LiFePO4 Capacity vs Temperature

Horsefly

Photon Sorcerer
Joined
Dec 12, 2020
Messages
1,439
Location
Denver, mostly
I'm wondering how much cell capacity is affected by cells being just a bit cooler. Here's my story:

In early Aug-2021, I received 8 Eve 230Ah cells, purchased via Michael B Caro (I was on one the lucky ones). I purchased one of the EBC-A40L testers (this one) and went about testing each cell. The tests were done in our basement, and although it was still pretty hot outside, the A/C kept the basement pretty cool (too cool for my wife's taste) at around 65°F-68°F. Of the 8 cells, one tested out at 237Ah, and the rest all tested at 238Ah or 239Ah. Obviously I was pleased.

Last week I received 8 Eve 280Ah cells, purchased from the famous @Amy Wan (Luyuan) at Shenzhen Luyuan. These were the new Eve LF280k cells that are supposed to have up to 6,000 cycles. Amy provides a report as the cells are shipped, and the report includes a cell capacity for each cell, although I know nothing about how the capacity is determined. The report indicated the capacities were 290.3, 292.0, 290.7, 288.5, 290.8, 289.5, 291.7, and 291.2. All good, with an average capacity of 290.6.

I used the same tester to do the capacity testing. I set it up for each cell with a multi-step program, so it would charge the cell to 3.65V at 25A, stopping the charge when the current dropped to 2.0A. Then the second step would start, discharging at 25A until the voltage reached 2.5V. I'm pretty sure this was the same process I used for the 230Ah cells I tested in the summer.

The test results were somewhat disappointing: 278.5, 284.0, 282.0, 275.7, 278.7, 281.8, 282.5, 283.5, 280.8 That's three cells that didn't even meet the 280Ah spec, and an average of only 280.8Ah, less than Amy's report by 9.8Ah.

I'll admit the results are not horrible, and I'm not likely to raise a stink. The person I'm building the two 4S packs for is perfectly happy with them. However, it is disappointing, and - given Amy's reputation here - pretty surprising. So I tried to figure out what could have somehow biased the test. It was the same tester and the same test methodology, so I don't think that's it. Then I realized that the spec for the LF280k states that the test conditions are at 25°C (77°F). Although it is December, here in the Denver area the daytime highs have been generally in the 50°s while I was doing the test, and our garage has been in the mid- to high-50°s. I'm guessing it may have gotten down into the mid-40°s at night, and I did run some of the tests overnight, since each test takes over 11 hours.

So here's the question: Can capacity testing at say 50°F - lower than the standard test conditions by 27°F - result in the capacity showing 10Ah lower? Can anyone think of something else that could have caused these results?
 

DLTooley

Solar Enthusiast
Joined
Jan 12, 2020
Messages
143
Controlling all conditions in a precise test is difficult. Easiest way to tell is to run the test again at the controlled temp you had previously.
 

bryn

Solar Enthusiast
Joined
Nov 4, 2021
Messages
89
I would say that the temperature makes a huge different for testing, there are a heap of data out there that shows the battery capacity at different temps.

Cold temp does decrease the capacity in Lithium,
 

robby

Photon Vampire
Joined
May 1, 2021
Messages
1,740
I would agree that it can definitely account for the drop.
There are a number of lengthy papers on this subject and how to calculate it out. The problem is that it's not just a simple table. It also involves how fast your discharging the batteries. The batteries produce heat during discharging, so the higher the rate of discharge is the lower the impact on the battery from lower ambient room temps.
 

fafrd

Photon Sorcerer
Joined
Aug 11, 2020
Messages
2,584
I'm wondering how much cell capacity is affected by cells being just a bit cooler. Here's my story:

In early Aug-2021, I received 8 Eve 230Ah cells, purchased via Michael B Caro (I was on one the lucky ones). I purchased one of the EBC-A40L testers (this one) and went about testing each cell. The tests were done in our basement, and although it was still pretty hot outside, the A/C kept the basement pretty cool (too cool for my wife's taste) at around 65°F-68°F. Of the 8 cells, one tested out at 237Ah, and the rest all tested at 238Ah or 239Ah. Obviously I was pleased.

Last week I received 8 Eve 280Ah cells, purchased from the famous @Amy Wan (Luyuan) at Shenzhen Luyuan. These were the new Eve LF280k cells that are supposed to have up to 6,000 cycles. Amy provides a report as the cells are shipped, and the report includes a cell capacity for each cell, although I know nothing about how the capacity is determined. The report indicated the capacities were 290.3, 292.0, 290.7, 288.5, 290.8, 289.5, 291.7, and 291.2. All good, with an average capacity of 290.6.

I used the same tester to do the capacity testing. I set it up for each cell with a multi-step program, so it would charge the cell to 3.65V at 25A, stopping the charge when the current dropped to 2.0A. Then the second step would start, discharging at 25A until the voltage reached 2.5V. I'm pretty sure this was the same process I used for the 230Ah cells I tested in the summer.

The test results were somewhat disappointing: 278.5, 284.0, 282.0, 275.7, 278.7, 281.8, 282.5, 283.5, 280.8 That's three cells that didn't even meet the 280Ah spec, and an average of only 280.8Ah, less than Amy's report by 9.8Ah.

I'll admit the results are not horrible, and I'm not likely to raise a stink. The person I'm building the two 4S packs for is perfectly happy with them. However, it is disappointing, and - given Amy's reputation here - pretty surprising. So I tried to figure out what could have somehow biased the test. It was the same tester and the same test methodology, so I don't think that's it. Then I realized that the spec for the LF280k states that the test conditions are at 25°C (77°F). Although it is December, here in the Denver area the daytime highs have been generally in the 50°s while I was doing the test, and our garage has been in the mid- to high-50°s. I'm guessing it may have gotten down into the mid-40°s at night, and I did run some of the tests overnight, since each test takes over 11 hours.

So here's the question: Can capacity testing at say 50°F - lower than the standard test conditions by 27°F - result in the capacity showing 10Ah lower? Can anyone think of something else that could have caused these results?
I saw a much more significant drop than 10Ah when testing my 280Ah EVE cells. Was just under 50F here in the Bay Area and I was getting ready to make a claim for replacement when I checked the data sheet and saw the same 25C test condition.

Rigged up a simple temperature-controlled test chamber using a cardboard box, a terrarium thermostat, and a carboy heater to maintain cell under test at 25C / 77F during both charge cycle and capacity test and low and behold, my capacity increased to within 99.5% or at most 99% of the 280Ah my cells were rated for.

So yes, any capacity test performed at temperatures below the temperature in the datasheet (for both charge and discharge) can not be taken at face-value as representing a deficiency in charge capacity...
 

Brett V

Solar Addict
Joined
Apr 5, 2021
Messages
310
I agree with the temperature affecting the capacity of the cells. Another thing to consider is the possibility of the resistance of the load in the tester changing with temperature.
 

Horsefly

Photon Sorcerer
Joined
Dec 12, 2020
Messages
1,439
Location
Denver, mostly
I would agree that it can definitely account for the drop.
There are a number of lengthy papers on this subject and how to calculate it out. The problem is that it's not just a simple table. It also involves how fast your discharging the batteries. The batteries produce heat during discharging, so the higher the rate of discharge is the lower the impact on the battery from lower ambient room temps.
Duh! I had specifically chosen 25A discharge because I thought it would have a minimal impact on the test. Wrong!

Thanks for that info!
 

Horsefly

Photon Sorcerer
Joined
Dec 12, 2020
Messages
1,439
Location
Denver, mostly
I saw a much more significant drop than 10Ah when testing my 280Ah EVE cells. Was just under 50F here in the Bay Area and I was getting ready to make a claim for replacement when I checked the data sheet and saw the same 25C test condition.

Rigged up a simple temperature-controlled test chamber using a cardboard box, a terrarium thermostat, and a carboy heater to maintain cell under test at 25C / 77F during both charge cycle and capacity test and low and behold, my capacity increased to within 99.5% or at most 99% of the 280Ah my cells were rated for.

So yes, any capacity test performed at temperatures below the temperature in the datasheet (for both charge and discharge) can not be taken at face-value as representing a deficiency in charge capacity...
You've made me feel better about these cells. Thanks.
 

Horsefly

Photon Sorcerer
Joined
Dec 12, 2020
Messages
1,439
Location
Denver, mostly
I agree with the temperature affecting the capacity of the cells. Another thing to consider is the possibility of the resistance of the load in the tester changing with temperature.
Interesting idea. I don't know how to tell if there is a resistance change in the tester, but I did do a clamp meter check and the current seemed to be still at around the 25A setting, so I don't think that was it.
 

Dzl

Perpetual Newbie & Unofficial Forum Librarian
Staff member
Moderator
Joined
Dec 10, 2019
Messages
5,089
Here is the temperature vs capacity test data for your cells (from the datasheet).
Screenshot 2021-12-22 at 22-54-36 22C-7-20210430142940 - LF280K (3 2V 280Ah) Product Specifica...png

One factor this doesn't address is what impact C-rate would have on this. If it were me I would test again at closer to 25*C.
 

Horsefly

Photon Sorcerer
Joined
Dec 12, 2020
Messages
1,439
Location
Denver, mostly
Here is the temperature vs capacity test data for your cells (from the datasheet).
View attachment 76840

One factor this doesn't address is what impact C-rate would have on this. If it were me I would test again at closer to 25*C.
Wow. I guess I remember seeing some curves at the end of the datasheet, but they were so small I didn't try to figure out what they said. It's a little confusing terminology: "Energy rate / %". But I guess it is telling me that at 25°C, 35°C, 45°C, and even 55°C (131°F!!) you can get 100%, but if the temp drops down to 10°C (50°F) you only get what looks to be around 95%. Fascinating.

Well, I think that proves it. You win the prize @Dzl !!
 

Steve_S

Offgrid Cabineer, N.E. Ontario, Canada
Joined
Oct 29, 2019
Messages
5,900
Location
Rural NE Ontario Canada
I have all of my packs setup in a "Stack" with the lowest 12" off the ground.
I heat the Powerhouse to 10C/50F and it keeps the heat very well. The powerhouse also has a Frost Protected Slab Foundation which is insulated with 4" of foam on the sides & underneath. (It also has Radiant Pex in it but not in use at this time)

My bottom battery is usually between 9-10C while the highest one may be 12-15C (The Inverter vents at that level) depending is charging or discharging the heat coming from it varies.

The Colder Packs take charge slower even at the same charge rate (Amps) (typically 20A or so just from solar on avg). The warmer packs take charge a bit faster but not that huge of a difference really. NOTE that while charging above 10A to Pack the cell temps do rise slowly and actually do come up to 15C temp while the hi packs (elevation wise) may reach up to 18C +/- a bit.

In summer the temps in there usually are no higher than 26C and no lower than 20C. The battery packs do charge up a bit quicker (as expected) and actually will settle at a Higher Voltage than when cooler (in winter). They will settle to 3.410Vpc (NB I bulk charge to 3.440) but at this time of year with it being 10C in there they will reach target voltage BUT will settle to 3.375. This is fairly consistent.

How the "Stack" is arranged.
24V/175AH (Used EV Grade Cells)
24V/175AH (Used EV Grade Cells)
24V/280AH (EVE K Series Matched & Batched) Luyuan
24V/280AH (EVE N Series Bulk) Luyuan
24V/280AH (EVE N Series Bulk) XUBA - from when Amy was still with that company.

The Cell Tests done by manufacturers are done at 25C Ambient Temps and is clearly stated in docs. General LFP Capacity retention is affected by Hi / Lo Temps. Doped LFP is not affected in such a way. Doped LFP ? Doped with Yttrium makes them useable to -25C (chargeable) but it is $significant add on... See Winston Cells for example.

Hope it helps, Good Luck
 

Horsefly

Photon Sorcerer
Joined
Dec 12, 2020
Messages
1,439
Location
Denver, mostly
I have all of my packs setup in a "Stack" with the lowest 12" off the ground.
I heat the Powerhouse to 10C/50F and it keeps the heat very well. The powerhouse also has a Frost Protected Slab Foundation which is insulated with 4" of foam on the sides & underneath. (It also has Radiant Pex in it but not in use at this time)

My bottom battery is usually between 9-10C while the highest one may be 12-15C (The Inverter vents at that level) depending is charging or discharging the heat coming from it varies.

The Colder Packs take charge slower even at the same charge rate (Amps) (typically 20A or so just from solar on avg). The warmer packs take charge a bit faster but not that huge of a difference really. NOTE that while charging above 10A to Pack the cell temps do rise slowly and actually do come up to 15C temp while the hi packs (elevation wise) may reach up to 18C +/- a bit.

In summer the temps in there usually are no higher than 26C and no lower than 20C. The battery packs do charge up a bit quicker (as expected) and actually will settle at a Higher Voltage than when cooler (in winter). They will settle to 3.410Vpc (NB I bulk charge to 3.440) but at this time of year with it being 10C in there they will reach target voltage BUT will settle to 3.375. This is fairly consistent.

How the "Stack" is arranged.
24V/175AH (Used EV Grade Cells)
24V/175AH (Used EV Grade Cells)
24V/280AH (EVE K Series Matched & Batched) Luyuan
24V/280AH (EVE N Series Bulk) Luyuan
24V/280AH (EVE N Series Bulk) XUBA - from when Amy was still with that company.

The Cell Tests done by manufacturers are done at 25C Ambient Temps and is clearly stated in docs. General LFP Capacity retention is affected by Hi / Lo Temps. Doped LFP is not affected in such a way. Doped LFP ? Doped with Yttrium makes them useable to -25C (chargeable) but it is $significant add on... See Winston Cells for example.

Hope it helps, Good Luck
Hey @Steve_S, sounds like your "About my system" is due for an update!;)

This set of 8 Eve 280k's is slated to go on a sailboat in Florida. So it will never see temps below 50°F/10°C. However, my 8 Eve 230's will be going up to our cabin in the mountains. The solar room is in the northeast corner of the basement, and even in the summer when it might get above 90°F/32°C outside, the current AGM batteries don't get much above 65°F/18°C. I'm now thinking they may have a little bit of capacity degradation.
 

Steve_S

Offgrid Cabineer, N.E. Ontario, Canada
Joined
Oct 29, 2019
Messages
5,900
Location
Rural NE Ontario Canada
It says (to be updated as it is out of date) and is presently in an Evolution mode and it's more than originally intended but will "button up" everything nicely. I am changing BMS' on everything and making sure all busbars are Identical and tapped in the centre for the BMS sense leads to be attached with M6 screws... There is also some minor structural changes for the building itself as I pull out almost 1200 lbs of Lead Battery and I have decided to no longer keep the Rolls Bank in place anymore.
 

HRTKD

Boondocker
Joined
Apr 24, 2020
Messages
7,692
Location
Somewhere South of Denver
Here in the Denver area, I discount the distance to empty number that my EV reports in the winter. Cold temperatures really do a number on the range. However, "cold" to me is anything lower than 40° F.
 

robby

Photon Vampire
Joined
May 1, 2021
Messages
1,740
Here is the temperature vs capacity test data for your cells (from the datasheet).
View attachment 76840

One factor this doesn't address is what impact C-rate would have on this. If it were me I would test again at closer to 25*C.
The problem with those curves is that it assumes a set C-Rate. As I was telling the OP there are a big set of equations available that take into account the Discharge rate as well as the Ambient Temperature. The Analysis based on electric vehicles using LIFePO4 is 130 pages long.
 
  • Like
Reactions: Dzl

Horsefly

Photon Sorcerer
Joined
Dec 12, 2020
Messages
1,439
Location
Denver, mostly
The problem with those curves is that it assumes a set C-Rate. As I was telling the OP there are a big set of equations available that take into account the Discharge rate as well as the Ambient Temperature. The Analysis based on electric vehicles using LIFePO4 is 130 pages long.
I was hoping that by using a low discharge rate (25A, so <0.1C) I would avoid issues that might be affected by C-rate. I hadn't thought about a temperature that isn't even very cold causing a drop in capacity. Heck, I was standing in the garage in a tee shirt doing these tests!
 

robby

Photon Vampire
Joined
May 1, 2021
Messages
1,740
I was hoping that by using a low discharge rate (25A, so <0.1C) I would avoid issues that might be affected by C-rate. I hadn't thought about a temperature that isn't even very cold causing a drop in capacity. Heck, I was standing in the garage in a tee shirt doing these tests!
Yep, it's not easy to figure it out and even if you use those curves it only applies so long as you remain within the parameters they tested under.
Your loss vs Temp to me is fully expected. I would not bother going through the test process again.
One time we went on an outdoor trip with some two way VHF radios that had been fully charged at base camp and we know would last for several hours in the summer. In the Sub zero temps of winter the batteries died so quickly that we got into an argument back at camp blaming one of the guys that he had forgotten to charge them.
 
Last edited:

RCinFLA

Photon Sorcerer
Joined
Jun 21, 2020
Messages
1,693
The cell's ion diffusion slows down as temp gets cold. This has the effect of increasing overvoltage terminal voltage slump with load current. This increases losses within battery. The effective cell impedance is greater than room temp. Below freezing the impedance can increase 5x what it was at 25 degs C.

The amount of capacity loss depends on load current and type of LFP cell. A thin electrode cell suffers less, a thick electrode cell suffers more. Most of the prismatic cells used by DIY'er like EVA, Listen, CATL are thick electrode cells. The cylindrical cells like used in BattleBorn batteries are thinner electrode so should do better at cold temps.

Any cell discharge curve chart that does not specify amount of discharge current for given curve(s) is pretty useless.
Unloaded, open circuit cell voltage does not change much with temperature.

Compare the cell terminal voltage drop, after 2-3 mins with load current, (try to load greater than 0.2CA). Since the effective cell impedance rises significantly at cold temps the terminal voltage slumps more. This results in less extractable capacity and more internal cell loss. At -20 degC with moderate current load you may only achieve 25% of capacity you typically see at 25 degs C. Capacity drop starts to accelerate below +10 degsC.

Roughly, for thick electrode cells discharged at 0.2 CA rate, extractable capacity relative to 25 degs C:
+40C -> 104%
+25C -> 100%
+10C -> 94%
0C-> 80%
-10C -> 57%
-20C -> 25%

The saving grace is at moderate current the extra cell loss creates more internal cell heating which warms up the cell so more capacity can be extracted. This assumes your inverter or BMS doesn't trip off at the beginning of load draw for undervoltage first. Older, more aged cells have greater internal impedance so they warm up faster but suffer greater initial terminal voltage slump at cold temp.

The extra self heating with moderate current during cold ambient temps makes capacity measurement curves very misleading. Normal real world useage may not typically be a moderate continuous load so there is less internal heating.
 
Last edited:
Top