Resting voltage after charging cell with 0.5P/C and terminate at 3.65V?

[cj0]

To validate that my, not exactly CB75 datasheet matching 0.5P, charging procedure is good enough, I would like to check the LiFePO4 cell resting voltage 30 minutes after charging ends (around 25ºC).

For the newer cell types like EVE MB30 and MB31, or Rept CB75, the datasheets mention to stop charging when cell voltage reaches 3.65V. So there is no constant voltage (CV) phase, where the current is tapering to 0.05C.

Who has experience with charging newer cell types with 0.5P and terminate at 3.65V/cell, and has measured the resting OCV 30 minutes after charge termination?

 

[RCinFLA]

As long as a cell maintains greater than 3.43v after charging it is fully charged. Open circuit voltage is not too temp dependent unless very low temps.

Between 3.43v and 3.65v depends on several factors.

First, going to 3.65v speeds up charging. There is an overpotential voltage overhead to drive the kinetics of ion migration within cell. The greater the cell current, the greater the overpotential voltage overhead. Higher bulk charging current means the terminal voltage is greater than the resultant OCV would be if you stop charging and let cell reach equilibrium.

As the cell approaches full charge, the cell current will drop, and associated overpotential voltage overhead will drop. By absorbing at a voltage higher than 3.45v you increase the overpotential voltage overhead and keep the charging cell current higher for a longer period of time. This gets the cell charge topped off quicker.

You could fully charge a cell with an absorb voltage of 3.45v but it would take very long (a day or two) for current to taper down indicating full charge. This is because as cell approaches 3.43v full charge voltage, the overpotential voltage will only be 3.45v minus 3.43v which is very low resulting in very low charging current.

Final factor is, at full charge, a layer surface charge capacitance will build up a static voltage, so you are effectively trading overpotential voltage for surface charge voltage build up.

The surface charge has almost no effective capacity and can be burned off quickly with a minor load in a couple of minutes. It is not part of lithium-ion capacity and is more like a supercap made primarily of the graphite anode electrostatic capacitance.

Surface charge will bleed off on its own with open circuited cell but the time to bleed down depends on leakage of given cell. It can vary from a couple of hours to a couple of days depending on given cell leakage rate.

This causes many folks to think their cells are not fully balanced in SoC after doing a top balancing charge to 3.65v. By the next day it is common to have cell open circuit voltages ranging from 3.45v to 3.60v after doing a full top balancing to 3.65v the day before.

As long as all cells stay above 3.43v at open circuit equilibrium state they are all fully charged and balanced in SoC at 100%.

 

[cj0]

@RCinFLA I'd wish a second source for the 3.43V.

However 3.43V is when compared to testing LiFePO4 cells, what you would reach :
- unknown charged "full" Rept 280Ah, discharging with 30A after 33 seconds
- charged 3.65V tapering to 15.3A MB30, discharging with 40A after 37 seconds
- charged 3.65V tapering to 15.3A MB31, discharging with 40A, after 61 seconds

The interest is in the 30m resting open circuit voltage difference between a 0.5P (3 till 4 times the charge current of a ZKEtester) teminated without current tapering down and termination with current tapering.

 

[RCinFLA]

3.43v is the chemistry of LFP cathode. You can easily verify it with a quick Google search.

The negative anode graphite gives the slight bumps in the discharge/charge curve. The bumps in the graphite anode potential are caused by the way the graphite lattice accepts lithium-ion intercalation (stuffing). There is a repeating sequence of six layer group of graphite lattice that have a regimented way of opening up the layers to lithium-ion intercalation.

The graphite anode strata levels in potential are a bit more pronounced for charging compared to discharging.

The LFP cathode voltage is almost dead flat between 98% and 5% SoC.

Cell voltage is positive cathode potential minus negative anode potential.

The cathode and anode each have their own overpotential vs cell current which is different but can be just combined as a single entity for the cell overall terminal overpotential voltage for discharge or charging current.

LPF are normally built with 10% to 15% more storage capacity in graphite then LFP cathode. This is to optimize the cycle life as for LFP cell the graphite is the dominate determiner of longevity of cell. (versus NCA, or Nichel based lithium-ion batteries, which longevity is dominated by cathode degradation).

 

[sunshine_eggo]

As long as a cell maintains greater than 3.43v after charging it is fully charged. Open circuit voltage is not too temp dependent unless very low temps.

I know you've stated this many times, but all of my test data indicates it is not correct. Here is an EVE 280Ah cell charged at 70A to 3.65V with a tail current of 14A (0.05C) per the cell specs:



A 20 minute rest immediately following charge:



In less than 4 minutes, it has dropped below 3.43V.

Immediately after the 20 minute rest, a 70A discharge to 2.5V:



The 280Ah cell delivered 302Ah.

All testing across 18 EVE 280Ah cells and 8 Topband 25Ah 3C cells has been consistent with the above. Same holds for 70+ CALB 40Ah cells, but they were used, and some were in pretty bad shape (below 80% SoH and 2X+ IR spec), so I don't put a lot of weight in their results.

Am I using calibrated test equipment? Gosh no. Have I correlated it to my Fluke VM and Klein Tools DC clamp ammeter? Yes. I do not think measurement error would account for this discrepancy.

 

[RCinFLA]

In less than 4 minutes, it has dropped below 3.43V.

I can guaranty you will not bleed off a full charged battery surface charge in 4 minutes unloaded. That is overpotential equilibrium recovery time. Cell is not fully charged.

You also cannot run an absorb voltage of 3.65v and fully charge a cell without getting surface charge that will take a lot longer than 4 minutes to bleed off on its own (open circuited cell).

 

[cj0]

I know you've stated this many times, but all of my test data indicates it is not correct. Here is an EVE 280Ah cell charged at 70A to 3.65V with a tail current of 14A (0.05C) per the cell specs:



A 20 minute rest immediately following charge:

View attachment 264627

In less than 4 minutes, it has dropped below 3.43V.

Am I using calibrated test equipment? Gosh no.

Which test equipment do you use for testing a single cell?

 

[sunshine_eggo]

If this happens the cell is not quite fully charged. The four minutes is equilibrium settling time of overpotential.
This is common when using higher charging currents that creates higher overpotential voltage.
Continue charging until taper current drops to lower level.

This would result in over charge as the cells specify a 0.05C cut off at 3.65V. Is the manufacturer wrong?

It is theoretically possible that the cell is a bad cell with very high self-discharge rate but this would be unusually rare.

I doubt it's the case. Eight of the first 9 I tested lost less than 0.7Ah over 5 months. The 9th one lost 3Ah. I did not test the second 9 for self-discharge.

Look at cathode and anode potential charts. Notice around 100% SoC what the graphite anode potential is. If graphite has a SoC potential of 0.06v then LFP cathode is at full charge there will be 3.43v (cathode) minus 0.060v (anode) = 3.37 vdc for cell.

The tolerance on the amount of extra graphite (typ. 10-20% extra graphite) for a given cell and its SoC alignment with LFP cathode SoC effects the 100% SoC voltage. Graphite degrades more than LFP over cell lifetime, which is why more graphite is provided to optimize cycle longevity of cell.

The bracketing of the graphite anode excess storage capacity and LFP cathode capacity may take several cycles on a new cell to settle down on their alignment.

Is it possible we're dealing with the difference between the theoretical and the practical application thereof? Chemical engineers might give different answers than battery engineers. I have 26 examples of cells charged according to the manufacturer's recommendations, but they do not behave as your references claim.

Which test equipment do you use for testing a single cell?

iCharger 4010DUO in synchronous mode, 70A charge/discharge on a single cell. I occasionally use a 3010 when 30A will suffice.

 

[RCinFLA]

This would result in over charge as the cells specify a 0.05C cut off at 3.65V. Is the manufacturer wrong?

Yes, 0.05C cut off at 3.65V is bad criteria unless bulk charging C-rate is also specified.

 

[cj0]

iCharger 4010DUO in synchronous mode, 70A charge/discharge on a single cell. I occasionally use a 3010 when 30A will suffice.

I suspect that the test was ran in some "regenerative mode"?

Can't it be the iCharger (4010DUO) that is somewhere down the road between charging and discharging, is (already) starting to source some current from what was the "target battery" and will become the "source battery" (when starting the discharge)?

 

[sunshine_eggo]

Yes, 0.05C cut off at 3.65V is bad criteria unless bulk charging C-rate is also specified.

It's a 1C charge, but how much does that really matter? Lower C rate charges hit higher SoC at any given voltage. Once a 1C charge tapers to 0.25C @ 3.65V is it going to be at a different state of charge than that the 0.25C charge @ 3.65V? I don't think that's the case.

 

[sunshine_eggo]

I suspect that the test was ran in some "regenerative mode"?

Yes.

Can't it be the iCharger (4010DUO) that is somewhere down the road between charging and discharging, is (already) starting to source some current from what was the "target battery" and will become the "source battery" (when starting the discharge)?

No.

There is a source and a target. It only draws operating power from the source, whether that's a power supply or a battery. To state it explicitly, for those 20 minutes, it was drawing power from the source (12V 10kWh FLA bank) while measuring the voltage of the target cell.

Technically, you have to draw some amount of current/power to measure voltage, but I don't think that's what you're talking about.

 

[cj0]

There is a source and a target. It only draws operating power from the source, whether that's a power supply or a battery. To state it explicitly, for those 20 minutes, it was drawing power from the source (12V 10kWh FLA bank) while measuring the voltage of the target cell.

Technically, you have to draw some amount of current/power to measure voltage, but I don't think that's what you're talking about.

For voltage measurement I'd expect a 1 MOhm resistance, thus 3.65 microamps at most. I think we can neglect the 3 to 4 microamps.

I'm more interested in whether some (even sub) milliamp current, for powering the iCharger display and microcontroller might be taken from the battery cell under test, while resting for 20 minutes.

 

[sunshine_eggo]

For voltage measurement I'd expect a 1 MOhm resistance, thus 3.65 microamps at most. I think we can neglect the 3 to 4 microamps.

Exactly.

I'm more interested in whether some (even sub) milliamp current, for powering the iCharger display and microcontroller might be taken from the battery cell under test, while resting for 20 minutes.

WHY would it work this way? I've tested literally thousands of batteries on regenerative chargers and never seen ANY evidence the chargers are drawing current from the test battery in a way that's inconsistent with voltage measurements. I've left them connected and powered for days and even a couple weeks at a time.

 

[cj0]

I've tested literally thousands of batteries on regenerative chargers and never seen ANY evidence the chargers are drawing current from the test battery in a way that's inconsistent with voltage measurements. I've left them connected and powered for days and even a couple weeks at a time.

Your resting graph has a totally different scale of voltage drop f.e. when comparing to Andy (OGG).

His MB31 charge video shows that it did take him at least 1 and a half minute (resting time) in the YouTube video (even ignoring the cuts) to move from the charge test, to the discharge test.

In a 1 minute rest your voltage drops to 3.526 volts.
In 2 minutes your voltage drops to 3.48 volts.
Andy is able to start run his discharge-at-least-1-minute-after-charge test with an OCV of 3.618 volts.

Andy's LF280K B-grade discharge test starts with: 3.436V. Unfortunately his LF280K certified was tested with 0 resting time.
His Yixiang EVE LF280K cell6 discharge test starts with: 3.589V. Cell12 discharge starts at: 3.564V. Cell15 starts at: 3.618V.

What do you think of these differences?

 

[sunshine_eggo]

Your resting graph has a totally different scale of voltage drop f.e. when comparing to Andy (OGG).

His MB31 charge video shows that it did take him at least 1 and a half minute (resting time) in the YouTube video (even ignoring the cuts) to move from the charge test, to the discharge test.

In a 1 minute rest your voltage drops to 3.526 volts.
In 2 minutes your voltage drops to 3.48 volts.
Andy is able to start run his discharge-at-least-1-minute-after-charge test with an OCV of 3.618 volts.

Andy's LF280K B-grade discharge test starts with: 3.436V. Unfortunately his LF280K certified was tested with 0 resting time.
His Yixiang EVE LF280K cell6 discharge test starts with: 3.589V. Cell12 discharge starts at: 3.564V. Cell15 starts at: 3.618V.

What do you think of these differences?

It's pretty obvious. He over-charged. His tail current was 1A. V2 datasheet specifies 14A cut off. The V2 cells specify an hour wait following charge. I was too impatient to wait 1 hour.

If he was testing V3 cells, he was grossly over-charged. They specify a constant POWER discharge to 3.65V with immediate termination, and a 30 minute rest.

 

[justgary]

Were any of the batteries kept under compression during the test? I suppose I could answer "no" for Andy right away....

 

[justgary]

batteries

Well, cells.

 

[fnnwizard]

I can 100% confirm the >3.430V resting V.

A coouple weeks ago I charged a 280LFK EVE @28A to 3.650 V, taper @ 14A.
It has rested >9 days now and Voc is at 3.4314V. I didn't allocate enough memory to the logger so it stopped 3 days after charging was complete.

I can do another one and show.

@sunshine_eggo It is 100% the charger. The ichargers are good but unless you have separate sense leads, there's going to be a V drop. Also, the ones I have tested all required some form of calibration. There is a way to do it with their older chargers and they can be off by ~20-40mV (generally reading high). I thought they did that on purpose to avoid over charging Lipos.

If you hook up the fluke to the battery themselves while charging you will see the delta.

 

[sunshine_eggo]

I can 100% confirm the >3.430V resting V.

A coouple weeks ago I charged a 280LFK EVE @28A to 3.650 V, taper @ 14A.
It has rested >9 days now and Voc is at 3.4314V. I didn't allocate enough memory to the logger so it stopped 3 days after charging was complete.

I can do another one and show.

@sunshine_eggo It is 100% the charger. The ichargers are good but unless you have separate sense leads, there's going to be a V drop. Also, the ones I have tested all required some form of calibration. There is a way to do it with their older chargers and they can be off by ~20-40mV (generally reading high). I thought they did that on purpose to avoid over charging Lipos.

If you hook up the fluke to the battery themselves while charging you will see the delta.

These are balance chargers. There are separate sense leads. I used them. I deliberately omitted the current voltage curves for clarity, and so I wouldn't have to explain it. I guess I shouldn't have as I'm having to explain it anyway. To state it another way, the green voltage curves are OCV, and they correlate to my fluke within 0.01V.

Have you ever overcharged that cell, i.e., you never top balanced it or charged at a lower tail current?

 

[fnnwizard]

These are balance chargers. There are separate sense leads. I used them. I deliberately omitted the current voltage curves for clarity, and so I wouldn't have to explain it. I guess I shouldn't have as I'm having to explain it anyway. To state it another way, the green voltage curves are OCV, and they correlate to my fluke within 0.01V.

Have you ever overcharged that cell, i.e., you never top balanced it or charged at a lower tail current?

Not this one, it's a brand new cell, with at least 1 cycle from factory, and 2 from me. The first one at 56A to tail of 15A just to verify capacity and this 2nd cycle at 28A to 13.998A all to 3.650V. I was going to discharge at 5A to find equivalent Voc of the cells, but haven't gotten time and it's just been sitting. The power supply I use is a lab grade Kikusui so it's very accurate.

I have a few unaccounted for NIB EVE 280LFK cells I've been meaning to cylce test, so perhaps can charge at 28A and 56A, with .05C tail current, let it sit and observe resting V behavior with 16bit logger.

 

[cj0]

It's pretty obvious. He over-charged. His tail current was 1A. V2 datasheet specifies 14A cut off. The V2 cells specify an hour wait following charge. I was too impatient to wait 1 hour.

If he was testing V3 cells, he was grossly over-charged. They specify a constant POWER discharge to 3.65V with immediate termination, and a 30 minute rest.

I don't dare to say whether Andy is over-charging or not.

You are correct for LF280K-S04-LF rev.A (apr. 2022) datasheet states "until the charging current is less than or equal to 14A". That datasheet might be V2. In V3 Eve datasheet LF280K-D04-01 rev.C (Jun. 2023) cycle test charge termination is done at 0.5P.

However in MB30 PBRI-MB30-D06-01 rev.A (Nov. 2023) a new item is introduced in the "Safety Limit Voltage Parameters" table: "Upper limit charging capacity": "The charging capacity shall be less than 113% of the nominal capacity.". That might give room for terminating with a CV phase until 113% of 306Ah = 345,77Ah are charged in to an MB30.

My guess is that you curves above are for an LF280K V2 cell?

What I found remarkable in Andy's test result, is that his "B-grade" cell test showed the lowest OCV (3.436V) just before starting the discharge test. Note: I've only checked Andy's results stored on Google Drive Dropbox doesn't work over here (empty pages).

 

[cj0]

I have a few unaccounted for NIB EVE 280LFK cells I've been meaning to cylce test, so perhaps can charge at 28A and 56A, with .05C tail current, let it sit and observe resting V behavior with 16bit logger.

Thanks for sharing your OCV at 100% = 3.43V results for tail current charged cells.

What is meant with the "NIB", New In Box?
Are yours the V2 or V3 LF280K cells?

 

[cj0]

The ichargers are good ...

@sunshine_eggo I don't know it the iChargers are good. The manual/manufacturer does not specify any ripple p-p voltage at all. And I can't find an online resource having tested the 4010duo output for p-p ripple in CV.

@fnnwizard Your Kikusui at least has a manufacturer assigned ripple noise specification, f.e. the nice low 50 mV p-p at 20Mhz for a PAV10-72, the older PVS series up to and including 20V are 75 mV p-p. A lot better then 120mV Mean Well HRP-600-3.3. For a single LiFePO4-cell my charging goal would be to stay below 100 mV p-p.

 

[fnnwizard]

Thanks for sharing your OCV at 100% = 3.43V results for tail current charged cells.

What is meant with the "NIB", New In Box?
Are yours the V2 or V3 LF280K cells?

I charged to 3.650V with 28A and ending with tail current @14A.

I also do know that if charged at high rates will same tail current, the cells do come down faster, to below 3.430V and I have seen them come down within minutes, so @sunshine_eggo is not incorrect, but the charger is giving a bit of false sense of 100% charge when the cells are not.

The .5C rate charge with tail of .05C does not give a true 100% charge state, but is a very safe place to end charging if fast charging is required.
Almost all my and recommended use case is with .20-.25C charge rates these days.

@sunshine_eggo I don't know it the iChargers are good. The manual/manufacturer does not specify any ripple p-p voltage at all. And I can't find an online resource having tested the 4010duo output for p-p ripple in CV.

@fnnwizard Your Kikusui at least has a manufacturer assigned ripple noise specification, f.e. the nice low 50 mV p-p at 20Mhz for a PAV10-72, the older PVS series up to and including 20V are 75 mV p-p. A lot better then 120mV Mean Well HRP-600-3.3. For a single LiFePO4-cell my charging goal would be to stay below 100 mV p-p.

The Kikusui I have is the PWX1500 80V one. It maxes out current at 58A though so I'm limited there. It's networked so can be remotely operated, I also put a Tyco Kilovac relay on it to completely isolate it if needed. Relay is controlled with 12V wallwart via smart switch.

I also have an electronic load (Ametek/Sorensen) with another relay attached controlled by another smart switch and wall wart so the same battery can be connected to both charger and load at same time. Then it's just a matter of opening and closing relays to cycle the cell remotely. Electronic load is currently programmed to pull current only if it detects V over 3.350V and stops at 2.501V but is not networked though, so I have to go in and manually adjust load settings if needed.

All of it is video monitored via a couple 4k NDI PTZ cam if I'm running testing for days.

Here a screenshot of the live video at the place. Can barely make out the power supply and electronic load. I had Osram Lightify lights but Osram discontiuned their IoT services so I'll need to replace all the lights there to remotely operate those. But it now has infrared switches so if there's movement there, the lights at least will automatically turn on.