Wiki Entry Review: BMS

Maast

Compulsive Tinkerer
Joined
Oct 31, 2019
Messages
770
Location
Washington State
I would add mention that an active balancers primary advantage is the ability to move much more charge around than what a passive balancers "burn off" can accomplish.
 

svetz

Works in theory! Practice? That's something else
Joined
Sep 20, 2019
Messages
4,902
Location
Key Largo
I would add mention that an active balancers primary advantage is the ability to move much more charge around than what a passive balancers "burn off" can accomplish.
Need some help with this one, a reference or some sort of comparison box around it for context. Otherwise it's too subjective.

For example, in my non-EE brain it immediately confuses me because resistors are dirt cheap and getting more energy burned off is just a matter of letting more watts/s burn off? So, couldn't any passive system easily match any active system?
 

svetz

Works in theory! Practice? That's something else
Joined
Sep 20, 2019
Messages
4,902
Location
Key Largo
After reading Steve's link I'm realizing the "when" to pick an active balancer isn't an easy thing.

Originally, I was thinking the primary benefit of an active balancer is saving energy (e.g., move energy instead of converting to heat). Pumping power uphill as active balancers do obviously takes energy and that also generates heat. As long as the active balancer is consuming less energy and generating less heat than a passive system, than you're saving energy.

Steve's link shows that active balancers are roughly 70% efficient and consume ~50 mW of standby power whereas passive are constrained by ohm's law.

So, for X watts of balancing on a passive system it's 1x, and for an active system it's .7x + 50 mW. The break-even point is 1x = .7x + 50, or 50 =.3x, or X=167 mW. If 0.01V represents a difference of 33 mW (average 18650 numbers for a 3300 mAh) then an active balancer would start to make sense if the cells were mismatched by .05 or more volts. That's pretty rough of course, in the link most of the examples had average waste heat under 1 watt with passive balancing (those probably represent well balanced packs) and for under a 1W you'd be hard pressed to recoup the costs.

Then it occurred to me what Will said was deeper than I thought. In a passive BMS the battery is still only as strong as the weakest cell. In a pack with well matched capacities you won't experience significant capacity losses.

But a battery with an active balancer isn't constrained by its weakest cell, because it works during the discharge and charge cycles it can pump power from the still strong cells back into the weakest cells. This is probably why pushing high amps around cells for active balancers is important (and relatively unimportant to passive balancers). This is also where the break-even point calculation above fell apart, because a passive system is only balancing at the end of the charge cycle so it's operational time is relatively short compared to the active balancer which is working all the time and therefore consuming power all the time. For example the 50 mW of standby power x 24 hours = 1.2 watt-hours; but you can't know the savings on active transfer without understanding the application and how mismatched the cells are.

So, I get Will's economic point that on a pack with well-matched cells an active balancer doesn't make sense...after all, the passive balancer doesn't burn much energy per cycle and there's no need to pump energy between cells during charging or discharging. I also get that adding another component is adding another failure point.

But what's still unclear is how mismatched a pack should be in order for an active balancer to make sense. The link is from 2010 and cites active balancers are about 10x the cost, possibly those prices have come down. From the link it looked like it made the most sense in heavy use industrial cases (e.g., public transportation, distributed power sources) that are extremely hard on batteries. Solar models are usually far gentler.
 

Maast

Compulsive Tinkerer
Joined
Oct 31, 2019
Messages
770
Location
Washington State
Need some help with this one, a reference or some sort of comparison box around it for context. Otherwise it's too subjective.

For example, in my non-EE brain it immediately confuses me because resistors are dirt cheap and getting more energy burned off is just a matter of letting more watts/s burn off? So, couldn't any passive system easily match any active system?
I've done experiments with active balancers, and they do indeed increase the effective capacity by boosting the low cells, but unless its low (10A or less) discharge rate the active balancers can't keep up. The end result was only a 100wh increase if I remember correctly -not much. However what I was referring to is the highest capacity passive balancer tops out at 1A of "burn off" where active balancers start there. The highest capacity active balancer I know of is a 10A JK BMS, also branded as a Heltec.

Once I corrected the capacity mismatches of the BYD modules to within 5-10ah on a 120AH cell the QNBBM balancer I was using kept things fairly even until I discharged too deep in the knees @ 2.95v on a 60A discharge for it to keep up and the cell voltages started to diverge. I posted a thread about my results about 6 months ago or so.

Where they're indespensible is on large banks with used or old cells with capacity mismatches. My anecdotal results show that you need a roughly a 1% capacity of the bank size in AH for the balancer.
 

snoobler

Solar Honey Badger
Joined
Jul 10, 2020
Messages
8,430
Location
HBR, AZ
I have a violently strong objection to the memory effect portion.

According to my understanding:

It does not strand capacity.
It alters the SoC to voltage response by... 3mV/cell.

Any concerns expressed were in the context of EV applications where SoC computations need to be precise.

From:


A co-author:

"The effect is in fact tiny: the relative deviation in voltage is just a few parts per thousand. "

Source paper:


1612935011951.png

Also of particular interest is Fig 4 (a) and (b), particularly the vertical axis scale.

Unfortunately, there are supplements that aren't included that sound very interesting.

IMHO, describing this as a memory effect is misleading based on the layperson's understanding and will only spread FUD. It should be summarized as "A study conducted in 2013 identified a phenomenon the authors labeled as a "memory effect" of LFP. This is not a memory effect as has been traditionally described for Nickel-based battery chemistries where cell/battery capacity is substantially reduced.

Failure to fully charge LFP results in an alteration of the voltage to SoC relationship by 0.003V/cell. This may slightly alter the target SoC when one is charging and discharging to selected target voltages and thus slightly reduce the capacity available between those target voltages, yet the cell/battery can still deliver its full capacity between rated voltage limits. Periodic charges to peak voltage will erase this discrepancy, but it will quickly return once cycling to reduced peak SoC resumes. For all but the most sensitive applications, "memory effect" of LFP can be completely ignored."
 

Hedges

I See Electromagnetic Fields!
Joined
Mar 28, 2020
Messages
8,329
Since you addressed high temp cutoff, low temp cutoff could also be added. Any LiFePO4 battery deployed in an environment that could see temperatures close or below 32° F needs low temp cutoff.

From the graphs and tables I've seen, a 32F/0C cutoff isn't sufficient.

The charge current that is allowed starts decreasing significantly above that temperature. A system which maxes out at a very low rate like 0.05C or 0.10C could be OK with 32F disconnect (so long as it also decreases current as SoC gets up to 90% or so; maybe CC/CV takes care of that?)

But one of the features of LiFePO4 is allowed high charge rate. If you have enough PV for 0.33C or 0.5C, either need to decrease that as temperature gets colder, or have a higher temperature disconnect than 32F.
 

Steve_S

Offgrid Cabineer, N.E. Ontario, Canada
Joined
Oct 29, 2019
Messages
5,402
Location
Rural NE Ontario Canada
I use QNBBM-8S on each battery pack, 2x-174 (Used LEV cells) & 2x EVE 280's, with Chargery BMS8T with V4.04 Firmware, ISO Board with External Power from Batt Terminals. They've gone through hard Thrash Tests which pushed the packs from 0%-100% and Back Hard & Fast to Slow and Easy and in various "combinations". Single packs with 225A Charge / Discharge to full bank and Low charge/discharge rates.

At first the cells would diverge quite a bit faster after reaching 3.400V and below 3.000V volts as well. But through the Thrashing Tests, the cells diverged a little less each time. After some tweaking (okay lots of tweaking and tinkering) I found a "Sweet Spot" by also using the Passive Balancer of the Chargery's.
Passive Balance ON for Charge & Storage, OFF during discharge. Start Passive @ 3.300, allowing for 20mv difference.
ON during Storage to discharge more when a runner cell hits HVD so the BMS will resume charging quicker once the cell settles and Hi Volts are transferred to Lo Volt cells by the Active Balancer.
I found that this in combo with the QNBBM, helps to control the divergence at both the Top & Bottom of the voltage Curve by hitting the "Runner Cells" allowing for a more even charge and higher intake of charge. IN general, throughout the Voltage Curve the cells diverge far less than without the Active Balancer as I can "see" the transfer from Hi to Lo cells. Average Difference during normal use 9-17mv. Passive really only helps for knocking down the runners allowing the other cells to gain more power before the BMS forces an HVD cut-off.

Memory Effect (long debunked for LFP) - A hangover from NimHi, NiCad and other rechargeables.
PLUS Numerous Scholarly Articles since 2018 to current disproving memory effect. Use Google-Foo.
 

ArthurEld

Photon Sorcerer
Joined
Aug 3, 2020
Messages
2,067
Location
Palm Harbor, Florida
LFP doesn't have the type of memory effect that the old batteries have.
The old batteries would lose capacity if you didn't completely discharge them all the way and charge them all the way once in a while.

But, to me the difference between state of charge and voltage is a memory effect. It's just not the same memory effect as the old batteries had.
The memory is which direction it came from and how far it came from that direction

By top balancing all of the cells remember that they went to the limit and that they are all coming from the same direction.

And if that isn't done, the states of charge can be so far off from each other that capacity is limited even more than it would be by the weakest cell.
 
Last edited:

svetz

Works in theory! Practice? That's something else
Joined
Sep 20, 2019
Messages
4,902
Location
Key Largo
First off, thank you for your feedback on this! Sorry I'm bit slow with the whole active balancing thing... zero experience with them so really appreciate it.

I've done experiments with active balancers...
<whoosh!>
That was the sound as most of your post went over my head! ;)

So, the first paragraph makes it sound like active balancers aren't very useful after all....? In my one and only pack build my passive BMS has teeny tiny resisters capable of burning mW ... certainly not handling amps ... but then the cells are well balanced, their C-Rate is about 3.3 Amps, and the resistors don't even get hot to the touch. So the thought of a 1 amp balancing burn is pretty amazing (I know your cells are 120 Ah, 36x mine)

Based on the second paragraph, would it be correct to add that active balancers may be warranted in packs with cells that have a greater than 8% mismatch or more (10 ah/120 ah = 8%)? And from the last paragraph, that the minimum current they can carry should be sized no less than 1% the C-Rate?
 

ArthurEld

Photon Sorcerer
Joined
Aug 3, 2020
Messages
2,067
Location
Palm Harbor, Florida
Active balancers are useful. Every time the cells get close to the knee the balancer evens out the state of charge as much as it can in the time it's there.
If I want to balance more, I just set my charger to float at whatever voltage I want to balance. It takes a long time but I just keep checking until they are within the range I want them to be. Not much effort just time.
 

svetz

Works in theory! Practice? That's something else
Joined
Sep 20, 2019
Messages
4,902
Location
Key Largo
I have a violently strong objection to the memory effect portion.
As well you should!

BMSes have no memory affect as far as I know. Fortunately, there's nothing in the BMS wiki entry on a memory affect. I suspect you're talking about the recent LiFePO4 entry, which has it's own thread so I'll respond to your post over there in a bit.
 
Last edited:

svetz

Works in theory! Practice? That's something else
Joined
Sep 20, 2019
Messages
4,902
Location
Key Largo
From the graphs and tables I've seen, a 32F/0C cutoff isn't sufficient.
Thanks for this feedback! I saw that in Will's interview with Battle Born, but couldn't see how to say it clearly and concisely while also providing guidance. The general adage of cutoff at 32° seemed simpler/less-confusing. Could you provide a link to the graphs/tables you mention? I'm thinking a reference to a more detailed follow-up would solve the dilemma.
 
  • Like
Reactions: Dzl

svetz

Works in theory! Practice? That's something else
Joined
Sep 20, 2019
Messages
4,902
Location
Key Largo
... I found a "Sweet Spot" by also using the Passive Balancer of the Chargery's.
Passive Balance ON for Charge & Storage, OFF during discharge. Start Passive @ 3.300, allowing for 20mv difference.
ON during Storage to discharge more when a runner cell hits HVD so the BMS will resume charging quicker once the cell settles and Hi Volts are transferred to Lo Volt cells by the Active Balancer.
I found that this in combo with the QNBBM, helps to control the divergence at both the Top & Bottom of the voltage Curve by hitting the "Runner Cells" allowing for a more even charge and higher intake of charge. IN general, throughout the Voltage Curve the cells diverge far less than without the Active Balancer as I can "see" the transfer from Hi to Lo cells. Average Difference during normal use 9-17mv. Passive really only helps for knocking down the runners allowing the other cells to gain more power before the BMS forces an HVD cut-off.
Will mentioned using passive in combination with active, but as I didn't understand it I left it out. I don't suppose you have a general reference to a when you should use both so I can take the cowards way out and just put in the link?

Memory Effect (long debunked for LFP) - A hangover from NimHi, NiCad and other rechargeables.
PLUS Numerous Scholarly Articles since 2018 to current disproving memory effect. Use Google-Foo.
I'll have to go through these... it'll take a while... I believe the original paper quoted was from 2010 and @snoobler's was 2013. So, if new articles from 2018 are debunking the original science that's great news! As with Snoobler's I'll link it up in the other thread as I review the tech articles.

Really appreciate you guys taking the time to share this wealth of information! Can't thank you enough!
 

Steve_S

Offgrid Cabineer, N.E. Ontario, Canada
Joined
Oct 29, 2019
Messages
5,402
Location
Rural NE Ontario Canada
From most of everything which I have read on the matter of using Passive & Active, it is STRONGLY discouraged as both can very easily result in conflict. I can indeed verify this to be the case when settings are not correct. Fortunately, "some" Smart BMS with Passive Balancing can be configured for their behaviours. Even how a Passive & Active system are wired into the pack will have an effect & counter effect. It is NOT without risks and potentials for major issues...

Again, the three weeks of Thrash & Bash was to answer just that.
During that process, I FRIED one Chargery BMS-8T and one QNBBM-8S which was a fairly expensive set of lessons but I "had to know" where edge cases and such were, I depend on my system for LIFE and I am not willing to risk unknows popping up out of the blue. It always & inevitable occurs during the worst possible times - you know, Murphy's laws Applied. Fortunely I have extra spare equipment on hand for emergency backup, some of which has now been sold off (new in box unused) for a hell'a'va deal.
 

svetz

Works in theory! Practice? That's something else
Joined
Sep 20, 2019
Messages
4,902
Location
Key Largo
...During that process, I FRIED one Chargery BMS-8T and one QNBBM-8S which was a fairly expensive set of lessons but I "had to know"...
Ouch! Okay, sounds like we shouldn't recommend the two systems in tandem. <sigh> The things you guys do for science!
 

Steve_S

Offgrid Cabineer, N.E. Ontario, Canada
Joined
Oct 29, 2019
Messages
5,402
Location
Rural NE Ontario Canada
Okay, sounds like we shouldn't recommend the two systems in tandem. <sigh>
Only if said Person is willing to Risk real money and goof it up... I do not suggest it because it is extremely easy to make one tiny oops to release scads of Magic Smoke in a hurry.
 

HRTKD

Boondocker
Joined
Apr 24, 2020
Messages
5,937
Location
Somewhere South of Denver
Is the purpose of the wiki entry to make people aware of the issues when selecting a BMS? Or, are we trying to make a comprehensive document that definitively answers all possible questions? For example, it's easy to say, "Don't charge below 32° F". At the very least, that's a safe statement. Perhaps a link to the thread that explains how the charging rate should be decreased as the temperatures fall, would then be good enough to provide further clarity.
 

Hedges

I See Electromagnetic Fields!
Joined
Mar 28, 2020
Messages
8,329
[From the graphs and tables I've seen, a 32F/0C cutoff isn't sufficient.]

Thanks for this feedback! I saw that in Will's interview with Battle Born, but couldn't see how to say it clearly and concisely while also providing guidance. The general adage of cutoff at 32° seemed simpler/less-confusing. Could you provide a link to the graphs/tables you mention? I'm thinking a reference to a more detailed follow-up would solve the dilemma.

Simpler, but not always sufficient.
Unfortunately we don't have graphs/tables for each brand. Ideally would have at least a few data points to characterize the curve.
Then a dumb system could remain within rectangular limits, and a smart system could taper off charge current when approaching the edges.



 

svetz

Works in theory! Practice? That's something else
Joined
Sep 20, 2019
Messages
4,902
Location
Key Largo
...Simpler, but not always sufficient....
Sweet! No time now, but will swing back later and beef up the entry with your links! Thank you!!!

... are we trying to make a comprehensive document that definitively answers all possible questions...
Definitely not! Anything that big no one would read. Ideally it's just the most important stuff with links to follow-ups, not restricted to BMS selection alone. But hey, your forums if you think it needs more or less let's do it! ;-)
 
Top