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

[witzgall]

#1, this is misleading and potentially dangerous. What happens when the alternator fails, and puts out over 15v?

#2 I cannot follow what you are trying to say.

Chris

Originally Posted by CatNewBee

1.) THE ALTERNATOR.
is usualy designed to charge a small LA start battery.
All the issius reported from burned alternators stem from this design goal an also apply, if you connect a big empty 1000AH LA bank to them. They are not made for house banks in the first place. Another issue is, EVERY charge source should be self-protective to high currents (short circuits), reverse polarity, overheat etc. If this is implemented, no worries with LFP FOR THE alternator, they are set around 14.2V, far below dangerous voltages for LFP too.

2.) Self-contained BMS
I simply not trust 1 fits all strategies. Either you limit the ability of the cells by to weak electronics (very common) or you spec them to handle 3C, than the electronics will be bigger and more expensive than the cells used. Its waste of money, to have 10 BMS, 10 Power regulators etc. Also in serial installation you may have to deal with the double voltage in case of a failure - opposite to a proper 24V system with a single BMS.

.

[Hoopla]

Originally Posted by nebster

I took it to mean that a series string of cells that are not bottom-balanced will start to diverge at the bottom of the discharge. As in, the voltages will start to run all over, like cats?

I can't imagine a small 4s pack with cells so out of whack that they would diverge that much at 25% SOC, however. I only have divergence data for top-of-charge (because I went with bottom-balancing), but at 75% SOC LFP is still well within the linear part of the voltage curve. It is closer to 90 or 10% SOC where the fun begins.

Nebster you are wrong about bottom balancing. I have never bottom balanced in my life and wouldn't know how as that approach doesn't fit with LFP on sailboats. But your right about the cats analogy.

As mentioned up thread that divergence at cats go running I expressed at 25% SOC is at around 0.3C and well above normal load in a sailboat. I referenced it to potential Delos loads with an Inverter driven Induction cooktop in conjunction with a Microwave. Hardly typical sailboat appliance's.

Having had the luxury of using in real life and testing LFP with the benefit of cell by cell data and with both old and new cells and capacity checking both over years, I think I have a pretty good idea of LFP chacteristics.

That said no one has a mortgage on intelligence.

[Delfin]

Originally Posted by nebster

Can you say more about this? I agree with everything else you wrote, but this doesn't ring true in my experience. I might be misunderstanding what you wrote.

I have an automated (not sure I would go so far as to call it "smart") charge system that works great in the real world, and it does not "trip to float" when loads change. I have my CV threshold set substantially lower than 3.525v, and this virtually guarantees that my pack will not charge above roughly 90-92% SOC.



Sure you can, that's what coulomb counters are for. They drift a bit, but they're plenty good enough to tell you whether you're at 20% or 50% or whatever. That data can be used to as an input to charge decision-making, too.

To your first...to test the recommendations of some to use low CV rates I set the Balmar bulk voltage and Trace charger to 3.45 v per cell. While running, the batteries acceptance rate was around 20% lower than my experience when CV was set to 3.6 v. Why? I have no clue, but that is what I measured with a Link 20 coulomb counter. I had a water maker running at the time, but that is typical and never has had any effect on the acceptance rate. Within 15 minutes or so, the batteries tripped to float since I have zero time allotted for absorption. Again, that is just my experience. With the CV set at 3.6 v, the acceptance of the batteries stays at 200 - 206 amps until the last 4 minutes of the charge cycle, at which point it plummets to 5% of C and I disconnect charging. This happens at around 3.54 v per cell.

I do use the Link 20 coulomb counter to gauge how much current has been used, more or less. My comment was related just to voltage, so you are correct on the use of the counter to get an idea of where you stand in terms of recharge, at least as reflected by the acceptance rate.

[nebster]

Originally Posted by Hoopla

Nebster you are wrong about bottom balancing. I have never bottom balanced in my life and wouldn't know how as that approach doesn't fit with LFP on sailboats. But your right about the cats analogy.

I can barely tell what you're talking about, but I think it's unlikely I'm wrong and much more likely that you're misunderstanding me.

A bottom-balanced string of cells will not diverge at 30% SOC or 0% SOC.

That is what I said. Nothing more.

As mentioned up thread that divergence at cats go running I expressed at 25% SOC is at around 0.3C and well above normal load in a sailboat. I referenced it to potential Delos loads with an Inverter driven Induction cooktop in conjunction with a Microwave. Hardly typical sailboat appliance's.

You probably know what you're talking about, but you're not expressing it clearly. I think you should say what you mean instead of "cats," because after this last post, I have no idea what you mean.

At very low SOC in a top-balanced series string of cells, the individual cell voltages will diverge. It is possible that one of them will end up very low even though the whole string's combined voltage is still okay. A system that does not monitor individual cell voltages will struggle to protect all the cells in this scenario.

This can be the case at 0.3C or 0C, so I am confused about what you are trying to get it, since it seems that you think the loads matter.

Having had the luxury of using in real life and testing LFP with the benefit of cell by cell data and with both old and new cells and capacity checking both over years, I think I have a pretty good idea of LFP chacteristics.

I'm sure you do, but more than one of us reading also have a pretty good understanding of cell behavior, and it's very hard to understand what you are writing right now.

[CatNewBee]

Originally Posted by nebster

I agree. However, there are no USA-made LFP storage cells anyway, so it's a moot point.

A more interesting question is whether "off brand" manufacturers' cells can be any good. My experience, a sample size of 1 (or 112, depending on how you want to count) says yes. However, I did have 2 out of 112 cells DOA.

Then again, someone else building a 32-cell CALB pack just a few weeks ago found 1 out of 32 DOA, too.

Some of the cells made by the lesser-known companies are more advanced technically than what you can buy from CALB or Winston -- both higher volumetric and higher gravitational density.

Where that leaves a buyer, I don't know. Depends on how much you feel like gambling, I suppose.

I do know you can get cells for right at $0.25/kWh right now FOB Shenzhen, so in my opinion it's probably a toss-up, since a DIY builder should be qualifying every cell anyway. Just buy some extra, throw out the bad ones, and carry a few extra cells on board in case something happens down the road.

Well, this is more supply strategy than technical.
In the EU you can get pre-selected, pre-balanced Winston cells, so the supplier checks them and sorts them out by making some profit.

You can buy the in China and ship them, than some spare cells is a good strategy. Just kepp them watched after in storage. charge them from time to time to a safe level, or sell the extra good ones on eBay.

[nebster]

Originally Posted by Delfin

To your first...to test the recommendations of some to use low CV rates I set the Balmar bulk voltage and Trace charger to 3.45 v per cell. While running, the batteries acceptance rate was around 20% lower than my experience when CV was set to 3.6 v. Why? I have no clue, but that is what I measured with a Link 20 coulomb counter.

Okay, thanks for the extra detail.

Are you able to see on the Balmar whether it is in CC or has reached CV? I'm assuming you can do that on the Trace unit for sure.

Your description of the behavior doesn't make sense to me. If a charger is truly in CC, it should be putting out its rated current (modulo thermal limiting, etc.).

How many times did you test this charge strategy? What was the SOC at the start of the charge, if you recall?

How are the two charging devices sensing voltage? Are they doing it on the charging wires directly, or do you have independent voltage sensing set up?

With the CV set at 3.6 v, the acceptance of the batteries stays at 200 - 206 amps until the last 4 minutes of the charge cycle, at which point it plummets to 5% of C and I disconnect charging. This happens at around 3.54 v per cell.

This is also fascinating, and weird. You're saying that, at 3.60v CV threshold, your system never reaches that threshold and thus stays in CC, but yet it manages to taper anyway?


From my perspective, the good news is that your system has something else going on that is more than just a simple CC/CV charger connected to an array of LFP cells. A basic setup definitely does not behave like this, and so I think I can stand by my refutation of your item 4; that is, in my experience it is perfectly viable to lower the CV threshold as a "poor man's" way of limiting the SOC.

[OceanPlanet]

Originally Posted by witzgall

#1, this is misleading and potentially dangerous. What happens when the alternator fails, and puts out over 15v?

#2 I cannot follow what you are trying to say.

Chris

FWIW, we always run the IGN input to the alt. regs (that tells them the engine is on and that it's time to run the alternator) through the BMS FCC control circuit. If at any time the BMS is going to cut off the charge bus (or the single bus if a single-channel system), it opens the FCC, which tells the alt. reg. to turn off.

Actually, you could also use the power to the alt. reg. or the field from it that goes to the alt. Any of the three will turn off the alt, which needs to be done before there is a cutoff from the battery.

[john61ct]

Originally Posted by Delfin

Here's another then.....https://www.navitassys.com/products-...t-systems-bms/

I do not see comms or coordination to external systems, nor user access to the 3.2V cells.

So yes these are normal drop-ins afaic.

[john61ct]

Originally Posted by Hoopla

As mentioned up thread that divergence at cats go running I expressed at 25% SOC is at around 0.3C and well above normal load in a sailboat.

I find this incomprehensible.

The SoC level - how full the bank is - has nothing to do with the discharge current.

LFP will handle .3C equally well at any point of SoC, if for a short time will not do anything unusual or even discharge the bank much. Same with .8C or even higher if the infrastructure is robust enough.

[john61ct]

I see 2000 cycles as a bare minimum lifespan from following the industry standard (too high voltage) charging profiles, barring "events" avoiding the shoulders staying between 2.99 and 3.48Vpc, should get many thousand more.

To me that includes not holding Absorb at all, hit the setpoint and stop, no valid reason to push past that, except when precise calibration of 100% Full SoC is required, in which case I use .02C for endAmps.

Also no Floating at all. Load devices coming online pull volts lower, so do not stop charging but keep it going longer.

[john61ct]

It is perfectly normal for a higher voltage to result in faster acceptance rate.

That does not justify pushing the voltage higher than the setpoint you think is best for the health of your bank.

Perhaps not harmful for the source to be "striving for" a higher voltage in the early stages, but as the bank voltage climbs, the setpoint should not be higher than the maximum you want it to actually reach.

[john61ct]

Automated charging and multiple layers of bank protection is relatively difficult and expensive, but simply a sine qua non requirement for many, if not most boaters.

[nebster]

Originally Posted by john61ct

It is perfectly normal for a higher voltage to result in faster acceptance rate.

Huh? Not in CC. The battery is constraining the voltage, not the configured CV threshold.

The amount of charge current delivered in CC should have nothing to do with the CV threshold.

[Delfin]

Originally Posted by nebster

Okay, thanks for the extra detail.

Are you able to see on the Balmar whether it is in CC or has reached CV? I'm assuming you can do that on the Trace unit for sure.

Your description of the behavior doesn't make sense to me. If a charger is truly in CC, it should be putting out its rated current (modulo thermal limiting, etc.).

How many times did you test this charge strategy? What was the SOC at the start of the charge, if you recall?

How are the two charging devices sensing voltage? Are they doing it on the charging wires directly, or do you have independent voltage sensing set up?



This is also fascinating, and weird. You're saying that, at 3.60v CV threshold, your system never reaches that threshold and thus stays in CC, but yet it manages to taper anyway?


From my perspective, the good news is that your system has something else going on that is more than just a simple CC/CV charger connected to an array of LFP cells. A basic setup definitely does not behave like this, and so I think I can stand by my refutation of your item 4; that is, in my experience it is perfectly viable to lower the CV threshold as a "poor man's" way of limiting the SOC.

The SOC for the test run was around 55%, based on amps used. I think, but don't know, that the reason the CAR of the Li bank dropped under this regime was simply because there was less current being generated as the lower bulk voltage settings that had to charge and handle house loads, so less was available to go to the Li batteries. I can't think of another explanation but if you have one, let me know.

See the attached graph for what CAR looks like underway with both Trace and alternator charging the Li bank with bulk set to 28.8 or 3.6 v per cell. As noted, I have no idea why, if I set the bulk voltage to 27.6 the Balmar and/or the Trace flip to float quickly, but they do. However, as you can see from the graph, once the batteries reach 27.6 volts for the pack or 3.45 volts per cell, you are within a few minutes of the CAR falling off the cliff. So, it would be easy enough to disconnect charging at that voltage and you would be almost full anyway. My question is whether that would have any material impact on longevity compared to not drawing the batts down to 20% SOC. Based on Lithionics data, they think you'll get 15,000 cycles at 50% discharge with a charging voltage of 3.6 v per cell and a minimal absorption time. Beats me.

And, I only tried the low bulk setting once, but it seemed to be pointless, so gave it up and went back to a charging regime that as I understand it, will provide long life and as you can see from the graph, means I can recharge very quickly.

[nebster]

Originally Posted by Delfin

The SOC for the test run was around 55%, based on amps used. I think, but don't know, that the reason the CAR of the Li bank dropped under this regime was simply because there was less current being generated as the lower bulk voltage settings that had to charge and handle house loads, so less was available to go to the Li batteries. I can't think of another explanation but if you have one, let me know.

How are your chargers measuring the voltage? Did you run dedicated voltage sense wires to each?

Are you 100% certain that both chargers are staying in CC and then going straight to float? Is it at all possible that one of the two does have an absorption interval?

My best guess, and it's a real stab in the dark, is that you have two things going on:

a) your chargers are reading voltage at their terminals, which will be high. Really high, at 200A. It's a "bad" place to use as a setpoint for charging, because what we care about is the voltage at the battery itself.

b) one or both of your chargers is, in fact, going into absorption (CV). That's the only thing that makes sense given the tailing current you see.

If (a) is true, it could explain why you see reduced charging when you lower the CV threshold: because, in fact, one of them is hitting the threshold immediately.

In addition to answering the questions above, if you want to get to the bottom of it, we need to know the voltages as measured by you at (1) the charger terminals and (2) the battery at approximately the same time. It is important to use the same, accurate meter for those three voltages.

Or, you may not care, which is fine! (But statement 4 in your earlier post is not.)