diy solar

diy solar

Question RE: Severity of Failure to Top-Balance

eagleray

ex-VR programmer, now DIY electric boat Youtuber
Joined
Dec 26, 2022
Messages
31
Location
Lake Havasu City
I was curious how my 4 CHINS LiFePO4 12V 100AH (non-smart) batteries would do in series without a top-balance. I know I should always top-balance them in the field, but on my test bench, I wanted to know how severe their imbalance would be if I didn't.

Before each test run, I charged each battery individually at a constant 18 amps until each BMS tripped overcharge protection. Post-charge, I deliberately did not put any absorption charge into any of them. After charge, I let them sit for at least 24 hours each to arrive at resting voltage (each landed at ~13.35V).

My test parameters:
Discharge 4 12.8V 100AH batteries in series until one BMS trips overdischarge protection
Load:
Test 1: ~15 amps (Hangkai 5HP marine motor, full throttle)
Test 2: ~1 amp (also Hangkai 5HP, lowest throttle)

I still have yet to top-balance them, but so far I've found that one out of the four batteries fully disrcharges much faster than the others. I'm actually not sure if the runt was the same battery in both tests because, like a true doofus, I neglected to label each battery to distinguish them (a mistake I have now fixed for the upcoming 3rd test).

In the 2nd test, I have 10.7V on the tripped battery and 12.8V on the other three. That's a sizeable difference, I think... something like 25% capacity remaining in 75% of the bank that becomes unusable.

So my question is this:

Is this giant imbalance from...
...failure to include absorption charge?
...failure to top-balance?
...faulty cell(s) in the runt battery?

In further tests, I'm hopefully going to answer my own question, but I'm wondering if the answer is just obvious to someone here who has lots of experience with LiFePO4.
 
The goal is to always operate within the operating range of the BMS.

In most cases, the danger is if one of the batteries triggers OVP while charging. This can be very disruptive and often prevents charging to a typical 48V system voltage and leaves unused capacity on the table.

You've discovered the issue on the other end.

The likely cause is simply imbalance in the battery as received. 25% is pretty big, so I wouldn't rule out the possibility that the battery has defective cells.

I would start by fully charging each to 14.4. Then parallel them and charge them to 14.4V and hold for 24 hours. The idea here is to provide plenty of opportunity for the battery to balance itself.

Repeat your test. If any battery fails to meet capacity spec, initiate an RMA.
 
Thanks for the tips!

I'm going to top-balance them after this round of intentionally imbalanced tests is done. I have my doubts that top-balancing will make that gap go away, but I'll report back.

It's curious that they all get charged to the same peak voltage, and they all fall to the same resting voltage, yet they perform very differently. I guess it's not too surprising, since we know LiFePO4 voltage never gives a reliable indication of capacity.

RE: charge voltage, constant current was set to 14.4 for each. Once the power supply gets them to 14.4, it quickly fails to be able to maintain constant current after that (within 60 seconds). It runs away to over 20 amps and trips overcharge protection. I don't intend to use the BMS as a charge cutoff in the future, but since it's the non-smart version, I was curious as to its default cutoff settings and I wanted to confirm cutoff is working for each. The CHINS BMS cuts off somewhere between 20-25 amps.

I'm also curious if anyone has tested their LiFePO4s with absorption, then without, to see how much absorption affects balance. For these first two tests, I tested non-top-balance and no absorption at the same time, but I think I'd like to test top-balancing and absorption separately. I have a feeling the imbalance may be coming partially from lack of top-balance and partially from lack of absorption.

And of course, once I confirm if the runt is the same battery every time, then I'll have lots of evidence showing three good batteries performing very similarly regardless of top-balance or absorption, which would suggest the runt cells are faulty. If, on the other hand, the runt is a different battery each time, the culprit has got to be lack of top-balance or lack or absorption or a combination of both.

I'm also wondering if placement in the series string has anything to do with performance. In test 1, the runt was 2nd from main negative, and in test 2, the runt was 3rd from main negative. That could be coincidental though, because I may have moved the same battery from position 2 to position 3. I'll track the configuration more carefully from now on.
 
Probably need to dismantle the cases and measure/monitor cell voltage when packs are 14.2-14.4V if want to know what is going on. ie. Non smart BMS on a closed system leads to battery capacity loss with low quality cells because you are flying by the seat of your pants because you can't measure cell voltages . You can't tell if cells are out of balance or balancing is working if you can't measure cell voltages.
 
Thanks for the tips!

I'm going to top-balance them after this round of intentionally imbalanced tests is done. I have my doubts that top-balancing will make that gap go away, but I'll report back.

It's curious that they all get charged to the same peak voltage, and they all fall to the same resting voltage, yet they perform very differently. I guess it's not too surprising, since we know LiFePO4 voltage never gives a reliable indication of capacity.

RE: charge voltage, constant current was set to 14.4 for each. Once the power supply gets them to 14.4, it quickly fails to be able to maintain constant current after that (within 60 seconds). It runs away to over 20 amps and trips overcharge protection. I don't intend to use the BMS as a charge cutoff in the future, but since it's the non-smart version, I was curious as to its default cutoff settings and I wanted to confirm cutoff is working for each. The CHINS BMS cuts off somewhere between 20-25 amps.

I'm also curious if anyone has tested their LiFePO4s with absorption, then without, to see how much absorption affects balance. For these first two tests, I tested non-top-balance and no absorption at the same time, but I think I'd like to test top-balancing and absorption separately. I have a feeling the imbalance may be coming partially from lack of top-balance and partially from lack of absorption.

And of course, once I confirm if the runt is the same battery every time, then I'll have lots of evidence showing three good batteries performing very similarly regardless of top-balance or absorption, which would suggest the runt cells are faulty. If, on the other hand, the runt is a different battery each time, the culprit has got to be lack of top-balance or lack or absorption or a combination of both.

I'm also wondering if placement in the series string has anything to do with performance. In test 1, the runt was 2nd from main negative, and in test 2, the runt was 3rd from main negative. That could be coincidental though, because I may have moved the same battery from position 2 to position 3. I'll track the configuration more carefully from now on.
Hi, any update on this? I’m curious if top balancing solved your problem or if you discovered a bad cell or battery.
 
From: https://marinehowto.com/lifepo4-batteries-on-boats/

Dangers of load dump's due to BMS disconnects...

Protecting the Alternator from a BMS Disconnect (LOAD DUMP):

One of the biggest obstacles of LFP is protecting charge sources such as alternators and inverter/chargers from a BMS disconnect. If the charge source is pumping out amperage, and the BMS decides to disconnect, the outcome can be fatal to your alternator. On many boats the DC bus is also connected to the charge bus and this means the voltage transient that is caused by the BMS/LFP battery disconnecting, causes a massive voltage transient. This transient can not only blow the rectifier in the alternator but can also destroy expensive navigation equipment.

Ideally your BMS should feature a charge control circuit. The OPE-Li3 batteries feature what Lithionics calls the FCC or (Field Control Circuit). This circuit disables the charge sources well before the battery is actually disconnected. The well before part here is critical as the large magnetic field needs to be de-powered before the load is disconnected. For a Balmar regulator this shutdown should be the Red wire in the regulator harness. Shutting down the brown or ignition wire is not fast enough to protect the alternator from a BMS load dump.

Unfortunately drop-in batteries and most aftermarket BMS systems are designed for electric car use and don’t always have the ability to shut charging down before the BMS open-circuits the battery.

Fortunately Sterling Power now manufactures a device called an Alternator Protection Device. It is used to prevent a load dump from causing a massive voltage transient. It is our belief that any system that does not have a way to shut charging down correctly needs, at a bare minimum, an APD.
A LiFePO4 drop-in batteries internal BMS can disconnect for the following reasons:

Cell Over Voltage
Cell Under Voltage
Cell Temperature
BMS Temperature
BMS Current Limits Exceeded

If there is a bad cell, temperature too high, too much charge current, a glitch in the charging voltage settings or a cell imbalance issue creating an over-voltage condition, the battery will physically disconnect itself from the vessel. Most drop-in LiFePO4 batteries can disconnect themselves with no advanced warning to the vessel occupants. This is called a load disconnect or load dump.

A load disconnect or load dump is something a lead acid battery can’t physically do on its own, so this, by definition makes “drop-in” LFP batteries not so “drop-in” because we now need a ways to ensure our alternator or inverter/charger is not suffering load dumps. Of course you don’t need to take our word for it. This is from Balmar, the worlds largest specialty marine performance alternator and regulator manufacturer.

Sure, many an owner has moved a battery switch with the alternator charging and had the destroyed alternator to show for it but the battery did not do this without warning, and the owner made a simple, and often fatal to the alternator, mistake. If a BMS disconnect / load-dump occurs, when charging with an alternator, or even a large transformer based inverter/charger, the resulting *voltage transient, can damage the charge source and also what ever is connected to the DC bus/system such as sensitive marine electronics.

*Voltage Transient – What occurs when a charge source such as an alternator is suddenly disconnected from the load (battery). The current now has nowhere to go sending the voltage through the roof. When the load (battery) is suddenly disconnected the voltage skyrockets to damaging levels in milliseconds.


How voltage drop murders charging speed:

At the charger end the charger sees 14.0V and enters CV mode or constant voltage mode. The charger now begins limiting voltage and current by controlling the output of the power supply, so as to not over shoot 14.0V. The problem is at the battery end, the voltage is below 13.6V and only so much current can flow into the battery at 13.6V, even an LFP bank. Voltage is the pressure that allows the charge current to flow into the batteries and LFP banks are not Ohm’s law exempt.

Attain a limiting voltage too early, due to voltage drop, and you have just extended your charging times and will have a longer current taper to get to 100% SOC, just like lead acid!

Dedicated voltage sensing at the battery terminals is critical to FAST CHARGING PERFORMANCE. If you use a generator to power an AC charger proper voltage sensing means less generator run time. If a charger or inverter/charger does not offer you this option please BUY ONE THAT DOES!

The Victron Inverter / Chargers represent an excellent value in an LFP capable inverter/charger, especially one that has dedicated voltage sensing leads. By the time you are done with the Outback, by adding FLEXNET DC (allows for volt sensing) and the MATE (remote control), you are well in excess of the cost of a Victron Multi-Plus Combi. Course if you are in the US the Outback is a US company and supports US jobs.

Choose your AC charger carefully.
0.4C Charge Rate & 12v 100Ah Winston Pack

Okay you’ve heard me discuss how to safely charge these batteries when used as house banks at factional “C” usage, and here’s a prime example of what I am talking about.

As we can see in this image the battery has hit full at just 13.88V (pack voltage) with a .4C charge rate. A .4C charge means a 40A charge on a 100Ah battery pack. Anything above this voltage point is technically over charging the battery, because the charging is done. If you stopped at this 13.9V level, and tested this pack for Ah capacity, you would see 99.5% to 100% of the capacity. I know this because I have conducted these tests, in our shops lab, many times.

TIP: Charge rate also plays a role in when your batteries are full. It’s not just voltage.

This over-charge can easily be denoted by the abrupt hockey stick rise in voltage once the cells hit “full”..

CONSIDERATIONS: On a DIY built bank, with impeccably matched cells, the only benefit to charging to voltages above 13.8V – 14.2V at a .3C to .5C charge rate, is a slightly shorter current taper at the top end of charge. On this 400Ah bank I allow the current to drop to 10A at 13.8V with a .35C charge rate before deeming the bank full enough to reset the Ah counter. Stopping the charge at 13.8V and ≤10A nets over 400Ah of capacity from these 400Ah cells.

Maximum Peak Charging Voltage: 14.2V to 14.6V (*No absorption at all – STOP once this voltage is attained)

*If you have drop-in batteries, or cells you don’t know match extremely well, you will want to charge into the “balancing range” with each cycle and hold this for the manufacturers specified duration.

Optimal Charging Voltage: 13.8V to 14.0V (*for a DIY built bank with impeccably matched cells)

Why do I suggest this? Take a look at what happens to the cell voltages as they get into the upper knee in this image. We have the lowest cell at 3.69V and the top cell at 3.81V which is now into the danger zone. The pack voltage may still look okay but we are now cutting into the cycle life of a couple of the cells by over charging them. While I have known, from testing, that lower charge voltages work fine for off-grid and fractional C use, and are arguably safer, research is finally coming out to back this up and to also shown that higher voltages lead to shorter cycling life too.

What Has Industry Learned?

Tesla and other Li battery research institutions have now been able to show that regularly pushing Li chemistry cells to high charging voltages results in the build up of electrolyte oxidation by-products which adhere to the negative plate. This eventually leads to a shut down of the cells. It has been shown that higher charge voltages, with all Li chemistries, results in shorter cycle life. Essentially higher charge voltages result in more electrolyte oxidation clogging the negative plate. While the capacity may look good for a period of time the cells eventually fall of the proverbial cliff. In NMC cells (LiNiMnCoO2) cycle life degradation was accelerated as charge voltages were pushed higher. Testing showed that a max charge voltage of 4.20VPC (these are not LFP) showed very little capacity fade where as a max charge voltage of 4.35VPC resulted in less than 200 cycles and a charge voltage of 4.45VPC resulted in less than 60 cycles. It is not however just max voltages that affect the cells it is time at voltage. In the tests above the cells were simply charge “TO” the upper voltage but with the lead acid chargers we use voltages are “HELD” at a steady voltage for a period of time. In the marine environment, in order to compensate and not over-stress the cells with CV charging, we can simply lower the max charge voltage and this can help accommodate the cells, if they are well matched to begin with. This serves to allow for the CV (constant voltage) stage to be safer for a slightly longer duration.

As you push into the upper knee the cells can rapidly run out of balance as one cell becomes more full faster than another. This is why cell matching is critical to any LFP battery build. As cells age Coulombic efficiency can change, especially when over-stressed by using high charge voltages and CC/CV charging equipment. The actual cell to cell capacity can also change as can internal resistance. By staying out of both knee ranges, voltage wise, the cells tend to cycle up and down with very little voltage drift. Regularly pushing into the upper knee often creates a need for more cell balancing and many of the BMS companies pray on this.
 
From:


Cell Balancing
Cell balancing is an extremely important aspect of LFP banks. When you have lead acid batteries in series they can be purposely over charged/equalized to a 15.5V pack voltage and they will, in a sense, self balance. With LFP banks this will not happen due to the knee ranges. As a cell becomes full the voltage all of a sudden skyrockets and the cells need to be in balance in order to charge and discharge at matched voltages.

TOP BALANCE vs. BOTTOM BALANCE:
There is much controversy over top vs. bottom balance mostly due to confusion over differing uses.

BOTTOM BALANCE:
A bottom balance simply means the cells are balanced at the lowest “safe” voltage and all cells will converge and match exactly at say 2.75 VPC. In the EV world bottom balancing is almost always the preferred method, and makes the most sense. With high loads, and frequent opportunities to completely drain the bank, a bottom balance is critical with an EV pack. In an EV the car is then brought back to the garage and charged with ONE charge source.

Bottom Balance:
1-Discharge cell using a 20-30A load to 2.50V
2-Let the cell rest at room temp for 24 hours and allow voltage to rebound
3-The cell will now be resting somewhere between 2.75V and 2.85V
4-Apply the load and stop discharging at exactly 2.65V
5-Allow voltage to recover for about 6 hours
6-Repeat load discharge to 2.65V until the resting stable voltage of each and every cell is 2.75V
7-As you get closer and closer to resting voltage of 2.750V a small resistor can be used as opposed to the large load.
Once all cells rest at 2.750V and stay there the cells are bottom balanced.

NOTE: A guy recently dropped off 4 cells he was having trouble “balancing”. He was attempting a bottom balance and intending on using these for fractional “C” use stopping at 70% DOD.. He had spent countless hours trying to bottom balance these cells, and he did.

So what’s the problem? The problem is that at a 14.0V pack voltage he had one cell at 3.65V and one cell still at 3.380V!!!! His cells tested at varying capacities and thus the cell with the lowest capacity was firing into the upper knee sooner than the rest, even at a pack charge voltage of 14.0V. These were cells with an absolute MAX cell voltage of 3.600V. With a bottom balance and used cells (I don’t suggest buying used cells) he was sending one cell into the dangerous upper knee even at just a 14.0V charge rate. I conducted a top balance for him and the cells now all remain well balanced at the upper charging voltage range. On the low end one cell will still fall off the cliff early, but at 70% DOD that does not happen.

TOP BALANCING:
On boats we have multiple charge sources, shore charger, alternator, solar, wind, hydro or even hydrogen fuel cells. Our risk of cell imbalance is more pronounced at the top end rather than the bottom end. We run a much higher risk of over charging imbalanced cells than we do by over discharging, like the electric vehicle (EV) guys do, but it can still be a risk.. For off-grid / marine use top balancing is quite often the preferred method so the cells converge or are in excellent balance at the top, when fully charged, rather than when dead or fully discharged…

In theory the BMS would always protect the cells at either the bottom or the top end but keeping the cells well balanced ensures an extra level of protection, just as keeping charging voltages out of the upper knee range does. Don’t discharge below 80% DOD and have a max charge voltage of 3.5VPC / 14.0V for a 12V bank, and your cells will be very happy.


Balancing – Wire The Cells In Parallel

2010: As mentioned, I first charged these cells, INDIVIDUALLY, to 3.75VPC and X current taper. The bench-top power supply allows you to set the voltage to 3.XX and let the cell become “full” at 3.XX VPC. For these cells, based on the data available at the time, late 2010, I held the voltage at 3.75V and allowed the current to tail off to 20A then stopped charging and moved onto the next cell.

Within seconds of wiring these in parallel only 0.59A was moving between cells which means the balance to 3.75VPC was pretty close.. Leaving them in parallel will get them in closer balance but this can take lots & lots of time.


Updated Cell Balance Process: Parallel Step-Method Top Balance

My goal when balancing cells is always the following: Keep the cells in the upper-knee for the shortest amount of time and still net a perfect balance

Trough testing and experimenting with numerous balancing processes I’ve found the “parallel step-method top balance” (PSMTB) has proven to be the absolute fastest method that also keeps the cells in the upper-knee the shortest. This means less upper-knee time for the cells. You will need a variable power supply capable of low voltage (3.6V) to do this. You will also want a model with the highest amperage you can source. Keep in mind that when we wire the cells in parallel the bank capacity grows tremendously. Four 400Ah cells become a 1600Ah 3.2V pack! Getting to 3.40V will take quite some time! The key with the PSMTB come from the fact that the cells are essentially full when you get to 3.40V and 0A. This 3.40V threshold is a perfectly safe voltage for the cells so no matter how long it takes to get there will not be causing damage to the cells. Once at 3.40V this means our steps to get to 3.5oV and then 3.60V are much, much shorter than the first step getting to 3.40V. The step to 3.50V is longer than the final step to 3.60V, which happens pretty quickly.

Parallel Step-Method Top Balance:
1- Wire the cells in parallel
2- Set the power supply to 3.400V and 80% or less of the rated amperage (80% to not burn it out)
3- Turn on power supply and charge cells to 3.400V
4- When current has dropped to 0.0A at 3.400V turn off the power supply & set it to 3.500V
5- Turn on power supply and charge cells to 3.500V
6- When current has dropped to 0.0A at 3.500V turn off the power supply & set to 3.600V
7- Allow current to drop to 0.0A (or very close) at 3.60V
8- Done, pack is balanced.
WARNING: Top each cell up, to a similar SoC level, prior to wiring them in parallel.

Balancing Via Parallel Resting Voltages???
Many often assume that by simply wiring the cells in parallel they will magically get themselves in balance. This is not entirely true, if you expect it to happen in a timely manner. When cells are wired in parallel, the the cell voltages attain a parity voltage rather quickly. Once a parity voltage is attained the transfer or movement of current between cells, in order to balance SoC, slows to a crawl. Ohms law is in control here and we are talking 0.0001A level movements of current. Attaining a true balancing, by letting cells sit in parallel, at a resting non-charging voltage, takes a very long time. You can let them sit for a week or more, but again, this may not be enough time. Balancing ideally requires a voltage differential to move current between or into the cells. When cells are at the same voltage this transfer of current = slow.

You can drastically speed the process by presenting the parallel wired cells with a charging voltage.. The PSMTB method is the fasted way we know of to attain a perfect balance. Once all cells are at the same voltage and no more current can flow into the cells they are all at the identical SoC.


Bench Top Power Supply

As I mentioned earlier I am a believer that if venturing into DIY LiFePO4 it should be done as a system. Part of that system should include funds for a bench top power supply and other equipment to test for capacity etc.. In my opinion a bench top power supply with variable voltage and current should be a pre-requisite for DIY LFP. Can you make do without? Sure, and I am certain Bode Miller could ski with only one leg, but why..? In the whole scheme of things they are inexpensive and they have multiple uses not just for charging or top balancing LFP.

The bench top power supplies I sometimes use are made by Mastech, specifically the Mastech EX series. We own a 3030EX and a 3050EX. These are not the fanciest or the most expensive power supplies but they work and they work pretty well, especially for the price. Years ago these devices would have run four figures each but today they are very reasonably priced. A Mastech 3020EX (30V X 20A) will run you just $219.95. It will save you $400.00 in your time fiddling with top balancing alone. You will be looking for a 0-30V and 0-10A or larger model. This is my 3050EX. The EX in the Mastech line signifies these units are specifically designed for charging batteries, usually Li batteries. The dial second from the left is EX knob or the over voltage protection dial. Set this dial and the power supply will protect itself.

TIP: When charging LFP cells or banks with a bench top power supply please dial the current back by about 20%. This will allow the power supply to run almost indefinitely and not cause undue wear and tear on the unit. I run my 30A model at 24A and my 50A model at 40A… I sometimes parallel them and charge at 64A when doing cycle testing.

Nothing makes top balancing easier than a bench top power supply:
#1 Charge individual cells to .05V below max top balance voltage and allow current to taper
#2 Wire cells in parallel and let sit, the longer the better.
#3 Charge cells to max top balance voltage Winston = 3.65V
#4 Allow current to go to 0.00A
#5 Turn off power supply and you’re done.
 
Forgive the two long threads above but I feel your pain, I'm dealing with it right now on a system so have recently been doing lots of reading, check out the links at the top of the articles for photos and more detail.

Biggest improvement I found was cranking back the charge current and voltage and slowing down my charge to let them internally balance without kicking the OVP and then tossing on some Victron balancers.
 
Last edited:
Hi, any update on this? I’m curious if top balancing solved your problem or if you discovered a bad cell or battery.
Not yet, I'm swamped with a whole bunch of video editing for the tests I did the past couple months, but as soon as the videos are cut, I'll be back to testing again. I'll update this thread after a top-balance ASAP, but it will probably a few weeks before I have results.
 
Forgive the two long threads above but I feel your pain, I'm dealing with it right now on a system so have recently been doing lots of reading, check out the links at the top of the articles for photos and more detail.

Biggest improvement I found was cranking back the charge current and voltage and slowing down my charge to let them internally balance without kicking the OVP and then tossing on some Victron balancers.
Great link, thanks! I've been reading about APDs too, although I mostly plan to play around with trolling motors and electric outboards, so my setup will stay pretty simple. All this info on upper charging limits was very helpful as well.
 
Back
Top