Older Radian system setup with SOK LiFePO batteries

[Rich V]

After ten years in service my AGM batteries gave up the ghost and I switched to SOK LiFePO batteries.
I would like to share my experience in setting up LIPO batteries and get some opinions on my settings.

Some info about my system and how it’s used.
Two Outback Radian GS8048 inverters (not the 8048A model)
Three Outback flexmate FM80 150VDC charge controllers
Two Outback GSLC load centers
Outback Mate3 system controller
Flexnet-DC
IBR-3-48-175 battery rack
Six SOK SK48v100-NCBT batteries ,48V/100AH each https://www.currentconnected.com/produc ... e-battery/
8.25 KW installed solar panels.

The system is grid tied and batteries are only used during a power outage. Power outages are not common so the batteries spend >99% of time in float. I tried to optimize my settings for float service with charging via the FM80s/solar panels. I’m located in central Arizona so lots of sun and I produce more power than I use on a yearly basis. I have net metering so selling to the grid is a break even proposition.

Installation and power up.
When received the SOK batteries BMS is off so I needed to provide 48V to the terminals to wake up the BMS. Simple to do, just jump the SOK terminals to the old AGM stack.
After wiring the SOK batteries in the IBR battery rack, 2 batteries per 175A breaker, I needed to power up the Radian inverters. At this point the system was powered down with all loads/inputs removed via the circuit breakers. I was aware of the large inrush current the Radians will draw on start up, this can cause the batteries to trip, so I used the method suggested by the SOK vendor, Current Connected. A 40watt 48VDC light bulb was used as the charge limiting resistor.

With the battery and IBR breakers in the off position I connected the bulb across one of the three 175A breakers installed on the IBR battery rack acting as a “jumper”. I then flipped the two battery breakers to on and powered up the first Radian by throwing both battery breakers on the GSLC load center. Unfortunately this did not work. The bulb flickered and the inverter would not power up.

I guessed the inverter needed more that the ~40 watts is was getting for power up so I needed a higher wattage resistor. I purchased a 1.5KW 48V water heater element https://www.amazon.com/dp/B01GPSL6IU?ps ... ct_details and wired it as per the light bulb. This worked with the inverter starting without issue. I then closed the 3 IBR breakers on the battery rack and turned on the second Radian. The water heater element worked perfectly and was a cheap ‘power resistor” for $32.

The settings I used for SOK batteries.
Radian settings
Absorb Voltage and Time 56.8 V / 0.1 hr
Float Voltage and Time 55.2/0hr
Re-float Voltage 51.2V
Re-bulk Voltage 48.8V not adjustable
AC Charger Limit (AAC) 30a
Low Battery Cutout 48 V
LBCO Delay 5 min not adjustable
Low Battery Cut-in 50V
High Battery Cutout 68v not adjustable
HBCO Delay Not adjustable
High Battery Cut-in <68V not adjustable
Sell_RE Voltage 54V

Charge Controller settings
Absorb Voltage 57.2V/0.1hr
Float Voltage 55.2V
Re-bulk Voltage 51.2v
DC Current Limit 80
Absorb End Amps 0A

FN-DC settings
Battery Ah 600
Charged Voltage 56.2V/5 min
Charged Return Amps 12A
Battery Charge Efficiency 98.00%

After running for a week the system is working normally. It goes thru a short MPPT bulk phase at sunrise then GT sell for the rests of the day. At night the batteries BMS shows all 6 units at 99-100% SOC and the 16 individual cells are at <0.01V differential voltage, nicely balanced, and open circuit voltage at 53.25V by DVM @ battery terminals (BMS shows 53.33V)

During the day with up to ~6-7KW of sell power into the grid the individual cells are at <0.15V differential voltage with in/out amps fluctuating at ~ 2-3A per battery. Voltage is ~55-55.4V in sell mode as reported by the FNDC.

I have not cycled the batteries but will do so soon.

I would appreciate any comments on my set up and suggested changes/improvements.

 

[G00SE]

This is awesome and thanks for sharing.
On the Outback forum it seems that many have a heck of a time getting various Lifepo4 batteries and radians to communicate
Oh and howdy neighbor. I’m just to the northeast of you based on your location of “central” az

 

[zanydroid]

Cool setup. I’ve heard that leaving LFP at the high SoC all the time is not ideal for calendar aging; along these lines I think the calendar aging specs are based on frequent enough cycling to avoid this effect.

Following to see what people recommend.

 

[Rich V]

This is awesome and thanks for sharing.
On the Outback forum it seems that many have a heck of a time getting various Lifepo4 batteries and radians to communicate
Oh and howdy neighbor. I’m just to the northeast of you based on your location of “central” az

I'm in the Prescott area so howdy back @ you neighbor.
Current connected steered me to the SK48v100-NCBT batteries over the SK48v100. Apparently they use a different BMS and the SK48v100 has a lower max short term current draw spec. The SK48v100-NCBT was cheaper and the Radian inverters won't talk to the batteries so a win win for me. I suspect the problems people have with Outback & LiFePO playing together is the very large current draw the Radians can have at start up or under a sudden heavy load. One Radian is rated @ 8KW continuous and 16KW peak. If the BMS can't handle the load spike it shuts down. Just my guess.

 

[Rich V]

I’ve heard that leaving LFP at the high SoC all the time is not ideal for calendar aging; along these lines I think the calendar aging specs are based on frequent enough cycling to avoid this effect.

Following to see what people recommend.

I have heard differing opinions on this and would like to hear from the experts as well. @HighTechLab
I also wonder if periodical discharge will be beneficial? This was the case for my AGMs.

 

[zanydroid]

With LFP the SoC tracking is done by coulomb counting in the shunt and not battery voltage. So some manner of cycling would be needed to recalibrate against self discharge (self discharge current would not go through the shunt so the shunt cannot know anything about it)

 

[Rich V]

With LFP the SoC tracking is done by coulomb counting in the shunt and not battery voltage. So some manner of cycling would be needed to recalibrate against self discharge (self discharge current would not go through the shunt so the shunt cannot know anything about it)

During the daily GT sell period there is a fluctuation of a few amps in and out of the batteries as the Radian is load matching/averaging (correct term?) the grid power sell state, i.e. when a cloud goes by it pulls power from the batteries to make up for the panel loss for a few seconds then reduces the power to the grid to match the panel output. On a good day I'll push > 30KWH to the grid. Also the Radians use ~0.5KWH at night to just be powered up. These "loads" may help the SOC tracking and keep calibration. Is it enough? Don't know and I hope the experts can speak to this.

 

[zanydroid]

I found this old thread while thinking about it. Since you are coming from AGM the LFP quirks like this are part of the learning curve you have.

Post in thread 'How does charge controller know SOC of LFP batteries?'
https://diysolarforum.com/threads/h...r-know-soc-of-lfp-batteries.50400/post-640272

 

[Rich V]

I found this old thread while thinking about it. Since you are coming from AGM the LFP quirks like this are part of the learning curve you have.

Post in thread 'How does charge controller know SOC of LFP batteries?'
https://diysolarforum.com/threads/h...r-know-soc-of-lfp-batteries.50400/post-640272

​

zanydroid, thanks for the link.
I have posted the same thread over on the Outback forum https://forum.outbackpower.com/viewtopic.php?p=102745#p102745
From the feedback there and here it looks like long term float @100% is not good for longevity of LiFePO batteries.
The question then is can the SOK battery charge parameters be set in my Outback system to float @ say 90% SOC max and still allow me to be in grid tied mode selling back to the grid?

As part of my research on Radian settings I put together a table of recommended settings that are published by Outback https://outbackpower.com/downloads/docu ... li-ion.pdf Simpliphi https://simpliphipower.com/wp-content/u ... -guide.pdf and for SOK from current connected technical expert. The first six columns are manufacturer recommended settings. The last two columns are what my lead AGMs were using and my modified SOK settings.
This shows a fair amount of latitude in charge settings for the listed LiFePO batteries. How they arrived at the settings and why they differ between brands is not clear, at least to me.

Any suggestions on how I could adjust the charge setting to get to a 80-90% max SOC during float (grid tied sell) would be welcome.

 

[zanydroid]

@Rich V you might want to get in touch with some of the other forum members here about Outback with non EnerSys batteries, EG

Post in thread 'Looking for feedback on my choices for inverter and batteries'
https://diysolarforum.com/threads/l...-for-inverter-and-batteries.67241/post-852640

That member says Outback is taking away capabilities for custom profiles on some hardware or taking away customer support resources

 

[zanydroid]

As another datapoint I noticed this OnePlus phone I just got (as a backup) has a feature that delays charging above 80% until right before the morning starts. That will also have the effect of reducing the time sitting at high SoC (since it will drain away from the high SOC once you wake up and start working the phone).

Granted that is LiPo and not LFP but there might be some similarities.

 

[Rich V]

zanydroid
Thanks for the posts.
My Outback Radian inverters are the older models. Outback already stated they don't support Lithium batteries for my model. Fortunately the flexmate FM80, Flexnet-DC & Mate3 combo I have gives me enough charger flexibility that I can adjust settings over a wide range.

My problem is finding settings that maintains grid tied sell functionality and allows for <100% SOC in float.
I'll play around with the settings and see if I find something that works.

 

[zanydroid]

There are also people on this forum that use a custom PLC to do whatever they want. IE sending APIs or other signals to the equipment to force them to change behavior. I’m not sure how much are doing it with Outback.

I know this has been done with Schneider XW a fair bit (which may not strictly speaking be legal for a unit with a UL listed ESS power control function but who cares as long as you aren’t doing it to abuse and cheat the net metering system).

 

[sfriedrich]

Hi all. First post on the forum after doing some reading while modifying my setup. Figured I would post here rather than start a new thread since equipment is similar.

My system is 5kW PV, 2 Outback GVFX3648 grid-tie inverters, 2 Outback MX60 charge controllers, and an Outback Mate/hub. I installed it back in 2006 with 8 sealed lead acid 100ah batteries. The batteries are only used during power outages, but at 17 years old were basically only good for an hour or two during an outage and were starting to swell.

So I did my research and ended up buying two SOK 48V100 server rack batteries that have built in BMS. Based on their manual, the recommended settings from SOK when connected to older open loop non-communicating systems like mine are:

Charge Current for normal use: Up to 63A (Per Battery)
Charge Current for fast charge: Up to 95A (Per Battery)
Absorption Voltage: 57.6v recommended; up to 58.4v max
Absorption Time: 15 Minutes
Float Voltage: 55.2v
Charging Target: 100% SOC
Low Voltage Shutdown: 48v

Under "Charging SOC%" they specifically recommend maintaining charge at 100% and not to be concerned about degradation due to maintaining at 100% SOC. My understanding is that the engineers at SOK are from the cell manufacturer Calb, so I expect they are knowledgeable about the characteristics of their cells. Also, in the newest Teslas that use LiFePO4 cells, they also recommend keeping the car at 100% SOC unlike their models that use NMC cells.

Based on the above, I have used the following settings in my system:

Inverter
Charger limit: 5 AAC (set low to maximize charge from solar when grid restored and reduce max current if charging from both inverter and CC)
Absorb set point: 57.6V
Absorb time limit: 0.3hr
Float set point: 54.4V
Float time period: 0.5hr
Refloat set point: 52V
Equalization set point: 57.6V
Equalization time period: 0hr (equalization disabled)
Sell RE_volts: 54.4V

MX60 CC
Charge current limit: 50A
Absorb volts: 57.6V
Float volts: 55.2V

A couple comments on the above settings. I initially tried setting Sell RE_volts to 53.2V but this caused the batteries to immediately start discharging once charging was complete, and suspect that it would cause the batteries to be cycled more than necessary. So I have now set the Sell RE_volts to be the same as the inverter float voltage and this seems to be working well. Also aligns with the resting 100% SOC voltage of 54.4V for LiFePO4.
I have set the MX60 CC float voltage higher than the inverter float voltage so that the CC tries to maintain a higher voltage than the inverter. This is what the MX60 manual recommends for grid tie systems, although in the inverter manual it says that when the MX60 is set to grid-tied mode that the inverter/Hub will automatically raise the MX60 float voltage to the absorb voltage. But that does not seem to happen based on what I see on the MX60 screen.

Happy to receive comments/suggestions.

 

[Rich V]

Hi all. First post on the forum after doing some reading while modifying my setup. Figured I would post here rather than start a new thread since equipment is similar.

My system is 5kW PV, 2 Outback GVFX3648 grid-tie inverters, 2 Outback MX60 charge controllers, and an Outback Mate/hub. I installed it back in 2006 with 8 sealed lead acid 100ah batteries. The batteries are only used during power outages, but at 17 years old were basically only good for an hour or two during an outage and were starting to swell.

So I did my research and ended up buying two SOK 48V100 server rack batteries that have built in BMS. Based on their manual, the recommended settings from SOK when connected to older open loop non-communicating systems like mine are:

Charge Current for normal use: Up to 63A (Per Battery)
Charge Current for fast charge: Up to 95A (Per Battery)
Absorption Voltage: 57.6v recommended; up to 58.4v max
Absorption Time: 15 Minutes
Float Voltage: 55.2v
Charging Target: 100% SOC
Low Voltage Shutdown: 48v

Under "Charging SOC%" they specifically recommend maintaining charge at 100% and not to be concerned about degradation due to maintaining at 100% SOC. My understanding is that the engineers at SOK are from the cell manufacturer Calb, so I expect they are knowledgeable about the characteristics of their cells. Also, in the newest Teslas that use LiFePO4 cells, they also recommend keeping the car at 100% SOC unlike their models that use NMC cells.

Based on the above, I have used the following settings in my system:

Inverter
Charger limit: 5 AAC (set low to maximize charge from solar when grid restored and reduce max current if charging from both inverter and CC)
Absorb set point: 57.6V
Absorb time limit: 0.3hr
Float set point: 54.4V
Float time period: 0.5hr
Refloat set point: 52V
Equalization set point: 57.6V
Equalization time period: 0hr (equalization disabled)
Sell RE_volts: 54.4V

MX60 CC
Charge current limit: 50A
Absorb volts: 57.6V
Float volts: 55.2V

A couple comments on the above settings. I initially tried setting Sell RE_volts to 53.2V but this caused the batteries to immediately start discharging once charging was complete, and suspect that it would cause the batteries to be cycled more than necessary. So I have now set the Sell RE_volts to be the same as the inverter float voltage and this seems to be working well. Also aligns with the resting 100% SOC voltage of 54.4V for LiFePO4.
I have set the MX60 CC float voltage higher than the inverter float voltage so that the CC tries to maintain a higher voltage than the inverter. This is what the MX60 manual recommends for grid tie systems, although in the inverter manual it says that when the MX60 is set to grid-tied mode that the inverter/Hub will automatically raise the MX60 float voltage to the absorb voltage. But that does not seem to happen based on what I see on the MX60 screen.

Happy to receive comments/suggestions.

sfriedrich
These are the settings I settled on.
Radian GS8048 Inverter (x2)
Charger limit: 30A
Absorb set point: 56.8V
Absorb time limit: 0.1hr
Float set point: 55.2V
Float time period: 0.0hr
Refloat set point: 51.2V
Sell RE_volts: 54.0V

FM80 CC (x3)
Absorb volts: 57.2V/0.1hr
Float volts: 55.2V/0Hr
rebulk 51.2
current limit 80A

Fn-DC settings
Battery Ah 600
Charged voltage 56.2V/1hr
Return amps 12A
Charge efficiency 98%

With these settings my system has worked well. All cells are at + 0.002v after resting overnight with a resting voltage @ 53.2V and SOC of 100%. (I do question the 100% SOC with the battery @ 53.2V resting) During the day my FM80 CC is showing ~55V max pushing power to the grid.
I try to do a pull down to ~50% then bulk absorb from the inverters once a month to keep the BMS readings accurate.
Hope this helps
Rich