18 cell "48v" Bank

[TheTerribleTriplet]

YES!!! YES!!! YES!!! YES!!!

The best Delta since the new cells were installed at this point:

1. NO bulbs were used.
2. All settings as per Andy, him also using big ah banks.
3. And Charge when Balancing was ON the whole time since last night.

Note: I have to watch it like a hawk from certain voltages … have this aversion for that thing called “DC Ripple” when you take your eyes off for a second.

The even BETTER NEWS!!!

NOW it can balance properly as it should, Victron pulled back as it always does - yes, running with no interface to the Cerbo.

Note: I still watch it, the low wattages Victron still passes, too low for the BMS to show.

At least I can switch on the geyser/oven on to keep far fr away from 3.6v … so yeah, still tied to the chair, but can walk away having switched the oven on with the door open.

Lets me see where this goes … verifying Andys result can now start.

Ps. Also stopped the NodeRED flow to allow the batts to recharge, at 100% I have to set the bank to Keep Charged so that the MPPTs don’t stop when left on Optimized.

Still going to get that Neey.

[TheTerribleTriplet]

And this is how it looks when Charge Balance ON when discharging. I can’t control it as the feature cannot be switched OFF.

Small Load:

BIG Load:

Small Load again:

So,at this point:
When discharging the passive balancer “burns off some heat” by trying the keep the cells within a 0.005v difference. Each “brake” results in energy retained, during which time other cells pick up the the “load”, as is reasonably deduced by what Plonk/Phil said above.

But now I wonder:
One:) The software determines when to balance a cell when discharging. We don’t what that logic is, or is it the charging logic in reverse … I cannot test that. Trust but verify.
Two:) 270mA is at play here, minuscule effect on a 280ah cell. Look at how fast the cells shoot out at <50w, BMS does not even register that low value, and the balancer “loses” control. 5a, that is a HUGE improvement, but still sometimes a challenge.
Three:) So if we agree that “this is not good” (Andy), to balance whilst discharging, as it does make sense, why did Andy see his results on 2 batts right next to each other, one Charge Balance ON, the other OFF?

Then it dawned on me … what do the BMS’es in the brand names batts do? You have no way to test, or even see yourself, the trust but verify again.
And what are their balancing current, mA or A?
Active or Passive?

Also realized, lithium batteries for common use are like when mankind discovered fire.
Lead acid is already at when we flew to the moon and back.

We don’t have verified facts, bar Andy.

FYI: Most active balancers, as per Andy’s videos, transfer energy from the highest to the lowest cell, yes CELL, just like balancing from the “outside” working towards the “center”, the happy place, cell by cell.

I’m going to run this Charge Balance ON for a few days.

[plonkster]

I can’t tell you about your own BMS, what I can tell you is that while the resistors does “burn off” a bit of energy, the intent is really to bypass a little bit of energy while a current is flowing, which does waste a little bit of it through heat. Passive balancers aren’t terribly efficient. If you’re firing them while there is no current flowing, you’re literally just turning energy into heat, and you aren’t really balancing at all (the low cells aren’t being pulled up), you’re literally just discharging the high cells. It may help to get in balance faster, but most batteries will rather stop the charge at this point and leave it as a job for tomorrow.

Second thing to mention: Interleaving. It doesn’t take a genius to realise that if all 18 cells are high, and the balancer turns on all 18 bypass FETs simultaneously… you’re going to see some smoke. BMSes therefore interleave the switching. Two adjacent FETs are never fired at the same time. That means if you have two adjacent high cells, it takes double as long to get them sorted.

Take a look at the pictures you posted. It never balances two adjacent cells. Neat init?

Edit: I suddenly realised something. Passive balancing is like Bubble Sort… but with coulombs. Active balancing is more like quick sort, you find two items that needs to be swapped (or in this case, have some charge transferred).

Bubble sort is O(n^2), order n-squared. The time it takes to do is proportional to the square of the number of items involved. Quick sort is O(log(n)), proportional to the log of the items. Mathematically, it means a few things. 1) The time taken to balance an 18-cell battery is 18^2/16^2 = 3 times longer than balancing a 16-cell battery, and 2) active balancers is WAY faster.

[TheTerribleTriplet]

@plonkster

The second thing to mention: Interleaving. It doesn’t take a genius to realize that if all 18 cells are high, and the balancer turns on all 18 bypass FETs simultaneously… you’re going to see some smoke. BMSes, therefore, interleave the switching. Two adjacent FETs are never fired at the same time. That means if you have two adjacent high cells, it takes double as long to get them sorted.

AHA, I always wondered why.
And the active balancer, when I saw “pictures” via Andy, suddenly made 100% sense to invest in, MORE SO with a big ah bank …

Knew from the start that mA is nowhere near enough. Actually, knew it with the first bank already.

Knew this is going to take a long while to get the bank settled and then to keep it balanced.

Open eyes I went in into it, sharing my non-scientific experiences here.

@plonkster

Mathematically, it means a few things. 1) The time taken to balance an 18-cell battery is 18^2/16^2 = 3 times longer than balancing a 16-cell battery, and 2) active balancers is WAY faster.

Let me be honest, the maths went like “whoosh” over my head, gone with the wind I just let go.

The part in black, now THAT made instant sense… something I never even knew to consider. Now I do.

Conclusion I’m reaching:
Neey with my big ass BMS, it is a MUST! … seeing I have 3 spares as I tend to blow things up, Note, happens fewer and fewer FWIW.

The new JBD with a Heltec active balancer is a perfect match going forward.

The other Active balancer BMSs, all in one units, on big ah cells, tend to also struggle to keep up, but still FAR better than a passive balancer.

Today I can say, an active balancer is a must on BIG aH cells. BMS technology is still in it’s infancy, as the world learns more, China improving their products.

Jinne, this is fun!
18 cell bank, then adding 3 cells after the fact, seeing in real life how it works, watching Andy doing similar, trust but verify, and more importantly, getting it all to work together nicely with insights from this forum to boot.

As Andy repeatedly says: We are here to learn.

Love it!!!

[plonkster]

Honestly, the math oversimplifies it. It is a worst case scenario. Bubble-sorting an already sorted list is very fast. What I am saying is that if you started with two random collection of cells (one a 16-cell bank, the other an 18-cell bank) the 18-cell one would take three times longer to balance with a passive balancer, despite being only two more cells.

In your case, you started with a bank that was already mostly balanced, and tacked the cells onto the end. So the impact would have been less severe.

When someone tells me the math went “woosh”, that’s when I realise that maybe the drilling in of all that time complexity stuff wasn’t for naught. I mean, someone has to think of this sh*t

Edit: On the topic of sorting. There is a algorithm called insertion sort. It’s like returning books to the correct shelf in a library. You walk until you find the exact spot where it must be, and then insert it into the stack. Repeat this for every book. This is insanely inefficient: In a worst case scenario, you have to compare each of n books with on average n/2 others, that is to say, the time complexity is O(n^2) (the big o just means “in the order of”). But if the stack is already sorted (as books in libraries usually are), the action of inserting one book (or a few books) into their locations is actually very efficient. It is worst case O(n) and on average n/2.

That’s the thing about algorithms, and in general about how to do things. Often it is about picking the best method for the circumstances, not the best method for ALL circumstances.

[justinschoeman]

@plonkster

I can’t tell you about your own BMS, what I can tell you is that while the resistors does “burn off” a bit of energy, the intent is really to bypass a little bit of energy while a current is flowing, which does waste a little bit of it through heat. Passive balancers aren’t terribly efficient. If you’re firing them while there is no current flowing, you’re literally just turning energy into heat, and you aren’t really balancing at all (the low cells aren’t being pulled up), you’re literally just discharging the high cells. It may help to get in balance faster, but most batteries will rather stop the charge at this point and leave it as a job for tomorrow.

Second thing to mention: Interleaving. It doesn’t take a genius to realise that if all 18 cells are high, and the balancer turns on all 18 bypass FETs simultaneously… you’re going to see some smoke. BMSes therefore interleave the switching. Two adjacent FETs are never fired at the same time. That means if you have two adjacent high cells, it takes double as long to get them sorted.

This is not entirely accurate. You can turn those FETs on at any time, and the effect will be almost identical (slightly slower when not under charge, as the voltage across the resistor - and therefore balance current - will be slightly lower).

The real reason you only balance when charging, is that it gives you the best measurement of the state of charge of the cell. When discharging, voltage is more dependent on load than on state of charge.

Also, passive balancers do balance adjacent cells (at least my Daly does). In fact to build a true passive balancer, just remove the FETs and replace them with a short circuit. This will effectively place a resistive ladder across the cells, and they will balance (although much slower and much more inefficient).

In fact, this is how we used to balance packs in the good old days, before active balancers became a thing.

Back in those days, a ‘passive balancer’ was just a resistive ladder. An ‘active balancer’ was a resistive ladder with switched resistors. Energy transfer balancers (what we now call ‘active balancers’) were just about unheard of.

[plonkster]

@justinschoeman

This is not entirely accurate

In fact… it is downright wrong in some respects. Thanks for the insights!

[TheTerribleTriplet]

Blow me down, I got all of that. Thanks guys.

Just goes back to the “feeling” I got the last few days, gutfeel long before that, me being a “picture” person, jip, that is how I remember/understand things, that this thing called Lithium, the balancing, all of it, “we have learned to make fire”, as it is a new general “tech” today with a LOT of permutations.

However, getting it all down pat, won’t take anywhere near as long from making fire to landing on the moon.

@justinschoeman

… a true passive balancer, …

At one stage an idea was discussed to build a thing (saw the OEM idea at Andy) that when a cell /s needs balancing, it automatically switches on the correlating 16/18 Halogen 12v car bulbs based on the cell’s volt reading in the BMS when you charge balancing, off when not.

But that would only be used one time, lots of effort goes into building the thing, software development for one-time use, my case, twice.

@plonkster

In your case, you started with a bank that was already mostly balanced and tacked the cells onto the end. So the impact would have been less severe.

I used the Victron settings to top balance the 15 cells from 3.45v to 3.6v in a week. Was so chaffed when that came to mind.

I did not see the EV-Peaks limitation heading straight for me.

Two things I learned here:

1. If you use an EV-Peak, bank must be <250ah cells. Sorted.
2. Thought to late, you can jippo the charging to 6a by using the LiLo setting, setting the ah on the EV-Peak to a lower aH setting so that EV-Peak thinks it has balanced a LiLo bank, at ±3.8v if memory serves, so that it stops at the expect 3.65v, that you a 280ah bank top balanced.

BUT!!! You better start all the cells at the EXACT same level, or you will blow it all to smithereens when 3 or 15 cells go ballistic.

At least I have an idea to do when I need to balance a bank VERY fast in an emergency WHEN they are brand new, batch matched, same 50% capacity … or buy a proper charger, .

Unless anyone can draw me a “picture” I can grasp why not, I seriously need a big-ass active balancer on my big ass cells.

[TheTerribleTriplet]

Update … with 99% less effort from before, Charge Balance OFF, here I sit today.

Now I drop the charge amps, currently at 2A, with 0-164w going in.
I watch the Delta. Been saying for years, keep the Watch the Delta, if it increases, slow down the charge current each time it increases.

Why, right now at 2A setting, 0 - 164w are still currently going in.

IF a cell/cells still wanna fight me … they get “The Bulbs”.

It that becomes too much effort, I switch off an MPPT. Going to try that now.

[TheTerribleTriplet]

Still balancing … getting there.

Currently draining a bit, then gooi-ing 3kw at it, drain again, gooi again.
Victron throttle back perfectly.

Now at 61.2v … once balanced here, up 0.5v, balancing again till I reach 62.2v (3.45v).
System is set ot Keep Charged and balancing set at 3.4v

On another note, a question:
If a passive BMS balancer, per cell, when CHARGING, turns the charge into heat, what happens when the bank DISCHARGES with balancing on?

It turns the higher cells balancer on, right, generating heat?

Correct?

So the higher cells drain minuscule faster … not so?

[Phil.g00]

Think of the lanes on an oval athletic track. The runners start in different places, but the finish line is the same for all runners.
Balancing when discharging is analogous to allowing the runners to change lanes so they all have the same starting line as well.
It is a no-no.

[TheTerribleTriplet]

I know, we spoke before on this.

But Andy’s test over 1-2 weeks between one with Balacner on, the other off, made me wonder.

So far, I don’t see the "oval’ track in daily viewing of the cell’s behavior.

Would have seen the volts getting worse.
They did not, actually looked better and better week after week with the balancer on all the time, from 3.2v

Currently at 3.4v before the balancer starts.

I get the suspicion that we don’t know the exact reactions in the cells/the BMS’es.

Like bulging, a scientist in Germany explained, the cells ARE going to bulge, you cannot stop that at the molecular level, unless you operate in a vacuum.

So bulging is going to happen, and how bad depends on the materials used, the purity as well as the manufacturing process.

[Phil.g00]

It is not the voltages you want equal when you are discharging. It is the actual SOC of the cells that must be equal.
The middle of the curve SOC is invisible to you, and the cell voltage is not a reliable indicator.
Where the voltage becomes a reliable indicator is as the cell voltage starts to climb.

So left alone, without balancing, each cell sees identical through-current, and the inter-cell SOC drift will be minimal, regardless of the individual cell voltage.
Balancing based on cell voltages at the same time ruins this arrangement.
The goal is to that they all arrive at a 100% state of charge together.

[TheTerribleTriplet]

@Phil.g00

The goal is to that they all arrive at a 100% state of charge together.

Correct. Keep in mind, I’m Top Balancing a unbalanced bank ruing the worst ever LS SA has experienced.

No time to take it apart and do it properly.

Hence the opportunity to witness what is actually happening.

Like here, actually up to about 3.40v already, the cells are perfectly in balance, getting better week after week, with balancing up and down.

The original 15 was balanced at 3.5v, the additions at 3.35v.

It is working, the balancing up and down, for a nood on Lifepo4 science.

[TheTerribleTriplet]

In other news!!!

My inverter is just under 4 years old.

Heard a whisper along the grapevine, and tried it today …

I extended my 4 year old Victron inverters existing warranty by another 5 years, for a couple of thousand ronts.

(drop mike)

[Phil.g00]

Way to spend £65.

[mmaritz]

@TheTerribleTriplet

I extended my 4 year old Victron inverters existing warranty by another 5 years, for a couple of thousand ronts.

Send us details on how please…

[TheTerribleTriplet]

Contact your installer/Victron dealer.

[TheTerribleTriplet]

Since yesterday:
Keep Charged, weather
Set to 3.4v instead of 3.45v
Balancing starts at 3.45v now.

Every now and then I draw big loads for a few minutes, geysers/oven, and drain it a bit.
Recharge at max amps
LS, recharged again afterward

Victron slows down nicely to here, where it then does the Victron thing.
Charges at ±20w to ±125w drain a bit, couple of watts, charge a bit, drain a bit, BMS does not show anything, too low for it.

When I get ±0.004v Delta, will increase the volts to 3.45v and start again.

Special Mention @Louisvdw driver, when I went to “approve the hospital services”, saved my bacon during that time. It stops the small charge dead on 3.5v, inverter is set on Inverter, not Inverter and Charge.

So have that if I need to be “away” again before the bank is Top Balanced.

Conclusion:
And that is how one Top Balance a huge bank when LS has no mercy.

Also means, if I ever have to install cell 19:
Using EV-Peak, set it to Lilo setting, drop the AH it has to charge to, and boost the spare cell carefully to 3.5v or some such. That will shorten the balancing period immensely.

[TheTerribleTriplet]

Some more insurance till 2029/12.