behavior of grid-tie hybrid (Outback Radian 8048A)

2»

Comments

  • inMichigan
    inMichigan Registered Users Posts: 51 ✭✭
    Had a great solar day for Michigan this past week:

    Everything has been working nicely through the winter.  We've consumed our 'credits' earlier than I planned, however, since early Nov I've been driving a Bolt EV.  As a result, our 'house loads' have been higher than planned.

    inMichigan

    42 SP-335's (14.1kw) ->   4 FLEXmax 80's /  5kWh using 17 CA100 CALB cells /  FLEXnet DC  /  MATE3  -> 2 Radian GS8048A and watched over by Vantage Pro 2+ PWS all running since 2015
  • Raj174
    Raj174 Solar Expert Posts: 795 ✭✭✭✭
    We did too! Good to see your system working so well.



    Outback Radians are pretty awesome inverters, especially with LFP's.

    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 ✭✭✭✭
    This is the first time I have read this thread. Greetings again InMichigan.

    Here are my comments for what they are worth.

    Regarding the 56.0 to 56.8 difference, where are the voltage readings being taken?

    You mention that your cells rest at 3.27 V/cell at night. At that sort of rest voltage our Winston cells would be at an SOC of about 45%

    One thing that concerns me is that if there is a grid fault the Radian's will stop exporting power instantly. If this happens your charge controllers could end up trying to feed >200A into the battery.

    If the grid is down and you can't export power your batteries could get charged to 3.41 V/cell which depending on the end current could mean an SOC of >97%. As you are bottom balancing I would check the individual battery voltages at 56 volts.

    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
     

  • karrak
    karrak Solar Expert Posts: 326 ✭✭✭✭
    I have had some further thoughts about your system and wonder if we are not missing the elephant in the room.

    I will preface the following comments by saying that my engineering career has been designing digital electronics and instrumentation, I am not an electrical engineer.

    The elephant in your case is the AC ripple current. Just as the AC voltage that your inverter supplies to the grid varies between 0 volts and a peak of ~+/- 340 volts 120 per second the current also varies, this variation in AC current is reflected in the current drawn by the inverter.

    If the average DC current drawn by the inverters is 200A, the actual current will vary between nearly 0A and ~280A 120 times a second. Your solar panels will only supply DC current at 200A, the difference (ripple current +80A to -200A) will be supplied by 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
     

  • inMichigan
    inMichigan Registered Users Posts: 51 ✭✭
    edited February 2017 #36
    karrak said:
    Regarding the 56.0 to 56.8 difference, where are the voltage readings being taken?

    You mention that your cells rest at 3.27 V/cell at night. At that sort of rest voltage our Winston cells would be at an SOC of about 45%

    One thing that concerns me is that if there is a grid fault the Radian's will stop exporting power instantly. If this happens your charge controllers could end up trying to feed >200A into the battery.

    If the grid is down and you can't export power your batteries could get charged to 3.41 V/cell which depending on the end current could mean an SOC of >97%. As you are bottom balancing I would check the individual battery voltages at 56 volts.
    Those voltages were what was being shown on the OpticsRE page coming from the FlexDC that is measuring the voltage across the bank where the cables attach to the battery terminals.

    Being grid-tied, I did not feel compelled to set my night time voltage at a particularly high value of SOC, as I am aiming for a partial SOC theme.   If I wanted/needed to go into the night with a higher charge, I'd use the Mate3 ability to set different operating modes based on time.   So, I could stop selling in the late afternoon and let the batteries top off.

    You have pointed out an interesting failure mode: Mid-day with fantastic sun, and the grid goes down.    I have 200+ DC fuse in the battery bank cable.  In the perfect situation, the sum of the 4 FM80's could be ~280 ADC.  For a very brief moment, the bank would see 2C maybe even 2.8C.   Not ideal, and although tempting, I'm not planning to intentionally test this mode.   This first set of 100 AH's was meant to be my 'trainer' set.  I plan to replace it with ~400 AH, but since it's running, and the popularity of lithium is rising, I'm torn with 'when' to pull the trigger.  How fast the FM80's can react to the rising voltage is the unknown question.   Whatever the case, I don't have a solution other than more batteries), do you have a suggestion?

    If the grid went down, with enough sun left in the day and if my house loads were less than the harvest, yes, my SOC would run higher than I normally would want to see on a regular basis.   I don't believe I have any values that would be dangerous, do you? 

    I've tested different values of V_sell and V_abs.  There has to be a large enough delta in the command values to let the system sell.   By the way, under the direction of tech support when first setting up, I was told to set my V_sell much, much lower that my current values.  It was sunny.  The battery bank was fully charged. When reconnected to the grid with sell enabled, after the 5 minute timer, the inverters tried to sell ~14 kW to the grid.   The batteries tried to supply this current until the fuse blew.   All happened so fast, I didn't have time to react.

    I've read about the AC ripple, and I understand the physics, but it's not my speciality.  I showed photos of the interior of the inverter unit (there are 2 inside the GS8048A) found on the internet to two controls engineers at work (the right kind of engineer).  From the photos, they believed the Outback engineers were trying to smooth that ripple out based on the components they saw.  I asked around to borrow an oscilloscope, but didn't find one (yet).   I did use my Fluke DVM in AC to see what AC signals might be 'on the batteries' and measured a few mV's.  However, in hindsight, I don't know if the Fluke is filtering to only show 60 Hz or not.  Anyway, the only solution is more batteries.  The rule of thumb might be 100 AH per 1kw of inverter, of which, I have 2x8, I can't image anybody have 1600 AH of lithium.  I suspect the 'rule of thumb' was primarily driven by charging values like C/11.   Also, the circuits in a GS8048A ought to be more advanced than a $300 unit.  Would surely like to see some "data" on this topic with LFP before I buy the next 'long term' bank.

    I normally measure each cell voltage every few months, usually at 'rest' to get a stable value.   I have checked them under a light charge.  I'll post in the future.

    inMichigan
    42 SP-335's (14.1kw) ->   4 FLEXmax 80's /  5kWh using 17 CA100 CALB cells /  FLEXnet DC  /  MATE3  -> 2 Radian GS8048A and watched over by Vantage Pro 2+ PWS all running since 2015
  • mike95490
    mike95490 Solar Expert Posts: 9,583 ✭✭✭✭✭
    the rule of thumb of 100ah per Kw of inverter (at 48V) is for Flooded Lead Acid, and it's generic Internal Resistance. If you can figure out how to scale the internal resistance of a lead-acid pack to the much lower resistance of a Li pack, then you have the size needed to handle the ripple for a small Li pack.
     But with the IR aside, at some point you start pumping a lot of power in/out of a small bank and that may shorten the pack lifetime.
    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 ✭✭✭✭
    You have pointed out an interesting failure mode: Mid-day with fantastic sun, and the grid goes down.    I have 200+ DC fuse in the battery bank cable.  In the perfect situation, the sum of the 4 FM80's could be ~280 ADC.  For a very brief moment, the bank would see 2C maybe even 2.8C.   Not ideal, and although tempting, I'm not planning to intentionally test this mode.   This first set of 100 AH's was meant to be my 'trainer' set.  I plan to replace it with ~400 AH, but since it's running, and the popularity of lithium is rising, I'm torn with 'when' to pull the trigger.  How fast the FM80's can react to the rising voltage is the unknown question.   Whatever the case, I don't have a solution other than more batteries), do you have a suggestion?
    The big unknown is how the battery will react to being charged at over 2C which is twice the recommended charge limit. You said earlier that if you stopped selling power the charge controller current dropped to 50A which is good. I would try the same test with 10%-20% SOC and see what you get. Would be nice if the FM80 could limit the output current depending on battery current not total output current, but I don't think this is possible.
    If the grid went down, with enough sun left in the day and if my house loads were less than the harvest, yes, my SOC would run higher than I normally would want to see on a regular basis.   I don't believe I have any values that would be dangerous, do you? 
    It could be a problem if your battery got to around 3.41 at a low charge rate which would equate to an SOC of >97% and the battery is more than 3% out of balance at the top end. The balance boards you are using might save the situation but they will ruin your bottom balance.
    I've read about the AC ripple, and I understand the physics, but it's not my speciality.  I showed photos of the interior of the inverter unit (there are 2 inside the GS8048A) found on the internet to two controls engineers at work (the right kind of engineer).  From the photos, they believed the Outback engineers were trying to smooth that ripple out based on the components they saw.  I asked around to borrow an oscilloscope, but didn't find one (yet).   I did use my Fluke DVM in AC to see what AC signals might be 'on the batteries' and measured a few mV's.  However, in hindsight, I don't know if the Fluke is filtering to only show 60 Hz or not.  Anyway, the only solution is more batteries.  The rule of thumb might be 100 AH per 1kw of inverter, of which, I have 2x8, I can't image anybody have 1600 AH of lithium.  I suspect the 'rule of thumb' was primarily driven by charging values like C/11.   Also, the circuits in a GS8048A ought to be more advanced than a $300 unit.  Would surely like to see some "data" on this topic with LFP before I buy the next 'long term' bank.

    I might be wrong but I think you will find that the ripple that Outback are filtering out is the ripple caused by the switching circuitry that generates the 60Hz AC output. This switching circuitry probably works at a frequency >50KHz. It is much easier to filter high frequencies. The inductors and capacitors only need to be 500th of the size to filter at 50KHz compared to 100Hz. A very quick back of envelope calculation tells me you would need >.1F of capacitance to filter the 100Hz. Capacitors are usually sized in uF (microF).

    This trace is the voltage across my battery shunt resistor which corresponds to the current drawn from my battery by my $3,000 4kW inverter.


    It would be really good if you could get a CRO to look at your system's voltage and current waveforms. It is the only way you will really know what is going on. Unless you have a 60Hz filter on your Fluke you should see any AC signal?
    I normally measure each cell voltage every few months, usually at 'rest' to get a stable value.   I have checked them under a light charge.  I'll post in the future.

    Battery imbalance only becomes apparent with LFP batteries within a few % of 100%SOC or 0%SOC unless your battery is very out of balance. See http://forum.solar-electric.com/discussion/comment/366658#Comment_366658 for more information.

    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
     

  • inMichigan
    inMichigan Registered Users Posts: 51 ✭✭

    In October 2015, I removed the balancing boards visible in those early pictures.  At that time, I wasn't having issues with nuisance breaker tripping so my daytime voltage was lower.   As a result, the balance boards weren't really balancing.  Their current draw was slowly pulling all the batteries away from each other.  I realized that saga was discussed on a different forum of which I'm no longer active.

    The tripping was a result of 12 Sunpower 335W panels on the one FM80 and 10 on each of the other 3 FM80's.  Raising the voltage to reduce the current stopped what was a 1 time per 7 days tripping on blue sky with puffy cloud skies.  The company I purchased the solar side of the equipment got a little too close to the cloud edge effect.

    I have some other info at http://outbackpower.com/forum/viewtopic.php?f=24&t=8088

    I'll summarize and update here as time permits.

    What's a CRO?

    About the potential exposure of the batteries to 2C, any estimate of what the FM80's reaction time might be to system rebalancing in the event of sudden stop-sell on a sunny day?   You mentioned 'charging at 2C', I had expected the exposure to be extremely brief, however, as you mentioned someday they will experience this (which is why I do plan to raise the bank size this Spring).   In ~18 months, so far, it hasn't happened.   Barring the AC ripple discussions, this failure mode alone indicates I need a minimum of ~300 AH to not exceed 1C momentary charging.

    Measured voltage charts.. I'll post that later tonight.  IE at work is not playing nice.

    At a similar solar situation, could you measure the voltage with your DVM in DC_mode and AC_mode across your battery bank?  I'd be really excited to see if your meter measures 60 mV 'riding' on top of your DC voltage, just like your Rigol measured.  (and what model of DVM do you have)

    Hopefully the sun has not set in your location!

    For systems with the FlexDC and a valid battery shunt, the Mate3 could use the battery current in it's control method.  Today, I could only limit the total current coming out of the FM's... which would not make a very useful system most of the time.

    inMichigan


    42 SP-335's (14.1kw) ->   4 FLEXmax 80's /  5kWh using 17 CA100 CALB cells /  FLEXnet DC  /  MATE3  -> 2 Radian GS8048A and watched over by Vantage Pro 2+ PWS all running since 2015
  • inMichigan
    inMichigan Registered Users Posts: 51 ✭✭
    I've never tried to explain this plot to anybody... it's for me... but I'll give it a shot.
    I have re-bottom-balanced 3 times. 
    a) The first time when I first receive the batteries and was setting up a single Radian and FM80 on a table to 'learn the Mate" 
    b) Again, when everything was moved into its final position 'on the wall' late June 2015
    c) After I removed the balance boards in Oct 2015

    The X-axis is the number of days since this rebalancing series: Blue, Gray, and finally Orange
    The Y-axis is the average voltage of the 17 cells.
    Radius of the sphere proportional to the standard deviation.... so, small bubbles are good.
    Don't worry, if you aren't into standard deviations, below, I've pasted in the Tabulated Values.

    Notice the blue and gray spheres 'grew' over time, that was the current draw of the balance boards, removed before the Orange.
    Also, notice in the first 30 days or so, I measured the voltages at different bank conditions, 
    I could never explain why that sphere around day 300 is so large.  I record the total bank voltage, and then each of the 17 cells.  This lets me sum up the 17 values and I better get the overall as a kind of double check against my poor penmanship.



    I've plotted this data in several different ways.  I have noticed, that the closer the state of charge is to the original voltage of the 'balancing event', the difference is smaller.... as it should be.   I have measured them during charge, but don't usually type it up.  I'll do it later this week.  Usually, they are all within a few mV's if they have had a chance to rest... except for 1 unexplained time last August.  In hindsight, I should have checked the Mate log to see if we had lost power for a short period.
    Row 16 is the "range" (max - min).
    Row 13 is how far away that groups average is from the balancing voltage from Day 0.




    If I'm damaging the cells (from the AC ripple), would they show more drift?    It's time to measure again, but back in Nov, at Day 401 the range was 4 mV across all 17 cells, and 2 mV across the best 15.   When the cells where new, use one of the Radian modes to slowly drain the batteries at a 'fixed' current draw. I did it night after night, and build up a discharge curve.  I plan do that once per year.

    Actually, in re-reading some old posts, I came across a work around to my 12 panel being a bit too much on a single charge controller (besides the obvious of turning off 2 panels)....   It is to change that FM80's maximum charge from 80 to some lower value (to be experimentally determined).   I will lose some harvest, but gain the safety of slightly lower voltage.  I won't make any changes until I can be home during the day.

    inMichigan
    42 SP-335's (14.1kw) ->   4 FLEXmax 80's /  5kWh using 17 CA100 CALB cells /  FLEXnet DC  /  MATE3  -> 2 Radian GS8048A and watched over by Vantage Pro 2+ PWS all running since 2015
  • karrak
    karrak Solar Expert Posts: 326 ✭✭✭✭
    inMichigan said:What's a CRO?
    I am showing my age, it is a Cathode Ray Oscilloscope
    At a similar solar situation, could you measure the voltage with your DVM in DC_mode and AC_mode across your battery bank?  I'd be really excited to see if your meter measures 60 mV 'riding' on top of your DC voltage, just like your Rigol measured.  (and what model of DVM do you have)
    I was measuring the voltage across the current shunt not across the battery. My shunt is a 200A 50mV shunt which means that with a current of 200A going through the shunt that there will be a voltage of 50mV across it. There should be no reason why you can't measure the DC and AC components with your multimeter. Different multimeters display the AC part of the signal in different ways. Check your multimeter manual. You can always get a real AC reading by putting a non polarised capacitor >= 0.1uF with a voltage rating greater than  the DC component plus the AC component of the signal in series between the multimeter and the signal being measured..

    For systems with the FlexDC and a valid battery shunt, the Mate3 could use the battery current in it's control method.  Today, I could only limit the total current coming out of the FM's... which would not make a very useful system most of the time.

    IMO battery voltage is a pretty blunt instrument to control the amount of current going into or out of a battery. I would use the FlexDC if it allows you to set the maximum charge current coming from the charge controllers and hopefully the amount that your inverter will export to the grid from 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
     

  • karrak
    karrak Solar Expert Posts: 326 ✭✭✭✭
    I've never tried to explain this plot to anybody... it's for me... but I'll give it a shot.
    You did a good job of describing your graph.
    If I'm damaging the cells (from the AC ripple), would they show more drift?    It's time to measure again, but back in Nov, at Day 401 the range was 4 mV across all 17 cells, and 2 mV across the best 15.   When the cells where new, use one of the Radian modes to slowly drain the batteries at a 'fixed' current draw. I did it night after night, and build up a discharge curve.  I plan do that once per year.
    All the evidence I have seen suggests to me that the more stress you put an LFP battery under the faster the loss of capacity, not sure about drift in balance. You can only see how well balanced an LFP battery is when you look at the difference in voltages with the cell rest voltages above ~3.4V or below ~3.0V

    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
     

  • inMichigan
    inMichigan Registered Users Posts: 51 ✭✭
    karrak said:
    All the evidence I have seen suggests to me that the more stress you put an LFP battery under the faster the loss of capacity, not sure about drift in balance. You can only see how well balanced an LFP battery is when you look at the difference in voltages with the cell rest voltages above ~3.4V or below ~3.0V

    CRO, ah yes, it's been awhile since I held a Polaroid camera up to one of those.  Yes, I will check my shunt. 

    I was kind of hoping somebody with a damaged set could provide some commentary...  

    Here is my hopeful logic:  If I were to over stress the bank causing damage, I can not image each cell suffers in the same manner as there is always variation in a manufacturing process.  Wouldn't the damage show up by changing each cell's voltage vs capacity relationship?  If that relationship was different, and I am charging them in series, each cell experiences the same current flow thru the series.   So, each cell's voltage behavior should show this damage.    Thoughts? 

    42 SP-335's (14.1kw) ->   4 FLEXmax 80's /  5kWh using 17 CA100 CALB cells /  FLEXnet DC  /  MATE3  -> 2 Radian GS8048A and watched over by Vantage Pro 2+ PWS all running since 2015
  • mike95490
    mike95490 Solar Expert Posts: 9,583 ✭✭✭✭✭
    If I were to over stress the bank causing damage, I can not image each cell suffers in the same manner as there is always variation in a manufacturing process.  Wouldn't the damage show up by changing each cell's voltage vs capacity relationship?  If that relationship was different, and I am charging them in series, each cell experiences the same current flow thru the series.   So, each cell's voltage behavior should show this damage.

    Gee, I thought the damaged cells show up as Fire. 

    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 ✭✭✭✭
    karrak said:
    All the evidence I have seen suggests to me that the more stress you put an LFP battery under the faster the loss of capacity, not sure about drift in balance. You can only see how well balanced an LFP battery is when you look at the difference in voltages with the cell rest voltages above ~3.4V or below ~3.0V

    CRO, ah yes, it's been awhile since I held a Polaroid camera up to one of those.  Yes, I will check my shunt. 

    A couple of thoughts about measuring the AC voltage. If you have an old computer power supply, electronic toaster or some other electrical equipment that has some electronics in it, it is more than likely that there will be a capacitor something like this in it that you could use as your series capacitor. If you have a true RMS meter, the peak voltage will be the AC voltage reading x 1.414.


    Here is my hopeful logic:  If I were to over stress the bank causing damage, I can not image each cell suffers in the same manner as there is always variation in a manufacturing process.  Wouldn't the damage show up by changing each cell's voltage vs capacity relationship?  If that relationship was different, and I am charging them in series, each cell experiences the same current flow thru the series.   So, each cell's voltage behavior should show this damage.    Thoughts? 

    I agree with your supposition that the degradation (I will use the word degradation  rather than damage to avoid any anxiety) will most likely not be the same across all cells in a battery. Having said that I haven't found any scientific studies that specifically address this subject so I would only say that it is a plausible. Other factors that I think would  add to the difference in capacity loss are that the cells in the centre of the battery will get hotter from self heating than ones at the ends and that some of the cells maybe closer to a wall or other object that is radiating or absorbing heat which will result in a temperature difference between the cells.

    I don't think the degradation will change the basic operating voltage of the cell but it will change the capacity of the cell and the impedance (resistance) of the cell. This change in impedance will mean that the voltage necessary to push current into the battery as it ages will increase and conversely the voltage drop when drawing current from the battery will increase as it ages. The resting cell voltage will not change as the battery ages.

    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
     

  • inMichigan
    inMichigan Registered Users Posts: 51 ✭✭
    edited February 2017 #46
    Spring is starting to show it's strength....


    But the really good news..

    After consulting with Outback Tech support, they reminded me where the battery limiting current setting is:

    It was set to 80 A, but I went ahead and dropped it to 50.

    In what charging conditions do your cells ever feel 'warm' to the touch?   Mine alway feel ambient, except for time when my bare arm made contact...

     A lot of internet thread voltage has been put towards keeping cells balanced... If degradation doesn't cause the resting voltages to 'drift' apart, then, what would cause drift? 
    42 SP-335's (14.1kw) ->   4 FLEXmax 80's /  5kWh using 17 CA100 CALB cells /  FLEXnet DC  /  MATE3  -> 2 Radian GS8048A and watched over by Vantage Pro 2+ PWS all running since 2015
  • karrak
    karrak Solar Expert Posts: 326 ✭✭✭✭
    edited February 2017 #47
    In what charging conditions do your cells ever feel 'warm' to the touch?   Mine alway feel ambient, except for time when my bare arm made contact...
    They never feel warm to the touch with my maximum charge rate of ~0.1C.
     A lot of internet thread voltage has been put towards keeping cells balanced... If degradation doesn't cause the resting voltages to 'drift' apart, then, what would cause drift? 
    I am not sure I know what you mean, I don't think the voltages drift apart, I think the storage capacity and the impedance of the individual cells diverge.

    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
     

  • inMichigan
    inMichigan Registered Users Posts: 51 ✭✭
    Time flies when you're having fun, and not hunting down problems...

    It's been 7 years since our system came on-line. I'm still running the same 17 CALB CA100 cells described above with over 2500 day/night charge/discharge cycles. Now being grid-tied, I'm am very lightly cycling the batteries, but, that was my intent all along. Maybe 1 or 2 times a year they get some real exercise. If the power is out at night, I raise the charging voltages closer to the cell limits to push in enough power to sleep thru the night without a generator running on my little 5kWh of energy. If we shed every practical load, we just barely make it these days, so, I have lost some capability over time (as expected or we have more things that we think need to be on).  To be honest, I never expected my 'original' set of batteries to last 7 years, so many forums implied its way to easy to crash that first set when while learning to fly!   (ok, I did blow that 250A fuse one time)

    In the years that have passed, I haven't studied the newer hardware systems in great deal. I was recently helping a friend review a different brand.  It was a kind of all-in-one inverter/CC as one unit. I did think of one advantage of having functions split into separate 'units' is long term upkeep like I have with FM's and GS's. I have 4 FM80's (plus I have a spare put away). I have two GS8048's which means I have 4 4kW inverters in the system. I can run on just 3 of the 4 which I did during an early failure that was quickly covered by warranty. I swapped the inverter sub-unit myself. I assume someday those will be available on eBay. Had all of this content been crammed into a single more complex unit, will it be as easy to maintain for 30 years?  The only feature I feel is missing in my Radians is more control over the low voltage cutoff. I cannot set it where I would like to, so when I go low, I get very nervous and try to avoid that situation. It's because I have FLP and 17 cells. Interesting, now that I have a battery tester and have seen how cells recover from a crash, I'm not quite as paranoid.  It's been a very long time since I've flashed any software updates, so, maybe that's changed. It's running, been running, and why rock the boat.

    Throughout 2022, I have collected dozens more CALB CA100 cells as these are now discontinued. Some are new, some are used (probably less than mine) and a few are early CALB vintage from their serial number.  
     I've also measured every cells AH capacity in discharge/charge using 40A 3.65V/2.50V cycle on my EBC A40L tester.  This has taken me months. 

    I ran a variety of test cycles before deciding how I would run my sorting hat program looking at repeatability of test results vs clock time.  I tried various 'wait' times between each step.   The results showed the 'waits' only added variation to the AH values and made the test longer.   In the end, I went with:
    • Charge to 3.65V @ 40A, immediately
    • Discharge to 2.50V @ 40A, immediately                 What was the AH pulled?
    • Then, Charge to 3.65V @ 40A                                 What was the AH injected?
    After the test ends at the 'charge to 3.65V', I measure the voltage the next day to see the 'drop'.  I have found a few cells that drop much more than their siblings.   I've measured the impedance at a low and high cell voltage.   I have found a few cells with 'odd' impedance values as well.  I plan to set these 'odd' cells aside.  If I have enough odd-balls, I'll build a bank kept in reserve as a 'reboot' hot-spare.   I would charge it up now and then, use a switch to isolate it.  Should something happen that low voltage crashes my primary bank, use that mini-set to get everything back on line. 

    In the next week or two, I'll take the whole system down, pull out and measure my old cells, and reconstitute all the cells into 17S6P.  I've been saving this work for the poor solar days of Dec/Jan.  They've arrived.  I wonder what the largest self-made prismatic LFP bank is?

    This is a photo of 
    cells in my kitchen (very forgiving spouse),

    As to why I'm going from ~5 kWh to ~25 kWh?  I don't have much longer on my original true "net metering" 1 to 1 agreement with the POCO. It appears as if we're heading into a TOU world.  I want to be able to use the batteries to avoid the high-cost period in the late afternoon to early evening.  There was an attempt to add a monthly "solar fairness fee" to have even the current non 1-to-1 metering program.  With a high enough fee, I'd just drop out of any solar program and not bother to sell power back to the grid. Store what I can, and burn thru any excess into a Bitcoin miner. 

    inMichigan
    42 SP-335's (14.1kw) ->   4 FLEXmax 80's /  5kWh using 17 CA100 CALB cells /  FLEXnet DC  /  MATE3  -> 2 Radian GS8048A and watched over by Vantage Pro 2+ PWS all running since 2015