LFP Battery Bank

Rev: 0.3

Author contact: paul.bouchier@gmail.com


Overview of Document

This document is comprised of a web page for the DIY-built LFP battery bank for Paul Bouchier's RV. It contains four sections, aimed at different audiences.

Audience

This document is intended for use by anyone trouble-shooting the battery system. It also describes changes from the standard RV electrical design shipped by the manufacturer. It is also intended to provide ideas for others designing and building their own battery systems. The reader is expected to have electrical knowledge at the level of electrician, electrical tech or electrical engineer.

License

This document is made available to anyone to use for any purpose. It is free of licenses and copyrights. It is  provided in the hope it will be useful but is without warranty or support of any kind. There is no assurance of correctness. The design may or may not be appropriate for any particular application. A best-effort was made to provide information at a level useful to someone skilled with electrical circuitry, but the user is responsible for their use of any design elements. 

File Locations

The source files for drawings, images etc are on Paul's laptop at ~/Documents/RVing/RVBatteryPower.

Section 1

Overview of LFP battery system for RV

Motivation

Design and construction of the battery system was motivated by two observations that suggested minimizing dependence on electrical grid hookups:

The project was scoped to be sufficient to achieve those goals, and didn't go much beyond. Solar was not included in the design because it adds a lot of cost and complexity to battery management. Solar likely would have doubled the cost, and we're not necessarily committed to this RV, so containing cost was important. An inverter was included because AC power adds a lot to the glamping experience: charging devices, microwave, TV, etc. LFP batteries were chosen because they have sufficient capacity, high charge/discharge rates, and much less danger of explosion or fire than Lithium Cobalt batteries. In short, they can operate safely at levels lead-acid batteries can't.

System overview

The system is buuilt around a 12V 280 amp-hour (Ah) LFP battery with a Battery Management System (BMS) to control when it is connected to the RV's 12V bus. An upsized power converter converts RV AC input power (shore power) to 12V for running the RV and charging the battery when connected to the grid or generator. An inverter converts 12V to 120V AC when switched on, and can power all AC electrical outlets when shore power is not available. A 2kW portable generator recharges the battery during extended off-grid camping. Fallback switchover to the original lead-acid battery was provided for resilience in case of system failure.

Summary of outcomes

The battery system delivered on the design goal of running for 2 - 2.5 days without recharging, and came close to the goal of recharging in 2.5 hours - it actually takes 3 hours. 

The battery system plus Honda EU2200 generator enabled us to boondock for 12-days straight, running the generator every 2nd day or so. The graph below shows State-Of-Charge (SOC %) over the 12-day off-grid period, and how it varied over time. SOC of 100% represents 280Ah stored, and 0% is completely flat batteries. An upward slope represents charging from generator, and downward slope represents discharging to supply electrical loads. The graph shows a full charge taking about 0.2 days, and a couple of partial charges that were necessitated by low battery but interrupted by trips out, or dinner. In an hour I could put in enough charge to last another day.

We were conservative with energy use, turning the inverter off when not needed, and using only necessary lighting. We preferred charging the laptop, iPad etc in the truck while driving around. We had available all the RV electrical mod-cons we needed (microwave, TV, hair dryer, electric mixer, etc), but not air conditioning. We preferred camp sites in the forest, where A/C wasn't required. In hotter areas we used RV parks with shore power for A/C.

We were also conservative with water use - we could last over a week on a tank of fresh water, and would dump and refill tanks when we moved.)

In summary, the battery system did enable camping at beautiful sites in desirable areas without much competition from other RVs, and I didn't have to run the generator very much while off-grid. At that level, the system achieved its goals.

System cost

The parts cost breakdown is shown below.

Total cost of battery system: $3000

The Honda EU2200 generator was an additional $1000, and was needed for boondocking anyway, so is not counted as a battery cost.

Comparison with Lead-acid Battery System

Many RV use lead-acid batteries. The table below compares the performance of the as-built LFP battery system with that of a hypothetical lead-acid battery system with equivalent usable capacity, built out of 6V 215Ah group GC2 deep-cycle golf cart batteries connected in series-parallel. In order to last the rated 500 discharge cycles, deep-cycle batteries should not be discharged below 50%. Therefore it is necessary to provision 600 Ah of capacity to deliver 300 Ah of usable capacity.

Discussion of parameters

From almost every perspective, LFP batteries deliver more performance for less money. The trade-off is the initial cost of LFP is higher, and they are more complicated to design to deliver the performance, because they need to be considered as part of a system.

As an example of LFP design complexity, there are many "drop-in lead-acid replacement" LFP batteries on the market, and they have a BMS built-in. But you need to carefully inspect their parameters for max charge/discharge rate etc. Also, BMSs for paralleled LFP battery systems typically don't communicate with each other, which can lead to unbalanced batteries.

Section 2 

User Manual

Operation Concept

The battery system is intended to provide power for basic RV use (fridge, lights, some TV, microwave) for 2 - 3 days of boondocking. When drained, it can be recharged from a 2200W generator in 2 - 3 hours, providing for extended boondocking with minimal generator use. It provides 120V AC for occasional light use when on battery power. Shore power may be provided from a 2-phase 50A source like an RV-park power stand, or house 15A outlet, or 2200W generator. The battery is intended to be disconnected when on shore power, unless charging. Charging shuts off automatically when complete and leaves the battery disconnected. The design does not accommodate solar charging, which would add much complexity. When shore power is disconnected, the battery automatically connects, to provide 12V for RV loads.

Controls

The following controls are provided:

Indicators

Inverter

A 2kW (4kW surge) inverter is available when the BMS has enabled discharge. (It is also available when shore power is connected but is not recommended for use.) When on, the inverter provides 120V AC to the power-center phase that feeds outlets, microwave, "General" loads like fridge, TV, and one AC unit. When on, the Automatic Transfer Switch (ATS) disconnects incoming power from that phase. The A/C is not recommended to be used when running on inverter - it will work but drain the battery in a couple of hours. When on, the inverter and ATS consume about 1 amp from the batteries; suggestion: only turn on when shore power is unavailable and 120V appliances are to be used, and turn off when done with 120V in order to economize on battery consumption. Monitor power consumption when on inverter, to avoid unintentional high loads like AC mode of fridge.

Inverter checklist

Before turning inverter on, check:

Lead-acid Battery Backup

If the main (LFP) battery system fails, or the battery runs flat, switch the battery-selector switch to the lead-acid position to draw power from the original lead-acid battery. The lead-acid battery is on charge whenever shore power is provided. The original 55 amp converter provides 12V power as it originally did if shore power is provided and the lead-acid battery is selected.

Section 3

Design Description

This section describes the requirements, constraints, and design of the LFP battery system. See the wiring diagram for implementation details.

High current 12V loads like jackstands operate off 12V from the converter when shore power is on, or from battery when off.

Requirements

Solar charging is NOT a requirement of this design

Use Cases

At home or RV park with AC hookups

Boondocking or RV park without AC hookups

Battery Use Model

RV Power

The Energy Sources table shows what kind of energy is used for each appliance, depending on whether it is being used while on shore power, or off-grid.

The AC Loads table shows the current drawn from shore power (120V) by each appliance, and it's equivalent watts and how many amps the inverter would draw from battery if powering it.

System Design

The pre-existing RV power design was changed in several significant ways.

12V charge power used to be brought into the RV on the trailer electrical connector, and used to charge the RV battery while towing. This is no longer possible, since the LFP battery runs at a different voltage and has a different charging philosophy. Therefore the charge power wire from the tow vehicle was unwired at the J-box on the hitch and is no longer connected to RV power.

The General AC circuit-breaker and GFCI AC circuit-breaker used to be on different legs of the 2-phase 50A input. The electrical panel was re-wired to move them both onto the leg powered by the inverter, so that the inverter could power anything plugged into an AC outlet.

Charge/discharge design approach

The way of dealing with charging and discharging LFP batteries is a key design decision. LFP batteries must not be left on "float" or trickle charge, unlike lead-acid, or their life will be dramatically shortened due to overcharging.

The gold standard in this space is the Victron Multiplus type inverter/converter, which can be aided by a REC BMS. In this type of system the BMS can ask the Muliplus to reduce charging or discharging current as the battery approaches the knee of the charge or discharge curve. When charging and the battery becomes charged the Multiplus will reduce current going into the battery, and draw all load current from shore power or solar, effectively isolating the battery. This comes at a cost: $1300 for the Multiplus on top of the roughly $600 for the BMS.

A lower-cost design approach was taken that involves semi-automatic charge management, using only a REC BMS. The user decides when to charge the battery, and activates charging with a button. The system automatically disconnects the battery and leaves it disconnected once charge is complete. When shore power is removed the system automatically connects the battery to provide 12V power. If shore power is reconnected the battery is again disconnected and holds its previous state of charge.

80% feature (doesn't work)

This feature was intended to keep a reserve charge available to run jackstands etc by shutting off discharge when SOC drops to 10%. It also would prolong battery life by preventing charging from driving the batteries into the knee of the charge curve. Once disconnected, hysteresis prevents the battery from being reconnected until some charge is added back. Override the 10% disconnect by switching this switch to 100%. However, battery balancing will not take place during charging unless the switch is in the 100% position. The reason it doesn't work in discharge is that disconnect can happen at SOC > 10% if significant current is being drawn (>5 amps), thus it disconnects when you run the microwave or slides or anything significant. The reason it doesn't work in charge is not well understood. The BMS-reported cell voltage fluctuates 200mV during charging for unknown reasons, and the voltage threshold for 90% is within the normal cell charging voltage, so it stops charging prematurely.

In order to implement a working 80% feature, voltage should not be used. A microcontroller should monitor the SOC reports from the BMS on CAN or RS485, and take action on that basis.

AC Power Design

The diagram below shows the AC wiring.

An Automatic Transfer Switch (ATS) is made from a 40 amp DPDT relay, with contacts paralleled to provide 80 amps of capacity. The input phase which feeds the off-grid-available AC loads is switched between shore power and inverter power. The inverter has a remote on/off button located in the RV power panel. When on, it activates the ATS to feed power from the inverter to one phase. The inverter only powers one phase, and the converter is on the other (unpowered) phase to avoid a loop where the inverter would power the converter which would power the inverter.

The ATS relay draws 12W from inverter power when the inverter is on. The relay is activated with inverter power for improved reliability: it seemed better for the default (unpowered) state of the relay to connect shore power to the AC phase. If the relay fails, or its coil wiring fails, the higher-reliability shore power source will be passed through. The design relies on the upstream breaker to protect the ATS relay in case of catastrophic failure.

A new 120 amp converter is added for charging the LFP battery and powering 12V loads when on shore power. Unfortunately, the converter can only actually deliver 90 amps, and consumes 18 amps at 120V while doing so, due to horrible efficiency and/or power-factor. The LFP battery, inverter and converter are tightly bound together with 1/0 AWG and 2 AWG welding cables to support the high currents between them. High current connectors enable disconnection and removal of each component. Wiring to the battery selector switch and onward is lighter gauge - 4 AWG - nothing else draws that level of current. The converter was switch-selected to constant voltage mode and set to 14.2V unloaded.

Automatic Transfer Switch (ATS) Construction

The high-current terminals for the 6AWG cables carrying main input power into the ATS were pulled from a hot-tub disconnect box and mounted with the ATS relay in a 100A electrical circuit-breaker box. Both boxes were obtained at Lowes. 6AWG wire was used to connect terminals to the ATS relay. The ATS box was cabled to the main power center box with 3 feet of NMC cable. The pictures below show the inside of the ATS box, and its cabling to the power center. The inverter AC was brought from the 3-pin 120V 15A plug on the inverter to the ATS box on a standard appliance cable (14AWG). (Inverter current is much lower - 2000W => 18 amps). 

The upper picture is the ATS box internals prior to inverter power and coil wiring. It shows the high-current terminals and the ATS relay mounting. The lower picture is the ATS box wired to the RV power center box. Also shown in the lower picture is a 3-outlet extension cable plugged into the converter outlet from the power center. The 3 outlets feed the two converters and the AC-On rely on the battery box. See the wiring diagrams for connection details.

Power-center Construction

The Power-center is the name for the OEM box housing AC circuit breakers and AC wiring terminations, and 12V fuses and their wiring terminations. It is physically located in the cabinet by the RV door, and backs onto a utilities space behind the basement wall, where the main battery system components were installed.

The Power-center was modified in two ways:

The added terminal was constructed from a 100 amp ground-terminal used in breaker panels mounted to a double-layer plastic plate such that the mounting bolt for the terminal does not exit the power-center. The picture below shows the added terminal and white mounting plates on the left side of the power center. The picture below that shows the power-center with front panel installed and new breaker labels.

Battery Box Design

The battery box is designed around four 280 Amp-hour prismatic LFP cells monitored by a REC Battery Management System (BMS) (victron-compatible version).

Battery power passes through a 400 amp fuse to a 400 amp master shutoff contactor, and from there to the converter, inverter, and battery selector switch. Current returning to the battery passes through a 500A/50mV shunt sensed by the BMS, which drives a coulomb counter based on measured shunt current, and displays SOC on an LCD near the RV control panel.

A pre-charge device drives power to the master shutoff relay coil to close the contacts; before activating the coil it puts a 66 ohm resistor across the relay contacts for a few seconds which pre-charges capacitors in the inverter and converter, thereby avoiding arcing when the contactor closes. This extends the life of the contactor contacts. The pre-charge device must supply the surge current required by the master shutoff relay coil (1.3A) through its control input. The BMS optocouplers cannot drive this, so the control input of the pre-charge is in turn driven from a pre-charge control relay which passes unswitched fused 12V into the pre-charge device control input.

The coil of the pre-charge control relay is wired-OR driven by the charge and the discharge signals, which cause the master shutoff relay to close under supported charge or discharge conditions and open to protect the battery otherwise,

The discharge signal is controlled by the BMS when shore power is not supplied (converterAC is 0V). In this condition the AC_on relay passes unswitched fused 12V to BMS main relay, which opens (deactivates discharge) when the lowest cell voltage drops to 2.50V (SOC 0%). Provided the lowest cell voltage exceeds 2.50V BMS main relay output is provided to Opto2 which opens (deactivates discharge) when minimum cell voltage drops to 3.18V (SOC 10%). The 100% relay, if active, shorts across Opto2, preventing it from deactivating discharge at SOC 10%, and allowing use of the last 10% of battery capacity. The discharge signal also drives an LED status indicator on the RV control panel via IO1.

The charge signal is controlled by the BMS when shore power is supplied (converterAC is 120V). In this condition the AC_on relay passes unswitched fused 12V to opto1, which opens (deactivates charge) when maximum cell voltage exceeds 3.60V (SOC 100%). Provided the maximum cell voltage is less than 3.60V BMS opto1 passes 12V to BMS relay 1 which opens (deactivates charge) when maximum cell voltage rises above 3.46V (SOC 90%). The 100% relay, if active, shorts across relay 1, preventing it from deactivating charge at 90%, and allowing charging to 100% and BMS cell balancing. Additionally, the charge signal cannot become active unless the charge momentary button on the RV control panel is pressed. When the charge signal is activated the charge-hold relay activates to hold it active until the BMS deactivates the circuit. The charge signal also drives an LED status indicator on the RV control panel via IO2.

The 100% relay which shorts across Opto2 and BMS relay 1 is controlled by the 100% switch on the RV control panel via IO3

The hibernate switch on the control panel puts the BMS into ultra-low power mode, with charge and discharge inactive, and removes power from the LCD.

See the diagrams below for Battery box wiring, and associated control panel wiring. The terminals labeled IO1 - IO9 connect to each other with a cable harness with Molex connectors at each end; the battery box is mounted behind a basement wall and the control panel switches and indicators are mounted in or near the RV master switch panel. 

Battery Box Mechanical Construction

The picture below shows the mechanical arrangement of battery box components. The cells are mounted in a box lined with  foam rubber, and mechanically isolated from each other with foam rubber. The box is made of 1/2" ply with all-thread rods providing some compression. Plastic side-plates brace the cushioning against the all-thread rods. Foam rubber floor tiles were cut to size and used as cushioning to isolate batteries and accommodate their size change. There is a little bit of flex in the diagonal series-connecting inter-cell power cables. The cells can expand or contract a little without stressing their terminals. (The battery cells are spec'd to change thickness 0.5mm between 30% and 100% SOC.) Most components are mounted on the side of the box for easy servicing. 

Battery Selector-switch Design

As shown on the AC wiring diagram above, the OEM lead-acid battery is retained, and charged from the pre-existing 55 amp converter. It provides a backup in case the LFP battery system fails, and can be switched in by a battery selector switch. In normal circumstances the lead-acid battery isn't used. The pre-existing battery-disconnect switch is used to isolate the lead-acid battery from parasitic loads for storage. This necessitated moving some loads (jackstands) that used to be connected to unswitched-battery over to the sys12v power line.

Battery Selector Construction

The picture below shows the front side of the battery OEM battery compartment wall. The red 300 amp marine battery-selector switch was mounted below the OEM battery disconnect switch, which was rewired to be a lead-acid disconnect per the wiring diagram above.

Converter Fan

As noted, the 120A converter came equipped with a noisy 80mm fan that either ran at full speed or not at all. The fan was removed, and a 120mm fan with adapter mounted outside the converter. The existing fan on/off power signal was wired to a thermostat fan speed controller, whose temperature sensor was in contact with the main heat sink in the converter. The thermostat was programmed to drive the fan at low speed below 30C and full speed above 40C, with variable speed in between. The picture below shows the converter with thermostat prior to mounting the thermostat in a box. The temperature sensor exits the converter on the left, where white caulk secures it to the converter frame. Also seen is the fan size adapter (red), and the high-current quick-disconnect connector for the converter.

BMS Battery Parameters

The battery charge/discharge curves showed 10% (25Ah) Depth-of-discharge (DoD) cell voltage should be 3.275V, and 90% (250Ah) DoD  cell voltage should be 3.15V. Cell resistance changes these during high-current charge or discharge:

Cell resistance is measured by the BMS at 0.4 milli-ohms. This implies 58mV drop at 123A, measured was 80mV

BMS Default Parameter Changes

The following BMS parameters were changed from their defaults to configure cutoffs for the purchased batteries (specs taken from EVE spec sheet), and to set the 10% and 90% cutoffs.

Battery Components Physical Installation

This section describes the installation of the battery system components into the RV.

There was a sizeable empty space behind the basement wall, into which the battery system components were installed. A hinged door was cut into the basement wall to provide service access to battery system components.  This space (and the basement) is within the temperature-controlled envelope of the RV, thus the batteries live at the temperature inside the RV. This would enable charging them when outside temperature is below 0C, as long as the RV interior was heated (meets charging temperature requirement).  The picture below shows the basement wall with the access door cut into it.

The picture below shows the space behind the door, and general arrangement of components. The base plate of the battery box and converter are secured to their respective mounting surface with removable pins with lanyards (meets removability requirement).

Control Panel Physical Installation

The picture below shows the RV electrical control panel after adding battery system switches and LEDs. Blank fillers were removed from pre-existing panel cutouts, and new switches installed for battery system operation. A cable harness runs down behind the wall to the battery box and inverter, alongside other control panel wiring. White labels indicate the battery switch function.

The inverter remote on/off button was installed outside the control panel housing, next to the thermostat. The BMS LCD touch-screen was Velcro'd to the wall to the right of the inverter button, where it is readily inspected for SOC.

The picture below shows a closeup of the battery control switches.

Section 4

Testing Results

Batteries

The four batteries were purchased from a no-name vendor on Amazon who had US stock. They arrived with no manufacturer labeling - only a serial number, and no spec. I used the Eve battery spec, hoping all these batteries come from the same place. I tested capacity when charging and discharging at some tens of amps, and they consistently showed a capacity of 275 - 285Ah - close enough to the 280Ah spec that I consider they met spec. The BMS reports their internal resistance in the region of 0.4 milli-ohms. They didn't seem particularly well matched - two of the batteries seem to reach full charge a little before the other two. The BMS balanced them to within 20mV, but they still reached full charge or empty at different times. The labels on two were somewhat different than the other two.

BMS

General observations

The REC BMS is very well documented. The manual is clear and contains sufficient details to design it into a system. The only really confusing thing is there are two variants: Standard, and Victron-compatible. It's not clear how to tell which one you have, other than reading the model # out of the serial port, but they run different communication protocols and have different manuals and some of the pins work differently. It looks to me like two different firmware loads on the same HW. REC would have done well to make them externally identifiable.

You have to buy parts to use the BMS that are built into some of the Chinese BMS's, e.g. a bluetooth transceiver, and a PC app to change BMS parameters such as battery capacity, and they're not cheap. It is a quality product and met all expectations, but REC should consider building-in more features to the base product to offer better value.

The documentation on the precharge unit was OK, but didn't seem to state the input current requirements. This tripped me up. The precharge unit cannot be driven by the BMS optocoupler outputs, and this was unclear.

Tech support from REC in Slovenia was always timely and of excellent quality. This is supremely important when building your own battery system, and is not often found with asian vendors.

Inverter

The inverter was tested up to 1900 watts. It worked fine, and internal fans started as expected. Internal resistance of the batteries caused the 10% cutoff to trigger at about 23% during high-power operation. The inverter has soft start/stop. It consumes about 1.8 amps when idle (21W) but a part of this is the ATS relay coil draw, which is active when the inverter is on . 

The inverter handled overload gracefully: when the microwave and coffee pot were both turned on, it shut off, then started back up again without intervention, but with an alert beep. It is rated at 2kW continuous, 4kW surge, but the surge rating must be short duration, since the microwave and coffee pot tripped it within a few seconds.

Converter

The Powermax PM3-12V LK series 120 amp converter disappointed on three fronts:

Idle current into the converter was 0.07A (no fan), 0.22A (fan) at 120 VAC.

See the picture below for evidence of poor conversion efficiency. The clamp-on ammeter is measuring 18.1A into the converter (2081 volt-amps). Line voltage was 115VAC. The BMS LCD display is showing 90.1A into the battery at 13.42V (1209 watts). Thus 58% conversion efficiency from VA in to watts out. I tried but was unsuccessful measuring power-factor, so don't know for sure how much of the inefficiency is power factor and how much is heat loss in the converter. However, the generator reports 1550 - 1650 watts delivered during charging, so I suspect about 480 VA is lost in power factor, and 400 watts is lost in heat.

System

Battery connections

I saw one instance of strangely-high voltage across one cell, and the BMS reported the cell's impedance as 4 milli-ohms (vs 0.4 for the others). Measuring voltage under load between terminal and lug showed 3 milli-ohms on one terminal, even though the lug was bolted directly to the terminal. Twisting the lug on the terminal immediately eliminated the resistance, and BMS-reported battery impedance returned to normal. The lesson is that with such high current and low resistances in the system, even direct metal-metal contact is not necessarily 0 ohms, and very tiny resistances matter.

Surprise loads

Some loads offered surprises.

The microwave when powered on but not running consumes only milliamps of AC. But now and then I'd see an additional 6 amps drawn when the inverter was on. It turns out the microwave counter-lights are incandescent, and when on draw 60 watts!

The 42" LCD TV also consumed only milliamps when off, and only around 100 watts when running - less than I expected. The Amazon Firebox also consumed only milli-amps - not enough to matter.

Charging the iPad with its AC adapter added a load of roughly 5 amps at 12V - larger than expected. 2 amps were the inverter overhead, and 2.5 amps were drawn by the Apple AC adapter when charging the iPad. The iPad is supposed to charge at 12 watts, but total draw is 60 watts. Considering that the iPad runs Candy Crush and Jigsaw puzzles about 20 hours/day, this is significant. I need to install some USB charger outlets that run directly off 12V to make this more efficient.

Charging the laptop consumed about 60 additional watts. Overall, charging devices degrades run-time before a battery charge is needed.


Failure Log

The purpose of this section is to document failures in the battery system, so that if they happen again they can be recognized and fixed the same way, or eliminated in some way.

Future Improvements

The following improvements may be made to the battery system at some point in the future:

References