diy solar

diy solar

Historical LFP early-adopter misinformation story (long)

Substrate

Solar Addict
Joined
Apr 28, 2021
Messages
1,545
Location
SoCal
A somewhat laughable (expensive) story of my early-adopter mis-information woes.

Or read as "What NOT to do!"

Early on, when the only cells we could get in any large capacity to drive DIY EV's and other hobbiest stuff, there seemed to be some confusion about LFP vs Non-LFP charging procedures. We were well aware of the operational differences and the super flat discharge curve of LFP, but still thought that somehow these lithium batteries were in the same family and could use the same charging techniques.

Quite a few of us took our cells to 4.2v, just like non-lfp. We even top-balanced up there. It seemed to work, albeit precariously, but we couldn't pin down why things went downhill so fast. We knew there was no real energy up so high in the knee, but the going theory was the higher you top balance at, so much the better. Even bottom-balancers would do a bank level trigger when the first cell reached 4.2v!

Part of the problem was brought on by early documentation of some ThunderSky prismatics, (1st gen blue ones, not yellow!) that spec'ed out as 4.2v ! Other manufacturers had 3.65, like A123 cylindricals, but you know - these prismatics are more advanced! And nobody was tack-welding A123 cylindricals or pouches on their own - given the problems with counterfeits, used or rejected stock, bad diy tack-welds ... So prebuilt prismatics was the way to go.

Things were going downhill fast after a few cycles. Top balance was hard to maintain, and was always busy. Bottom-balancer guys were doing better, but still..

In the end, after a LONG period of trying to pin down WHY some manufacturers had spec sheets showing 4.2v as the max CV voltage to charge to, it turned out to be a "customer is always right" thing, and not a scientific one. "Sure, you can take these to 4.2v, go ahead - they're your cells!" I don't think it was a case of dishonesty either - I think that many early on, just didn't know for sure.

Anyway, can you say plating from secondary reactions due to high-voltage? We can now, but not then when the cells were NEW and seemed ok for the first handful of cycles out!

This kind of reminded me of the whole gel vs agm CV voltage charging fiasco from many years past! Even today, many don't know the difference.

So when I built my first 4S 12v learner bank, I bought 5 cells. The one with the lowest capacity or highest internal resistance would be the odd man out. I chose the more normal 3.65v as my top-balance CV limit. Whew.

So with that one cell left out, I did a destructive experiment on it in a SAFE ENVIRONMENT:

1) Charged the cell like normal at 3.65v CV, and let it absorb or saturate to full until there was very little tail current flowing.
2) Let it rest for an hour to settle.
3) Changed my CV to 9v to see what would happen in a simulated HVD charging failure.

I put my tracking multimeter on it, and set it to be audible, so at every 0.1v change it would beep.

So with a CV of 9v on this already charged cell, we let loose. Nothing happened. No current flow. No change in voltage.

WOW! Sure looks like they are somehow self-protecting! WRONG. Things were fine for about 15 minutes.

Beep!

Huh, ok. Back to the adult beverage. (not recommended for safety)

10 minutes later:

beep!

Another 10 minutes go by, but now

beep beep

You get the picture. In a very short amount of time:

beep beep beep beep beep beep beep beep.....

Then I disconnected everything and stopped. What was fascinating was that despite ZERO CURRENT flowing to charge, just having a high CV voltage applied caused a change in terminal voltage!!

My best guess here is that the rise in terminal voltage with no current flowing was not a charge reaction, but a secondary chemical reaction (destructive, along with plating) happening. Fascinating.

To circle back to the early days when some were top-balancing at 4.2v, I had to laugh. We were trying to balance at voltages caused by secondary reactions! Egads, no wonder it went so wrong so fast. :)

It was educational to say the least, and worth taking a spare cell to destruction (safely) to learn and watch that happen.
 
Last edited:
No problem - it was eye opening to see what I think was the transition from normal charge reactions, to the secondary chemical reactions occuring with no current flowing! (used a Fluke ammeter, not just my bench supply readouts)

Which brings back to the old "don't float lfp". Well certainly not necessary. And covered in other threads..

If my float voltage is lower than my battery terminal voltage, and hence no current flowing - it is only an application of voltage and not of current.

But unlike the extreme example above where the opposite was true about voltage (abusive test), I'm not so sure that those who do employ a low-voltage float for whatever reason, and no current flowing - being lower than the battery terminal voltage are actually causing secondary reactions.

Since I don't have a lifetime to prove it, my "hunch" is that lower-voltage floating with it's inherent non-current condition, is actually doing harm. Wish I was 20 years old again to prove this in the field. :)
 
So, before Will came along, we had stuff like this to watch. Taking Thundersky's to 4.2v

Feel free to skip directly to 48:30 in. This is historical please, so no need to rehash all that stuff covered long ago.


We very quickly learned to walk ours down first to no more than 4.0v CV. Many builds later, we joined the 3.65v CV is good enough club.

Our major mistake, since it was never indicated, or you were relying on word of mouth, was that taking them to 4.2v was only meant to be done ONCE! And just for this specific brand. Here, if you can watch long enough, he indicates that he's only charging to 4.2v / cell once.

Lurkers: This is a historical personal story - not technical advice to secretly double your capacity! For one thing, the values have changed. And remember where we are - the DANGER ZONE! sub-forum. 'Nuff said.

It was easy to burn through $$$ in those days. :)
 
Last edited:
A similar scenario is occurring as we speak with the current trend of trickle current parallel top balancing. The results will be on display in a few years - until then because someone made a cool youtube it must be right!

So many people complaining unaware that many of us have been building LiFePO4 packs for over a decade, and have observed many characteristics that still aren’t well documented.

As for Chinese battery specs, the manufacturer of my cells published detailed charging specs only once. Within days his exact specs were included with certain chinese cells that has previously been quite different. (ever wonder how Winston cell lifespan went from 3000 to 5000 cycles with no chemistry change?)
 
Most likely from reducing that initial max CV rate from 4.2v down to 4.0, and even the sometime later to something like 3.9v, so punters like me who didn't realize you only do this once, and then revert back to a more normal 3.65v max for the rest of life. I was working with pre-yellow ThunderSky's gen-1's.

What I was told back at the time, was that this one-time initial higher CV charge on a new cell was to ensure that all the material was active and evenly charged. Otherwise, in high current motive-power applications, you could develop "hotspots" where some of the material was doing more work than the rest. The problem of this one-time deal is that of time and not paying attention to the full saturation stall in charge current, and leaving the room to do something else. Secondary reactions await!

I never got to test that, since my application was mostly sub-c storage, where hotspots, if they exist, would be for the most part hidden and not revealed. Well, actually I did later, but that was purposely destructive, where it didn't reveal hotspots, but sure revealed secondary reactions.

I mentioned in another thread where bottom-balancers would just pour on the smoke with high-amperage charging right after taking their cells well down into the bottom of the knee, and LFP doesn't intercalate quick enough when hammered like that in the deep depths, causing damage.

Essentially, without enough knowledge we were hurting them on the top end, and if not patient enough at deep discharge to take it easy until getting out of the knee, hammering them down there too. :)
 
A similar scenario is occurring as we speak with the current trend of trickle current parallel top balancing. The results will be on display in a few years - until then because someone made a cool youtube it must be right!
What exactly are you referring to? Do you mean it is unsafe to top balance at low current? I ask because I am currently top balancing 32 cells using a 3.65 volt charger it starts at 14 amps and then tapers down as it gets closer to the 3.65 voltage. 14 amps would be considered trickle charging on a 6400 a/h bank.
 
Yeah, confused a little here too. I think what toms was warning about was the factor of time trying to achieve perfect balance, when spending too much time doing so is more harmful than having a bit of slop. I call it balance-addiction.

Imaginary video:
"So guys, it's been 4 days of balancing my 400ah cells at 3.65v, and I'm going to let it go for another week to see if I can get the balance down to 0.0001v difference! Be back in a week with the results!"

Which doesn't matter anyway since unless they are lab perfect in both capacity and internal resistance, you'll have a bit of a ragged edge at the bottom no matter how 0.0001v difference you are at the top.

(For me, a difference of no more than 0.05v (maybe even 0.07v) and I'm done. Much longer takes more time, so 0.05v is my practical tradeoff between time and perfection.)

So toms - was this what you were trying to express, or did I get it wrong?
 
Last edited:
Thanks for the great thread full of interesting historical insights.

I’m addicted to cell balancing; calculating it as an average or median over delta cell Voltage, seems like the most “representative” method if it is available. Ok.. maybe as a 3D function balanceScore(cellVoltage,cellSoC,cellAmps).

Highest score achieves very low cell voltage delta over entire used state of charge range (e.g. 20~80%) and all amp draws.

Need arduino to log voltage of every cell repeatedly and derive deltas and generate histogram with bins based on cell Voltage.

Random thoughts. Diagnostic tools are fun
 
Good enough to me is < 12.5 mV delta among all cells in a 100Ah pack of 4S/8S. Or something. Chasing anything much less than 8mV feels like effort I should focus elsewhere.

I’m fond of active and passive balancing to push a pack firmly into balance, but currently in practice am only using passive activated above 3.425 Volt per cell.

Having passive balancing enabled >3.425Vpc essentially at all times has side effect of reducing time spent at elevated voltage.

edit:
Yak shaving is also fun.
 
Last edited:
Good enough to me is < 12.5 mV delta among all cells in a 100Ah pack of 4S/8S. Or something. Chasing anything much less than 8mV feels like effort I should focus elsewhere.

I’m fond of active and passive balancing to push a pack firmly into balance, but currently in practice am only using passive activated above 3.425 Volt per cell.

Having passive balancing enabled >3.425Vpc essentially at all times has side effect of reducing time spent at elevated voltage.

edit:
Yak shaving is also fun.
Shaving can be fun... :) I am floating my 8 cells at 3.385, 3.384 3.384 3.383 3.385 3.385 3.384and 3.384 volts per cell. My delta is .002mv's.

I accomplished this using the Overkill BMS. It took awhile and now I can be proud of myself......until I cycle the pack and I am sure my delta will go back to where it was....lol
 
I'm gonna do like Hannibal Lecter and rent a storage unit prepaid for years, set up my PSU and top balancing and lock the door. Then return to find my cells balanced to within .0000001v of each other. I'm gonna come back with screenshots and win the prize and you slackers can eat it!
 
Heh, this kind of goes back to the historical bottom-balance vs top-balance wars before anything was smart!

Bottom balancing was promoted as a once-n-done concept as being superior to those trying to top-balance all day / week long with random loads like headlamps, coils of wire sitting inside buckets filled with water etc. And it was!

Now we've just scaled back to using arduino's and such to do the same thing! :) <running and ducking>
 
Charge each cell individually to 3.65V until current drops to 0.05C

Then assemble pack.
You know that can't be done using the 10 amp power supplies most are using. If charging each cell individually the C rate is only .0357 to begin with. Charging until the current drops to .05C is about the same as 15 amps if charging a 280ah cell. What size power supply would one need to buy to accomplish what you are suggesting?

If parallel top balancing four 280ah cells with a 10 amp power supply the C rate is only .009.

Let me rephrase my question. If parallel top balancing four 280ah cells to 3.65 volts with a 10 amp power supply until the current of the power supply reaches zero how much damage is done to the cells?

A similar scenario is occurring as we speak with the current trend of trickle current parallel top balancing. The results will be on display in a few years - until then because someone made a cool youtube it must be right!
I parallel top balanced my eight 280ah cells with a 12 amp power supply. That's a C rate of only 00535. What results can I expect to see in a few years compared to what I see now? Loss of capacity, cycles or both? Should I or anyone else be concerned?
 
Last edited:
Back
Top