Seeking Normally Closed Latching Contactor and Q&A about balancing my Li Ion battery bank

123578

Comments

  • karrak
    karrak Solar Expert Posts: 326 ✭✭✭✭
    BTW - I have been talking to boB at Midnite about updating the controller to handle two inputs. Doesn't look like it's going to happen soon, but there is a solution developing using a Rasberry Pi and talking to the Classic via the Ethernet jack and MODBUS. Here is the thread if you want to have a look: http://midniteftp.com/forum/index.php?topic=3973.0. i am '5 minutes to midnite'.

    So I don't know if your temperature sensor trick or the Pi would be the quickest path. Need to do something though.


    Using the Pi would give you more flexibility. If you got one of the cheap Chinese BMS's you could hook it up to the Pi and run my BMS software on the Pi.

    What experience do you have running Linux? Do you have any experience programming computers?

    I would say that now you have the battery balanced that there is not so much urgency to be able to stop the charge controller remotely.

    Simon
    Off-Grid with LFP (LiFePO4) battery, battery Installed April 2013
    32x90Ah Winston cells 2p16s (48V), MPP Solar PIP5048MS 5kW Inverter/80A MPPT controller/60A charger, 1900W of Solar Panels
    modified BMS based on TI bq769x0 cell monitors.
    Homemade overall system monitoring and power management  https://github.com/simat/BatteryMonitor
     

  • LostinSpace
    LostinSpace Registered Users Posts: 97 ✭✭
    Even though I did an IT career (now semi-retired) I never got to deep into programming. And minimal Linux. I was making too much money fixing MS Windows. ;>

    I think since my cellogs have the cell over-voltage trigger I would like to keep the electronics to a minimum and see if the Pi script being provided in the Midnite forum will work. You code requires the BMS? Won't work with the cellog output relay. How is the BMS triggering the Pi that your code is running on?

    I was thinking too of putting the balancers back on. I also have to now charge and balance the other 16 cell bank, and then reconfigure back to that array you provided the diagram for.


    Qty 24 Sharp 153W poly panels, Midnite MNPV6 combiner, Magnum MS-4448-PAE inverter with RC50 remote, Magnum MMP Mini panel, Midnite Classic 150 charge controller, Qty 32 LiFePO4 72AH cells in 2 string 48V array. BMS.
  • LostinSpace
    LostinSpace Registered Users Posts: 97 ✭✭
    Simon, one correction. The iMAX charger does indeed display AHs. The reading appears when the charge cycle is complete. Just hadn't noticed it yet with all the other telemetry on the LCD display.

    The AH reading does vary a lot. These are 72AH rated cells, and the iMAX found one cell to be 83AH, and the next one to be 134AH. Of course this is not after a full discharge and charge cycle, which as noted earlier I am not attempting with these hobby chargers. I'm curious, does your charger require a full discharge/charge cycle to render an accurate AH number?

    ------------------------------------------------
    And by the way, now that I have one bank of 15 cells in good shape (still fine tuning a bit) I will be moving on to the second bank of 16 cells. I'm going to repeat was done on bank 1, but then the question of forming all 32 cells into a pack per the earlier diagram you provided.

    Here's the question: When pairing up cells to be configured in parallel (2 paralled cell X 16) would you be more likely to select two cells in close voltage proximity, or might it be batter to pair cells that have the widest voltage difference?

    For example take 4 cells: 3.38, 3.39, 3.41, 3.42

    How would you pair those cells? Am I making sense?


    Qty 24 Sharp 153W poly panels, Midnite MNPV6 combiner, Magnum MS-4448-PAE inverter with RC50 remote, Magnum MMP Mini panel, Midnite Classic 150 charge controller, Qty 32 LiFePO4 72AH cells in 2 string 48V array. BMS.
  • LostinSpace
    LostinSpace Registered Users Posts: 97 ✭✭
    Raj174 said:
    A reasonable analogy might be to imagine each cell as a long neck wine bottle, all with a little different capacity, all proportional in shape but varying in size ranging from 30 to 34 ounces. When charging in series, all receive almost exactly the same amount of liquid at the same rate, and as the liquid  reaches the tapered neck of the smaller bottles the level will rise faster than the others. So obviously the smaller bottles will overflow first. To balance the liquid level a tube is connected between the bottles allowing flow from bottle to bottle. This is connecting them in parallel. If this is done with all the bottles full, the smaller bottles will overflow.  :*   Not good. So you really want to fill the bottles until the smallest bottle is almost full while in series, then connect them in parallel and wait for the fluid levels to even out. Once they are level, you'll want to feed one bottle slowly until the even level reaches the top of the smallest bottle and congratulations, your case of wine is now balanced.   :)

    The thing is, you have a lot of cells to monitor. And as you know the higher the voltage gets the faster it will rise. So you may get busier than a one armed wallpaper hanger.  :)  

    Rick

    I really like your wine bottle analogy. But I am learning something to add as well. Consider this my 3rd (maybe 4th) theory as to why the balancers may not have been doing a great job:

    THEORY: THE NECKS OF THE WINE BOTTLES ARE NOT ALL THE SAME VOLUME.

    OK? For example, today while charging I had 3 cells that hit the 3.6v alarm threshold I had set on the cellogs. I stopped charging, and placed a bulb on each of the three cells to drop them down a bit. I stopped when their voltage was several millivolts below the other cells.

    Upon turning on the charge controller again these three cells were first to go high again. In fact they shot up from about 3.4 to 3.6 FAST. just a minute or two. Maybe less.

    So employing the metaphor, the wine bottles for the three cells were less full than the other bottles, yet when the 'liquid' started being poured in they very quickly filled to the neck, then to the top of the neck, nearly spilling over, while the other bottles were filled only to the neckline (or so).

    I already knew capacity varies from one cell to another. But the evidence here is that charge rate -the ability of a cell to take a charge faster- (at least when near the top of the charge curve), also seems to be a variable.

    I don't know if this is common among LFP cells, or perhaps just the brand I am using. But this tends to explain why the balancers can't keep up. Again, those cells went from 3.4x to 3.6x (2/10ths of volt) really fast.

    Therefore, at least with these cells, an argument is made for a BMS that stops charging on a per cell basis. To justify the expense and trouble of a full blown BMS however, one would have to quantify how much additional capacity could be gained. I am tempted for now to simply back off the pack voltage a bit to keep the errant cells from shooting high.














    Qty 24 Sharp 153W poly panels, Midnite MNPV6 combiner, Magnum MS-4448-PAE inverter with RC50 remote, Magnum MMP Mini panel, Midnite Classic 150 charge controller, Qty 32 LiFePO4 72AH cells in 2 string 48V array. BMS.
  • Raj174
    Raj174 Solar Expert Posts: 795 ✭✭✭✭
    edited February 2018 #126
    Raj174 said:
    A reasonable analogy might be to imagine each cell as a long neck wine bottle, all with a little different capacity, all proportional in shape but varying in size ranging from 30 to 34 ounces. When charging in series, all receive almost exactly the same amount of liquid at the same rate, and as the liquid  reaches the tapered neck of the smaller bottles the level will rise faster than the others. So obviously the smaller bottles will overflow first. To balance the liquid level a tube is connected between the bottles allowing flow from bottle to bottle. This is connecting them in parallel. If this is done with all the bottles full, the smaller bottles will overflow.  :*   Not good. So you really want to fill the bottles until the smallest bottle is almost full while in series, then connect them in parallel and wait for the fluid levels to even out. Once they are level, you'll want to feed one bottle slowly until the even level reaches the top of the smallest bottle and congratulations, your case of wine is now balanced.   :)

    The thing is, you have a lot of cells to monitor. And as you know the higher the voltage gets the faster it will rise. So you may get busier than a one armed wallpaper hanger.  :)  

    Rick

    I really like your wine bottle analogy. But I am learning something to add as well. Consider this my 3rd (maybe 4th) theory as to why the balancers may not have been doing a great job:

    THEORY: THE NECKS OF THE WINE BOTTLES ARE NOT ALL THE SAME VOLUME.

    OK? For example, today while charging I had 3 cells that hit the 3.6v alarm threshold I had set on the cellogs. I stopped charging, and placed a bulb on each of the three cells to drop them down a bit. I stopped when their voltage was several millivolts below the other cells.

    Upon turning on the charge controller again these three cells were first to go high again. In fact they shot up from about 3.4 to 3.6 FAST. just a minute or two. Maybe less.

    So employing the metaphor, the wine bottles for the three cells were less full than the other bottles, yet when the 'liquid' started being poured in they very quickly filled to the neck, then to the top of the neck, nearly spilling over, while the other bottles were filled only to the neckline (or so).

    I already knew capacity varies from one cell to another. But the evidence here is that charge rate -the ability of a cell to take a charge faster- (at least when near the top of the charge curve), also seems to be a variable.

    I don't know if this is common among LFP cells, or perhaps just the brand I am using. But this tends to explain why the balancers can't keep up. Again, those cells went from 3.4x to 3.6x (2/10ths of volt) really fast.

    Therefore, at least with these cells, an argument is made for a BMS that stops charging on a per cell basis. To justify the expense and trouble of a full blown BMS however, one would have to quantify how much additional capacity could be gained. I am tempted for now to simply back off the pack voltage a bit to keep the errant cells from shooting high.
    I think what you are seeing with the rapid rise in voltage at the peak of the charge curve after a discharge from 3.6 to 3.4 is caused by the fact that a cell with a load or even for a while after the load is removed will not read the correct voltage.  Once the load is removed the cell voltage will rise for some time until it comes to true resting voltage. So putting a charge on it will cause a rise far more rapid than if charging from a resting state. This is why, when balancing the cells in parallel, once 3.6 volts has been achieved I terminate the charge and walk away. Leave them 10 to 12 hours, 24 hours would be better. There is considerable voltage drop with 3.6 volts, even through buss bars, over 5 feet. It takes a lot of time to push the milliamps around with such low voltage.

    Rick
    4480W PV, MNE175DR-TR, MN Classic 150, Outback Radian GS4048A, Mate3, 51.2V 360AH nominal LiFePO4, Kohler Pro 5.2E genset.
  • karrak
    karrak Solar Expert Posts: 326 ✭✭✭✭
    LostinSpace said:I think since my cellogs have the cell over-voltage trigger I would like to keep the electronics to a minimum and see if the Pi script being provided in the Midnite forum will work.
    The outputs of the Cellogs are not isolated so you can't link the outputs from the Cellog8  that monitors the lower 8 cells to the Cellog8 that monitors the upper 8 cells directly to the Midnite. You will have to electrically isolate the two alarm outputs with a dual optocoupler something like this. Unfortunately this PCB commons the two separate  input grounds on pins 2 and 4 of the D213 optocoupler IC together. You will have to cut the copper trace between pins 2 and 4 or desolder and lift one of the legs of the IC and wire each separate pin to the separate cellog GND outputs or you could buy two of the optocoupler PCBs. I had a look at the Classic150 manual and couldn't find any information as to whether or not there is an inbuilt pullup resistor on the AUX2 input.
    You code requires the BMS? Won't work with the cellog output relay. How is the BMS triggering the Pi that your code is running on?

    I originally built my BMS around ADS1115 analog to digital converters and the Beaglebone Black microcontroller. I got the cheap BMS units for my electric push bike conversions and thought I would look at interfacing them to my BMS system. My BMS obtains the raw voltage and current data from either my ADS1115 circuit or the Bluetooth BMS and uses this. Click on the link in my signature for more information.
    I was thinking too of putting the balancers back on. I also have to now charge and balance the other 16 cell bank, and then reconfigure back to that array you provided the diagram for.


    If you use a ring crimp connector like this connected to the battery terminal

    connected to a short length of cable with a connector something like this on the other end

    you can quickly and easily connect and disconnect things like the balancing boards to the battery.

    Simon

    Off-Grid with LFP (LiFePO4) battery, battery Installed April 2013
    32x90Ah Winston cells 2p16s (48V), MPP Solar PIP5048MS 5kW Inverter/80A MPPT controller/60A charger, 1900W of Solar Panels
    modified BMS based on TI bq769x0 cell monitors.
    Homemade overall system monitoring and power management  https://github.com/simat/BatteryMonitor
     

  • westbranch
    westbranch Solar Expert Posts: 5,183 ✭✭✭✭
    edited February 2018 #128
    LiSpace....  this is so complex |I am wondering if that is why people like  PNJunction are recommending BOTTOM  Balancing?

    You empty them all and then fill the tank till the smallest on is full and then stop...


     
    KID #51B  4s 140W to 24V 900Ah C&D AGM
    CL#29032 FW 2126/ 2073/ 2133 175A E-Panel WBjr, 3 x 4s 140W to 24V 900Ah C&D AGM 
    Cotek ST1500W 24V Inverter,OmniCharge 3024,
    2 x Cisco WRT54GL i/c DD-WRT Rtr & Bridge,
    Eu3/2/1000i Gens, 1680W & E-Panel/WBjr to come, CL #647 asleep
    West Chilcotin, BC, Canada
  • Raj174
    Raj174 Solar Expert Posts: 795 ✭✭✭✭
    edited February 2018 #129
    LiSpace....  this is so complex |I am wondering if that is why people like  PNJunction are recommending BOTTOM  Balancing?

    You empty them all and then fill the tank till the smallest on is full and then stop...


    Did he really? If he did, I wouldn't think it would be for a solar setup. Maybe he was converted in some EV cult.  :*

    Rick
    4480W PV, MNE175DR-TR, MN Classic 150, Outback Radian GS4048A, Mate3, 51.2V 360AH nominal LiFePO4, Kohler Pro 5.2E genset.
  • karrak
    karrak Solar Expert Posts: 326 ✭✭✭✭
    Simon, one correction. The iMAX charger does indeed display AHs. The reading appears when the charge cycle is complete. Just hadn't noticed it yet with all the other telemetry on the LCD display.

    The AH reading does vary a lot. These are 72AH rated cells, and the iMAX found one cell to be 83AH, and the next one to be 134AH. Of course this is not after a full discharge and charge cycle, which as noted earlier I am not attempting with these hobby chargers. I'm curious, does your charger require a full discharge/charge cycle to render an accurate AH number?
    This doesn't make any sense. 83Ah is just plausible, 134Ah makes no sense. Is the iMAX summing the charge from multiple charge cycles?

    The only way to measure a cell's capacity accurately is to discharge it to 2.5V then charge it up to 3.6V with a termination current of less than C/50 and measure the Ah needed to charge the battery.

    And by the way, now that I have one bank of 15 cells in good shape (still fine tuning a bit) I will be moving on to the second bank of 16 cells. I'm going to repeat was done on bank 1, but then the question of forming all 32 cells into a pack per the earlier diagram you provided.

    Here's the question: When pairing up cells to be configured in parallel (2 paralled cell X 16) would you be more likely to select two cells in close voltage proximity, or might it be batter to pair cells that have the widest voltage difference?

    For example take 4 cells: 3.38, 3.39, 3.41, 3.42

    How would you pair those cells? Am I making sense?


    You can't use voltage to pair the cells. You can only use capacity obtained by doing a capacity test.

    If I were in your situation I would discharge your battery connected to the Midnite by about 20%SOC. Then reduce the charge voltage on the Midnite to 3.4V/cell (54.4V) and float to 3.3V/cell. Then do a normal charge with the Midnite and make sure all the cells are balanced at the 3.4V charge voltage.  Charge all the other cells to 3.4V with the IMAX. Then disassemble the battery connected to the Midnite and put all the cells from this battery and all the separate  cell in parallel and charge all the paralleled cells to 3.6V using the iMAX. Ideally you would want to leave the cells charging at 3.6V until the charge current reduces to around zero and then leave the cells on charge at 3.6V for another ten hours or so but it doesn't look like you can do this with the iMAX. I couldn't find any information on what conditions cause the iMAX to terminate the charge. This being the case doing a Balance Charge at 3.6V will have to suffice.

    Simon


    Off-Grid with LFP (LiFePO4) battery, battery Installed April 2013
    32x90Ah Winston cells 2p16s (48V), MPP Solar PIP5048MS 5kW Inverter/80A MPPT controller/60A charger, 1900W of Solar Panels
    modified BMS based on TI bq769x0 cell monitors.
    Homemade overall system monitoring and power management  https://github.com/simat/BatteryMonitor
     

  • LostinSpace
    LostinSpace Registered Users Posts: 97 ✭✭
    edited February 2018 #131
    westbranch Interesting, I have more entries from PNJunction posts in my LPF notebook prior to purchasing, than any other member. Very high credibility.

    Qty 24 Sharp 153W poly panels, Midnite MNPV6 combiner, Magnum MS-4448-PAE inverter with RC50 remote, Magnum MMP Mini panel, Midnite Classic 150 charge controller, Qty 32 LiFePO4 72AH cells in 2 string 48V array. BMS.
  • LostinSpace
    LostinSpace Registered Users Posts: 97 ✭✭

    karrak said:
    This doesn't make any sense. 83Ah is just plausible, 134Ah makes no sense. Is the iMAX summing the charge from multiple charge cycles?

    The only way to measure a cell's capacity accurately is to discharge it to 2.5V then charge it up to 3.6V with a termination current of less than C/50 and measure the Ah needed to charge the battery.



    Ha. I think it is accumulating. I ignored the reading after the first two cells. But just checked the number at the end of a charge cycle, several cells later, and the number is now huge.

    So obviously I am clearing the last status by hitting the STOP button in between cells, but the charger is holding the cumulative total in memory. Good call.


    You can't use voltage to pair the cells. You can only use capacity obtained by doing a capacity test.

    If I were in your situation I would discharge your battery connected to the Midnite by about 20%SOC. Then reduce the charge voltage on the Midnite to 3.4V/cell (54.4V) and float to 3.3V/cell. Then do a normal charge with the Midnite and make sure all the cells are balanced at the 3.4V charge voltage.  Charge all the other cells to 3.4V with the IMAX. Then disassemble the battery connected to the Midnite and put all the cells from this battery and all the separate  cell in parallel and charge all the paralleled cells to 3.6V using the iMAX. Ideally you would want to leave the cells charging at 3.6V until the charge current reduces to around zero and then leave the cells on charge at 3.6V for another ten hours or so but it doesn't look like you can do this with the iMAX. I couldn't find any information on what conditions cause the iMAX to terminate the charge. This being the case doing a Balance Charge at 3.6V will have to suffice.

    Well, in for a penny, in for a pound. I may just wait to invest in the proper charger, which I assume will allow for the full Monte you are describing.

    But let me play this back to you: The final step in your scenario is to charge all 32 paralled cells to 3.6... then wait for charge current to drop to near zero.... then continue for about another 10 hours?
    Qty 24 Sharp 153W poly panels, Midnite MNPV6 combiner, Magnum MS-4448-PAE inverter with RC50 remote, Magnum MMP Mini panel, Midnite Classic 150 charge controller, Qty 32 LiFePO4 72AH cells in 2 string 48V array. BMS.
  • karrak
    karrak Solar Expert Posts: 326 ✭✭✭✭
    The gold standard for balancing cells is charge them all in parallel to 3.6V-3.65V, wait for the current to drop to nearly zero and then charge for another ten hours or so.

    Since the iMAX charge current is so low, in your case C/384,  just charging them to 3.6Vwith the iMAX until it terminates the charge will do a good job. Might take a few days though.

    I don't think I can get my Turnigy 300W to charge the extra ten hours but I can get it to keep charging until the current gets to nearly zero. If I was going to balance the cells in parallel I would use my 20A bench power supply after charging the cells in series to ~3.4V/cell with the Turnigy to save time.

    Simon
    Off-Grid with LFP (LiFePO4) battery, battery Installed April 2013
    32x90Ah Winston cells 2p16s (48V), MPP Solar PIP5048MS 5kW Inverter/80A MPPT controller/60A charger, 1900W of Solar Panels
    modified BMS based on TI bq769x0 cell monitors.
    Homemade overall system monitoring and power management  https://github.com/simat/BatteryMonitor
     

  • LostinSpace
    LostinSpace Registered Users Posts: 97 ✭✭
    I am going to give the 'gold standard' a try with the iMax charger. Will report findings.


    Qty 24 Sharp 153W poly panels, Midnite MNPV6 combiner, Magnum MS-4448-PAE inverter with RC50 remote, Magnum MMP Mini panel, Midnite Classic 150 charge controller, Qty 32 LiFePO4 72AH cells in 2 string 48V array. BMS.
  • mike95490
    mike95490 Solar Expert Posts: 9,583 ✭✭✭✭✭
    karrak said:
    The gold standard for balancing cells is charge them all in parallel to 3.6V-3.65V, wait for the current to drop to nearly zero and then charge for another ten hours or so.
    But what's the recommended current to do this charge at   1C ?  c/20?  c/40 ? 
    Powerfab top of pole PV mount | 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 ,

  • karrak
    karrak Solar Expert Posts: 326 ✭✭✭✭
    mike95490 said:
    karrak said:
    The gold standard for balancing cells is charge them all in parallel to 3.6V-3.65V, wait for the current to drop to nearly zero and then charge for another ten hours or so.
    But what's the recommended current to do this charge at   1C ?  c/20?  c/40 ? 
     Recommended maximum charge current is whatever the battery manufacture recommends. I haven't come across any information that suggests a minimum charge current. It depends on how long you want to wait.

    In Jim's case he has 32x72Ah batteries. If they are charged in parallel this equates to 2304Ah, charging from 0% SOC at 6A would take 384 hours (16 days). New LFP batteries when purchased are usually at an SOC of ~50% so would take ~8 days to charge. If you charged the cells to 3.4V/cell in series which would take the SOC to > 90% SOC then did the final balance charge it would take less than two days.

    Simon

    Off-Grid with LFP (LiFePO4) battery, battery Installed April 2013
    32x90Ah Winston cells 2p16s (48V), MPP Solar PIP5048MS 5kW Inverter/80A MPPT controller/60A charger, 1900W of Solar Panels
    modified BMS based on TI bq769x0 cell monitors.
    Homemade overall system monitoring and power management  https://github.com/simat/BatteryMonitor
     

  • LostinSpace
    LostinSpace Registered Users Posts: 97 ✭✭
    edited February 2018 #137
    And that is what I have started on (as of yesterday). The 32 cells are all configured in parallel. I am letting them sit that way for 48 hours to rest and equalize. Charged both banks first to ~3.41x volts to give them a head start.



    Qty 24 Sharp 153W poly panels, Midnite MNPV6 combiner, Magnum MS-4448-PAE inverter with RC50 remote, Magnum MMP Mini panel, Midnite Classic 150 charge controller, Qty 32 LiFePO4 72AH cells in 2 string 48V array. BMS.
  • LostinSpace
    LostinSpace Registered Users Posts: 97 ✭✭
    edited February 2018 #138
    DAY 0: Have now started charging the pack. Pack voltage 3.327v.


    Qty 24 Sharp 153W poly panels, Midnite MNPV6 combiner, Magnum MS-4448-PAE inverter with RC50 remote, Magnum MMP Mini panel, Midnite Classic 150 charge controller, Qty 32 LiFePO4 72AH cells in 2 string 48V array. BMS.
  • Raj174
    Raj174 Solar Expert Posts: 795 ✭✭✭✭
    Awesome! It's going to take quite a while at 5 amps, but patients will be rewarded.

    Rick
    4480W PV, MNE175DR-TR, MN Classic 150, Outback Radian GS4048A, Mate3, 51.2V 360AH nominal LiFePO4, Kohler Pro 5.2E genset.
  • LostinSpace
    LostinSpace Registered Users Posts: 97 ✭✭
    end of Day 3 = 3.340

    zzzzzzzz....

    Note: The charger did turn off on day 1 for an unknown time. I had disabled the Max Charge Time timer, but not the Capacity Cut-off timer, which was set at the maximum setting of 50,000mAH (50 AH).

    Thankfully I don't need to have my system powered on for critical items like Satellite internet. Nor less-critical items like refrigeration, heat, and lights.


    Qty 24 Sharp 153W poly panels, Midnite MNPV6 combiner, Magnum MS-4448-PAE inverter with RC50 remote, Magnum MMP Mini panel, Midnite Classic 150 charge controller, Qty 32 LiFePO4 72AH cells in 2 string 48V array. BMS.
  • Raj174
    Raj174 Solar Expert Posts: 795 ✭✭✭✭
    edited March 2018 #141
    Yep, this is why it is important to get as many amp hours pumped into the bank with the initial series charge as possible. It reduces the time charging in parallel. When I balanced my 195AH cells with a 10 amp charger, it took about 10 hours in this phase. Of course, it was a rebalance so cells were closer to start with.

    Rick
    4480W PV, MNE175DR-TR, MN Classic 150, Outback Radian GS4048A, Mate3, 51.2V 360AH nominal LiFePO4, Kohler Pro 5.2E genset.
  • LostinSpace
    LostinSpace Registered Users Posts: 97 ✭✭
    Day 7. Sick in bed for 3 days straight. Upon checking on Day 7 found that the charger had completed the cycle.

    The hr timer only seems to go to two digits so the total time calculates as 114 hours, or nearly 5 days. And of course some hours were lost day 1 when the charger turned itself off.

    So not too bad, really. So now I am going to disconnect the parallel rigging. Not sure whether to then let them sit for a couple of days and see what individual voltages settle down to, or just reconfigure into a 144AH array and start cycling.




    Qty 24 Sharp 153W poly panels, Midnite MNPV6 combiner, Magnum MS-4448-PAE inverter with RC50 remote, Magnum MMP Mini panel, Midnite Classic 150 charge controller, Qty 32 LiFePO4 72AH cells in 2 string 48V array. BMS.
  • karrak
    karrak Solar Expert Posts: 326 ✭✭✭✭
    Sorry to hear about you being sick. I was beginning to wonder if I had miscalculated something. Thanks for the update.

    I wouldn't worry about waiting for a few days.

    When you start cycling I would start with a charge voltage of 3.4V/cell (54.4V) and end current of C/50 or absorb time of 1/2 an hour and see what the balance is at the end of charge and work up to 3.45V/cell (55.2V).

    Simon
    Off-Grid with LFP (LiFePO4) battery, battery Installed April 2013
    32x90Ah Winston cells 2p16s (48V), MPP Solar PIP5048MS 5kW Inverter/80A MPPT controller/60A charger, 1900W of Solar Panels
    modified BMS based on TI bq769x0 cell monitors.
    Homemade overall system monitoring and power management  https://github.com/simat/BatteryMonitor
     

  • LostinSpace
    LostinSpace Registered Users Posts: 97 ✭✭
    Check.

    Qty 24 Sharp 153W poly panels, Midnite MNPV6 combiner, Magnum MS-4448-PAE inverter with RC50 remote, Magnum MMP Mini panel, Midnite Classic 150 charge controller, Qty 32 LiFePO4 72AH cells in 2 string 48V array. BMS.
  • LostinSpace
    LostinSpace Registered Users Posts: 97 ✭✭
    edited March 2018 #145
    OK. System is online and looking happy. I indulged myself by waiting for 24 hours after removing the parallel bus bars from the array. I wanted to see what cell voltages would do after resting for a day. Today I saw something I had not yet seen since purchasing the system. All cell voltages remained in the 3.5.x range. See photos

    Every other time I have checked the cells when I charged them as a PACK to the 3.6 the cells fell back to the high 3.3x range up to the 3.4.x range.

    But they have never held their charge like they did this time after charging in a parallel config with a 3 volt LFP charger.

    There's only one possible explanation. The cells have never been truly charged since I bought them.

    I also took photos of cell voltages right after turning on the inverter and loading the array down with the load from my freezer. Was happy to see that although voltage of course dropped, the cell voltages are staying pretty close.

    Pretty good. Pretty, pretty good. (That's my Larry David imitation).

    Before connecting to inverter


    Inverter ON

    Qty 24 Sharp 153W poly panels, Midnite MNPV6 combiner, Magnum MS-4448-PAE inverter with RC50 remote, Magnum MMP Mini panel, Midnite Classic 150 charge controller, Qty 32 LiFePO4 72AH cells in 2 string 48V array. BMS.
  • Raj174
    Raj174 Solar Expert Posts: 795 ✭✭✭✭
    @LostinSpace
    Great job! That's just what it should be. Really takes time and patients. You have a nice high balance, so an absorb setting equivalent to 3.45 volts per cell should work well. 
    Thanks for the continued updates.
    Rick
    4480W PV, MNE175DR-TR, MN Classic 150, Outback Radian GS4048A, Mate3, 51.2V 360AH nominal LiFePO4, Kohler Pro 5.2E genset.
  • LostinSpace
    LostinSpace Registered Users Posts: 97 ✭✭
    Thanks Rick. Sure have appreciated your input.

     For "scientific" purposes I am going to post a bit more on this just in case someone happens along. I want to take some readings near the discharged state, then read cell voltages again when charged to 3.4v, and see how well the pack is staying balanced. I will likely post on pack balance now and then for informational purposes.

    I say 3.4, because the scenario here is that this is primarily a standby system for the time being. We have grid power. But I don't want to simply set cell voltages at the storage voltage and turn the system OFF until needed, because in my experience all things atrophy from disuse. ;>

    So I will continue to run my ~350w stand up freezer, and find a sweet spot where I can charge the pack enough to run the freezer yet maximize cell life (open to suggestions). I will say that during the winter, with only 6 panels mounted so far, on cloudy/stormy days the system may draw close to the full range of the charge. But with the pack properly charged now I will take a fresh look at how much standby time is needed.

    I have the inverter set to turn off at 48v (3.0v). That should not be anywhere near where degradation might occur. It's weird because the specs for these particular cells specifies minimum voltage at 2.0. But from everything I've read that is dangerville.






    Qty 24 Sharp 153W poly panels, Midnite MNPV6 combiner, Magnum MS-4448-PAE inverter with RC50 remote, Magnum MMP Mini panel, Midnite Classic 150 charge controller, Qty 32 LiFePO4 72AH cells in 2 string 48V array. BMS.
  • Raj174
    Raj174 Solar Expert Posts: 795 ✭✭✭✭
    I see, so after running all your tests at higher absorb settings, around 3.4 volts, it might be interesting to test absorb slightly  above 3.35 (float) setting. This in my opinion, would be less stress on the cells, and I think the freezer wouldn't be too heavy a load that the bank wouldn't cycle well. The SOC would likely stay between 75 and 90%. A lot of experimentation possible.
    I agree, low voltage cut out at 3 volts per cell is fine. I'm sure you will be monitoring the balance at that time and I am interested in the results as well as the SOC.

    Rick

    4480W PV, MNE175DR-TR, MN Classic 150, Outback Radian GS4048A, Mate3, 51.2V 360AH nominal LiFePO4, Kohler Pro 5.2E genset.
  • LostinSpace
    LostinSpace Registered Users Posts: 97 ✭✭
    edited March 2018 #149
    OK. I just did a full cycling of the pack and here are some numbers:

    DISCHARGE
    One hour into discharging cell voltages were 27mv apart (3.324 - 3.351)

    Three days (60hrs) into the discharge cycle cell voltages were 14mv apart (3.169 - 3.183)

    The inverter is set to turn off at 48.0v (3.0v) At that event cell voltages were 202mv apart (2.872 - 3.074)

    CHARGE
    Upon recharging to 54.4v (3.4v) cell voltages were 171mv apart (3.352 - 3.523)

    Only one cell was particularly high at 3.523. if that cell was dropped from the averages the range would have been 3.352-3.465) or 113mv.

    ----------------------------
    I'm wondering if perhaps setting the cutoff voltage such that no one cell dropped below 3.0 might yield better results for keeping the pack balanced.


    Qty 24 Sharp 153W poly panels, Midnite MNPV6 combiner, Magnum MS-4448-PAE inverter with RC50 remote, Magnum MMP Mini panel, Midnite Classic 150 charge controller, Qty 32 LiFePO4 72AH cells in 2 string 48V array. BMS.
  • karrak
    karrak Solar Expert Posts: 326 ✭✭✭✭
    I am surprised that you are getting such a variation in cell voltages after the balance.

    Discharging the battery so far, the fact the cells are new and might not have had enough SEI forming cycles at the factory might all be factors.

    I think it is better for the battery to have more little charge/discharge cycles rather than fewer large cycles. With your current setup I would let the solar controller charge the battery each day and have the loads discharge it at night.

    For battery life if you are not concerned about running out of power due to cloudy weather it is better to cycle it between ~30%-~75%. I would look at the last 20% of charge (0%-20%SOC) as an emergency reserve and only go there occasionally. I doubt our battery goes below 25% more than ten times a year.

    I would put your balancing boards back on. Hopefully your battery will become more stable after a few cycles.

    Do you have the Wizzbang Jr. working so you have an SOC meter?

    Simon




    Off-Grid with LFP (LiFePO4) battery, battery Installed April 2013
    32x90Ah Winston cells 2p16s (48V), MPP Solar PIP5048MS 5kW Inverter/80A MPPT controller/60A charger, 1900W of Solar Panels
    modified BMS based on TI bq769x0 cell monitors.
    Homemade overall system monitoring and power management  https://github.com/simat/BatteryMonitor
     

  • LostinSpace
    LostinSpace Registered Users Posts: 97 ✭✭
    edited March 2018 #151
    karrak said:
    I would look at the last 20% of charge (0%-20%SOC) as an emergency reserve and only go there occasionally. I doubt our battery goes below 25% more than ten times a year.



    Simon I want to put a voltage number to those percentages if you don't mind. Are you defining 0% charge as 3.0 volts?

    If so, are we calling 3.45v 100% charge, and 3.0 volts 0%?

    Thus 3.23 would be 50% discharged.

    Thus 80% discharge would be 3.09v (or rounded to 3.1v).
    Qty 24 Sharp 153W poly panels, Midnite MNPV6 combiner, Magnum MS-4448-PAE inverter with RC50 remote, Magnum MMP Mini panel, Midnite Classic 150 charge controller, Qty 32 LiFePO4 72AH cells in 2 string 48V array. BMS.