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

[Jd1]

For the time being the lithium pack will be a completely separate setup and all the lead acid batteries will stay and be used as normal. The lithium will be used for longer trips or when power demands are higher. As it sits, I don't expect to be operational for 6 months or so by the time the boat has been modified (shelf under Vberth epoxied in), an enclosure has been built, a separate box for the electronics has been built and so on.

[evm1024]

Originally Posted by Ex-Calif

What would be wrong with a DC to DC charger for Aux batteries.

I could see a LA for windlass and for starter battery and isolating them from the LiFePO system seems like a good idea.

In fact an AC charger up front for a LA placed up front makes sense as the wire run can be lower gauge.


SNIP

I have been toying with the idea of running charging bus on the boat at fixed voltage a volt or few greater than the battery charging voltage. Then using a uP controlled synchronous switching power supply to do the charging. The advantage would be having a much less complex power generation subsystem (the alt, solar etc gives a fixed 16 or 24 volts) that is distributed where needed. with the switchers doing the isolation of batteries and taking care of any charge curves desired.

Getting a 50 amp switcher is easy, it gets a bit more complex for a 100+ amp switcher.

I do like the idea of using a 24 volt alternator being off the shelf and haveing a smaller conductor to the windless. Maybe is is time for crowd funding...

I do run 120 VAC to the windlass battery along with 12 volts on a #6 cable for the echo charge.

Regards

[OceanSeaSpray]

Originally Posted by evm1024

I have been toying with the idea of running charging bus on the boat at fixed voltage a volt or few greater than the battery charging voltage. Then using a uP controlled synchronous switching power supply to do the charging. The advantage would be having a much less complex power generation subsystem (the alt, solar etc gives a fixed 16 or 24 volts) that is distributed where needed. with the switchers doing the isolation of batteries and taking care of any charge curves desired.

Getting a 50 amp switcher is easy, it gets a bit more complex for a 100+ amp switcher.

I do like the idea of using a 24 volt alternator being off the shelf and haveing a smaller conductor to the windless. Maybe is is time for crowd funding...

I do run 120 VAC to the windlass battery along with 12 volts on a #6 cable for the echo charge.

Regards

Remember that the voltages of all the power sources found on boats are essentially held down by the loads and without load they would stabilise somewhere well above useful voltages, i.e. 21VDC for many solar panels etc.
There would be no benefit if you needed to somehow pre-regulate each of those down to 16VDC first and then regulate lower again to battery voltage using some big PWM-type unit.

The alternator is always going to remain a special case, because it is not regulated by switching its output like most other sources. Many wind generators can't be open-circuited like solar panels, same for tow generators. Otherwise you could conceivably feed all your sources into a single large PWM regulator as you say and be done with it.

As soon as you start hooking AC chargers to inverters etc, you have lost the plot in my opinion, because of all the conversion losses along the way taking the voltage up and down and extra gear. Keep it simple.
To me, a windlass seems to be a prime candidate to run off the house lithium bank in spite of the longer heavy cabling and keeping the engine starting battery as lead-acid is easier and makes more sense:

• Independent power source to start the biggest power generator on board if ever needed
• This battery is normally always full, there should be zero drain on it other than crank and recharge, so no cycling, leading to very long life
• It can share the lower alternator charging voltage of the lithium bank for "bulk" charge if ever needed (split over diodes and compensate for the voltage drop), and have its own solar regulator to bring it up to full and float it there

It is by far the simplest installation. As soon as you introduce cycling lead-acid cells in the system, you create a problem that is much more difficult to resolve in a clean and tidy way, because then you need to supply a lot of charging current with a completely different voltage profile.

[evm1024]

Thanks for the thoughtful response. For my first pass I am using an AGM start battery to go with the 700 AH house LiFePO4 battery. Much like you suggest.

I am looking to the future for the "fixed voltage" distribution system. Some back of the napkin calculation shows the losses on voltage regulation are offset by the charge-in/charge-out ratio that LiFePO4 gives over FLA.

These are real and interesting problems to resolve. KISS is an interesting goal. Something like the Balmar mc-614 us a KISS solution that takes some very in depth understanding of alternators and battery chemistry and turne it into a black box if you take my meaning.

We are lacking a LiFePO4 black box.

In years gone by I've had a number of interesting things to work on. Everything from automating sawmills to programming SCADA/RODS systems on power grids (BPA anyone?) It is interesting to me that some of the stuff I once worked on applies to my boat is I shift my thinking.

In any case I'll be installing the lithium cells in my boat next weekend. I've been going back and forth on the design for the last week. Some simple things and some complex.

REgards

[OceanSeaSpray]

Originally Posted by evm1024

Thanks for the thoughtful response. For my first pass I am using an AGM start battery to go with the 700 AH house LiFePO4 battery. Much like you suggest.

I am looking to the future for the "fixed voltage" distribution system. Some back of the napkin calculation shows the losses on voltage regulation are offset by the charge-in/charge-out ratio that LiFePO4 gives over FLA.

These are real and interesting problems to resolve. KISS is an interesting goal. Something like the Balmar mc-614 us a KISS solution that takes some very in depth understanding of alternators and battery chemistry and turne it into a black box if you take my meaning.

We are lacking a LiFePO4 black box.
...
REgards

You can try and think some more about regulating a charging bus, but there are technical difficulties. I wouldn't want a system only as efficient as FLA, I want the cycling efficiency of the lithium chemistry.

A LiFePO4 black box is something I have given much thought in recent months. I think a solution is best engineered as a few black boxes, rather than one. It would be too much of a central point of failure and some flexibility is needed around differences between boats and systems.

First, there is no simple "fit and forget" decent BMS available for marine applications, so this is the starting point. The HPBMS is an EV product, it can't handle bistable relays, it is not fail-safe... possible, but not suitable.
I have a board design here that can self-test, it has redundant hardware so it won't damage the bank if there is a failure, it can handle various contactor types, it doesn't need to hold power to protect like the HPBMS, it can automatically control cell rebalancing when needed without going over-voltage etc. Manufacturing the first few is just around the corner and the firmware is not too far away either.

Alternator control could be another building block down the track. The regulator needs to be able to do current limiting and stay within the thermal envelope. The problem with alternator solutions is that you would really want to package the regulator and alternator so you can guarantee the outcome, but then it means also replacing the alternator and it becomes expensive and more awkward.
I don't like aftermarket external alternator regulators, because most of the time they are installed without any (or any correct) alternator temperature sensing and this is the best way of burning out the unit. And what is correct alternator temperature sensing? The hot spot is in the windings. It is for the factory to determine.
There is hardly a limit to how much power an alternator will try to put out if you keep going up in revs.
One pathway I have been thinking about is building a module to control the voltage sensing input of a stock standard alternator. Easy. It would do current limiting and voltage control for lithium and keep the factory protection and much higher reliability of the built-in regulator.

Another component would be a small boost converter/charger to properly charge the FLA starting battery from the lithium bank. Just regulate everything for the lithium and don't worry about the lead-acid.

So I think it would possible to come up with a standard diagram and a few dedicated building blocks to make a really simple and perfect system. As long people do keep it simple and as designed!

Regards,

Eric

[BillyDoc]

I've read this thread, but I don't remember anyone experimenting with LiFePO4 charging from a PWM source. I know it works well with LA batteries, but the basic nature of PWM is a "too high" voltage during the "on" part of the cycle. Is this voltage instantly dropped to a "proper" voltage by the battery load? Are there any turn on or off effects? Has anyone done this experiment?

I'm hoping that the PWM approach works as well with LiFePO4 as it does with LA, because it could simplify things quite a bit while retaining excellent efficiency.

Love this thread, by the way, and many thanks to all who have contributed!

Bill

[OceanSeaSpray]

Originally Posted by BillyDoc

I've read this thread, but I don't remember anyone experimenting with LiFePO4 charging from a PWM source. I know it works well with LA batteries, but the basic nature of PWM is a "too high" voltage during the "on" part of the cycle. Is this voltage instantly dropped to a "proper" voltage by the battery load? Are there any turn on or off effects? Has anyone done this experiment?

I'm hoping that the PWM approach works as well with LiFePO4 as it does with LA, because it could simplify things quite a bit while retaining excellent efficiency.

Love this thread, by the way, and many thanks to all who have contributed!

Bill

Bill,

It shouldn't really be of concern with lithiums, because you want to connect the source continuously until the target voltage is hit and then disconnect. There shouldn't be any PWM because it is not worth bothering with the CV absorption tail at modest charging rates. There is no need for voltage "regulation" as such. Lithiums are really easy to charge.

There is a fair bit of inertia in the system with a battery bank hooked up, but PWM does cause a ripple.
I built PWM controllers in the past and adding a capacitor on the output of the PWM worked as a low-pass filter and averaged the current quite a lot, but it also made the PWM switch work harder due to the inrush current into the capacitor each time. A good place for a capacitor is in fact at the battery terminals.

Regards,

Eric

[evm1024]

After letting my 700ah bank sit in parallel for a few days as the end of top balancing I wired them up in series. As a first test I wired up a shunt and a 700 w inverter. The inverter was connected to an oil filled type space heater set to low (600w). The inverter did not have the oomph to do more.

This photo is after about an hour of driving the heater. It shows the pack voltage at 13.14 volts (3.285 vpc) and 63 amps draw. 63 ah removed plus or minus.

After removing the load the pack voltage bounced up to 13.28 (3.32 vpc) with all cells within 1 mV of each other. That measurement was within an hour of removing the load. I'll have to measure tonight to get the resting voltage 24 hours later.

Clearly the top balance placed the pack in the upper knee. You could watch the pack voltage drop at a constant rate then the drop tapered off and held steady at 13.14 for the remainder of the test.

I should record the pack and resting voltages at various discharges to generate a DOD curve. Could take a lot of time....

[Jd1]

I discovered a rather interesting an alarming problem in my potential lithium battery circuit. Looking at the Tyco LEV200 relay, the spec sheet gives a 1A holding current. Needing a HV and LV cutoff as well as a combined emergency cutoff, that is 3A holding current and if my math is correct that works out to 72 Ahr per day (!!!!) which is nuts! The EV200 has a much lower holding current at 0.13A or about 9 Ahr which is still significant although acceptable.
Being able to use bistable relays would be nice but I am concerned about failure mode on something that requires a signal to shut down a circuit compared to just dropping out if the control circuit looses power for some reason.

[simat]

I have made available the design information and software I have used to make a flexible battery monitoring system using the Beaglebone Linux single board computer. The project can be found here https://github.com/simat/BatteryMonitor.git

Rather than design a purpose built board I have gone for the very capable Beaglebone and added a simple battery interface comprising a resistor divider network and a few multichannel 16 bit A/D chips to convert the battery cell voltages to a digital signal, this gives a resolution of around 1mV for a 24 volt battery bank. Battery current can also be monitored with the addition of a shunt and two small resistors.

Currently the program just logs to files, but it would be easy to extend this to generating web-pages for display over the internet. It is also a fairly trivial exercise to have the Beagleboard produce alarms to drive safety relays or maybe even send texts to your mobile phone.

My primary use of this device is to monitor my LiFePO4 battery bank and provide me with information via the internet on the battery state when I am away from home, and also log the battery data over a long period of time. It is more than likely that this project will develop into a full power management system.

I have had this up and running for a little while and found it to be very useful, hopefully it will be useful to others.

Here is the running summary file Attachment 88284, and the daily summary file Attachment 88285 produced from the program, the circuit diagram Attachment 88282 of what I built and some photos

[BillyDoc]

Thanks for the reply OceanSeaSpray, and of course you are right. I was thinking, however, of using PWM to lower the current draw from sources like unmodified alternators to a level that would not be in danger of overheating them. High switching currents would be involved, but MOSFET transistors are easy to parallel to just about any desired current capability, and could also serve as a "cutout" relay with an extremely low "on" current draw that would fail in the off mode if the failure was the gate signal dropping off. I also think an array of MOSFETs is much more reliable than a mechanical relay.

Bill

[OceanSeaSpray]

Originally Posted by BillyDoc

Thanks for the reply OceanSeaSpray, and of course you are right. I was thinking, however, of using PWM to lower the current draw from sources like unmodified alternators to a level that would not be in danger of overheating them. High switching currents would be involved, but MOSFET transistors are easy to parallel to just about any desired current capability, and could also serve as a "cutout" relay with an extremely low "on" current draw that would fail in the off mode if the failure was the gate signal dropping off. I also think an array of MOSFETs is much more reliable than a mechanical relay.

Bill

Bill,

In the case of an alternator I think it would be a bit risky, because the voltage would spike up each time the "switch" turns off and you would end up nuking something in fairly short order. Alternators really need to be controlled on the excitation side and the output must remain solidly wired to a load that holds the voltage down.
MOSFETs are incredible devices indeed, but they are easily damaged by voltage transients, so switching circuits must be well protected. When they fail, they almost invariably end up conducting through. Heating and voltage losses usually limit the practical current per device to around 10-20A, so paralleling many on a heatsink is the way to go for high current applications.

A good high-current, fail-safe solid-state relay to go along with the BMS is on my list of projects indeed, but when you do the numbers it doesn't end up very cheap and making it fail-safe can double the cost...

Eric

[BillyDoc]

Originally Posted by OceanSeaSpray

Bill,

In the case of an alternator I think it would be a bit risky, because the voltage would spike up each time the "switch" turns off and you would end up nuking something in fairly short order. Alternators really need to be controlled on the excitation side and the output must remain solidly wired to a load that holds the voltage down.
MOSFETs are incredible devices indeed, but they are easily damaged by voltage transients, so switching circuits must be well protected. When they fail, they almost invariably end up conducting through. Heating and voltage losses usually limit the practical current per device to around 10-20A, so paralleling many on a heatsink is the way to go for high current applications.

A good high-current, fail-safe solid-state relay to go along with the BMS is on my list of projects indeed, but when you do the numbers it doesn't end up very cheap and making it fail-safe can double the cost...

Eric

I agree it isn't easy, and voltage spikes from dropping the load suddenly would certainly be a problem. Many people have alternators, however, that do not lend themselves easily to controlling the field voltage, so I think it's worthwhile to try to come up with an approach that can effectively do this from the load side, if it's possible. That's what I'm exploring here, and I thank you for your excellent help!

If, for example, an array MOSFET switch was between an alternator and a LiFePO4 battery there could also be a capacitor and a reverse biased diode between the alternator and the switch (across the lines). If the switching frequency is high enough, and the capacitor big enough, the capacitor should integrate the "off" pulses enough so the alternator only "sees" a load with a slight ripple. This should allow limiting the amps to a level the alternator can sustain, and of course the internal regulator will still provide thermal, etc. protection. The problem then is to turn the load off gracefully, and it seems to me that this could be done by simply ramping the PWM off over a few milliseconds, thus giving the alternator's internal regulator time to shut itself down.

The data sheet for IRLB3036PbF MOSFETS (for example) rates them with an on resistance of 0.0019 ohms and a package heat dissipation of over 300 watts (if you believe it). Digikey will sell you ten for $26.38 and they have logic-level inputs. An entire array of these could be controlled with a single optical coupler which is driven by an Arduino, or the like. Ten of these MOSFETs in an array would only need to dissipate 1.9 watts controlling 100 amps, if I didn't lose a decimal point somewhere. This would surely require a heat sink, but nothing too fancy.

Does this seem feasible? Worth doing?

Bill

[OceanSeaSpray]

Originally Posted by BillyDoc

I agree it isn't easy, and voltage spikes from dropping the load suddenly would certainly be a problem. Many people have alternators, however, that do not lend themselves easily to controlling the field voltage, so I think it's worthwhile to try to come up with an approach that can effectively do this from the load side, if it's possible. That's what I'm exploring here, and I thank you for your excellent help!

If, for example, an array MOSFET switch was between an alternator and a LiFePO4 battery there could also be a capacitor and a reverse biased diode between the alternator and the switch (across the lines). If the switching frequency is high enough, and the capacitor big enough, the capacitor should integrate the "off" pulses enough so the alternator only "sees" a load with a slight ripple. This should allow limiting the amps to a level the alternator can sustain, and of course the internal regulator will still provide thermal, etc. protection. The problem then is to turn the load off gracefully, and it seems to me that this could be done by simply ramping the PWM off over a few milliseconds, thus giving the alternator's internal regulator time to shut itself down.

The data sheet for IRLB3036PbF MOSFETS (for example) rates them with an on resistance of 0.0019 ohms and a package heat dissipation of over 300 watts (if you believe it). Digikey will sell you ten for $26.38 and they have logic-level inputs. An entire array of these could be controlled with a single optical coupler which is driven by an Arduino, or the like. Ten of these MOSFETs in an array would only need to dissipate 1.9 watts controlling 100 amps, if I didn't lose a decimal point somewhere. This would surely require a heat sink, but nothing too fancy.

Does this seem feasible? Worth doing?

Bill

Bill,

Remember that the power you store in the capacitor(s) needs to go somewhere eventually and at 100A or so, it wouldn't take long to charge even a huge capacitor array.
Also - assuming that some kind of switch mode circuit could directly throttle the output - imagine the damage if it suddenly stopped switching. It would be highly stressed.
You can download Tina-TI or LTSpice free and try simulating a few circuits. I think you would be horrified by the wave forms.

10 paralleled MOSFETs like the one you mention could certainly handle very high currents. Carrying the current to and from them requires careful design and construction however. Also remember that in order to switch on the high side with N-channel FETs, you need to produce around 10V more than the battery voltage to drive their gates and if you start switching fast, then you need quite a bit of current at that voltage.

Now, thinking about it differently, your best bet could be using a linear circuit instead, forget about switching. If you can drop a few tenths of a volt in a controlled way between the alternator output and the battery, then you can effectively force the alternator output voltage up to its regulation limit: say it regulates at 14.5V and you drop 0.5V to keep LiFePO4s happy with 100A charging current, you need to get rid of 50W in heat. That is achievable with transistors on a good heatsink and maybe a fan.
That would be easier to do and a lot more reliable.

Best regards,

Eric

[OceanSeaSpray]

Originally Posted by simat

I have made available the design information and software I have used to make a flexible battery monitoring system using the Beaglebone Linux single board computer. The project can be found here https://github.com/simat/BatteryMonitor.git

Rather than design a purpose built board I have gone for the very capable Beaglebone and added a simple battery interface comprising a resistor divider network and a few multichannel 16 bit A/D chips...

Neat project! Actually, I had also kicked around the idea of using some generic CPU board last year.
I moved away from it because of the high power consumption and also because there is in fact quite a lot of hardware to add around it to get a truly hardened system with all the input channels and power outputs.
Since I was going to have to make a board with all that on it anyway and ultra-low power consumption was essential, integrating the processor was the way to go and only a very small step further.