LiFePO4 Batteries: Discussion Thread for Those Using Them as House Banks

[CharlieJ]

OceanSeaSpray in #5383 makes an excellent case for voltage sensing and Coulomb counting to control the end of charge cycle. I agree.

The points I was making to #5380 are:
1. Do not top balance each cycle as the required 3.65VPC puts undue stress on the cells.
2. When a bank has reached its end of charge, however the operator defines and controls that point; stop charging.

[OceanSeaSpray]

Originally Posted by thomasow

The concerns of voltage-only charge termination is not restricted to LFP deployments, FLA/SLA technology also presents several compromises when using voltage-only as a termination method, chiefly chronic undercharging. Traditionally charging sources would get around this via extending time-based charging cycles – relying on the battery to accept slight overcharging with minimal damage (perhaps increased water consumption in the case of FLA – SLA needs to be careful here). But of course LFP technology cannot do this.

Truth is, for most all battery technologies monitoring the acceptance amps in combination with voltage is truly one of the best way to reliable charge a battery w/o damaging it. It is why 3 years ago I developed the open source Alternator Regulator that includes current monitoring/regulation capabilities.

Which brings up another interesting topic: There is the question of ‘What to do about Float’ with FLP batteries. Many point to ‘Disable everything’, which is certainly in line with the common EV usage model of charge then discharge, but that usage pattern is not how house batteries are used. We do not often have the pure Charge and then Discharge states. Example: supporting a washer/dryer while underway via an inverter; is nice to let the alternator carry that load vs. depleting the ‘fully charged’ battery before we anchor for the evening.

Recently have been looking at active regulation of the battery current. Rather then a voltage set point for ‘float’ regulate instead to a current set point. Specifically 0A. So when the FLP battery reaches its fully charged state, we then manage the alternator to a 0A net current into/out of the battery – while still being able to support existing house loads. Of course voltage is still monitored, but it typically is well under the ‘recommended float voltage’. Early tests seem promising, and I am curious of others thoughts on this approach.

The difference with SLA is that you can actually obtain a working compromise using voltage only because the chemistry is tolerant to overcharging through the shuttle reaction. It is far from perfect, but not completely inadequate.

Using LFP cells, you never get to a truly working compromise using SLA chargers because they invariably end up trying to charge into a full battery if they get a chance. The usual idea that "if I set my float voltage low, it will be ok" terminates the charge initially and dodges the first bullet, but short of cycling the installation actively, all these controllers sooner or later kick back in again and try to charge regardless. Just stop the engine and restart it for example.

I also built an alternator controller that can regulate current and terminates absorption based on voltage and residual current, precisely to charge my lithium cells. If I start the engine with a charged battery, the cycle is over in matter of seconds.
The funny things also is that - because it has access to the basic information that actually matters for charging - it doesn't need 47 obscure user-configured settings to try and approximate a solution.

With lithiums, regulating at 13.35V at the battery after the charge essentially means that it can't really discharge (but won't charge either unless quite low), so it is ultimately more flexible than controlling battery current at zero.
I use this strategy with solar power control to extend the discharge duration, but it is not the universal answer: it will never overcharge like all solar controllers, but under the wrong circumstances, it can keep a fully charged battery charged forever - just the wrong thing. It is currently performing very well because I always consume some energy overnight and solar power disappears at night.
Dropping down to 13.2V instead would mean no output until significant capacity has been depleted. It would increase cycling frequency and potentially waste a lot of good sunlight hours. In winter, it would become very detrimental.

This is why trying to dissociate charging from battery management in this type of application is ultimately hopeless. Energy availability, energy usage, individual cell voltages, charge acceptance, temperature and SoC are all contributors into the decision of whether we should charge, hold or discharge the bank at any one time.
It is far more demanding than simple EV-type service.

The outcome of all this is that, after two years, I am now about to build a completely new and much more powerful BMS solution. It will be a couple of weeks before all the parts get here and the fun starts.

[thomasow]

Originally Posted by marujo.sortudo

Regulating to 0A is an interesting idea. I imagine it would work better if the battery was only 95% charged. Harder to monitor charge acceptance on a full battery, perhaps? I do a similar thing by floating to 13.3V with our LiFePO4 bank. Not quite the same, but what my regulator is capable of and we rarely do long runs anyway.

It seems there are perhaps 3 concepts floating around here that might be fun to explore deeper: “What to do with Float, or How to Park, When to Park, and Where to Park.”

How to Park: The idea of parking a battery can perhaps be restated as a state where only the amount of energy needed to account for self-discharge is being placed into the battery. (very low value for Li based technologies).

Voltage set points have traditionally been one approach, but in an indirect way: By characterizing a typical battery pack a voltage is selected – ala the cited: 13.3v It will be interesting as more experience with Li based technologies is gained how definitive a fixed voltage will be (recalling that FLA/SLA battery can to some extend deal with getting it wrong). To see what impact imbalance in cells, cell drift over time, cycles of the battery, temperature and perhaps other influences has on a predetermined voltage setpoint’s long term stability. 13.35v might be a good value, but the bottom line is: If energy flowing into a battery exceeds any self-discharge its power level (SOC) will continue to increase.... Likewise, even a lower energy outflow will drain the battery. And of course, this is not just about the battery, but the whole DC system - and being able to support ongoing loads even when the battery is 'parked'.

When to park: There appears to be a lot of benefits with keeping a guard-band at both the top and bottom %SOC ends with Li technologies. Less stress to the battery resulting in longer life. 5% - 95% appear to be reasonable extreme limits? As had been pointed out, there is also evidence to the benefit of using a lower charge voltage combined with a properly monitored and terminated acceptance phase. By not needing to drive the battery pack to a higher voltage levels stress is reduced to the battery, as well as increased cell overvoltage margin in the case of an unbalanced battery pack. (Is interesting to note the charge profiles used in true high-reliability deployment of Li based batteries, ala NASA)

Where to park: Now here is something new! No surprise that many boats spend the VAST majority of their time at the dock, patently awaiting their masters. And during this time the batteries are in effect ‘in storage’. FLA/SLA batteries like to be kept near fully charged. But Li based technologies do not like long term high SOC storage. So, are there really two ‘parking places’ that should be considered?
95% SOC (or such) when operationally – we are underway and looking to anchor for the night.
50% SOC (or such) when coming back to port after a long weekend, preparing to leave the boat unused for a period of time.

Is 50% always the right value? What about considerations such as reserves for bilge pumps, or?? How to determine when to select Storage vs. Operational parking. A human pressed button? GPS association with the dock? ??? Lots of questions here ---

[thomasow]

Originally Posted by OceanSeaSpray

I use this strategy (13.35v / 3.34vpc) with solar power control to extend the discharge duration, but it is not the universal answer: it will never overcharge like all solar controllers, but under the wrong circumstances, it can keep a fully charged battery charged forever - just the wrong thing. It is currently performing very well because I always consume some energy overnight and solar power disappears at night.
Dropping down to 13.2V (3.3vpc) instead would mean no output until significant capacity has been depleted. It would increase cycling frequency and potentially waste a lot of good sunlight hours. In winter, it would become very detrimental.

Are you saying you see problems using the 3.34vpc float approach, and still caution is needed? Not sure I am completely following the issue here.

Originally Posted by OceanSeaSpray

…it doesn't need 47 obscure user-configured settings to try and approximate a solution.

Ha, I kind of refer this as ‘refining the guess’

Will be interesting to see what all you have planned for your BMS.

[OceanSeaSpray]

Originally Posted by thomasow

Are you saying you see problems using the 3.34vpc float approach, and still caution is needed? Not sure I am completely following the issue here.

Will be interesting to see what all you have planned for your BMS.

I don't like using the term "floating" because it refers to maintaining the voltage in a range by charging intermittently after a LA has been fully absorbed and this makes no sense with lithium, because once charged we don't want to go back to voltages still capable of charging unless the battery has been depleted...

Yes, I see major problems with holding at 3.35V/cell indefinitely in the sense that it won't allow a fully charged bank to discharge again. Now, should the battery not be fully charged for a start, 3.35V isn't going to achieve that, because it is just the stabilised OCV of a full cell and not sufficient to charge, i.e. shift electrons, past a point.

From a battery point of view, leaving it connected to a supply of a given voltage means nothing if this supply is no higher than where the battery would be on its own (and there is no ripple on the supply etc). So pick the OCV of a battery at 50%, slightly less than 13.2V, and hold there, you are maintaining the SoC more less indefinitely and can draw power off the supply. It is no longer cycling and essentially in storage. It is completely transparent to the battery.

As you point out, when keeping a SLA fully charged, the desire to achieve maximum reserve capacity and maximum battery life are fully compatible and coincide perfectly.
This breaks badly with LFPs. The answer of the commercials to this has been to go for maximum reserve capacity regardless. By the time it has
ruined the battery, it is well and truly out of its short warranty and you don't make money by selling packs that last forever: no problem there. One of them is selling little "lithium" solar controllers that just hold 14.2V indefinitely. That is a simple strategy alright.

Here we try doing better and succeed up to a point, but it is a fragile equilibrium that requires a human-BMS around. It is a technology to play with, not a true robust solution.

I want to move the whole problem into the firmware and actively manage the operation of the battery. No more game of settings. The problem is very dynamic and it requires dynamic responses. Obviously, I can't see into the future, so if you leave the boat in storage for a time and then come back, you won't find maximum reserve capacity available, but even half a bank should be enough to live for more than a day on a sensibly designed system.
At the moment, pushing reset on the BMS forces a new charge cycle to begin - which we never really do. The new code will detect changes in demand and adapt, so it will be more proactive.

[Dockhead]

Originally Posted by OceanSeaSpray

I don't like using the term "floating" because it refers to maintaining the voltage in a range by charging intermittently after a LA has been fully absorbed and this makes no sense with lithium, because once charged we don't want to go back to voltages still capable of charging unless the battery has been depleted...

Yes, I see major problems with holding at 3.35V/cell indefinitely in the sense that it won't allow a fully charged bank to discharge again. Now, should the battery not be fully charged for a start, 3.35V isn't going to achieve that, because it is just the stabilised OCV of a full cell and not sufficient to charge, i.e. shift electrons, past a point.

From a battery point of view, leaving it connected to a supply of a given voltage means nothing if this supply is no higher than where the battery would be on its own (and there is no ripple on the supply etc). So pick the OCV of a battery at 50%, slightly less than 13.2V, and hold there, you are maintaining the SoC more less indefinitely and can draw power off the supply. It is no longer cycling and essentially in storage. It is completely transparent to the battery.

As you point out, when keeping a SLA fully charged, the desire to achieve maximum reserve capacity and maximum battery life are fully compatible and coincide perfectly.
This breaks badly with LFPs. The answer of the commercials to this has been to go for maximum reserve capacity regardless. By the time it has
ruined the battery, it is well and truly out of its short warranty and you don't make money by selling packs that last forever: no problem there. One of them is selling little "lithium" solar controllers that just hold 14.2V indefinitely. That is a simple strategy alright.

Here we try doing better and succeed up to a point, but it is a fragile equilibrium that requires a human-BMS around. It is a technology to play with, not a true robust solution.

I want to move the whole problem into the firmware and actively manage the operation of the battery. No more game of settings. The problem is very dynamic and it requires dynamic responses. Obviously, I can't see into the future, so if you leave the boat in storage for a time and then come back, you won't find maximum reserve capacity available, but even half a bank should be enough to live for more than a day on a sensibly designed system.
At the moment, pushing reset on the BMS forces a new charge cycle to begin - which we never really do. The new code will detect changes in demand and adapt, so it will be more proactive.

When do you expect to bring these devices to the market?

I'm still watching with interest and waiting for this technology to mature. I've just replaced my lead acid battery bank again but sure hope that the next set will be LFP.

[Maine Sail]

Originally Posted by OceanSeaSpray

I don't like using the term "floating" because it refers to maintaining the voltage in a range by charging intermittently after a LA has been fully absorbed and this makes no sense with lithium, because once charged we don't want to go back to voltages still capable of charging unless the battery has been depleted...

Yes, I see major problems with holding at 3.35V/cell indefinitely in the sense that it won't allow a fully charged bank to discharge again. Now, should the battery not be fully charged for a start, 3.35V isn't going to achieve that, because it is just the stabilised OCV of a full cell and not sufficient to charge, i.e. shift electrons, past a point.

From a battery point of view, leaving it connected to a supply of a given voltage means nothing if this supply is no higher than where the battery would be on its own (and there is no ripple on the supply etc). So pick the OCV of a battery at 50%, slightly less than 13.2V, and hold there, you are maintaining the SoC more less indefinitely and can draw power off the supply. It is no longer cycling and essentially in storage. It is completely transparent to the battery.

As you point out, when keeping a SLA fully charged, the desire to achieve maximum reserve capacity and maximum battery life are fully compatible and coincide perfectly.
This breaks badly with LFPs. The answer of the commercials to this has been to go for maximum reserve capacity regardless. By the time it has
ruined the battery, it is well and truly out of its short warranty and you don't make money by selling packs that last forever: no problem there. One of them is selling little "lithium" solar controllers that just hold 14.2V indefinitely. That is a simple strategy alright.

Here we try doing better and succeed up to a point, but it is a fragile equilibrium that requires a human-BMS around. It is a technology to play with, not a true robust solution.

I want to move the whole problem into the firmware and actively manage the operation of the battery. No more game of settings. The problem is very dynamic and it requires dynamic responses. Obviously, I can't see into the future, so if you leave the boat in storage for a time and then come back, you won't find maximum reserve capacity available, but even half a bank should be enough to live for more than a day on a sensibly designed system.
At the moment, pushing reset on the BMS forces a new charge cycle to begin - which we never really do. The new code will detect changes in demand and adapt, so it will be more proactive.

This ^^^^^ Bingo !!!

Currently a multi-charge source off-grid LFP installation ultimately requires both human intervention and a BMS. This is why all our charge sources can be manually turned off, with the flip of a switch, including the alternator, solar and shore charger.

[transmitterdan]

The way the laptop designers are doing this might be instructive. They can't deduce the future needs based on past usage so they provide a user interface to the charging function. Basically there are two settings.

One user setting is for the case where there is sufficient charging resource to keep the battery from depleting (ala dock queen). On that setting, the BMS tries to keep the Li bank at some mid range % SOC (near 50%). They call this the "battery life extension" mode or something equally silly.

The other user setting is for the case where the full battery capacity is desired (ala on the hook liveaboard). In this case, the BMS charges up to 100% SOC any time there is a charging source available. The idea is that in this mode the charging source(s) are intermittent enough that the bank never sits for a long time at 100% SOC.

They haven't gotten rid of the "human BMS" but it is pretty easy for "dummies" to choose the correct setting. I suppose it would be easy enough to automate the switch between modes based on recent usage.

[exMaggieDrum]

One issue I need to think through (i think some others have and I need to review their solutions and discussions) is the small parasitic draws that you can have even after "parking" a battery. E.g. battery monitors, relays that always use power, etc. that are connected "directly to the battery". That is, they are not connected downstream of the main load disconnect should the batteries get too low. To avoid this completely you would have to put the battery sensors and these other "always on" circuits downstream of the main battery disconnect used to park the batteries. Or you could take the battery leads off.

This shouldn't be a problem for alternator regulators where the B+ is turned off at park, nor the solar chargers when turned off (I assume). The BMS would have to be turned off and relays downstream of it would have to be off. I suspect many installation would have some battery monitors directly on the batteries and some of them are always on unless you put a switched contact on the sensor feed.

The result could be a completely depleted battery after an extended park. Bilge pumps are a separate issue. I have electronic bilge switches currently and I need to see if they have a parasitic draw when operational. Otherwise I will have to go back to mechanical switches or ensure I have a recharge on the battery before it gets too low. Running the bilge pump is a tradeoff - lose the boat or lose the battery (which probably will result in losing both anyway at some point unless action is taken before the boat sinks).

[OceanSeaSpray]

I think "parking" is the wrong concept, unless you have zero charging ability and therefore can't afford any consumption. Even a tiny amount of solar can keep a system healthy and fully functional for years without any activity: it will only get used in the rarest of occurrences.
If a bank goes down too far, the BMS opens the load bus, giving it a chance to recharge. If it keeps going down, there could be no charging source or something draining power from the charge bus, so it also isolates the charge bus, cuts most of its own power off (which is very, very small) and goes into a deep sleep. At this point, the consumption is so low that physically disconnecting the module wouldn't make any difference.
I can handle simple contactors, but I personally think that latching relays are a much more sensible option for most installations and this is what I have been using.

I am thinking that "application templates" could be the way to go, partly because they would allow configuring the inputs/outputs for typical installations (for charge and load management), but the idea is to make the solution essentially self-adjusting, i.e. no intervention required. A small user interface could allow requesting maximum reserve capacity or standby in advance.

If you live in an area with high sunlight and have low consumption, there is no reason whatsoever to charge to a high SoC, except occasionally to prevent memory effects from setting in. Power is always easy to find. If winter comes, days get short and grey, you want to charge to full if you get a chance and capture as much power as possible. If you start an engine or genset, you might want to capture the power while it is there.

My first BMS module has worked quite nicely, but there were challenges with making it widely available. First the firmware needed to be proven in the field, and this can take time. It would be ok now, but at the start the idea of having to recall the units to reprogram them if an issue needed to be addressed wasn't great, so I didn't let them go too far initially.
Next, it isn't as good as I want it to be now, because some functionality was excluded at the time it was designed and I am also nearly out of program memory. I can't improve it further without building completely new hardware, so I quickly decided not to accumulate too much of a legacy in this direction.

This time I decided to add a USB port to allow configuration with a small utility and data logging. I also want the firmware to be user-flashable via USB, so we can add functionality and improvements over time. I added a display port as well to take full advantage of SoC tracking.

I need to build the prototype and validate the hardware now. Once I can read and control everything on the board, I will get the USB interface working and migrate some of the existing BMS/charge control code over to get a first working unit.
Making it user-flashable will require a bootloader and a Windows configuration utility, and documentation will need to be written, so my time on this side of Christmas is pretty well taken care of, but the idea is making some early units available as soon as possible and then keep working on the more advanced functionality.
Other people run their installations differently and feedback is always hugely interesting.

[missourisailor]

Originally Posted by OceanSeaSpray

....
I can handle simple contactors, but I personally think that latching relays are a much more sensible option for most installations and this is what I have been using...

OceanSeaSpray,
What are your thoughts about using Blue Sea's 7700 RBS (remote battery switch) https://www.bluesea.com/products/770..._-_12V_DC_500A for the main contactors?

They are latching and have zero current draw once switched.
The problem with them is that they take a pulse to switch states.
Will the RBS work with your new BMS?

[rvlandlubber]

The use of latching relays for load/ battery disconnect function seems to be preferred over conventional normally open relays with a lot of the systems mentioned here.



From the fail-safe design viewpoint these devices can be problematic.



I would be interested to know what the thinking is when deciding to use them.



Is the low power consumption the most important factor?



How are users managing the possible situation where the latching relay is faulty and won't change state?



Thanks in advance.


Sent from my iPad using Cruisers Sailing Forum

[missourisailor]

I'll wait to see what others say...But I'm using them for several reasons.
1) No current draw once switched.
2) They can be manually switched when all else fails. Once manually tripped, they must be manually reset before normal operation.

[OceanSeaSpray]

Originally Posted by missourisailor

OceanSeaSpray,
What are your thoughts about using Blue Sea's 7700 RBS (remote battery switch) https://www.bluesea.com/products/770..._-_12V_DC_500A for the main contactors?

They are latching and have zero current draw once switched.
The problem with them is that they take a pulse to switch states.
Will the RBS work with your new BMS?

All latching relays need to be pulsed indeed and it is what I do with the Tyco. The BlueSea 7700 needs to be pulsed on the positive side with a lot of current. At the moment, it wouldn't be compatible, because I switch the negative side, which is a lot more efficient from an electronic point of view.

Changing the design around could address that, but the drivers for these outputs include monitored redundant transistors to be fail-safe and there is a bit in it.

I prefer the Tyco unit, which is industrial/automotive instead of marine-branded, can be wired on the high or low side, and also costs less than half the price - unsurprisingly.
The BlueSea is also a "violent" device to control, with a much higher inrush. I don't think it was quite intended for doing this kind of thing, i.e. electronic control.

Still, interesting query, thanks.

[SV THIRD DAY]

One thing to keep in mind here is that we are talking about a battery and control system for a Cruising Boat, not a remote unmanned cell tower. You can tie yourself up in knots with the assumption that your bank and control system needs to function completely without human interaction, monitoring, and oversight. With any system, if you think you will be able to "set it and forget it" like the Ronco Rotisserie Chicken Cooker, then you could have a surprise coming....not just with your battery bank/control system but with life aboard a cruising boat in general.