Might have a bad cell, looking for advice on deciding

[zamboni]

Because i didn't get a logging charger, i don't have my own curves to consult - that would certainly be a nice thing to have as i suspect these cells have some "personality". The manufacturer's curve in the datasheet is a little crazy, it has a much shallower 'knee' than usual and goes up to 3.9V at fully charged. The flat part is also shifted up, with 3.4 landing closer to the 50% point. Not that i really believe it.

IMO it is not worth trying to keep a battery any better balanced.

Thanks for the feedback, it didn't occur to me to look at it from that perspective. I wasn't so much concerned that the balance isn't good enough, more like i thought i balanced at the wrong voltage...i thought maybe the curve for my cells was still too flat at 3.4V and my balance was imprecise.

Why are you charging to 14.6V? Is it something to do with the low temperature you are charging at?

The unusual winter weather has passed and with the trailer back up and running the cells stay nice and warm even when it does get close to 32 overnight. So temps aren't part of the equation right now.

I chose 3.65VPC charging voltage based on what i saw while working with the individual cells: Initially i tried charging with 3.5V, but the rested voltage was settling around 3.3, and i was getting extremely low capacity back out of them - like half. I figured they must have kinda high internal resistance and needed higher charge voltage. When i started charging with 3.65V they did come to 3.4 resting. I carried that forward into my solar charge controller settings.

Now that i've typed that out, i think i made a mistake by forgetting the low temps during individual cell testing -- i should probably leave the balance alone and just dial the charge voltage back. Tomorrow will be good sun so i'll try your suggestion of charging at 3.45 and check balance, then i'll try increasing charge voltage in steps and watch what happens both with balance and how much more energy the pack accepts. Maybe i just aimed too high.

-Jerud

[karrak]

Low temperatures have a large impact on the charge/discharge curves for LFP cells. Look at the discharge curves for Winston LFP cells at different temperatures in the top left hand graph. I would think it is more than likely that you would have to raise the charge voltage at low temperatures. Not sure it is good for battery life.

Charge and discharge current also has a large impact. The figures you give are quite plausible with charge currents of greater than or equal to 0.5C. Usually these tests are done at 0.5C. Do you have the link to the charge discharge curves you are talking about?



If I get the time in the next few weeks I might try running some tests at 0C^o.

Simon

[zamboni]

When i was testing individual cells the charger was set to max 40A for both charge and discharge, that's 0.2C.
At the pack level my in-use charge/discharge rates are also pretty gentle; the Classic maxes out at 99A which is only 0.1C, and my inverter breaker is 250A though i rarely draw over 200A; 0.2C - 0.25C. So high Vdrop while charging due to high currents isn't what i've got going on. I'm remembering that Ri increases as the cells get full (hence the knee), so if i'm having to set charge voltage higher then either my cells have a shallower knee than typical or higher Ri than typical.

I charged at 13.8V today, but went into float right as the sun hit the treetops so didn't get to try higher charging voltages. With the charge controller off and nothing running, voltage dropped almost immediately to 13.2. This mirrors what i saw in single-cell testing. Tomorrow is rain but Wednesday i'll see how many more Ah the batter will accept with progressively higher charge voltages - that will tell me if i'm trying to push too close to "full".

I have only been able to find one datasheet on my cells, and the curves are not very well made. I'll try to attach screenhots of the curves, never done images before so hopefully don't mess it up:

discharge at 3C perhaps.png
charge-discharge.png

[karrak]

I'm remembering that Ri increases as the cells get full (hence the knee), so if i'm having to set charge voltage higher then either my cells have a shallower knee than typical or higher Ri than typical.

Not sure that this is correct. All batteries are far more complex than just an ideal voltage source plus internal resistance. This following graph comes from a good paper on LFP batteries.



Simon

Off grid 24V system, 6x190W Solar Panels, 32x90ah Winston LiFeYPO4 batteries installed April 2013
BMS - Homemade Battery logger github.com/simat/BatteryMonitor/wiki
Latronics 4kW Inverter, homemade MPPT controller

[PNjunction]

I have access to free shore power for another couple weeks but then i'll be on the road and fully off-grid again. So now is the time to rebalance to a higher voltage (maybe 3.5?) if that's what i need to do.

3.4v per cell is too LOW to do proper balancing at. 3.5 would get you closer. 3.55 (with an attendant taper to .05C maybe .025C) would be my preference with the PL8. Yep, that'll take time, but it has to be done.

Remember, 3.45v gets you into the absorb knee, and the higher you can get the cells into the knee without going over the top (3.65v maaax individually monitored) will get you closer to a good top-balance.

So no big deal - you're close, but no cigar yet. Not unexpected when you balance at too low a voltage.

For the notebook next time: When you start doing massive parallel cells like this (5p I think?), what really should happen is that you measure each cells capacity and internal resistance, and match them up as best as possible as mates for each other when you parallel them.

In some cases, the quick method is to charge each cell to full, let them sit for a month or two, and charge them back up noting how much current was used to get them back to a full charge. Pair up the cells as close to each other depending on how much current was used to recharge. It won't be much.

At this stage, let me point you to something similar - the custom wheelchair guys who purposely use very high-current cylindrical Headway cells (prismatics aren't good enough):

http://www.wheelchairdriver.com/boar...php?f=2&t=1813

You'll find this just as interesting as the marine LFP threads, and you may find it useful since you are paralleling so many cells. It's a PITA compared to using larger prismatics for our relatively low-demand application. Op Burgerman has some great PL8 programming files for LFP.

[zamboni]

Thanks again for the info...sorry i'm always late to respond; for some reason i never get emails on my own threads even though i'm "subscribed"? I get emails on other subscribed threads but not ones i start.

So much rain recently...been taking lots of advantage of the whole PSOC situation with lithium: ) Just got sun again but it'll take a couple days to reach full since my array is undersized vs. battery capacity. I plan to step-up charge voltage and check resting voltage/balance at each step, to get an idea of where i am right now. Then once the pack is as full as i dare, i'll plug in the balance charger and let it run for a few days (or whatever it takes!) to get up to 3.5VPC balance point. That is, assuming i don't see any counter-indications along the way.

@Karrak: Thanks for the paper, i'm definitely going to read it, though my LFP learning is currently on hold while i deal with taxes and some other unfortunate "real life" stuff. I do want to understand more about how batteries really work - as you say, nothing is ever as simple as the idealized simplifications we normally use...

@PNJunction: sadly i squandered my opportunity to match these cells better. When i picked the cells up from the supplier, he let me use his shop to do some tests on them, including using his 4-wire resistance tester (HIOKI BT 354-01). I paralleled them all overnight then picked out the lowest-resistance cells from the batch -- but stupidly did not record the resistance of the individual ones for matching. But i recall the meter only read out to one decimal so maybe it wasn't even accurate enough? Sounds like if i were to do this again i'd be better off following your described procedure anyway...is there any such thing as a meter which can directly measure Ri accurately enough to match cells?

Thanks for the powerchair link, i will definitely spend some time in there learning more.

[Sunking]

Not sure that this is correct. All batteries are far more complex than just an ideal voltage source plus internal resistance.

Of course you would say that, you are a fraud, and do not know what you are talking about. No battery is any more complex than an open/loaded voltage and resistance.

[Sunking]

is there any such thing as a meter which can directly measure Ri accurately enough to match cells?

Yes several Hobby Chargers made for RC Planes and Toys will read down to .001 Ohms. Or you can buy an ESR meter. You can even do it with some power resistors and a DMM. Click on this link and there is a wealth of information on Davids website. He wrote the bible for lithium batteries. You will find a ton of white papers, and every lithium battery and BMS ever made. Once you read through the info you will discover how big a fraud Karrak is. Example this link from David's site on Internal Resistance.

[zamboni]

Well, as soon as i upped the charge voltage from 13.8 to 13.9V, bank "B" voltage shot up quickly so that made re-balancing at 3.5VPC my new priority. I put the pack on the GT1200 and it worked all day yesterday to get the other 3 banks up to 3.430VPC after drawing bank "B" down to 3.500. Today i resumed the balancing...and then the magic smoke came out of the charger!?!?!?! I'm shocked, there was zero change to the charging setup, i checked all my connections both before and after the smoke display, the PSU is fine, the cooling ports of the charger were not obstructed (it wasn't even hot). I'm really bummed; i thought Revolectrix was supposed to be good quality stuff. Will see what they say when i call support tomorrow AM.

In the meantime, my conclusion is that i simply did a poor job balancing my pack the first time through, which would explain my most recent voltage-related observations.

SunKing, you mention that some hobby chargers will measure Ri accurately, but i think i've also seen you say that Revolectrix uses a poor measurement method in their chargers. They make no claims at all to the accuracy of their Ri measurement, but then again i don't see that for other brands, either. How can i know if an Ri measurement is to be trusted?

[karrak]

This is interesting that you are finding the battery going out of balance so quickly after doing an initial balance. Lostinspace who posts on the WindSun forum is having a similar problem with cheap LFP cells that I can't remember the name of purchased from Electric Car Parts Company in the US. I think that either some batteries leave the factory before they have undergone enough testing or that certain charge/discharge regimes especially when batteries are new can lead to a rapid deterioration in balance or it could be a combination of these factors.

I have been asked to make up two new off grid LFP based systems and rather than doing the balancing by putting the cells in parallel I have assembled one of the batteries as a 2p15s battery and when we get some sun will set the charge controller to 3.35V/cell and see what the balance is at this voltage. If there is any imbalance I will adjust the individual cells by taking charge out or putting charge back into the cells that are out of balance. I will then slowly increase the charge voltage to 3.6V/cell adjusting the balance as I go. As a safety measure of course I have a BMS connected to the cells which is connected to a Raspberry PI running my battery monitoring software.

I wouldn't get too hung up with battery RI. For the two systems five year old systems that I am responsible for I have never found this to be a problem. Both systems with a load of around 0.5C-0.6C only have an imbalance of ~40mV at these high discharge rates. I haven't done any cell matching with both these systems. Sure if I wanted to win the world speed record for a wheelchair or golfcart I would look at this but not for an off grid system.

Simon

Off grid 24V system, 6x190W Solar Panels, 32x90ah Winston LiFeYPO4 batteries installed April 2013
BMS - Homemade Battery logger github.com/simat/BatteryMonitor/wiki
Latronics 4kW Inverter, homemade MPPT controller

[zamboni]

I don't think the cells are getting "out of balance", it's that i did a poor job balancing them in the first place -- my balance voltage was 3.4 which wasn't an adequate "top balance", so one bank of cells (probably due to just 1 cell within it) is weak and is ramping up before the rest of the pack. Until that happens, the cells are within 10mV of each other, usually less.

A major factor here - and it's not like i wasn't warned about this - is simply how much time is required to correctly and thoroughly work with a pack this size. It's "easy" work but time consuming especially when the primary tool is a hobby charger that can only put out 40A. It simply consumes a lot of calendar days. I have ended up stuck here much longer than originally planned (not due to battery issues) so i now know i _could_ have spent a couple weeks longer and measured each cell individually. But when i was doing the testing i didn't think i had that kind of time.

- Jerud
------------------------------------------------------------
1220W array / 1000Ah LFP house bank
MidniteSolar Classic, Magnum MS2812
ME-RC, Trimetric, and JLD404
2001 Fleetwood Prowler 5th wheel 25 foot, self-rebuilt
Full-time 100% electric boondocking (no propane, no genny) since 2015

https://www.livesmallridefree.com

[zamboni]

A little closure: At least I hope so; this thread is pretty dead but it seems like whenever I open my mouth, a foot flies in...

While rearranging the cells in my pack to physically match Ri is not practical, I did get my hands on a PL-8 (after much drama) and performed a 4s balancing charge on the pack. The "bank" which had been "taking off up its knee" was clearly much farther up its SOC curve than the others. This was definitely causing the voltage spike seen on the CellLog8.

I re-balanced to 3.55VPC with a taper to C/20. Then I rested for 4 hours, and did a balanced discharge from there at 250mA to 3.50VPC. Another rest (only for an hour but it was a pretty gentle discharge after all) and cells were at 3.508, 3.515, 3.509, 3.509 (per my Amprobe).

For the last week I've been using 14.0V to charge and having no trouble getting the coulomb-counter to return to 0Ah (that was not happening before). The bank voltages are staying within 0.02V (per CellLog) at all times. I feel a lot better about all this now. If I let the controller go to float (13.7V) then disconnect the pack and rest it for an hour, I get bank voltages around 3.43V. Yes, I know that's not a real rest. I'm only trying to use 700Ah out of a 1000Ah pack and stay well away from both the knees - so this makes me feel like I'm accomplishing what i set out to do.

-Jerud