LiFePO4: what voltage?

[tanglewood]

Originally Posted by rgleason

PCM "Protection Circuit Module"
sorry for the Acronym daze, perhaps TAIFP (turn about is fair play)?

PCM for Lithium battery pack - Evlithium

Other balancers:

https://www.electriccarpartscompany....tery-Balancers
https://www.electriccarpartscompany....ERS-EQUALIZERS
https://www.electriccarpartscompany....ERS-EQUALIZERS

https://www.ev-power.eu/Battery-Management/?cur=1


Why use balancers https://www.electriccarpartscompany....ime_p_699.html




Thanks for confirming over 3S can cause balance problems.

The multiple-strings-in-parallel "issue" I think really relates to just that, string in parallel. Not so much for cells in parallel which is the preferred way to build up LFP banks. So, for example, the following are all the same capacity and voltage, but built up differently.


4P4S is not an issue because it is one string. All the paralleled cells are at the same voltage. Cells can exist at only 4 different voltages.



2P4S2S is two strings in parallel. The sum total voltage of the strings will remain in sync, but the cells that make up each string can drift differently. Cells can now exist at 8 different voltages.


1P4S4S is four strings in parallel. Again, the total string voltages will be the same, but all the cells can be at different voltages. Calls can exist at 16 different voltages.

[john61ct]

No, going past 3-4 parallel anything is the issue, even if 1S6P.

[john61ct]

But just like a 3.45Vpc charge setpoint, or "don't go below 50%", this is not some hard & fast rule where things explode, but a useful guideline, "rule of thumb" in striving for maximum longevity.

To the extent you must violate it, the probability of problems increase.

Thousands of people buy say 6 regular 12V batts to create a high-AH 12V bank, and are happy to get their 5-6 years service out of them,

blissfully unaware they may have got more with 3 larger AH pairs of 6V.

[nebster]

Originally Posted by john61ct

No, going past 3-4 parallel anything is the issue, even if 1S6P.

Please give us references or at least your own explanation of the nature of the concern, because I think you have this pretty much backwards.

The science and the math suggest the opposite: higher block-level parallelism improves balancing performance. Large numbers of cells in parallel act to average out the cell-level differences, bringing each block then closer to the others in a series string. The closer each unit in series is in capacity to the others, the easier it is to maintain balance.

This is why you see the harvested, homebrew 18650 used cell ESS packs performing so stably when they are series-ed up: they put 30 or 60 or 100p and take advantage of the law of large numbers.

It's also a part of why Tesla chooses topologies like 74p6s in their EV modules.


However: ANY time you parallelize cells, if you want equivalent thermal safety to a series-then-parallel topology, you must fuse between each cell in each parallel block. Almost no one does this*with prismatic builds, and we don't hear of a lot of LFP fires, I suspect because LFP is inherently very safe to begin with. But if you want the highest level of safety, add intra-cell fuses on parallel blocks. (Just like Tesla does.)

[Delfin]

Originally Posted by nebster

Please give us references or at least your own explanation of the nature of the concern, because I think you have this pretty much backwards.

If you are able to spring a reference loose, you will win an award.

[john61ct]

In any LFP discussions, I am only talking about using large prismatic cells as the basis for a House bank.

How Tesla engineers their very robust and complex systems has nothing to do with our simple DIY House bank setups.

And if I bought six Victron packaged systems and wanted to parallel them, I would take their word as to how to do so safely, but again, not what we are talking about here.

The "don't go past 3-4 paralleled strings or there may be balancing problems" guideline has been repeatedly stated across many forums by several members, with in-depth knowledge and extensive professional experience whose opinions and advice I value.

I have not kept records of the threads where they went into the details, which were more technical at the time than I was willing to pursue.

I am happy to simply follow that guideline, with 180+AH cells readily available it is not an undue burden.

As always in the end each owner is free to do what they like with their gear, but for those who are apparently unaware of the issue I will continue to raise it.

[nebster]

Originally Posted by john61ct

In any LFP discussions, I am only talking about using large prismatic cells as the basis for a House bank.

How Tesla engineers their very robust and complex systems has nothing to do with our simple DIY House bank setups.

You're really hurting your credibility with sentences like this, John.

Tell us about your simple DIY house bank setup, if you would?

And if I bought six Victron packaged systems and wanted to parallel them, I would take their word as to how to do so safely, but again, not what we are talking about here.

This must be because a trusted vendor has magic pixie dust that makes it okay, but we individuals are unable to achieve something similar, eh?

John, which packaged or home-built system do you have installed?

The "don't go past 3-4 paralleled strings or there may be balancing problems" guideline has been repeatedly stated across many forums by several members, with in-depth knowledge and extensive professional experience whose opinions and advice I value.

Yeah, that's what I thought. You have no idea where you got this idea, you just "heard it," and so it must be true.

I am happy to simply follow that guideline, with 180+AH cells readily available it is not an undue burden.

You can follow whatever vaguely-recollected guidelines you'd like. For the rest of us, we might prefer to use actual data and reasoning and experience to drive decision-making.

Which cells and topology are you using on your boat today?

[tanglewood]

I don't think my earlier examples were understood. Paralleling cells is fine in large numbers. Paralleling strings becomes problematic.


So 4P4S is fine, but 4S4P would be problematic. And it's not problematic because lots of people I know and trust told me so. It's problematic because you have many more cells able to drift in voltage and SOC increasing the likelihood of cell imbalance. And with LFP, it's not only theoretically true, there is a good study with good old lab tests demonstrating how paralleled cells comes into balance on their own, and do so very quickly with no fireworks or other drama.

[nebster]

Originally Posted by tanglewood

Paralleling cells is fine in large numbers.

Yes, definitely fine.

If you want maximum safety equivalent to what you get with the other topology, though, you need overcurrent protection on every single cell in each parallel block. Almost no one does that with LFP , and no one's boat burns down because it's still super safe. But, it's still not as safe for the pack as choosing to do it.

I think a big part of the reason no one does it is because it is very difficult to safely and inexpensively fuse large prismatics. You need big, annoying fuses, it makes the interconnect really complex (and I believe working with the cell interconnect is the most dangerous part of large DIY LFP packs), and consequently it does drive up cost in money and time.

Paralleling strings becomes problematic.

Here's where I have to disagree, unless by "problematic" you just mean "more expensive."

It is necessary to instrument every string in a series-then-parallel pack to the same level you would instrument the one string in a parallel-then-series pack. And that instrumentation can get expensive, especially if you insist on using fancy BMSes with active balancing features and whatnot.

But, if you are willing to provide cell-level voltage monitoring, string-level disconnect on HV and LV, and overcurrent protection at the string level, plus you are willing to take steps to ensure that the resistance is large enough and close enough between each string and the paralleling bus, then you can totally have a very manageable, safe multistring pack.

I know because that's what I've got! (Six of those pesky strings, currently, with an option to add more.)

Now, I'd actually really like to debate this further, because it's always possible that I've missed some failure mode that I didn't anticipate. If any of y'all want to propose a way that such a pack (let's use mine, if you like? 16s-then-6p) is at risk, unstable, or subject to unique problems or issues, I'll gladly take a stab at it and see if I can, err, defuse things.

[kcj]

I did 6p4s for two reasons—first, my understanding was the smaller the cell, the stronger and more stable the ‘internals’ of the cells were. Less room for the pouches to shift. Second, the 100ah Calb Cells fit best in my compartment. Larger, 200ah cells, would have had to be installed on their side, in non optimal configs.
Furthermore, I was unaware that going to 6p was an issue... how does 2p of 100ah differ than 1p of 200ah? Is the 200ah fused internally?

The bank has been working fine for two years. I only see big deltas at the knees.... the cell spread is only 30mV at 3.45v, which is where I usually quit charging. I have been pushing it a little lately trying to test my OVP circuits, an interesting exercise in itself. In full disclosure, i never did a ‘proper’ top balance as I did not have a 3.6v power supply.

Cheers,
John

[tanglewood]

Originally Posted by nebster

Yes, definitely fine.

If you want maximum safety equivalent to what you get with the other topology, though, you need overcurrent protection on every single cell in each parallel block. Almost no one does that with LFP , and no one's boat burns down because it's still super safe. But, it's still not as safe for the pack as choosing to do it.

I think a big part of the reason no one does it is because it is very difficult to safely and inexpensively fuse large prismatics. You need big, annoying fuses, it makes the interconnect really complex (and I believe working with the cell interconnect is the most dangerous part of large DIY LFP packs), and consequently it does drive up cost in money and time.



Here's where I have to disagree, unless by "problematic" you just mean "more expensive."

It is necessary to instrument every string in a series-then-parallel pack to the same level you would instrument the one string in a parallel-then-series pack. And that instrumentation can get expensive, especially if you insist on using fancy BMSes with active balancing features and whatnot.

But, if you are willing to provide cell-level voltage monitoring, string-level disconnect on HV and LV, and overcurrent protection at the string level, plus you are willing to take steps to ensure that the resistance is large enough and close enough between each string and the paralleling bus, then you can totally have a very manageable, safe multistring pack.

I know because that's what I've got! (Six of those pesky strings, currently, with an option to add more.)

Now, I'd actually really like to debate this further, because it's always possible that I've missed some failure mode that I didn't anticipate. If any of y'all want to propose a way that such a pack (let's use mine, if you like? 16s-then-6p) is at risk, unstable, or subject to unique problems or issues, I'll gladly take a stab at it and see if I can, err, defuse things.

I don't disagree. And you have identified the steps to mitigate issues, in particular per cell-block monitoring. I think much of this comes from LA thinking where there is no cell monitoring, and return to full charge is so important. There, cell voltage/SOC drift had greater consequences. With LFP, cell block SOC drive is largely inconsequential until a cell starts approach the low or high boundaries. A system will active balancing a-la Victron makes the problem go away, but of course bring with it the issues associated with balancer faults and possible battery damage.


But all that said, the more cell blocks you have, the more you need to instrument, and the more likely you will have a block that starts to run too high or too low, and requires intervention. Then this is balanced against the increased power availability of redundant strings and BMSs that can keep power running even if a cell block in one string gets wonky and the string needs to be shut down. On a cruising boat in remote areas, this is pretty appealing and something I will likely do when I build up the system for my boat.


And by the way, don't you mean "ensure that the resistance is low enough"?

[nebster]

Originally Posted by tanglewood

I don't disagree.

Darn. I was hoping you were going to identify some issues that I could either debunk or start to worry about.

There are a couple of interesting papers and slide decks that discuss concerns with paralleling strings.

My opinion is that most of the issues discussed within are either overblown or acceptable tradeoffs. But then there are a couple of them, particularly the "eddy currents" concern, that I can't reconcile with the physics. In particular the Orion guys' explanation for how eddy can affect strings doesn't make sense to me. I am still trying to understand if there is a potential problem lurking here that I need to mitigate. I certainly can't measure any problem within the limits of my test gear.

But all that said, the more cell blocks you have, the more you need to instrument, and the more likely you will have a block that starts to run too high or too low, and requires intervention. Then this is balanced against the increased power availability of redundant strings and BMSs that can keep power running even if a cell block in one string gets wonky and the string needs to be shut down. On a cruising boat in remote areas, this is pretty appealing and something I will likely do when I build up the system for my boat.

Agreed: much moreso on a boat, but even on our RV, I look at the power system as being critical.

One thing to look at is, should a string drop in a two-string system, can the other string handle the load-doubling that it is about to experience? Here, having more strings is a big plus. The advantage is somewhat analogous to having a large block of paralleled cells.

And by the way, don't you mean "ensure that the resistance is low enough"?

It turns out having some resistance in the circuit makes some aspects of the battery easier to manage, or at least I'm pretty sure it does.

The extra resistance imposed by the interconnect (fuses, wire, bus bars, etc.) creates a "buffer" that helps equalize the current in parallel paths. Otherwise, fluctuating cell impedance could drive huge current imbalances and stressful transients.

[rgleason]

Nebster wrote:
"..science and the math suggest the opposite: higher block-level parallelism improves balancing performance. Large numbers of cells in parallel act to average out the cell-level differences,..."

bringing each block then closer... in a series string. "

Comment: This is plausible, but do you have white papers for reference or is this based on experience?

I do appreciate John's brief comments and his experience and I don't think it helps to be adversarial, particularly when we are all learning, admittedly some have actually done more than others, but time and more experience will tell.

"This is why you see the harvested, homebrew 18650 .. series-ed up: they... take advantage of the law of large numbers."

I note that very long strings in series or parallel require many connections and lots of wiring. We've seen videos of this process and the chances of getting a bad cell, bad connection or using undersized wire or weak wiring or connections is quite high. We are depending on the Chinese manufacturer to have high standards in an unregulated competitive environment.
Mainesail cautions about this in his LiFePo4 article and one of his many suggestions is to have vibration testing done. He also suggests that larger cells be used. Also as Mainesail points out, marine & rv is not a targeted market, so we may not be getting the best cells in the diy market. As the industry matures and experience is gained quality should get better. So perhaps marine packagers are a surer bet.

[nebster]

Originally Posted by rgleason

"..science and the math suggest the opposite: higher block-level parallelism improves balancing performance. Large numbers of cells in parallel act to average out the cell-level differences,..."

bringing each block then closer... in a series string. "

Comment: This is plausible, but do you have white papers for reference or is this based on experience?

I don't think we'll see anyone write a white paper on this, because it's just basic math. I certainly don't have any I can point you to.

If one cell in 100 loses half its capacity because of a flaw, the entire block still has 99.5% capacity. That improves that block's ability to stay balanced amongst a series string of blocks.

Likewise, if you have 400 cells to distribute amongst 4 blocks (for a 100p4s pack), each with a capacity along a normal distribution, you're likely to average out those variations when you put 100 random ones together.

There is a whole forum dedicated to people building ESS packs out of recaptured 18650 cells, though, and you can use their results as an example if you like. Paralleling large numbers of cells is also the way almost all EV packs are built.

It is silly for John to say that there is an arbitrary, small number ("3-4") above which paralleling is inappropriate.

I do appreciate John's brief comments and his experience and I don't think it helps to be adversarial, particularly when we are all learning, admittedly some have actually done more than others, but time and more experience will tell.

John occasionally writes some great stuff. But mixed in with it is a bunch of wrong stuff, and much of it is presented as if it were gospel. This is almost the worst case scenario, because to a casual reader it seems like he knows what he's talking about. But, when it comes to building a substantial lithium bank and running it in real life, it is obvious to those of us who've done it that he has not. (And, as you may have noticed, he won't share with us what his experience is, for some strange reason.)

To make progress as a community of learners, we have to separate the wheat from the chaff. I am trying to help do that by pushing back on some of the nonsense that I see being propagated... by John or anyone else.

I note that very long strings in series or parallel require many connections and lots of wiring. We've seen videos of this process and the chances of getting a bad cell, bad connection or using undersized wire or weak wiring or connections is quite high. We are depending on the Chinese manufacturer to have high standards in an unregulated competitive environment.

I agree. But, what choice do we have if we want a large pack? Cells above about 200Ah are pretty hard to come by these days. We see a retreat from their availability in the marketplace over the last few years, probably because they were not hard to make in the factory but very hard to make last in the field. None of the commercial assemblers build their units out of larger cells, either.

What I don't agree with is that assembling a large battery has to be high risk. It is incumbent upon the pack builder, though, to spend the time verifying the cells and then using the right wire and tooling to ensure a good build. And then everything needs to be tested.

Mainesail cautions about this in his LiFePo4 article and one of his many suggestions is to have vibration testing done. He also suggests that larger cells be used. Also as Mainesail points out, marine & rv is not a targeted market, so we may not be getting the best cells in the diy market.

I couldn't figure out a way to vibrate 700 pounds of cells in a test, so I just put them in the vehicle and started driving around on bumpy roads.

Larger cells are very helpful, but as discussed above, with the exception of Winston, it seems like there is nothing out there above about 600Wh.

So perhaps marine packagers are a surer bet.

For a lot of folks, I think so. I think we will see the RV market start to demand and drive more commercial offerings, too. That should be a good thing.

[rgleason]

Nebster
I agree with much of what you are saying, and I think you'd best stay off some of the bumpier roads (I had a laugh). Will you help me design a moderate sized bank to be charged by alternator or FLA and BB charger.