LiFePO4 Batteries: Discussion Thread for Those Using Them as House Banks

[fallingeggs]

Originally Posted by SV THIRD DAY

Using the Old Lead is Dead charge terms of Bulk, Absorption, and Float for LiFePO4 batteries should be a crime punishable by having to scrub out the bilge because it traps people right back into the old LA way of thinking and it confuses people.

There is something to be said about keeping the terms around because we have to adapt to a world of charge controllers that use them (MC-614, Morningstar in my case). We are forced to program something into the Bulk, Absorption and Float phases after all. Use the phrase, but understand what it means for LiFePO4.



Sent from my iPhone using Cruisers Sailing Forum

[exMaggieDrum]

This may have been brought up in past posts but I don't recall seeing it.

I just read that ABYC is now recognizing the fact that more and more boats are going to "Lithium" batteries and they have set up a committee to look in to them and the safety aspects with the goal to promulgating "standards" for LFP on boats. This will be a positive move provided they put some common sense in it with real facts and analysis and not hear-say and fear-mongering.

It will take some away some of the actual stigma that owners who have LFP systems face when they sell their boats. And, perhaps it will result in some good safety "guidance" for LFP installations and usage. My first concern would be that they don't recognize that many DIY installations are very safe and capable, i.e. that only complete LFP batteries from known manufacturers with built in or "sold as a package" BMS systems, system monitoring, and fail-safe devices will meet their specs.

Remember, any who might want to get up in arms over any restrictions in your right to kill yourself how you see fit, everything that ABYC puts out are recommendations and are not laws and regulations. They may seem like that because many boat builders want to advertise that their boats "meet" ABYC specs. In truth there are very few boats that manage to meet the standards 100% in any case.

[JBPohle]

Top balancing and charging 8 CALB CA180's @ 180 ah batteries all 8 are in parallel and then I've got 7 single batteries in series with this large cell. This is so that I also create 24V nominal battery. Total of 15 batteries. Two experiments at one time. Trying out the CellLog 8S at 24V and also top balancing 8 of the batteries in parallel. Using a Volteq HY3030EX.

I've been charge the 8 batteries in parallel CC/CV 30.6 A, 3.6V off and on for a week or more. Only when I am on the boat. I live aboard so that is every evening at least. Batteries received from EVTV at 3.28V. I recorded the last 8 hours of charging with the CellLog and produced the graph below.

It is interesting because the charging rate is very small, 30.6A/8 batteries, so only 3.825A per battery. Very low charge rate and voltage held steady at 3.35V forever. Then you can see a clear increase in voltage and the amps also began falling off. I discontinued charging at 2.25A/Battery and 3.519V. It is pretty clear to me that we have reached the knee. Now we'll leave these parallel for a while and bring the other 8 batteries up to a similar voltage, then parallel all of them for while.

I know this is not the most efficient way to do this time wise, but it is a learning process anyway. For the other 8 batteries we'll charge in series for a little while to put the initial amps in quicker. 30 amps an hour and we'll set the power supply for something on the order 27.2V or 3.4 volts per battery for the CV portion. The CellLog will give me some idea how things are going and we'll then top up the last of it with cells in parallel similar to the first 8. Then parallel all the cells and see what we stabilize at.

On thing is abundantly clear now, charging at very low rates 2A requires VERY low voltages... (Mainesail has mentioned this too) don't know how low yet, but I'll probably go to 3.25 or 3.3V per cell for a float voltage.

Just posted this for anyone who is interested... no questions or tears today

[JayH]

Originally Posted by JBPohle

Top balancing and charging 8 CALB CA180's @ 180 ah batteries all 8 are in parallel and then I've got 7 single batteries in series with this large cell. This is so that I also create 24V nominal battery. Total of 15 batteries.

John,

Perhaps I've misunderstood, but by your description and pictures it appears to me that you're trying to build a 24V pack using 8 cells in parallel, and then 7 more cells in series, for a total of 15 cells.

That isn't going to work. You can't mismatch the capacity of the cells in series that way. If one cell in the string is 8 times stronger than the others, your pack is inherently unbalanced. You will very quickly overcharge the 8, or over discharge the 7.

For 360Ah@24V, you need 16 cells in 2P8S (two cells in a parallel group, 8 groups in series). With 15 cells, you have enough for only single cell groups, which is 180Ah@24V in 1P82.

Apologies if I misunderstood your post. Just trying to save you an expensive mistake.

[JBPohle]

JayH

Thanks for taking the time write this caution. This was just an experiment. Cell log and BMV needed more than 3.4 volts and I wanted get some graphs and check calibration of all of them. Don't worry this is not the final configuration. As a matter fact last night all were paralleled and we're going to let them equalize for the next week, topping them up in the evening. Final configuration will be 2P8S.


Sent from my iPhone using Cruisers Sailing Forum

[OceanSeaSpray]

Originally Posted by JBPohle

As a matter fact last night all were paralleled and we're going to let them equalize for the next week, topping them up in the evening.

Leaving cells in parallel for ages believing they are going to equalise is a misconception. They don't. Almost immediately after paralleling, there is no voltage difference left to drive any current and whatever imbalance there may be just remains.
All it does is harm the cells by having them sitting around near 100% SOC doing nothing.

They need to be charged in parallel at low current at the end, so the current can go to the lower SOC cells first as they are all coming up together.
It takes some voltage to begin moving the ions and once this starts happening with all cells seeing exactly the same voltage, then they can balance in relation with each other.

It is also very important to have good connections, proceed with low current only to minimise voltage drop effects and - for large numbers of cells - charge symmetrically from the opposite ends of the + and - rails so the current travels the same distance through the connections to get to each cell.

If the cells have been pre-charged up to the point where balance problems start showing, it is usually the matter of an hour or so to balance in parallel and finish the job from there.

[ebaugh]

I finally completed (mostly) the BMS project I've mentioned here long ago. It's been installed now for a couple weeks and looks like it will work out well.

It's based on a Triangle Research Nano 10 PLC and HMI Edit. All of the logic is in BASIC code installed on the PLC. HMI Edit is strictly a user interface running on IOS, collecting data via MODBUS TCP and isn't required for essential functions. The PLC uses external devices on a RS485 bus for sensing analog voltage readings and controlling external relays.

I attached a picture of the installation and some screen prints from the User Interface. If you look closely at the RELAYS user interface page, the primary BMS logic may become clear. Each relay is ON if all conditions are met, OFF if not. Each relay can have its own set of values.

What HMI Edit brings to the table is a very powerful analytical tool. It constantly logs string voltage, bank current, bank A/H and SOC. All of these are read from registers in the PLC. A/H looks pretty accurate, it simply samples current every second, and applies that to a counter. It's a coulomb counter. SOC is an attempt to track bank charge state solely on voltage. This is not acceptably accurate yet, and I doubt it ever will be.

But what HMI does that's pretty neat is the attached history screen. You can scroll and zoom in or out on any desired timeframe. 99% of the programming for this is part of HMI Edit, you basically just have to layout the screens. You do need to dedicate an IOS device. Probably a iPad mini would be optimal, but I'm using an extra full size iPad I had laying around.

The component costs are roughly:

PLC $230 (includes $80 software charge)
ADC $100
Relay Box $60
HMI App $150 (ouch)
Current Sensor $100 (Hall Effect)

Call it $700 plus an IOS device and lots of time....

If someone wants to duplicate this, it won't be plug and play, but I will provide my HMI and PLC programs as samples. Along with a wiring schematic for my installation.

It is probably possible to upgrade the PLC from a Nano 10 to a FMD88 and eliminate both the ADC and Relay Box. Saves some and reduces footprint. However you go from 16 bit resolution on the ADC to 12 bit, and all voltage references must be to bank B- requiring voltage dividing resistors, further reducing accuracy. The relay box has firmware that turns all relays OFF if communications with the PLC stop. This is an integral fail safe part of the design in case the PLC fails or freezes.

Not counting the IOS device charger, this BMS system draws about 450 mA, higher than I'd like. In the event of a low voltage event with no one on the boat, it does completely shutdown after a period of time and requires a manual restart to get back online. But that eliminates all current draw from the bank.


Attachment 125229

Attachment 125226Attachment 125227Attachment 125228


Sent from my iPad using Cruisers Sailing Forum

[ebaugh]

GBS Cells after 4 years
-----------------------

In June/July 2012 I installed 48 100Ah GBS cells as our primary house bank. Ordered in the US, delivered by container ship to Grenada for install.

I've posted before that we are seeing about a 10% per year degradation in capacity. Unfortunately, that still seems about right, but the data collected isn't completely solid. Especially the baseline, which I don't have at all. The original 1200 Ah is estimated at 700-800 today. My gut instinct is the rate of degradation has slowed. A little.

Further, we've had one cell failure. This was noted by an imbalance at top of charge. The first time this happened, it was a failure of a BMS cell board. One of the 4 boards suddenly tripled the parasitic current use dragging down a string. I assumed it was the same thing, so that's what got me to finish the BMS I just posted about. However, that wasn't the problem. One cell developed an internal problem and was dragging down the string.

Worthy of note, two untouched strings remain in perfect balance after 4 years. The other good string had 4 cells removed for testing off the boat, I have no reason to suspect they would be different if left in the bank.

Since September of 2013, this bank has been at dock with shore power 99% of the time. The charging regime a bit odd due to the logic available in my Magnum charger. They charge to 13.6V, about 50-60% SOC targeted, then discharge to 12.7-12.8V where the charger kicks in and takes them back to 13.6V.




Sent from my iPad using Cruisers Sailing Forum

[toddedger]

[QUOTE=ebaugh;2132745]GBS Cells after 4 years
-----------------------

In June/July 2012 I installed 48 100Ah GBS cells as our primary house bank. Ordered in the US, delivered by container ship to Grenada for install.

I've posted before that we are seeing about a 10% per year degradation in capacity. Unfortunately, that still seems about right, but the data collected isn't completely solid. Especially the baseline, which I don't have at all. The original 1200 Ah is estimated at 700-800 today. My gut instinct is the rate of degradation has slowed. A little.

Further, we've had one cell failure. This was noted by an imbalance at top of charge. The first time this happened, it was a failure of a BMS cell board. One of the 4 boards suddenly tripled the parasitic current use dragging down a string. I assumed it was the same thing, so that's what got me to finish the BMS I just posted about. However, that wasn't the problem. One cell developed an internal problem and was dragging down the string.

Worthy of note, two untouched strings remain in perfect balance after 4 years. The other good string had 4 cells removed for testing off the boat, I have no reason to suspect they would be different if left in the bank.

Since September of 2013, this bank has been at dock with shore power 99% of the time. The charging regime a bit odd due to the logic available in my Magnum charger. They charge to 13.6V, about 50-60% SOC targeted, then discharge to 12.7-12.8V where the charger kicks in and takes them back to 13.6V.




I'm curious, why the 5 degree difference in bank temp? I have heard about simular GBS capacity loss due to operating in high temperature environment. When changing location (motor home) and lowering charging voltage, capacity loss was greatly curtailed. Are your cells located in the engineroom? What kind of cycling numbers would you estimate your bank has, and what is the max charging voltage that you have used?

[ebaugh]

[QUOTE=toddedger;2132989]

Originally Posted by ebaugh

GBS Cells after 4 years

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


I'm curious, why the 5 degree difference in bank temp? I have heard about simular GBS capacity loss due to operating in high temperature environment. When changing location (motor home) and lowering charging voltage, capacity loss was greatly curtailed. Are your cells located in the engineroom? What kind of cycling numbers would you estimate your bank has, and what is the max charging voltage that you have used?

There is some instrumentation/calibration error in that 5 degree difference. There are 3 different thermistor arrangements. Two go direct to PLC analog inputs, three to the external ADC. On the ADC, on is a 100K Thermistor, the others are 10K with additional balance resistors. No attempt to calibrate these yet. 4 are on ring terminals one inside the balancing resistor coil.

The bank is in the engine room. Normal temps are in the 80's, short running days <110, long running days <120.

Off grid, we use 600 Ah a day. Usually charge 2x a day. From installation in 2012, the next 13 months were full time cruising, the ABC islands 1/2 at anchor, Colombia all at marina, Panama all at anchor, San Andres all at anchor, Roatan and Mexico (at marina), then back home. There were probably 400 cycles year 1. Since then maybe 50 a year. So call it 550 cycles since new.

Outside of checking for balance, they were never charged beyond 3.4V per cell. Checking for balance, we have been to 3.55-3.6V maybe a couple dozen times, and never held there longer than a few minutes.


Sent from my iPad using Cruisers Sailing Forum

[toddedger]

Ebaugh, T1 Terry made a post on memory effect loss of capacity from continually undercharging that you might find interesting. Page 315 post #4720

[nixsee]

There has been a lot of discussion lately (and throughout the entire thread history) about stages of charging, cutoffs, full, empty, etc... Regardless of charging stage, current tapering, etc..., it seems fairly unanimous that a reasonable HVC is around 3.45-3.5 VPC, which should get you at least 90% full.

However, there is a lot more discrepancy about LVC. People talk about 80% DOD, though many manufacturers Specs (like my CALB CA cells) say 10-90% is the safe range. Regardless, we know that it is hard to measure SOC even with the limited Peukert effect, so we (and most BMS) tend to rely on voltage. I know that some people will discharge to 2.5-2.6VPC in a capacity test (just as we charge to 3.6-3.8), but what VPC is considered a reasonable LVC for regular discharging? 2.9VPC? Is that too into the knee? Too conservative?

One related question is on "cycles". What constitutes a "cycle"? Our charging is somewhat erratic given solar availability, alternator usage etc..., so the SOC doesn't just go from full to empty as an Electric car would. But is there an "ideal" charging profile? If you were to simply cycle between ~60-80% SOC (with the odd charge to top it off for "memory"s sake), rather than a wider ranging, but still conservative 30-90%, would this improve your cells' lifespan? Or should we try our best to mimic an electric car's full to empty cycle - charging to full (3.45?) and not charging again until empty (2.9?)

[OceanSeaSpray]

Originally Posted by toddedger

Ebaugh, T1 Terry made a post on memory effect loss of capacity from continually undercharging that you might find interesting. Page 315 post #4720

Those were also my thoughts, thanks for looking back and finding the post...

3.4V is way too low an absorption voltage, it just reaches the reaction voltage for LiFePO4.
A lot of confusion was created with charging voltage as people were trying to find a voltage they could charge at "indefinitely". This doesn't exist, unless the battery simply doesn't charge properly any more and it led to suggesting lower and lower values.

The big deal is not about the voltage, it is about terminating the charge. Bob has all the data he needs to do that in his PLC, it is just a matter of reprogramming and putting the PLC in control of the charging, rather than scattered little chargers all with a mind of their own and no idea about what they are really doing.

[ebaugh]

Originally Posted by OceanSeaSpray

Those were also my thoughts, thanks for looking back and finding the post...

3.4V is way too low an absorption voltage, it just reaches the reaction voltage for LiFePO4.
A lot of confusion was created with charging voltage as people were trying to find a voltage they could charge at "indefinitely". This doesn't exist, unless the battery simply doesn't charge properly any more and it led to suggesting lower and lower values.

The big deal is not about the voltage, it is about terminating the charge. Bob has all the data he needs to do that in his PLC, it is just a matter of reprogramming and putting the PLC in control of the charging, rather than scattered little chargers all with a mind of their own and no idea about what they are really doing.

Yes, thanks! The old post interesting, but don't know yet if I can "retrieve" capacity by cycling. I think there is a "formation" charge at the end of cell manufacture. This some number >3.6V. I wonder if after a period of time this should be repeated? While rebalancing this bank, I did observe far more charging going on in the 3.5V range than on install. When new, at 3.45-3.46V it was done, maybe another 20-30Ah to 3.6. When rebalancing it was easily 100Ah, possibly as much as 200Ah, I didn't watch that close.

I attended a several day conference in Ft Lauderdale on marine electrical propulsion, mostly commercial interests using other lithium chemistries. But several of the battery guys suggested several "full" cycles, including an engineer from Saft, the only LFP company there to recover capacity. The Europeans are promoting Lithium Titanate, a subject worthy of further discussion.

Attached is the first "full" discharge cycle over a several day period. And then a zoom in on the "charge". String/Cell 2 is low, having only 11 of 12 cells connected. And now that I have the visual, we have been in fact living the last 2 1/2 years at dock in the 10-30% SOC window, lower than I thought.

Without a different charger (or a new board in the Magnum), I can't change the 12.7 to 12.8V where the charger switches from "FULL" charge back to Float. I will raise the Float setting to get the high end SOC higher. But how high? This cycle is non-configurable logic by the charger to not overcharge FLA.

I have to think more about PLC/BMS control of the charger. I always liked the idea it didn't really "control" anything, but was simply the fail safe monitor. Putting the BMS in control creates a single point of failure since to actually control the process, it would need the ability to charge to a pretty high voltage. You could add a CC/CV charger limited to 14.4V which normally wouldn't be an issue, but could be if control was lost.

But let's say I can do exactly what I want....what is that? Is it different at anchor than at dock?

One of the original advantages of LFP over FLA was you could keep it anywhere between 10-90% SOC and then shift your worries to your next cold beer....

Attachment 125354Attachment 125355


Sent from my iPad using Cruisers Sailing Forum

[Maine Sail]

Originally Posted by ebaugh

But several of the battery guys suggested several "full" cycles, including an engineer from Saft, the only LFP company there to recover capacity. The Europeans are promoting Lithium Titanate, a subject worthy of further discussion.

[/ATTACH]


Sent from my iPad using Cruisers Sailing Forum

Bob,

I recently ran this test on my own bank out of curiosity.

For this test I chose to use an intervalometer (see video below) to actually make a movie of the discharge. The first frame is the charge end point of 13.8V and 7.5A flowing into the bank. The charger then turns off and the constant load discharge tester turns on. The bank was discharged at 30A constant current and I have been using this tester for its repeatability. At a .5C or 1C load the Ah capacity would be slightly less but with the low Peukert of LFP not by a huge margin.

As mentioned Terry suggested that LFP banks can develop a situation where you may not see all the true capacity after repeated light discharges.

I had also noted this in an over-discharged Mastervolt marine battery where I was able to see continual capacity increase in each cell as I kept cycling them to 2.7V then back to 100%. I decided to test this theory on my own bank by conducting multiple 100% discharges followed by 100% recharges.

This test was conducted at cycle # 772 and done at a .075C discharge rate (nowhere near .5C or 1C). This rate is limited by my constant load dischargers ability (it can do 40A but I don't like to push it at that for 10+ hours).

Here's the video: (if you click for full screen you can more clearly see the capacity testers screen.)

VIDEO: Go full screen to see the discharger data:
https://youtu.be/fGBmEh72UlY

These cells have:

--Never been floated, they get charged, then discharged
--Only absorbed to a net 8A - 10A at 13.8V -14.0V
--Not charged above 14.0V unless for testing purposes (I now have a few other banks for that)
--Max charge rate at approx .3C
--Stored at 50% SOC when not being used or cycled
--Stored in 45-60F temps when not being used or cycled
--Only very rarely exceeded 80F
--Highest voltage they have ever seen was 3.8VPC while top balancing initially.

On these last few capacity tests I ran a total of seven 100% DOD tests back to back and noted that I gained nearly 15.3 Ah's of capacity from test #1 to test #7. I meant to do five tests but the video quality messed with me a few times and I am still not entirely happy with it. This resulted in two extra tests. There certainly seems to be some truth to what Terry had been suggesting about light loads diminishing/masking some of the real true capacity.

My last published capacity test was at cycle #550 (I generally conduct them every 50 cycles) and it delivered 419.2Ah. Most tests since I began using the constant load tester have been within a few Ah's of this 419.2Ah cycle 550 mark. These however have been just 1 capacity test and to a lower cut off voltage of 2.8 volts per cell. I now use 2.9 volts per cell because I noted very little capacity below 2.9V and voltage on my 2009 Thundersky cells starts to drop off rapidly below this (lower knee).

I suspect I did not see as much gain with the back to back 100% DOD cycles as Terry had because I am normally discharging this bank to 80% DOD & then recharging at .23 - .3C to less than a 10A tail current.

With a total of seven back to back Ah capacity tests at a .075C discharge rate I did see an increase of about 15.3 Ah before the Ah capacity stabilized and stopped increasing. What does this mean? I don't know, but I did see an increase of 15.3 Ah by cycling this bank from 100% to 0% seven times back to back.

This LFP bank is an n=1.... In science an n=1 is not sufficient data. Cells vary and every boater will treat them differently. I have customers with over 12 years on GEL batteries (one at 16 years) and others who murder them in 2 years. I know of a number of owners who've murdered LiFePO4 banks. I have a dead & severely swollen $10,000.00 factory made "Marine LiFePO4 battery" in my shop as evidence of that (regularly charged at 3.65VPC).

I am treating this LFP bank how I suspect & have surmised they may perform well with and it seems to be working. An n=1 really tells us little in the whole scheme though. I really do wish more LFP owners had a constant discharger so baselines > forward could be better established.

The frustrating part is that very few of us in the LFP community are actually capacity testing, with repeatable equipment (even if .3C to 1C can't be attained), to know what is actually happening with our LFP banks as they age. Until more fractional C data comes out, it's still a learn it as we go game...