"Low current overcharge"

[venquessa]

Part of the work (hobby horse) I am doing with my DIY lithium system involves testing the "top and bottom end" responses of the system as a whole.

Questions are basically: In the setup, what happens at the top of charge with the sun fully on the panel? Also. What "should" or "should not" happen.

I can answer some of that, I think, but there are "shadow" areas of uncertainty which worry me.

One of them is basically the term of "low current overcharge". I'm trying to understand it and honestly been getting frustrated with the less than ideal explanations or any data for it.

My understanding gets me to there being a higher 'quiescent' current through the cell at 3.650V than say at 3.4 or 3.35. The question is, how much damage is or is not occurring at 3.650V quiescent versus a lower 'float' setting?

On one hand I can let the MPPT hold 14.40V for 3 hours and make sure any balancing or tapering has occurred. Or I can restrict it significantly. I can, given time and weather test both to see if there is/or how much benefit there actually is, but sometimes getting 1% more capacity could be costing you 1% your cycle life in addition to all the other things eating into that already.

 

[sunshine_eggo]

Part of the work (hobby horse) I am doing with my DIY lithium system involves testing the "top and bottom end" responses of the system as a whole.

Questions are basically: In the setup, what happens at the top of charge with the sun fully on the panel? Also. What "should" or "should not" happen.

Just establishing that sun on the panel doesn't have a driving influence other than whether or not the PV can sufficiently power the charger - the charge controller is a load. It draws from the solar panel based on its programming. It only draws what it needs up to the maximum of PV. If the PV can provide more, it simply doesn't do anything.

I can answer some of that, I think, but there are "shadow" areas of uncertainty which worry me.

One of them is basically the term of "low current overcharge". I'm trying to understand it and honestly been getting frustrated with the less than ideal explanations or any data for it.

My understanding gets me to there being a higher 'quiescent' current through the cell at 3.650V than say at 3.4 or 3.35. The question is, how much damage is or is not occurring at 3.650V quiescent versus a lower 'float' setting?

I think low current overcharge is something of a false boogeyman for 2 reasons:

1) LFP used to be charged to 4.2V. It was found to be suboptimal. Cycle life was notably improved by reducing to 3.65V while only losing a few % of capacity. This was done for years. This is more of a mitigating position, i.e., we have an example of excessive overcharge. It wasn't great, but it wasn't catastrophic. By comparison, floating to 3.40V forever is almost certainly less stressful than regularly cycling to 4.20V.

2) When one looks at all of the flavors of server-rack LFP batteries that can communicate with AiO or Victron or other ecosystems, they operate in one way and one way only in normal operation:

Charge to this voltage and don't exceed this current.

That voltage is ALWAYS higher than 3.45V.

If charge current is available, it will hold the charge voltage until that current is no longer available, i.e., absorption and float are the same voltage.

In other words, battery manufacturers claiming many thousands of cycles of life are charging based on elevated/identical charge/float voltages, some as high as 3.65V.

On one hand I can let the MPPT hold 14.40V for 3 hours and make sure any balancing or tapering has occurred. Or I can restrict it significantly. I can, given time and weather test both to see if there is/or how much benefit there actually is, but sometimes getting 1% more capacity could be costing you 1% your cycle life in addition to all the other things eating into that already.

Per just about every datasheet, LFP is fully charged at 3.55-3.65V/cell and 0.05C, i.e., a 100Ah battery is full if a 5A or lower current can hold the voltage at 3.55-3.65V/cell. There is no point in pushing more current as there will be no increase in usable capacity, so the only reason you would need to hold 14.40V is if you needed to allow additional balancing, and that shouldn't be needed if regularly cycling to full charge. Typical taper is rarely more than 15-30 minutes.

All that said...

Since LFP holds its charge, why risk it?

Options:

1. Charge to 3.55-3.65V/cell and hold it there until the PV can't sustain it.
2. Charge to 3.55-3.65V/cell with a 30 minute absorption and 0.05C tail current. Float at 3.375V to hold near full and prohibit LCO.
3. Charge to 3.45V/cell with a 2-3 hour absorption and no tail current (or a really really low one). Float at 3.375V to hold near full and prohibit LCO.

#1 is what the server rack folks do.
#2 is what the typical "12V" battery manufacturer does, though they tend to specify 3.375-3.45V/cell float.
#3 is what most folks do if they want to maximize cycle life. The lower peak charge voltage and correspondingly lower current over a long duration is ultimately less stressful to the cells. Claims of negligible capacity loss after five years of this practice have been made. Additionally, most batteries will have an opportunity to balance if 3.45V/cell is held. In fact for those chasing 12V cell imbalance issues, absorbing/floating at 13.8V has been demonstrated to improve balance for those having trouble with BMS's cutting off charge due to single cell over voltage protection.

 

[venquessa]

Thank you. Makes sense. More sense than half the "parrots" scare crowing about it without explanation or qualifiers.

I too figured it was a a bit of "better safe than sorry" and if you can avoid it entirely, why not?

I researched elsewhere too and on this particular topic, it seems even the scientific research is still using words like "May", "Some studies have suggested", "Other studies have suggested", "We need to learn more.", and many, many "it depends". It would seem there is still a lot of active research into the long term maintenance of LFP energy storage even today. Although they seem to agree time above VoC isn't helping your cycle life. It just seems to be "how much", "how quickly" and by what mechanisms etc. that are less sure or harder to find data on.

To be honest I was browsing around on Ali* and looking at battery packs. It was when I was drooling over a set of 8 300Ah Eve cells for £2000 or something I thought, "Do you want to play top end charge games with that money?" It does make one think a little, compared with a pack that cost me £160. Well that was the idea anyway. Start with some no-name cheap cells, learn my lessons, make my mistakes and don't waste a BIG battery.

My present settings are for 14.40V ~3.6Vpc. BMS HVC is 3.65V. Balancer power is 0.4A, but I have an upgraded version with more when my second set of 4 cells arrives. The "holding time" does relate to the available balancing time. Which as you say shouldn't be a problem when the cells are all nice settled and cycled a few times with a top balance each time. They are not yet in that state and are still settling.

Current MPPT (EPEver Trace 4210) can put out 40A, but the panel currently tops out at 22 Amp into the battery. 0.22C charge max. The total duration of that is going to 4 hours or so (insolation angle). I do realise that is slightly over-powered but another 4 cells are on the way which will then half that current.

I knew of low current overcharge in a different frame of thinking with LiPo high discharge cells and charging them to 4.20V until taper cut off. It was made very clear that time spent "at the danger zone line" is time spent in danger. (Duh right?). It wasn't so much cycle life (although you did try), it was more about not setting fire to something or having your LiPo pouch cell turn into a fire spitting baseball. Charge them to 75% the evening,morning before, then top them up to 4.20V just before you use them, with a small cooling period. If you don't end up using that pack, discharge back to 75% (or less) before you leave and NEVER leave fully charged LiPos in your car. It will not stay at 4.20V as it heats up and cars in the sun can get pretty damn hot. Back then, most of us learnt just how serious the warnings were by the sad posts from teenagers who burnt half the house down, the burnt out cars, the gutted garages etc. Still to this day I will not charge one of these packs unattended. If I do more than go for a bathroom break, I stop the charge and fully disconnect the whole pack. It's not about cycle life, it's about house insurance

Having 4 and soon 8 of it's much, much larger cousins charging unattended, lets say I prefer they are in the garage on a concrete floor with nothing but themselves and brick wall to burn and nowhere near 4.20V!

I was hoping to try and assess what magnitude of damage an hour at 3.65V and quiescent would do, every day. On one hand it could cost a cycle. So a 4000 cycle pack ends up lasting only 2000 cycles until its 80% rated (they still use that dont they?).

However, I'm resolving to the fact that finding that out, to any certainty is near on impossible. Unless I want to offer my pack up as test, split it in 2 to give a control and abuse the hell out of it.... for years. And that is then only valid for "my" pack.

The other option, which I think most people are taking is to consider it "could" be that bad. Unlikely, very unlikely in the extremes, but as we can't really get a good "order of magntiude" on it. It's probably best to assume it's bad and avoid it if we can. Sleep easy.

Making that all work with the particular MPPT and BMS I have is probably a topic for another thread.

 

[upnorthandpersonal]

8 300Ah Eve cells for £2000

You can do better than that. You can get those from a pretty reputable seller for $184 each including shipping.

DIY LiFePO4 Cost Analysis - Almost February 2023 Edition

This is a recalculation based on the situation today, of what I did last year. Disclaimer: I mention vendors below and use them as an example and as reference. This should not be considered an endorsement of said vendors! Let's start again with leveling the playing field. The EG4 is a 5kWh...

diysolarforum.com

You can even get them cheaper...

 

[Robbert]

Thank you. Makes sense. More sense than half the "parrots" scare crowing about it without explanation or qualifiers.

I too figured it was a a bit of "better safe than sorry" and if you can avoid it entirely, why not?

I researched elsewhere too and on this particular topic, it seems even the scientific research is still using words like "May", "Some studies have suggested", "Other studies have suggested", "We need to learn more.", and many, many "it depends". It would seem there is still a lot of active research into the long term maintenance of LFP energy storage even today. Although they seem to agree time above VoC isn't helping your cycle life. It just seems to be "how much", "how quickly" and by what mechanisms etc. that are less sure or harder to find data on.

To be honest I was browsing around on Ali* and looking at battery packs. It was when I was drooling over a set of 8 300Ah Eve cells for £2000 or something I thought, "Do you want to play top end charge games with that money?" It does make one think a little, compared with a pack that cost me £160. Well that was the idea anyway. Start with some no-name cheap cells, learn my lessons, make my mistakes and don't waste a BIG battery.

My present settings are for 14.40V ~3.6Vpc. BMS HVC is 3.65V. Balancer power is 0.4A, but I have an upgraded version with more when my second set of 4 cells arrives. The "holding time" does relate to the available balancing time. Which as you say shouldn't be a problem when the cells are all nice settled and cycled a few times with a top balance each time. They are not yet in that state and are still settling.

Current MPPT (EPEver Trace 4210) can put out 40A, but the panel currently tops out at 22 Amp into the battery. 0.22C charge max. The total duration of that is going to 4 hours or so (insolation angle). I do realise that is slightly over-powered but another 4 cells are on the way which will then half that current.

I knew of low current overcharge in a different frame of thinking with LiPo high discharge cells and charging them to 4.20V until taper cut off. It was made very clear that time spent "at the danger zone line" is time spent in danger. (Duh right?). It wasn't so much cycle life (although you did try), it was more about not setting fire to something or having your LiPo pouch cell turn into a fire spitting baseball. Charge them to 75% the evening,morning before, then top them up to 4.20V just before you use them, with a small cooling period. If you don't end up using that pack, discharge back to 75% (or less) before you leave and NEVER leave fully charged LiPos in your car. It will not stay at 4.20V as it heats up and cars in the sun can get pretty damn hot. Back then, most of us learnt just how serious the warnings were by the sad posts from teenagers who burnt half the house down, the burnt out cars, the gutted garages etc. Still to this day I will not charge one of these packs unattended. If I do more than go for a bathroom break, I stop the charge and fully disconnect the whole pack. It's not about cycle life, it's about house insurance

Having 4 and soon 8 of it's much, much larger cousins charging unattended, lets say I prefer they are in the garage on a concrete floor with nothing but themselves and brick wall to burn and nowhere near 4.20V!

I was hoping to try and assess what magnitude of damage an hour at 3.65V and quiescent would do, every day. On one hand it could cost a cycle. So a 4000 cycle pack ends up lasting only 2000 cycles until its 80% rated (they still use that dont they?).

However, I'm resolving to the fact that finding that out, to any certainty is near on impossible. Unless I want to offer my pack up as test, split it in 2 to give a control and abuse the hell out of it.... for years. And that is then only valid for "my" pack.

The other option, which I think most people are taking is to consider it "could" be that bad. Unlikely, very unlikely in the extremes, but as we can't really get a good "order of magntiude" on it. It's probably best to assume it's bad and avoid it if we can. Sleep easy.

Making that all work with the particular MPPT and BMS I have is probably a topic for another thread.

Just for my understanding, what is the difference between a regular overcharge, like charging to a voltage that is too high, and a low current overcharge?

 

[venquessa]

Just for my understanding, what is the difference between a regular overcharge, like charging to a voltage that is too high, and a low current overcharge?

Anywhere beyond what the manufacturer states as "Max charge voltage" is considered "over charging" in the classic sense.

The "low current overcharge" phenomena relates to holding the cell above it's "VoC" or open circuit voltage. A fancy name for it's unloaded rest voltage. Usually taken 10-12hours after a full charge. If instead of terminating the charge at, for example, 3.65V and the current has fallen away, you just keep on supplying 3.65V for the foreseeable, such as if the MPPT controller or battery charger does not support termination. Does it damage the battery? "Low current overcharge" is the suggestion/claims it does.

I think that summarizes it?

 

[venquessa]

You can do better than that. You can get those from a pretty reputable seller for $184 each including shipping.

I was coming around to this shortly. When I first went looking to buy lithium for the solar system I think I backed away two of three times, because picking which cells from which seller and all the horror stories out there. In the end after putting it off for 2 years I just picked what looked popular by orders and a price that looked nice and winged it. The jury is still out, but I bought with no expectations of quality and was even happy to accept they turn out to be used.

However. Now I got that out of the way and my toes in the water, and that I'm now stuck with buying 8 at a time, looking back at the market, I feel I should do a bit more research this time. Paying £1000 for 90% of what you ordered if it should have cost you £2000 is one thing (B Grade etc.), it's if you paid the full £2000 and get screwed over with 90% and they will only take them back if you pay the £350 shipping.

 

[Robbert]

Anywhere beyond what the manufacturer states as "Max charge voltage" is considered "over charging" in the classic sense.

The "low current overcharge" phenomena relates to holding the cell above it's "VoC" or open circuit voltage. A fancy name for it's unloaded rest voltage. Usually taken 10-12hours after a full charge. If instead of terminating the charge at, for example, 3.65V and the current has fallen away, you just keep on supplying 3.65V for the foreseeable, such as if the MPPT controller or battery charger does not support termination. Does it damage the battery? "Low current overcharge" is the suggestion/claims it does.

I think that summarizes it?

That is a clear explanation.
So if I understand it correctly, this only occurs with higher voltages per cell.

So charging up to e.g. 3.4x and floating somewhat lower (or even a bit higher if following the recommended settings as posted by Will Prowse in the DIY battery section) wouldn't be an issue.

 

[venquessa]

I think the general concensus is the higher above VoC the more likely the damage.

The 3.4xVpc float voltage is based around that 3.4xV being the true rest voltage of the cell. At that point, in theory, the only current it will consume is it's self discharge which we know is tiny.

My 4 cells to take me up to 24V arrived literally as I typed this message 3.22V each. I was going to match them up and put them straight into service, but given where we are, I might as well bench charge one and play. I suppose a proper capacity test of a single cell might be in order too as "jury is still out" on these "Varicore" cells.

 

[sunshine_eggo]

That is a clear explanation.
So if I understand it correctly, this only occurs with higher voltages per cell.

So charging up to e.g. 3.4x and floating somewhat lower (or even a bit higher if following the recommended settings as posted by Will Prowse in the DIY battery section) wouldn't be an issue.

Funnily enough, your example voltage is the voltage of concern. There are claims that 3.40V permits low current overcharge.

I have many cells that will fall into the 3.3XX range while retaining 100% charge.

The generally accepted, "it won't happen" voltage is 3.375. Not that it's definitive by any means, but this is the default float voltage then one selects the LFP profile for Victron MPPT.

 

[Ampster]

1) LFP used to be charged to 4.2V. It was found to be suboptimal. Cycle life was notably improved by reducing to 3.65V while only losing a few % of capacity. This was done for years.

Thanks for that. I had always wondered about that when I saw it it specs. In the past I was charging to 3.45 V per cell.

There are claims that 3.40V permits low current overcharge.

Recently my SolArk has had a blip where after a few hours of charging the current rises and Voltage goes up. I have a ticket with SolArk and now I have a good reason to want to insist that SolArk finds a solution. In the meantime I have lowered the CV setting to 3.38 per cell so when that blip happens I do not risk low current overcharge.
For context my pack is 840 Ahs and the current when this happens is only 35 Amps. My loads are almost the same but as far as I can tell the solar is covering the loads, separately sending some to my batteries and exporting the rest to the grid .

 

[venquessa]

The generally accepted, "it won't happen" voltage is 3.375. Not that it's definitive by any means, but this is the default float voltage then one selects the LFP profile for Victron MPPT.

I suppose one could measure the VoC of their cells. However, that could be subject to age, cycles and temperature and will probably change based on how well "fed" it was on it's last charge.

I am running a "pigeon" experiment. Not hoping to show anything but for my own quick eye balling I want to "guage" the current at various hold voltages. It's a good job you spoke up as I was about to start at 3.40V! It's just about to blow through 3.350V. It will be very rough as I'm not going to be waiting the appropriate amount of time, and I'll probably cut some corners to save time. All I hope to see is, does 3.650V draw 10% more or 100% more than say 3.4.

 

[Ampster]

I am running a "pigeon" experiment. Not hoping to show anything but for my own quick eye balling I want to "guage" the current at various hold voltages.

I would expect over time you would see current taper consistent with a typical charge curve. If you do not wait enough time to observe the current taper, I would guess that at any point in time the current would be inversely correlated to the SOC of the cell. Of course you would need a good Coulomb counter to know what the SOC is unless you wanted to use a typical charge curve mentioned above. Let us know if you find anything meaningful and what you think we might be able to do with that information.

 

[wattmatters]

Let us know if you find anything meaningful and what you think we might be able to do with that information.

That's the crux. Getting actionable intelligence.

Homebrew tests/experiments can be a good learning exercise but for validity and wider applicability they need a lot of controls put in place and to be repeated over a large number of cells and test conditions.

In the meantime...

#1 is what the server rack folks do.

Ultimately my LiFePO4 batteries are purchased for doing regular work, so keep voltage in check at either end and just cycle them. Calendar ageing is probably more the enemy than cycle count will be, so I figure just put 'em to work. That's what they are for.

 

[Ampster]

Homebrew tests/experiments can be a good learning exercise but for validity and wider applicability they need a lot of controls put in place and to be repeated over a large number of cells and test conditions.

I agree, I am happy to use a typical charge curve and stay out of the knees. I have been messing around with Lithium batteries since 2011 and not much has changed. There has been some change in the value equation. I started with Automotive grade for an EV conversion at $300 per kWh and now have a pack of Grade B cells that cost less than $125 per kWh.

 

[wattmatters]

I don't expect to ever get to the lower end of voltage (rarely do I go below 50 V, only in long grid outage scenario would that occur).

At the upper end I go to 56.4 V (3.525 V/cell) as I operate a hybrid LiFePO4 + SLA battery. My SLA specs are to be charged up to 2.35-2.40 V/cell at 20°C (56.4 - 57.6 V).

So I settled on 56.4 V as not too high for the LiFePO4 and not too low for the SLA. If I were using only LiFePO4 then I'd probably drop back to 55.2 V (~3.45/cell).

The SLA does very little cycling. It's my backup reserve.

So when my battery sits at 56.4 V for longer periods (>1 hour) in reality the charge current is pretty much all going into the SLA.

I thought about putting a separate shunt on each but CBA*. It just works. The past seven days:



15 Apr in more detail as that has the longest period at higher voltage:



And the period sitting at higher voltage:



This is into a battery with ~ 400 Ah capacity. 1-2 A. Mostly SLA. I've checked it with clamp meters (to the extent they are able to give a reasonable reading). The LiFePO4 might see a current in the order of a dozen mA, or often zero.

* I will however be setting it up so that such an addition is easy to do.

 

[sunshine_eggo]

* I will however be setting it up so that such an addition is easy to do.

I thought you were going to define CBA here...

 

[venquessa]

Homebrew tests/experiments can be a good learning exercise but for validity and wider applicability they need a lot of controls put in place and to be repeated over a large number of cells and test conditions.

Agreed. I wasn't hoping to prove anything, just experimenting for my own interest mostly.

CC @ 20A to 3.350 and CUT         - 5 minute settle VoC: 3.307 -43mV
CC @ 20A to 3.350 and taper to 5A - 5 minute settle VoC: 3.329 -21mV 
CC @ 20A to 3.400 and taper to 5A - 5 minute settle VoC: 3.354 -46mV
CC @ 20A to 3.450 and taper to 5A - 5 minute settle VoC: 3.371 -79mV
CC @ 20A to 3.500 and taper to 5A - 5 minute settle VoC: 3.392 -108mV
CC @ 5A  to 3.600 and CUT         - 5 minute settle VoC: 3.435 -165mV
CC @ 5A  to 3.600 and CUT - recovery time:  4 mins (72Wmin)                                                                                                                                               
CV @ 3.600V to 0.01C, 1 Amp       - 5 minute settle VoC: 3.521 -79mV

Rest at +10 hours = 3.347

Not recorded was the timings. Subjectively however all of the first 3 charge phases took a considerable time. Going up from 3.400 took very little time. Nothing really learned there that I didn't know, but it was still nice to get a feel for the "elasticity". Again subjective (partially), but compared to Lipo/LiIon the draw at the upper end is higher than I expected. They seem to have a lot more "momentum" if that makes sense. Maybe that is just the size of the cell, the next size of cell down I have tested was 8Ah!

In the 3.500-3.600V test I noticed that it actually took a significant time to get back up to 3.500V even. So I repeated that test so see how much energy was simply "spilling back out" so to speak. It took it 4 minutes at 5 Amp to get back up to 3.600V. That's 72W minutes, or more worryingly a apparent/pseudo 4 Amp self discharge rate!

Pushing it to 0.01A at 3.600V took quite a long time. The PSU I had was not accurate enough in it's CC/CV modes and I had to manage the tapering manually to keep it pinned on 3.600V +/- 2mV. I don't think there was much further to go before quiescent as sitting at 1 Amp it stayed there for nearly an hour oscillating between 1.10A and 1.01A. When I switched to constant current 1Amp, it then sat at 3.599V, 3.600V breifly and back down again. I waited until it touched 3.601V once and called it.

Rest voltage looks a little lower than expected, but it is a brand new cell.

I was hoping to record some graphs and stats, but unfortunately the charger wouldn't play ball on USB.

The most interesting outcome was seeing this little fella when I opened the scope. It's inductor ringing from the PSU. The little (~130Mhz) ring on the crest of the larger (600kHz?) ringing peaks way, way over the charge voltage. The cursors in the second are trying to show how much of that spike and how long is it over 3.650V. This is also why I did not push up to 3.650V as the spikes on the scope at that point would be over 4V!

It is a very, very short pulse which breaches the 3.65V line. I'm not that concerned, however I will be scoping the MPPT to see how good/bad it is! Should be noted that 130Mhz is outside of my scopes nominal bandwidth, so there could potentially be another +3dbV on the top of that!

 

[venquessa]

The setup.

The assistant:

 

[venquessa]

I started with Automotive grade for an EV conversion at $300 per kWh and now have a pack of Grade B cells that cost less than $125 per kWh.

I was reading up on Grade B and as I understand it, they are the cells off the same production line which failed one of the far higher standards tests required for EV use and then get grade "B" and resold to the likes of Ali* I don't know the validity of this, but I have seen at least one person claiming Eve don't distribute any grade A cells via Ali* and all Ali* cells are B grade. Eve are supposedly going to laser etch the QR code with a "B" so it can't be used before distributing them. Of course the resellers will just re-modify the QR code and claim it's grade A. I actually searched for "Grade B Cells" on AliE and how surprising was it to find almost nobody selling them. Yet if you go to a supposed authorised local dealer (and pay the premium), they do offer both A and B grade cells.

I have no references for that, so consider it "heresay" and "rumour mill".

Anyway. I am considering buying grade B cells from an supposedly authorized European supplier, they are more expensive than AliExpress cells, but they also arrive in a week, not 8-10 weeks! It's also far more likely you will get a coherent response to questions around the cells validity.