Pack Cycle Testing

rev0

Member
Joined
Oct 3, 2017
Messages
372
Perhaps this experiment has been done before, but I couldn't really find a good answer on bottom vs. top balancing, BMS systems with balancing, etc., lots of conjecture and anectode and not much hard data. So I built a basic test setup to measure the cell voltages of each cell in an 8s1p pack and run it through charge/discharge cycles. The pack in question is made of Samsung 18650-22P cells from the Hoverboard packs on eBay, I picked my lowest capacity set of 8 cells for this since I'm using them for a project already. I started by bottom balancing all the cells to 3.15V and starting the cycles from there. Charging is done at 2A (~1C) and discharging at 1A (~0.5C) which makes the capacity math easy (hours = amp-hours).


image_wgrppp.jpg

image_iiakpa.jpg

image_dhalin.jpg


So far the capacity hasn't changed much over the 6 cycles, but the cell voltage difference is slowly creeping up, especially at high load(charging at 2A in this case) and termination. So it seems to me it's still necessary to keep the pack consistently balanced via BMS, or at least be able to monitor cell voltages to make sure none drop below/above your safety limits. I'll let this continue to run and post data at 10 or 20 cycles if there's interest.

The project I will be putting these cells into is a range extender pack for a Nissan Leaf, basically following the procedure of Leaf Xpack on Youtube. So far I've only bought 200 cells (~1.4kWH) which is enough for a 96s2p pack to add hopefully around 5 miles of range.

And for the nerds out there, here's the data in real-time (no offset/gain correction, channel 2 of the ADC is wonky so it needs to be corrected by 0.75): https://thingspeak.com/channels/308182
 
Interesting.
What equipment are you using for your test rig?
 
It's a custom board with an ESP8266 running Arduino code as the brain, and a Maxim MAX11629 8-ch 12-bit ADC with a bunch of resistor voltage dividers to sense the voltages at each cell in the series string. The ESP8266 also controls a power MOSFET which switches on the constant current load (bench equipment) or a relay to switch inthe power supply (set to 2A current limit, 33V). Here's a picture before I changed it to a relay (only had 1 power MOSFET, the smaller MOSFET couldn't handle the voltage/load:

image_wpwrlp.jpg
 
The bit of info that I can't spot in your report is how well your now-serial cells were matched for capacity before you started.

If the capacity matching is at all off, AND you are doing full-voltage-range cycling, then the runt in the serial pack is going to get found out.

My understanding of the "bottom balance and you are good" argument is that it also includes the requirement to limit the charge/discharge range to suit the capacity of the weakest cell in the serial stack.
Lets say you have one 1500mAh cell in series with 6 2000mAh. And bottom balance the cells at a conservative final voltage of say 3.2v.
Now if you limit the pack to cycles of say 1250mAh, the runt can keep up, and so the pack should stay quite well balanced (though the cell voltages willdiverge after charging, putting in and then draining out a charge that all can handle comfortably will see them return to their bottom balance point).
But if you run cycles of more than 1500mAh, the pack will inevitably become progressively more unbalanced.
You have to limit things so that the weakest link in the serial chain isn't over-driven.
If you do that, whether you top or bottom balance shouldn't make a difference.
Any difference comes when you exceed the capacity of the runt, and whether you are then over-charging or over-discharging it (as you cause different damage by the different actions).

I have deliberately chosen a distinct cell capacity difference to demonstrate the point, but it applies whatever the difference might be.


Out of interest, what current do you expect the cells to be expected to deliver in this application? Running 96s 2p there's nowhere for a weak cell to hide. What cells have you chosen?
 
They are matched in capacity to within 31mAH, seems close enough for me.


image_ouhrjt.jpg


I amcharging thepack to a voltage below 4.2V/cell (4.125V/cell in my case), and discharging until any cell hits 3.2V. Should the capacity in/out then be limited by the weakest cell? Or because I am not also limiting the charging at cell level will there still be an unbalance?

For the pack I am using the same cell; Samsung 18650-22P, they're a lot higher current capacity than I need, but I'll be using only cells between 2075-2200mAH (125mAH max delta). I also plan to use a Nissan Leaf BMS standalone with LeafSpy to balance/monitor the cells in use.
 
Awesome test m8! I would love to see how it looks up to 100cycles and above. This clearly will show the importance on equal packs/cells and/or a bms that deal with the out of balance that "will" occour after some time.
 
rev0 said:
Here's the update, now at 15 cycles, starting to see a degradation trend as well as the cells getting ever more out of balance:


image_ibnkem.jpg

image_bhygfs.jpg

Fantastic testing!
Thanks for sharing!
I am going to keep watch on this.
 
Here's the cycle testing update, finally got an interesting phenomenon; the voltage delta between cells seems to have plateaued after 25 cycles, and the capacity loss as a result of imbalance (having to stop discharge when the lowest cell hits 3.2V) has also plateaued. I also noticed the highest voltage cell is getting very close to 4.2V, so if the trend continued further it may have caused damage with the way the test is set up currently, or I would need to modify the code to cut off charging when the first cell hits 4.2V as well.


image_fmqhdu.jpg


image_ornnnm.jpg


Here's what a full cycle looks like after the voltage delta has flattened out:


image_nilwsf.jpg
 
Thank You for sharing Your findings with me :) and the other users as well :)

Can You fill me in on the following: did the low capacity cells reach the upper voltage limit first in the first few cycles before the high cap cells took over that position?

Initially I thought that the lowest capacity cell would reach the "full" state first, but Your data showsthat all this changes, when You keep cycling the cells, and I think I know why. (If Im wrong, pse let me know why).

I think that the "overhead", the ratio between how many AmpHours You have to charge in return for the AmpHours You take out, will increase as the level of Discharge increases, so the high cap cells will not need as many Ahs to reach the "full" state as the lower cap cells.

Seems that we could get a few more Ahs out, if the 2 high cells were bleeded a bit so we could charge the rest a bit longer.

It would be fun to watch the difference if a BMS were added.

Tnx again for a nice experiment :)
 
ChrisD5710 said:
Thank You for sharing Your findings with me :) and the other users as well :)

Can You fill me in on the following: did the low capacity cells reach the upper voltage limit first in the first few cycles before the high cap cells took over that position?

Initially I thought that the lowest capacity cell would reach the "full" state first, but Your data showsthat all this changes, when You keep cycling the cells, and I think I know why. (If Im wrong, pse let me know why).

I think that the "overhead", the ratio between how many AmpHours You have to charge in return for the AmpHours You take out, will increase as the level of Discharge increases, so the high cap cells will not need as many Ahs to reach the "full" state as the lower cap cells.

Seems that we could get a few more Ahs out, if the 2 high cells were bleeded a bit so we could charge the rest a bit longer.

It would be fun to watch the difference if a BMS were added.

Tnx again for a nice experiment :)

It looks like cell 7 and 8 are getting the highest voltage in the charge cycle, cell 8 has the lowest capacity in the pack, and cell 7 has the highest internal resistance.. otherwiseI'm not seeing much correlation. There's a table with the capacity/IR data a few posts back, the cell number is it's number in the series chain (cell 1 is the bottom most cell from 0-3.7V, cell 2 is 3.7-7.4V etc.)
 
This will be the last update of this experiment without a BMS; one of the cells has exceeded 4.2V on the charge cycle so I've stopped it. Here's the final data, interesting that the imbalance started increasing again after cycle 43:


image_ckylcc.jpg


image_tfxngd.jpg


image_sagrbf.jpg


Hopefully in the next day or two I will be hooking this pack up to a Nissan Leaf BMS which I intend to use with my final 96s2p pack. It only can provide 10mA of balance current, but that may be enough since the voltage drift over time is pretty low.
 
Is there any interest to repeat the experiment with the pack top balanced or mid balanced to see if it lasts more cycles than the bottom balanced result?

Turns out the BMS can't balance only an 8 cell pack with the stock controller, gotta wait until I can buy a WolfTronix microcontroller to replace it with.
 
I think some usefull data might be how much energy is lost to Ballancing over time. So you start with a balanced pack and keep track of the energy dissipated per cell per charge.
 
This brings a question for me. I believe that my system will only be as big as my smallest 80p pack. Is this true? I'm not capacity checking every cell in my packs, but I am at least capacity checking at least one battery per pack and assuming the rest in that pack are close to the same. What I'm afraid of is in my 14s 48v system, my capacity might be sacrificed by doing it this way, because the smallest capacity 80p pack will hinder the capability of the entire system. Am I wrong? I'm thinking I will have one longmon consistently burning excess from my strongest pack, rather, biggest pack in Amp hours.
 
shonalex said:
This brings a question for me. I believe that my system will only be as big as my smallest 80p pack. Is this true? I'm not capacity checking every cell in my packs, but I am at least capacity checking at least one battery per pack and assuming the rest in that pack are close to the same. What I'm afraid of is in my 14s 48v system, my capacity might be sacrificed by doing it this way, because the smallest capacity 80p pack will hinder the capability of the entire system. Am I wrong? I'm thinking I will have one longmon consistently burning excess from my strongest pack, rather, biggest pack in Amp hours.

Yes that's my understanding. Using a passive balancerwill make sure the weak cell doesn't overcharge but won't give you any extra capacity.
 
shonalex said:
This brings a question for me. I believe that my system will only be as big as my smallest 80p pack. Is this true? I'm not capacity checking every cell in my packs, but I am at least capacity checking at least one battery per pack and assuming the rest in that pack are close to the same. What I'm afraid of is in my 14s 48v system, my capacity might be sacrificed by doing it this way, because the smallest capacity 80p pack will hinder the capability of the entire system. Am I wrong? I'm thinking I will have one longmon consistently burning excess from my strongest pack, rather, biggest pack in Amp hours.


Yes that's correct. A smaller capacity pack(total Ah of cells in parallel) will reach the extremes quicker than the bigger packs when a current is drawn from the pack series as they all will see the same current.

That's why it's so important to try and build the packs with the same capacity as then your lowest packs capacity is much closer to all the other packs and your overall capacity of the whole pack series will be higher as no single pack limits the overall capacity.

The only real way to test this is to do a load test and then keep monitoring the cell (parallel packs) voltages over time and then stop the discharge when the lowest cells reach the low voltage threshold you have decided on.

If one particular cell pack is very low then either replace the cell pack or add a smaller 'booster' cell pack to make up for the lower total capacity of that cell pack but that can get real messy depending on the physical layout of your battery bank and available space.
 
I'm not capacity checking every cell in my packs, but I am at least capacity checking at least one battery per pack and assuming the rest in that pack are close to the same

My first question is why you are only testing a few cells capacities in a pack? Are these new cells you are using or reclaimed. If reclaimed, you really should capacity check every cell to make sure you don't get a dud mixed in the pack. If you have a self discharging cell in the pack, it will really make keeping the pack balanced extremely difficult if not impossible.
 
Korishan said:
I'm not capacity checking every cell in my packs, but I am at least capacity checking at least one battery per pack and assuming the rest in that pack are close to the same

My first question is why you are only testing a few cells capacities in a pack?

My process right now is to charge up each battery, let it sit for a week to verify it's not self discharging, then capacity check as much as I can with my one opus during all that time. Once I get my load testing done I'll be able to see the weak packs, then replace if needed. Perhaps capacity check those 80 batteries.

They are reclaimed cells out of laptops and medical packs. I have a supplier that sells them to me at $1.00/lb.


Is there a way too sign up for notifications? I would have replied sooner.


Lol, nvm. I see it. I'll subscribe.
 
The medical pack cells are probably fine. But the laptop cells probably should all be checked regardless, unless they are newish packs.


Dont worry about the reply time. Some take weeks to reply ;)
 
Back
Top