Announcement

Collapse
No announcement yet.

Might have a bad cell, looking for advice on deciding

Collapse
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • Might have a bad cell, looking for advice on deciding

    I'm top-balancing my pack of 20x, 200ah LFP prismatics to 3.5VPC. The pack is built in 5p4s configuration for a 12V pack at 1000ah, but of course i have all the bus bars off for the balancing. I'm using the Revolectrix GT1200...similar to a pL8 but with fewer features. The pack has been in service for 9 months. I had been monitoring cell voltages manually and realized one of the parallel banks was running higher than the others, that's when i began to understand i had not correctly balanced in the first place. Yes, vampire boards were (briefly) involved...they are gone now...please no lecturing on how dumb i was. I'm just trying to make the best of things now.

    I first wired the cells in parallel and charged with 3.5V, then rested them for several days (scheduling issue, not procedural). Then i made a second pass, charging each cell individually. Charge voltage was 3.6; max current 40a and cutoff current was C/20 (too low; i failed to update a setting). I'm checking behind the GT1200 with my Amprobe 560 for voltage measurements and the two readings agree well. I plan to make a third pass, using the discharge feature on the cells that are coming in too high if necessary.

    The rested OCV for all the cells before individual charging ranged from 3.315 to 3.541. After my second pass, the OCV (after 9 hours resting) ranged from 3.415 to 3.565...except for cell #5, which remained at 3.358. #5 had started out at 3.316 - almost the lowest of the pack. This would not bother me too much except: While no other single cell took more than 15 minutes to complete its individual charge, #5 took 3.5 hours. I did not log the Ah each cell accepted but evidently #5 took a lot more than the rest.

    None of my other observations indicate any special trouble with #5 except the sheer number of Ah it accepted without breaking 3.4V. I'm trying to decide if this indicates an actual concern, or if it falls within normal (albeit kinda unlikely) variation. With the flatness of the charge curve, i suppose it could simply have been far more discharged than the others? I'd be grateful for insight or advice, thanks.

  • #2
    Right - quick review

    Any *during charge* voltage from 3.45 to 3.6v per cell will result in a fully charged cell given enough time. 3.45 takes a long time, and 3.6v the shortest absorb to .05 or lower. Just make sure that each cell stops at the same current cutoff.

    *AT REST* a fully charged cell will settle to anywhere from 3.38 to 3.45v. SO, if your cell which has rested for at least 12 hours is higher than 3.45v, then it is OVERcharged, and needs to have some sort of discharge applied to it as soon as possible. If your cells are just a bit mismatched at this point, then a light bank / pack level discharge will get those overcharged cells back down into a normal range.

    So, let's start over with KISS. You've already charged to 3.6v with a .05C absorb current cutoff. That's good enough to ensure that the electrolyte has spread evenly over all the internal surfaces so that you don't create any hot-spots. (This can happen to those who never take their cells to 3.6v at least once in their lifetime!).

    So, no need to go to 3.6v for the time being - especially so since some are overcharging. Charge each cell individually at 3.5v, but still with the ..05C (or whatever the revo stops at) absorb current cutoff. It''ll take a bit longer at 3.5v, but not much.

    The thing we haven't discussed is your wiring!!

    What sort of interconnects are you using between the cells? Are they snug? Did you CLEAN the terminals before tightening them down? That is, did you take a NON-METALLIC scrub pad to the aluminum contacts and nickel-plated jumpers, or are you using something else for interconnects? It is SUPER important that these are clean. Many apply a very *light* coating of no-alox or other anti-oxidant conducting compound after cleaning to the interconnect surfaces. Because the internal resistance is so low with LFP, any poor connections with just slightly varying resistance of their own can throw your bank out of whack. Did you man-handle any of the interconnects - over-torque them possibly?

    Seriously, with a 5P/4S bank, that a LOT of scrubbing!! And, have you done any sort of external strapping to keep the terminals from getting side or vertically tweaked in torque by being hit by something accidentally?

    Ok, if that is good, you can also do an individual capacity test with your Revolectrix. Unlike lead-acid, you can hammer them with a big load if you have to and get nearly the same results in much quicker time, but be sure you have your LVD set in the Revo!

    You could start out in a bank/pack mode, and babysit like the dickens. Or do an individual cell discharge test if the Revo can count coulombs.

    Thing is, what about the REST of your 5p/4s cells. Are they uneven too?

    You can do it if you take your time. Admittedly, you bit off a HUGE bit of LFP with that large of a bank with no learner-bank experience. But we'll get there.

    And please tell me you aren't dumb like 99% of the install videos where the ops are using bare-tools! For safety's sake, TAPE ALL YOUR TOOLS TO THE HEAD when working on LFP, (or any battery really). All it takes is a slip of the handle across two terminals to create a nightmare.
    Last edited by PNjunction; 01-30-2018, 04:17 AM.

    Comment


    • #3
      Originally posted by zamboni View Post
      I first wired the cells in parallel and charged with 3.5V, then rested them for several days (scheduling issue, not procedural).
      When you said "rested them for several days" were they still connected to the charger at 3.5V? To do a proper commissioning balance you should leave the cell charging in parallel at a voltage of at least 3.5 volts (preferably 3.6V) overnight or longer. This will give you the best results. After this you should assemble the battery in its final 5p4s configuration.

      None of my other observations indicate any special trouble with #5 except the sheer number of Ah it accepted without breaking 3.4V. I'm trying to decide if this indicates an actual concern, or if it falls within normal (albeit kinda unlikely) variation. With the flatness of the charge curve, i suppose it could simply have been far more discharged than the others? I'd be grateful for insight or advice, thanks.
      I think you are right that cell #5 could have been more discharged than the others. The question is why is it more discharged? Is it suffering from higher internal leakage than the other cells. This can occur if the cell is damaged at some stage by taking it below 2 volts.

      If you have the time I would be temped to to do a capacity test on four or more cells including cell #5 to see what the variation is. First I would charge the cells you are going to test to 3.6V in parallel and let the charge current taper down to less than C/100. I would then disconnect the cells from each other and leave them for a week or so. I would then do the capacity test by discharging them individually down to 2.5V and then recharge them.

      Simon
      Last edited by karrak; 01-30-2018, 07:20 AM.
      Off-Grid LFP(LiFePO4) system since April 2013

      Comment


      • #4
        Originally posted by PNjunction View Post
        So, let's start over with KISS. You've already charged to 3.6v with a .05C absorb current cutoff. That's good enough to ensure that the electrolyte has spread evenly over all the internal surfaces so that you don't create any hot-spots. (This can happen to those who never take their cells to 3.6v at least once in their lifetime!).
        Thanks for that interesting bit of information PNJ. I haven't heard of that before. I would be interested in more information on this. Do you have a link to a scientific or technical paper on this or the source of this information.

        Simon

        Off-Grid LFP(LiFePO4) system since April 2013

        Comment


        • #5
          PNJunction, you obviously see that i'm trying to follow your KISS post. I stayed up late last night reading through yet more of the forum which i had not previously found or had not been able to understand the first time i read it. More is making sense, i think. Other things that i thought i understood are getting fuzzier...

          What i think i've learned is that the charge curve for LFP cells is a roadmap but not an absolute; any individual cell's curve will look very similar to that shape but with shifts in voltage - much like the way individual cell capacity may differ by 10% from rating. Or in other words, voltage - even OCV after several days of resting - is not an "absolute" SOC indicator. You charge, you absorb, you taper...and what you end up at is "full" for that cell. If you want to reliably hit, say, 90% for that cell, you have to do a coulomb-counting discharge to measure 100% in Ah, then recharge by 90% of those Ah and now you know the 90% voltage for that cell. I never thought i'd miss my hydrometer this much : )

          What i also think i learned is that how you get to "full" affects what the "full" voltage is - whether you creep up at 3.4V and current slowly tapers out, or stuff in a lot of current at 3.6V and then quickly taper. This is mind-blowing stuff for a mechanical engineer, to have these path-dependent quantities.

          So i can accept that, after charging my cells to "full" at 3.6V with taper, they will all register different voltages. I can accept that only one cell out of 20 was substantially more discharged than the rest, because my interconnects were suboptimal and contact resistance could have been playing havoc. It's still kinda hard to accept this one cell coming to rest at 3.358 when being charged with the same program that took every other cell over 3.4 and in fact "overcharged" several to 3.5 (i discharged those all down to 3.45). I ran a second charge cycle on #5 and let it rest overnight, it's still only at 3.366V, and that charge cycle lasted less than 5 minutes.

          I feel that i must be making some simple, bad assumption that is hamstringing my full understanding though. Most charge curves i see go to around 3.65V. If we're saying a "full" cell will take 3.6V+current taper and end up around 3.4V...then how does that cell ever reach 3.65V for the curve? Am i hearing "full" and (incorrectly) thinking "100%SOC"? To me, "overcharged" means the cell is holding all the energy it can, and you kept trying to charge it so the voltage was forced upwards, but the chemistry isn't going to give you any more capacity as a result because you're damaging the cell. That's not what's happening at 3.5V is it? Are we using "overcharged" to mean "charged to a higher SOC than you probably ought to"? Are charge curves showing voltage while under charge or OCV?

          Having said all this, now i am confused about how top balancing works (dammit, i thought i knew that one). Bottom-balancing involves doing a few discharge/rest cycles (at maybe 2.4V) until resting OCV is closely matched across all cells at some value near 2.5V and calls that "empty". That's not really 0%SOC for each individual cell, but the purpose is to stay off the knee when discharging the pack to "empty", so that works. The fact that "full" voltage will vary across the cells is fine because it's bottom balanced and that would happen anyway. This seems clean and easy.

          I had imagined that top balancing would work the same way: Several charge/rest cycles at 3.6V until all the cells line up close to 3.5 OCV, and call that "full", which lets you charge to "full" and then float around (since we're solar) without the risk of taking any cell to the knee, because the balance point is just past the beginning of the knee so your pack voltage is most accurate where you most need to rely on it. Discharging to "empty" rarely happens if ever, and in my case the CellLog will alarm if any of the 4 "5p" banks hits 2.75V (in addition to a 12V pack-level LVD). In actual use, 3.5VPC is the HVD setting; the charger is set for 3.4-3.45VPC; we don't actually want to charge the cells that close to truly 100% because we want to float and preserve cell life...

          This all seemed really clear until i actually tried it...But now i can't reconcile that with the 3.38V - 3.45V guidance for "full" OCV, nor with stating 3.5V as overcharged. Balancing at the bottom is defined by cell voltages and consequently the pack voltage is going to be super close to 10.0V at "empty". But balancing at the top is defined not by voltage but by having charged a certain way, and it's anybody's guess what the pack voltage will look like when it's bolted up because the individual cell voltages are all over the map between 3.38 and 3.45. If i have not actually brought all the cells to +/-10mV at 3.5VPC before connecting them together, how do i know that a pack voltage of 14V actually means 3.5VPC - the way that i know 10V means 2.5VPC in a bottom-balanced pack? Am i thinking myself in circles?


          When you said "rested them for several days" were they still connected to the charger at 3.5V? To do a proper commissioning balance you should leave the cell charging in parallel at a voltage of at least 3.5 volts (preferably 3.6V) overnight or longer. This will give you the best results. After this you should assemble the battery in its final 5p4s configuration.
          No, this was a totally disconnected rest because i was looking for OCV and was planning on making subsequent passes with the charger on a per-cell basis. The first pass of charging was done as 4 separate charges on each bank of 5 in parallel. My charger/power supply combo won't push enough current to parallel all 20 at a decent rate and i didn't want to run a 3-day long charge when i knew i'd have to leave the premises a few times.

          You may recall that i was originally hoping to individually capacity-test all 20 cells before rebuilding the pack...that dream has obviously evaporated due to the time involved (this insane winter on the east coast is not helping). But i do lean towards the suggestion of discharge-testing a few (including #5) to see if there's more going on. Paralleling 5 of them and tapering to C/100 @ 3.6V before discharging...what is that accomplishing? I thought such a small taper current was a bad idea? Why does it matter if the "capacity" i'm going to measure is the re-charge Ah from 2.5V to our definition of "full" (3.38 - 3.45 after taper)? And the week of resting...that's a long time. What does that buy compared to say, only 24 hours' rest?

          I really don't think this cell went down to 2.0V with my pack LVD set to 12.0V...but i realize it is possible because it was paralleled with 4 others. What is the test for internal leakage?

          Thanks for all the concerns and input about the pack and safety. Yes, i definitely tape my tools. The cells are strapped together and clamped down with allthread, with 1/8" spacers between each cell for airflow. The previous interconnects were poor; i had sourcing issues at the time and couldn't get what i wanted to use. The reassembly will be an improvement: Yes, nylon brush the terminals. Yes, noalox immediately after. Yes, nickel-plated copper bus bar and SS fasteners. Where i previously had a series of 2-hole copper bus bars layerered over each other at each terminal like shingles, i am instead using 1-piece bus bar with multiple holes drilled along it. I have Erico "flexibar" which i will crimp a slight bend into so it can take up physical movement without stressing the joint. I have been using external-tooth lockwashers and have yet to have any back off on me; i torque until i feel the teeth bite and then stop. Nordlocks would be cool but for now i'll settle for regular torque inspections.

          And yes, that is a LOT of scrubbing and assembly. I had to originally drill and tap every one of the 40 battery terminals. Thankfully i had a drill press for the drilling but didn't trust the press to tap aluminum so i did both taper and bottom-tapping by hand. It is not an ideal "learner" bank for sure, but i had an opportunity and this is what it looked like : )

          Comment


          • #6
            I am trying to think of a physical analogy of a lithium ion battery. The best I have come up with is two sponges, one representing the anode and one the cathode and water representing the lithium ions. Charging the cell is like putting the cathode sponge under a vacuum which will suck the water out the cathode and then transfer that water into the anode sponge. The higher the vacuum (analogous to voltage) the higher the water flow (current). Discharging is a process of putting the anode sponge under a vacuum and transferring the water to the cathode sponge. The higher the load the higher the vacuum and the higher the water flow.

            When the sponges are nearly empty the flow will naturally slow down unless we increase the vacuum. The porosity of the sponges is analogous to the impedance of the cells and is the thing that limits the water/current flow. This impedance/porosity between and within different areas of each cell will vary which is why you get a slightly different relationship between voltage/vacuum and current/water flow between cells and at different SOC.

            When we charge a cell the voltage/vacuum that we place the cathode at will define the current/water flow. When the cathode is nearly empty the slower the current/water flow will become. To get the last of the lithium ions/water out of the cathode-sponge we can either increase the voltage/vacuum or wait for the lithium ions/water to drip out at the lower rate.

            It is very difficult to quantify what 100% full means. If we charge a cell to 3.45V/weak vacuum and wait for the cathode sponge to stop dripping we will find that there is still a small amount of charge/water left in the cathode sponge. If we raise the charge voltage to 4.0V/strong vacuum we will be able to extract more charge/water out of the cathode sponge but the extra voltage/vacuum will damage the cell if it is subjected to this voltage for any length of time. The difference in SOC from charging a cell to 3.45V compared to 4.0V and letting the current taper to zero will be less than 1%. It is usual to define 100% full as say 3.6V at an end current of C/20.

            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
            Off-Grid LFP(LiFePO4) system since April 2013

            Comment


            • #7
              Originally posted by zamboni View Post
              No, this was a totally disconnected rest because i was looking for OCV and was planning on making subsequent passes with the charger on a per-cell basis. The first pass of charging was done as 4 separate charges on each bank of 5 in parallel. My charger/power supply combo won't push enough current to parallel all 20 at a decent rate and i didn't want to run a 3-day long charge when i knew i'd have to leave the premises a few times.
              The way around this is to do a balance charge using your charger with the cells in series with the balance leads connected to 3.4V x number of cells in series first. Then put them in parallel and charge the last couple of % to 3.6V. Leave at 3.6V overnight to let all the cells get to the same SOC.

              You may recall that i was originally hoping to individually capacity-test all 20 cells before rebuilding the pack...that dream has obviously evaporated due to the time involved (this insane winter on the east coast is not helping). But i do lean towards the suggestion of discharge-testing a few (including #5) to see if there's more going on. Paralleling 5 of them and tapering to C/100 @ 3.6V before discharging...what is that accomplishing? I thought such a small taper current was a bad idea? Why does it matter if the "capacity" i'm going to measure is the re-charge Ah from 2.5V to our definition of "full" (3.38 - 3.45 after taper)? And the week of resting...that's a long time. What does that buy compared to say, only 24 hours' rest?
              I think the idea of letting the the current taper being bad comes about because of doing fast charges at high charge rates without any BMS. To charge at high rates you have to raise the the charge voltage to say 3.6V/cell to get over the impedance of the battery limiting the current. If you charge at say C/2 and terminate the charge when the voltage reaches 3.6V you will get an SOC of ~98%. If you let the current taper off this will very rapidly approach 100% SOC. If your battery is not very accurately balanced this will mean that the voltage on one or more cells may go higher then 4.5V in a very short space of time.

              If the leakage in #5 is larger than the other cells the longer you leave it the greater the difference in SOC between the cells will be and the easier it will be to measure. DifferenecInSOC=DifferenceInLeakage x Time
              I really don't think this cell went down to 2.0V with my pack LVD set to 12.0V...but i realize it is possible because it was paralleled with 4 others. What is the test for internal leakage?
              If the cell is paralleled with others it cannot get to 2.0V unless all the other paralleled cells get to less than 2.0V. They all share the same voltage. One advantage of paralleling cells is that it balances out any variation between the individual cells.

              If you don't have the time to let #5 sit you can set your battery up in its final configuration of 5p4s and keep an eye the individual block of cells voltage. If #5 does have higher leakage you will see the voltage of the block of cells that #5 is in get less relative to the other blocks over time. Be Warned! This difference will be amplified when the battery is nearly fully charged or discharged.

              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
              Off-Grid LFP(LiFePO4) system since April 2013

              Comment


              • #8
                I am trying to think of a physical analogy of a lithium ion battery. The best I have come up with is two sponges,
                I applaud your effort, i think it does describe the results of varying charge voltage and taper...and makes sense in light of some other conversations about making use of solar during precious peak hours by charging at higher voltages. Maybe not a 'universal' analogy but it helps with this concept at least. If the "suction" applied to the sponges is the charge voltage, and the amount of water in the anode sponge represents SOC, then when you remove the charge "suction" what quantity represents the cell's OCV? When it comes to battery analogies, i have yet to hear one that really nails it, but would love to discover a truly good one. The old "bucket of water" concept has done more harm than good. I'm beginning to think that water and electricity just aren't similar enough to even keep barking up that tree...

                Karrak, i've seen you mention a couple times about paralleling cells and letting them rest so they reach the "same SOC". This is confusing me because i thought we were defining "full" not by OCV but by charging @3.6V+tapering current...and whatever the resultant voltage is for each cell, that is SOC="full". The underlying truth being (i thought) that SOC is not strictly tied to voltage. If paralleling all the cells has the effect of equalizing voltage across them, but they were all individually "full" separately, isn't this actually slightly "un-balancing" the cells in the name of equal voltages? In my case the cells will eventually get paralleled anyway in the assembled pack...so what is the implication of that for my "top balance"?

                I feel like i'm being really difficult here but these details are keeping me from feeling like i actually understand.

                If the cell is paralleled with others it cannot get to 2.0V unless all the other paralleled cells get to less than 2.0V. They all share the same voltage. One advantage of paralleling cells is that it balances out any variation between the individual cells.
                In that case there's no way this cell came remotely near 2.0V.

                I discharged #5 at 2.4V with taper to 2A and it went from 3.361OCV (rested) to 2.768 OCV (rested) after the first pass, giving up 144Ah. At first this seemed bad...but my battery compartment is hovering in the mid 40s*F right now (i have a thermometer in there and constantly watch for freezing)...normally it's warmer because the cells are always under load or charge, but with the pack busted down it's chilly. Manufacturer's spec sheet derates capacity to 75% at freezing, so for a 200Ah-rated cell, 144Ah doesn't seem out of line. Evidently this cell just maxes out slightly below that 3.38V figure which is so often quoted. Am i putting my head in the sand or does this seem believable? Cell #5 will get taken the rest of the way down to 2.5 tomorrow AM and then charged up to "full". We'll see how many Ah it accepts.

                Sorry my posts are always so damn long; i'm just not a concise person.


                - 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
                www.livesmallridefree.com

                Comment


                • #9
                  Originally posted by zamboni View Post
                  If the "suction" applied to the sponges is the charge voltage, and the amount of water in the anode sponge represents SOC, then when you remove the charge "suction" what quantity represents the cell's OCV? When it comes to battery analogies, i have yet to hear one that really nails it, but would love to discover a truly good one. The old "bucket of water" concept has done more harm than good. I'm beginning to think that water and electricity just aren't similar enough to even keep barking up that tree...
                  I agree that the bucket of water doesn't work well when you think of a battery as a bucket full of water that you pour through a load on to the ground or charge by filling up from a water main. I think it does a reasonable job if you think of it as two bodies of water at different heights. The anode being the higher body of water and the cathode being the lower body. Charging is equivalent to pumping the water to the higher reservoir from the lower reservoir and discharging is releasing the water back to the lower reservoir through a turbine. The OCV is equivalent of the gravitational potential of the water at greater height. The reason for the sponge is that unlike an open body of water the lithium ions are within a crystal lattice. Maybe a better analogy for the sponge would be an underground water reservoir where the permeability of the aquifer emulates the impedance of the cathode.

                  Karrak, i've seen you mention a couple times about paralleling cells and letting them rest so they reach the "same SOC". This is confusing me because i thought we were defining "full" not by OCV but by charging @3.6V+tapering current...and whatever the resultant voltage is for each cell, that is SOC="full". The underlying truth being (i thought) that SOC is not strictly tied to voltage. If paralleling all the cells has the effect of equalizing voltage across them, but they were all individually "full" separately, isn't this actually slightly "un-balancing" the cells in the name of equal voltages? In my case the cells will eventually get paralleled anyway in the assembled pack...so what is the implication of that for my "top balance"?
                  You have to remember that we are measuring the voltage at the battery terminals which is different to the voltages at the points in the crystal lattice in the cathode and anode where the lithium ions migrate to. The voltage difference between the battery terminals and different points in the anode and cathode is dependant on the current flow and the impedance of the various parts of the battery. This impedance varies not only between different cells but within the same cell. To further complicate matters the cathode and anode are not one amorphous block but made up of gazillions of particles, the cathode is LiFePO4 particles coated with carbon and the anode is particles of graphite pressed together.

                  By letting the current reduce to zero and giving the battery time to come to some sort of equilibrium gives us the most accurate measurement of SOC. Having said this you are probably only going to have an error of less than 1% if you don't let the battery come to equilibrium.

                  I feel like i'm being really difficult here but these details are keeping me from feeling like i actually understand.
                  I don't think you are being difficult. I think it is important to try to get some understanding of the underlying principles.

                  I discharged #5 at 2.4V with taper to 2A and it went from 3.361OCV (rested) to 2.768 OCV (rested) after the first pass, giving up 144Ah. At first this seemed bad...but my battery compartment is hovering in the mid 40s*F right now (i have a thermometer in there and constantly watch for freezing)...normally it's warmer because the cells are always under load or charge, but with the pack busted down it's chilly. Manufacturer's spec sheet derates capacity to 75% at freezing, so for a 200Ah-rated cell, 144Ah doesn't seem out of line. Evidently this cell just maxes out slightly below that 3.38V figure which is so often quoted. Am i putting my head in the sand or does this seem believable? Cell #5 will get taken the rest of the way down to 2.5 tomorrow AM and then charged up to "full". We'll see how many Ah it accepts.
                  I would be horrified if I only got 144Ah out of a 200Ah LFP battery but my battery doesn't work in such cold temperatures. I would try to test your cells at the same temperature preferably around ~25C (77F) otherwise you wont be able to accurately compare them.

                  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
                  Last edited by karrak; 02-01-2018, 09:29 AM.
                  Off-Grid LFP(LiFePO4) system since April 2013

                  Comment


                  • #10
                    I would be horrified if I only got 144Ah out of a 200Ah LFP battery but my battery doesn't work in such cold temperatures. I would try to test your cells at the same temperature preferably around ~25C (77F) otherwise you wont be able to accurately compare them.
                    Yeah, normally mine doesn't either, and no matter what the chart in the spec sheet says i'm still a little horrified...but I don't have any way to get that compartment up to STC so i'm going to soldier on. I recharged #5 yesterday but have no idea the Ah it accepted because evidently my charger automatically clears the info screen about 15 minutes after the cycle ends That would be a nice tidbit to include in the manual! Next time i won't go for a ride while the charger is working... So now i'm discharging that cell AGAIN, so i can re-charge it AGAIN and get a good Ah figure. Then i'll try another "good"-seeming cell. I'll keep testing cells until i run out of other projects to work on in parallel...evidently i'm going to get a whole lot of long-postponed repairs done on the trailer this weekend.

                    Is comparing Ah removed during discharge vs. Ah replaced during re-charge apples and oranges due to battery efficiency?


                    As for the battery analogy...i believe a battery's behavior could be modeled with any number of mechanical systems -- but they would all end up being relatively complex. The "one bucket" analogy is frequently invoked when talking with laypersons who would be just as disoriented by discussing the porosity of an aquifer as they are about anodes and cathodes. For me, it's about finding an intuitive sense of what's happening, the way i can look at a bunch of tanks, pumps, valves, engines, gearboxes, etc. and immediately grok how they work together. I'm beginning to accept that i need to get better educated/experienced with electricity so i can learn that intuition without the crutch of an analogy.


                    - 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
                    www.livesmallridefree.com

                    Comment


                    • #11
                      Just thought i'd supply a little closure:

                      I capacity tested several other cells in my pack and came up with very close numbers to what #5 was showing...but #5 just comes to rest at a lower voltage than the other cells, given the same charge profile. Everything else seems to be okay. So i brought all the cells to 3.3V, wired them in parallel, applied a gentle charge with taper, and then let them sit for what ended up being nearly a week as i dealt with some other fun issues that came up.

                      Then i connected the pack up with new bus bars and capacity tested. It gave up 702Ah going from 13.38V (rested) down to 11.4V (rested). That's not a full discharge...i didn't go lower because it was about 2am by then, bad timing by me. So on a 1000Ah-rated pack, removing what ought to have been about 80% capacity, i only got about 7/8 of expected. The unusually cold winter blast had subsided by then but it was still close to freezing, so cell temps were below STC as well. Given that the cells are not top-quality, i'm okay with all this. If they don't degrade too rapidly from this point, i'll have gotten a fair value. Throughout all this, the group of cells with #5 in it didn't manifest any voltage discrepancies. I also got about a 30% reduction in Ri by making better bus bar connections, which is nice... All thanks to great info and support from this forum, so: Thanks!

                      -Jerud

                      Comment


                      • #12
                        Glad you came up with some stable numbers and it's amazing what effect good connections has.
                        Powerfab top of pole PV mount (2) | Listeroid 6/1 w/st5 gen head | XW6048 inverter/chgr | Iota 48V/15A charger | Morningstar 60A MPPT | 48V, 800A NiFe Battery (in series)| 15, Evergreen 205w "12V" PV array on pole | Midnight ePanel | Grundfos 10 SO5-9 with 3 wire Franklin Electric motor (1/2hp 240V 1ph ) on a timer for 3 hr noontime run - Runs off PV ||
                        || Midnight Classic 200 | 10, Evergreen 200w in a 160VOC array ||
                        || VEC1093 12V Charger | Maha C401 aa/aaa Charger | SureSine | Sunsaver MPPT 15A

                        solar: http://tinyurl.com/LMR-Solar
                        gen: http://tinyurl.com/LMR-Lister

                        Comment


                        • #13
                          Hi zamboni - glad things are going well so far ...

                          Just remember that for the first 5-10 cycles or so, things will walk around a little bit as the SEI layer stabilizes to what it will be for most of the rest of it's life. So keep an eye on things and check it out again after the 5th cycle or so and look for any indications or trends going askew. Worst case would be a rebalance, or perhaps just another manual charge to full individually again before placing into the standard 80/10% SOC service - or whatever you decide.

                          I'd hate to be a guy just coming into the thread(s) because it looks like it takes a rocket-scientist to maintain these things. Eventually, you'll be pleasantly surprised at how easy it is - once things become as old-hat as Pb-think.

                          Re: charging to full at least once. Sure enough, this is needed if you want the best / reassurance out of them. Some unfortunate EV'ers who just pulled cells half-charged from the factory crate and did a bottom balance immediately may have wondered why their "new" cells failed under heavy load. That's because they weren't properly prepared. So that's one thing - no matter which balance procedure one decides to use.

                          TIME - fortunately for our application, unlike an EV'er or custom wheelchair application, we can afford slight imbalance (no more than .1V, preferably less, and assuming cells are reasonably the same in capacity and IR). The saving factor here is TIME. The question one has to ask themselves is time spent balancing at full-charge, or nearly so, worth the wait for another to catch up to a microscopic voltage difference in balance? Probably not!

                          Example: If someone just randomly slaps cells together, and then it takes 3 days to finally top balance all the cells to no more than .0001v difference, is the 3 days spent at full charge more harmful? YES. Many LFP threads all over the place overlook the time vs perfect balance issue - why? Because you don't notice it when cells are new. Give it some time, and then it becomes apparent when one actual says after 4 years "geez, I just timed it out, and my cells actually spent 8 MONTHS under full charge basically!" They were too concerned staring at the microscopic cell voltage differences and not the clock! So one strikes an um, balance, between time and perfection.

                          Example 2: although a little bit different, if I use my PWM solar charge controller (which is nothing more than a fancy and fast on/off switch to my panel terminals), if you put a scope on it and saw the 22v pulses at 300khz or so, is that killing my 12v LFP? NO - due to the *time* average of pwm which brings voltage down on a time average to my setting of 14.2v. Scope sees 22v pulses - but overall, the fast switching makes the lfp "see" 14.2v on average. Anyway, getting on a tangent here...

                          I'll have to re-read some of the threads, but I think you've got the hang of it! Trust me, after awhile this will all distill down to just a handful of parameters.
                          Last edited by PNjunction; 02-28-2018, 08:59 PM.

                          Comment


                          • #14
                            Well, in case someone is worrying that it's rocket science, let me just remind them i broke rule #1 and started with suboptimal cells. So i've made it a lot harder on myself. This definitely should be pretty straightforward for someone who is careful and paying attention to details (and the advice here). I like it better this way though, because working through these issues helps me get closer to understanding -- and i still get a better battery than i had with golf-cart cells. Even a bad LFP pack is a huge improvement over Pb.

                            I just noticed something that makes me consider rebalancing though: We've had lots of rain and hit the low teens SOC a few times, with only 2 charges to full before that. The cells have been holding balance within 100mV throughout. Yesterday we got great sun and back to 99% - but when that happened, one bank of cells took off up it's "knee". By one "bank" i mean one paralleled set of 5 cells (ie there's 4 'banks' in my 4S pack). This bank does not contain cell #5 that i originally wrote this post about. Since i did not measure the capacity of every individual cell, i can't know for sure - but my interpretation is that my weakest cell must be in this bank, and it's hitting the knee before the others. Isn't this exactly what top-balancing should be avoiding? I thought all my cells should start up the knee about the same time...?

                            The cells were balanced to 3.4VPC but they had lots of rest throughout that process. In-use, of course, they never rest so voltages are higher. What i saw yesterday, was that they were charging at 14.6V, and just before the 3% taper was done the banks measured 3.582 / 3.736 / 3.590 / 3.565. Float is set to 13.5 so once at 100% bank voltages fall in-line. I have one CellLog monitoring bank voltages and set to trigger my LVD at 2.5VPC but i don't have a per-cell HVD trigger, only pack-level (i thought with top balancing and frequent manual monitoring this would be an acceptable risk).

                            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. Rebalancing will take many days with the Revolectrix, which is fine now but expensive if i have to pay for an RV park just to do it. Thoughts?

                            Note I have programmed my battery monitor to read on a 0-100% scale based upon the portion of the pack capacity i'm actually using, with 100% defined as 3.5VPC+3% taper and total Ah set to 700 (out of 1000 rated), which puts 0% around 2.7VPC rested.

                            Comment


                            • #15
                              Looking at the charge/discharge curves, the difference between 3.565 and 3.736 volts at ~C/30 (~35) is less than 1%. IMO it is not worth trying to keep a battery any better balanced.


                              I would look at what the voltage spread is at 3.45V/cell (13.8V) with end current of ~C/30. If you have a voltage spread of more than 0.050V then you could remove ~0.5%SOC from the highest cell by placing a load across it. 3.3 Ohm 5W resistor or a 50W 12V halogen light or old car headlight will all do the job. Just calculate the amount of time you need to connect the load to the high cell to remove the 0.5%SOC.

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

                              Simon
                              Off-Grid LFP(LiFePO4) system since April 2013

                              Comment

                              Working...
                              X