don't see the knee. Your slow rising voltage isn't in the knee part of the curve where the voltage goes up fast. Am I missing something?
If you carried on charging to 3.65V/cell I am sure you would see a far more rapid rise in voltage.
On the curve I posted, 3.5v is at about 80% charged.
You have the same charge rate as me 0.05C

 

[vas]

in the above graph you see all 8 cell values and you may ignore the top two lines (yellowish and green) which are the shunt and bank voltage respectively (bank is calculated before the shunt, there's always a small discrepancy, don't really care...)


the second graph is SOC and amps, you can see that it drops to around 1A for the time that it's passive balancing which I have to cut down as currently it's over 2h.

that was yesterday's balancing second proper one after 5 months. Now I've configured the BMS to not charge before going down to 40% SoC and will add another limit on charging to 60% so for the next few months till I use the boat again it will be between 40-60 (could get it lower as well, may do as the weather gets better and there's lots of solar.
The two drops in Amps at 10 and 10:15 was obvs some clouds over the panels. More clouds came when balancing but 600W solar could provide 1Amp no problems
Greg that's grafana with data coming from the diyBMS. More data coming from VenusOS to the same influxDB in my office and they are all presented together.
GHA, I'm sending 30s as I don't want to spend too much data (not that it makes any sense actually. I'm not keeping the db onboard, poor rpi3B+ has enough of a job running and managing the integration of Victron inverter/mppt and the lifepo4 BMS.

cheers

V.

 

[vas]

I don't see the knee. Your slow rising voltage isn't in the knee part of the curve where the voltage goes up fast. Am I missing something?
If you carried on charging to 3.65V/cell I am sure you would see a far more rapid rise in voltage.
On the curve I posted, 3.5v is at about 80% charged.
You have the same charge rate as me 0.05C

agree, in my graphs above you can see the knee over 3.4V per cell. GHA keeps it to the first smoother voltage lift to 3.4. TBH, if it's OK it's not like you're getting much more out of the batteries going from 3.4 to 3.45. In my graphs above that is done within 1/2h and it gives circa 2% SoC, from 98 to 100 (REPORTED!) obvs not actual, but again obvs don't care, doesn't make any difference.

V.

 

[GHA]

I don't see the knee. Your slow rising voltage isn't in the knee part of the curve where the voltage goes up fast. Am I missing something?

Yep, short timescale & a load of noise from solar & looks like laptop was on. It's the start of the knee for sure, just stops earlier than most you`ll see.

 

[GHA]

Now I've configured the BMS to not charge before going down to 40% SoC and will add another limit on charging to 60% so for the next few months till I use the boat again it will be between 40-60 (could get it lower as well, may do as the weather gets better and there's lots of solar.

I fiddled for ages to try & get the smartshunt as close to reading accurately each cycle without resetting soc & left it a tiny bit pessimistic so when it hit the same place if the knee each day it would loose maybe 0.1%. When it finally got back up to the knee a week or so ago it read think it was just shy of 60% instead of 95% which I call full. Still tweaking efficiency & peukerts but can't get it bang on, always drifts a tiny bit. Not surprising really. If you aren't drawing much prob not an issue though, doubt if you'll be cycling much with the boat asleep?