DIY Electric Car Forums banner
1 - 20 of 35 Posts

·
Registered
Joined
·
58 Posts
Discussion Starter · #1 ·
I am trying to figure out whether or not I will need to replace the batteries in a used EV conversion I purchased. Everyone is telling me I can't get battery life from the voltages since they have such flat voltage curves. The batteries are: (qty. 34) TS-LYP160AHA Thundersky Lithium Iron Phosphate 3.2V 160 Ah. I am seeing the voltages drop to around 2.9v and a little lower when driving the truck and drawing around 100A, so it doesn't seem promising. Is this enough evidence, or should I remove a cell or cells and bench test?

It is an 08 Ranger, converted in 2009 and used as a city vehicle.
The controller shows 1540 miles on it but I have no idea how the batteries were treated since the original BMS was recently replaced with an Orion.
 

·
Registered
Joined
·
916 Posts
Yes voltages tell nothing about SoH.

You need to test Ah capacity.

Ideally this gets benchmarked with a standardized protocol at commissioning time

and you repeat using the same specs each time

so your residual capacity State of Health SoH% is objective apples to apples.

Otherwise use the original rated Ah capacity as your 100% and a low discharge rate like 0.1C for testing.

Ideally each battery gets tested separately, using a CC load, precisely timed, rather than using a coulomb counter, much less accurate.

Winston's old specs were charge to 4.25V and discharge to 2.5V OMG I hope your PO didn't follow those or they likely are dead!


I would use 100% SoC defined as up to 3.60V maximum, then holding Absorb there until current trails to 5A

And 0% at 2.99V no lower and start recharging right away.

So 160Ah

0.1C or 16A discharge rate, a dummy load tester is maybe $100 will automate everything regulate the CC draw as voltage drops.

Or you could rig up a bank of light bulbs, resistors even the guts of a hair drier depends how handy you are with electronics

and keep the load at 16A precisely

In theory each test will be close to 10 hours, the percentage of 600 minutes it actually takes

gives you SoC%

Anything below 70-75%, ready for the scrap yard IMO.
 

·
Registered
Joined
·
58 Posts
Discussion Starter · #3 ·
Thank you for the detailed response John. I am handy so I’m confident I could do the test. I could get my hands on a heating element and I have a DC ammeter and multimeter. I uploaded a few screenshots from the BMS utility. I tried to get a screen video to upload but I’m working with a broken laptop running Windows 7. The average voltage is about 3.3 at 95% SOC. When I plug in the charger it climbs to 3.7V (average is 3.57V) within a minute or two and cuts off(BMS limit is set at 3.7V)
 

Attachments

·
Registered
Joined
·
58 Posts
Discussion Starter · #5 ·
I was planning to test the cell life by taking 4 cells out, connecting them in series and powering a 12v inverter so i could plug in and run a 120v appliance for the test. I was going to fully charge the 4 cells to 3.6V for the test since a bad cell in the car could be limiting the charge of the other cells. I have a variable DC 5A power supply. Is it ok to charge at 5A? Would this be considered trickle charging?
 

·
Registered
Joined
·
916 Posts
I was planning to test the cell life by taking 4 cells out, connecting them in series and powering a 12v inverter so i could plug in and run a 120v appliance for the test.
Fine for an initial ballpark estimate, but really testing at the per cell level and accurately would be better.

Even if you don't decide to keep them, you'd be amazed what idiots pay for even well-worn name-brand used cells, and of course you want to be honest, so a per-cell load test report / graph will make them fly off the shelf!

> I have a variable DC 5A power supply. Is it ok to charge at 5A?

Yes in fact the low C-rate will give a more precise SoC benchmark even without holding any CV/Absorb time, just stop at a precise termination point using a HVC.

In normal cycling 3.45V or 3.50V would be better for cell longevity, but for infrequent test-benchmarking purposes going higher is fine.

But for apples to apples, use the same current (available) level and non-manual charge termination mechanism for every test.

Same as above for the discharge / LVC half of the cycle, important to keep the load current precisely CC constant, and precisely time the discharge period from (your definitions of) 100% to 0%.

Using a coulometer instead is easier but much less precise.
 

·
Registered
Joined
·
392 Posts
The voltage can’t go higher than the battery so it doesn’t matter what voltage you power source is as long as you shut it down when you reach 3.6v.

Batteries are charged with a procedure constant current until the desired voltage is reached and the constant voltage till the current drops to your desired level. Just skip the constant voltage stage you will be 95% charged by then.
 

·
Registered
Joined
·
916 Posts
The voltage can’t go higher than the battery so it doesn’t matter what voltage you power source is as long as you shut it down when you reach 3.6v.
The charge voltage toward the end is **always** higher than the full battery at rest, otherwise no current could flow in CV.

LFP full at rest is 3.33-3.35V, depending on factors.

At a low enough current rate you can get to 100% at a 3.43-3.44 setpoint.

At a high enough rate stopping at a CV of 3.65V may only get you to 85% SoC.

Voltage by itself while current is flowing says next to nothing wrt SoC.

And to get to a consistent precise SoC point, e.g. benchmark testing

you do need to use a consistently precise endAmps as well as CV setpoint.

Everything else you wrote there is spot on correct.






Batteries are charged with a procedure constant current until the desired voltage is reached and the constant voltage till the current drops to your desired level. Just skip the constant voltage stage you will be 95% charged by then
 

·
Registered
Joined
·
392 Posts
Everything I said was correct. You took parts of it out of context.

And yes I have tested many cells and you get around 95% SOC by skipping the constant voltage stage. You can also confirm that by calculating the area under the curve. But in any case for a capacity test it won’t make much difference. I suspect they have lost 50% based on what was said thus far.
 

·
Registered
Joined
·
392 Posts
Everything I said was correct. You took parts of it out of context.

And yes I have tested many cells and you get around 95% SOC by skipping the constant voltage stage. You can also confirm that by calculating the area under the curve. But in any case for a capacity test it won’t make much difference. I suspect they have lost 50% based on what was said thus far.
I have never tested Thundersky but it looks like they charge to a higher voltage. This was some time ago so maybe experience caused them to lower they endpoint voltage. It seems he charges the CALB cells above 4v too which I wouldn’t do.

http://media.ev-tv.me/1Cbatterytests.xls
 

·
Registered
Joined
·
916 Posts
Winston Chung has claimed several unique qualities result from his Yttrium doping

they are fine quality cells, but all 3rd party testing I've seen, says their performance responds to treatment factors the same way CALB, GBS, Sinopoly LFP do.

In an EV context the high C-rates likely wipe out the benefits of the finer points in coddling, but in a lower rate context of getting 4-5000 cycles out of them, these other factors can double lifespan.

The vendor spec'd maximums are very stressful, **not** healthy for your expensive bank.
 

·
Registered
Joined
·
58 Posts
Discussion Starter · #13 ·
John thanks for the reply. Yes I want to do a ballpark test. The Orion BMS is reporting 91% state of health and the controller odometer shows 1541 miles so those are positive signs, but I realize if the batteries were not treated well They could be bad. Not sure if anything else would cause a voltage sag below 2.9v on a number of cells when the motor is drawing around 100 amps other than bad batteries.

For the testing, I charged the batteries for 3 hours and 48 minutes at 5 A until the power supply got to 14.4 V. Cells settled to 3.33V. I started the test with the 12 V inverter powering an air scrubber with variable speed, so I adjusted the speed until the Ammeter read 16 A. Ran for 4hr 8min yesterday and I’m continuing the test now.
 

Attachments

·
Registered
Joined
·
58 Posts
Discussion Starter · #15 ·
I finished the test.

The 16 amp load ran for 8 hrs and 2 min before the first battery hit the 2.99v limit. So it looks like they have roughly 80% life based on this test.
 

·
Registered
Joined
·
916 Posts
If that's typical can def get a bit more life out of them, just reduced range and a bit less oomph.

Or, if budget is no issue, sell on and buy new, I'm amazed at what people pay, sometimes more than 50%.

Shipping is a real hassle though, best to sell locally
 

·
Registered
Joined
·
1,498 Posts
You've only half-done the testing. You don't know the capacity until you charge the depleted cell back up to some fixed reference. You charged the cell before testing from an unknown state.

Plus what is the low voltage used for the 160 A-hr determination? i'll bet they took the cells down to 2.5 or something, and probably started at higher than 3.33.

For good longevity operating on 80% keeping the cells off both the ultimate top and bottom limits is a good idea. The cells may very well be good for 120 A-hr operated in this middle 80%, just plan range based on that instead of the marketing spec of 160.
 

·
Registered
Joined
·
916 Posts
Repeating
Winston's old specs were charge to 4.25V and discharge to 2.5V OMG I hope your PO didn't follow those or they likely are dead!

I would use 100% SoC defined as up to 3.60V maximum, then holding Absorb there until current trails to 5A

And 0% at 2.99V no lower and start recharging right away.
The first set are flat-out damaging, to firmly set aside.

The rest are still stressful, to be used for later benchmarking only.

Of course in an emergency could go that low but for normal cycling stop at 3.1V (higher C-rates) or even 3.2V if more gentle.

Charging to 3.45V if holding Absorb

or 3.5V if CV only at say below 0.4C charging.

If you have to fast-charge, only in warm temps, then CV to 3.6 will be OK.

Resting OCV at 3.34V or so is as Full as is useful.

You may get even a few years out of them this way.

*If* they all test like this, are still decently matched no big imbalances.

The pack is only as good as the weakest members.
 

·
Registered
Joined
·
58 Posts
Discussion Starter · #19 ·
You've only half-done the testing. You don't know the capacity until you charge the depleted cell back up to some fixed reference. You charged the cell before testing from an unknown state.
kennybobby, Right! I'm jumping the gun. I'm recharging it now with the 5A power supply and I'll rerun the load test. I guess I can at least say that they aren't scrap!

Plus what is the low voltage used for the 160 A-hr determination? i'll bet they took the cells down to 2.5 or something, and probably started at higher than 3.33.
Good point. So they could have more life if tested to the manufacturer's testing procedure, which I don't plan to do based on advice from everyone here.
 

·
Registered
Joined
·
58 Posts
Discussion Starter · #20 ·
Charging to 3.45V if holding Absorb

or 3.5V if CV only at say below 0.4C charging.

If you have to fast-charge, only in warm temps, then CV to 3.6 will be OK.

Resting OCV at 3.34V or so is as Full as is useful.

You may get even a few years out of them this way.

*If* they all test like this, are still decently matched no big imbalances.

The pack is only as good as the weakest members.
Right now the pack charges at 26 A from the 220V house power. I'll have to check the charge settings I think it currently cuts off at 3.7V.

I read some last night and someone mentioned that Lifepo4 batteries lose power more than capacity, so I'm wondering if this is why the voltage drops so low even though the batteries MIGHT have around 80% life left.
 
1 - 20 of 35 Posts
Top