Battery monitor--where in the system do you put it? Is monitoring every battery overkill?

RV situation and a bit of a gadget guy. I currently have a single 12V 100Ah LiFePO4 battery and want to add 1, maybe 2 more, in parallel.

With Coulomb-counting volt/amp meters being fairly cheap these days, I'm considering putting one on every battery.

I'm thinking it's nice to know what each battery is doing. If, for example, 1 battery fails, I might not even notice if I have a meter at only the output of the battery bank. In that position and if I didn't use the bank close to its full capacity, the meter wouldn't help me tell if there were 1 or 10 batteries behind it.

Is monitoring every battery overkill? I'm talking about 1 to (in the future) 4 batteries. If the battery bank was big, like for a home solar situation, then maybe it's too much.

Related, for those who have huge banks, how can you tell if one battery is going bad? Where did you put your monitor(s)?

Ample said:

Related, for those who have huge banks, how can you tell if one battery is going bad? Where did you put your monitor(s)?

Click to expand...

I can tell by watching my cell voltage reports from my BMS. If one cell goes bad, even if I have several in parallel, the deltas increase at the top of the charge cycle. I can monitor from a cloud based site on my phone or laptop.

Adding a shunt sampler to the negative lead for each string makes sense to me.
As does adding an MRBF fuse to the positive lead for each string.

Putting shunt on each parallel battery leg is good especially with batterries with different age and/or have been used under different conditions. You can check their amperage to understand any imbalance that you can use to avoid overcurrenting a single battery.

You may find the balance variance changing over discharge (and charging) cycle where one starts out dominating then is will fade and another battery will dominate as discharging/charging progresses.

Again, the primary key object is to know how much you can push discharging and charging peak current to avoid damaging perfectly good battery. Just because you have two batteries in parallel does not mean you can draw twice as much peak current.

RCinFLA said:

Just because you have two batteries in parallel does not mean you can draw twice as much peak current.

Click to expand...

Can you expand on that?
I would have thought that as long as the strings are well matched up to the aggregation point that you could expect close to 2x the draw.

Even small differences in wiring resistance can drive significant current imbalance in parallel batteries. At high currents, this causes one battery to supply more than 50% of the current. The result is that one battery can drop out due to BMS overload while you are still under the theoretical combined maximum. Paying close attention to balancing the wiring resistance for each parallel pack is critical. Even then ~90% of theoretical is all I would guess is possible.

Luthj said:

Even small differences in wiring resistance can drive significant current imbalance in parallel batteries. At high currents, this causes one battery to supply more than 50% of the current. The result is that one battery can drop out due to BMS overload while you are still under the theoretical combined maximum. Paying close attention to balancing the wiring resistance for each parallel pack is critical. Even then ~90% of theoretical is all I would guess is possible.

Click to expand...

Could small things such as bus bars (copper vs steel), use of washers, etc., make such a difference?

What about mixing batteries from different manufacturers? I don't mean different chemistries but say mixing Brand A vs. Brand B, and perhaps vs. a DIY battery, all of the same chemistry, voltage and Ah capacity. I'm thinking internal resistance could be different due to different gauge and/or the number of wires between the cells and the BMS. BMSes could be different. (Good thing @Will Prowse has done battery teardowns! I know how some of them are made now!)

And what of a new model from even the same manufacturer in the same chemistry, voltage, and capacity? Presumably, they'd make sure it was compatible with a former model so that people can add more to their strings. But having worked in high tech before, I've seen that it's not always the case.

Luthj said:

Even small differences in wiring resistance can drive significant current imbalance in parallel batteries. At high currents, this causes one battery to supply more than 50% of the current. The result is that one battery can drop out due to BMS overload while you are still under the theoretical combined maximum. Paying close attention to balancing the wiring resistance for each parallel pack is critical. Even then ~90% of theoretical is all I would guess is possible.

Click to expand...

Getting similar wiring resistance is a good idea for maximum pack efficiency, but in real life the paralleled batteries (strings) have other factors that affect their current flow much more, in my experience.

On top of that, there are LOTS of battery installs where people are just paralleling a group of drop-ins, each with a discrete BMS, and these installs are not always hewing to the principle of bussing everything up perfectly evenly. (Daisy-chaining them is really common.) And they still work okay, just the strings with the higher resistance get to do a little less of the work and the wear will be a bit uneven.

I have 6 strings paralleled up, as evenly as I think is humanly possible, and especially at high and low SOC, I can see large variations in the current coming from (or to) each string. If I draw 200A and therefore expect 33A average per string, I might see 50A from one string for a few seconds at a time, and then it settles back or even goes below average. Another string is then "catching up" at that point and supporting the load. Sometimes they will all be about the same for minutes at a time, and then there'll be a flurry of little perturbations across the pack. It's kind of mesmerizing.

This variation is too high to be accounted for by a (fixed) difference in resistance. Instead, it must be the chemistry itself shifting and equilibrating inside the cells.

I have fuses that stop a string at 100A, and they've never blown, so swings of 3x don't seem possible in my pack at least. But it's true that the pathway to the bus bar must be able to support a fair bit more than the average, under peak conditions, in order to accommodate the fluctuation.

nebster said:

Getting similar wiring resistance is a good idea for maximum pack efficiency, but in real life the paralleled batteries (strings) have other factors that affect their current flow much more, in my experience.

On top of that, there are LOTS of battery installs where people are just paralleling a group of drop-ins, each with a discrete BMS, and these installs are not always hewing to the principle of bussing everything up perfectly evenly. (Daisy-chaining them is really common.) And they still work okay, just the strings with the higher resistance get to do a little less of the work and the wear will be a bit uneven.

I have 6 strings paralleled up, as evenly as I think is humanly possible, and especially at high and low SOC, I can see large variations in the current coming from (or to) each string. If I draw 200A and therefore expect 33A average per string, I might see 50A from one string for a few seconds at a time, and then it settles back or even goes below average. Another string is then "catching up" at that point and supporting the load. Sometimes they will all be about the same for minutes at a time, and then there'll be a flurry of little perturbations across the pack. It's kind of mesmerizing.

This variation is too high to be accounted for by a (fixed) difference in resistance. Instead, it must be the chemistry itself shifting and equilibrating inside the cells.

I have fuses that stop a string at 100A, and they've never blown, so swings of 3x don't seem possible in my pack at least. But it's true that the pathway to the bus bar must be able to support a fair bit more than the average, under peak conditions, in order to accommodate the fluctuation.

Click to expand...

The packs that draws more current its always the same?... in that case, it could be possible to "balance" the system by adding a little more resistance to the more current drawing strings?... like adding a little longer cable?.

nebster said:

This variation is too high to be accounted for by a (fixed) difference in resistance. Instead, it must be the chemistry itself shifting and equilibrating inside the cells.

Click to expand...

Could these be describe as eddy currents? I first heard that term in a white paper by my BMS vendor when discussing the challenges of using Tesla modules.

mrdavvv said:

The packs that draws more current its always the same?... in that case, it could be possible to "balance" the system by adding a little more resistance to the more current drawing strings?... like adding a little longer cable?.

Click to expand...

It's not always the same, no. It shifts around.

One way to tune the resistance is just to change the torque on one of the many lugs in the string that you're interested in. Even a small amount makes a difference, once the current is high enough.

Ampster said:

Could these be describe as eddy currents? I first heard that term in a white paper by my BMS vendor when discussing the challenges of using Tesla modules.

Click to expand...

I think so, or at least I think the phenomenon might also be the source of what your vendor's write-up described.

When I first read someone (a BMS manufacturer) expressing concern about "eddy currents," it was in a write-up where they were advocating against having multiple, parallel battery strings. I found their argument unpersuasive in light of the observations on my pack, mostly because I don't see current moving hardly at all once the pack is isolated from a load or charger. If the individual strings were spending time at rest constantly passing energy back and forth between each other, I could see that being a big problem, because it would be a way to amplify wear and tear by creating a lot of extra, partial cycles. But I don't think they really do that in real life.

I actually called them and spoke with an engineer, and when I asked for more details about these "eddy currents," he basically said that he didn't have any hard data and that he thought running several of their BMSes in parallel would be fine. So, I really have no idea what they were getting at.

nebster said:

I actually called them and spoke with an engineer, and when I asked for more details about these "eddy currents," he basically said that he didn't have any hard data and that he thought running several of their BMSes in parallel would be fine. So, I really have no idea what they were getting at.

Click to expand...

You know, I believe this is a "Left over" ideation that originated with Lead Acid Batteries. With FLA doing Parallel batteries can become a nightmare if things are just not 100% bang on. I started with Big FLA and ran parallel and it can be quite a PITA and one learns to rotate batteries within the parallel set, as there is always one that has to be different.

With anything Lithium Based, be is 18650 NMC or LFP, the BMS is the key and having the setting properly matched for the cells and packs, so no pack over or under the limits set. If the BMS' have differing settings for each pack within the bank the results will be less than stellar and you will have an imbalance and likely other issues too.

I cannot understate that t is very important to ensure your system is Voltage Calibrated and corrected. Even small voltage drops, increased resistance can have negative effects which with any Lithium Based battery system is essential. FLA can take much more abuse and variations and won't bother them, but with LFP for example, overcharge by .5V and damage could result, or if cutoff happens at 2.0V instead of 2.5, the result won't be what you want. see link in my signature regarding calibration.

Steve_S said:

You know, I believe this is a "Left over" ideation that originated with Lead Acid Batteries. With FLA doing Parallel batteries can become a nightmare if things are just not 100% bang on. I started with Big FLA and ran parallel and it can be quite a PITA and one learns to rotate batteries within the parallel set, as there is always one that has to be different.

Click to expand...

That certainly seems plausible. There are real issues with chronic undercharging of lead acid, such that I suspect poor paralleling ends up doing a lot of early damage to batteries that are getting the short end of the charging stick. This is almost a non-issue with lithium, except insofar as it might underutilize one battery versus the next.

Since no two batteries will ever have an identical internal resistance, there will often be oscillation between the batteries in parallel at high currents. Typically this will manifest as a gradual shift of 10% or so in current in a well balanced system. One battery will supply more current for a while until its voltage starts to sag every so slightly, then the imbalance can shift to the other battery(s) until they drop a bit, then it shifts back. This part of the reason that you want 20-30% overhead for your highest continuous current.

Another cause of cycling load shifting is heating and cooling of conductors. If you have a contact point that's heating up it can affect current balance. In severe cases it can actually increase the resistance of a connection by mechanical means (decreasing contact or mechanical stresses).

This is a big part of the reason why EV packs are typically 1 logical battery, with each logical cell group being made of numerous paralleled cells. By reducing the wiring and variance between each logical cell, current variance due to temperature, interconnects, etc, can be minimized. This means the parallel circuits are restricted to a single cell block, where quality and temperature control are much easy to control.

nebster said:

That certainly seems plausible. There are real issues with chronic undercharging of lead acid, such that I suspect poor paralleling ends up doing a lot of early damage to batteries that are getting the short end of the charging stick. This is almost a non-issue with lithium, except insofar as it might underutilize one battery versus the next.

Click to expand...

The flat discharge curves and lack of partial SOC degradation eliminates much of the concern with LFP packs compared to lead. You still need to take peak current issues into consideration, but that's fairly easy if you leave some overhead above your expected max current levels.

Since LFP cells have such a flat discharge curve, a current imbalance can cause sudden current shifts near the bottom 20% of SOC. If you have a 10% imbalance, then as the higher rate battery approaches the bottom knee, it can suddenly start supplying less current. This could in theory cause it to drop out when the other battery(s) have plenty of capacity. If you are near the current limits, having a battery drop out due to SOC could result in a overcurrent cascade taking all the packs offline. Either due to BMS overcurrent protection, or physically blowing fuses.

Now, the wiring resistance can help to balance this out, as an increase in current from one battery will see increased voltage drop in the wiring, thus limiting the current imbalance in most cases. Assuming well balanced packs.

With regards to mixing different age/brand drop-ins in parallel. For lower current applications I don't see any harm, just be design for and be prepared for the eventuality of one dropping out during service. If you are mixing different capacities, Its a good idea to adjust your paralleling wiring according to capacity. So a 200Ah battery would get half the resistance of a 100AH battery. This serves to balance current flow at high rates (if present). Though each batteries internal resistance will help that. As currents exceed 0.2C, design differences in the packs could see a smaller battery delivering more than its share of current during the flat middle of the discharge curve, causing some odd behavior as it hits the knee. Not necessarily dangerous, but could be annoying if the system goes off line due to low voltage or over current with plenty of capacity left.

smoothJoey said:

Can you expand on that?
I would have thought that as long as the strings are well matched up to the aggregation point that you could expect close to 2x the draw.

Click to expand...

You sort of answered question. 'if well matched'. Battery discharge and charging profiles are dependent on their use history. Slight variations grow over successive cycling. To get perfect balance they must have same voltage at all drawn current levels, over all discharge or charging state of charge level, and little variation in temperatures between cells. This includes any interconnect straps and terminal connections. A little difference in effective series resistance upsets the balance. That is a lot of things to have perfectly matched.

This is why I am very leary of buying used cells with unknown and likely different use history. Putting used cells in parallel is high risk. If you must use used cells don't trap multiple series sub-string cells in parallel. Keep the series strings of single cells separate then parallel at top level. At least you can then keep tract of imbalance with current monitoring on each series leg.

Both of these last two posts by Luthj match my understanding/experience exactly.

One thing I would add is that the overcurrent cascade failure risk is reduced when there are more strings in play. Then, the contribution of a string that suddenly drops out can be distributed across more sibling strings with less of an impact.

Another way to build is to make sure every string can handle the full design load of the system. I would probably go that route in a two-string design, but it becomes really expensive as you scale up.

I’m about to move from six strings to eight strings, and the two new strings will have 3x the capacity of the old ones. It’s super convenient to already be on a multi-string topology when it comes time to make additive upgrades.

n+1 redundancy as applied to batteries.
Fascinating!

Yeah, if you are going with multiple parallel strings/batteries for a mission critical application, you should size to handle at least half your strings going down. This is typically how data centers do it. For very critical ones the UPS may be triply redundant with all three capable of handling loads. Though most would likely have just 2, with only critical servers connected to both.