Testing large amounts of cells quickly

Solar, wind, hydro and other non-fossil sources of electricity.
Post Reply
BYqSXt8Z   100 W

100 W
Posts: 119
Joined: Jul 03 2014 8:29pm
Location: GTA, Ontario, Canada

Testing large amounts of cells quickly

Post by BYqSXt8Z » Jul 19 2019 7:35pm

Hi, I got some second hand batteries (a123 26650) that I need to sort by capacity... 28850 of them. Does anyone know of a tool that can extrapolate battery capacity from things like ir, voltage sag, and such in a few seconds or even minutes?

Thank you

Sent from my CLT-L04 using Tapatalk


flat tire   1 MW

1 MW
Posts: 1592
Joined: Feb 26 2014 12:20am

Re: Testing large amounts of cells quickly

Post by flat tire » Jul 19 2019 7:54pm

No. You have to individually test each battery.

User avatar
Matador   10 kW

10 kW
Posts: 889
Joined: Jun 29 2016 10:00pm

Re: Testing large amounts of cells quickly

Post by Matador » Jul 19 2019 8:48pm

I wish... But to do so in such short time, you'd need to use supercapacitor rather than lithium cells. The bottleneck is the discharge rate for measuring discharge capacity...
DCIR can be measured in seconds however... You can thus extrapolate voltage sag by ohm law and knowing how much load you wanna pull: Sag (aka delta V) = DCIR (in ohms) x load (in amps).

Matador

User avatar
Bison_69   100 W

100 W
Posts: 133
Joined: Mar 09 2009 2:18pm
Location: Quebec, QC / Canada

Re: Testing large amounts of cells quickly

Post by Bison_69 » Aug 31 2019 5:30am

Hi guys,

Battery University came up with a quick way to analyze lithium cells using pulse frequencies...

If someone could reproduice this electronic device to adapt to many individual channels (multiplexing circuit)
then many cells could be checked in a very short time.
https://batteryuniversity.com/learn/art ... _batteries

They actually have the tool (technology) but not made for large scale testing, may be one day
someone could reverse ingineering it and build a version for the hobbyist.
https://www.cadex.com/en/technology/our ... /quicksort
CityTrackPro 2015 eBike http://kobois.com
QS MOTOR 205 H50 V3 hub-motor
KELLY 7230S 90V 120A (300A peak) Controller w/Bluetooth
Large screen CycleAnalyst
Batteries 18650 VTC4 Li-ion 21S10P
SMART BMS 16-32S / 200A w/Bluetooth and LCD Display
SolidWorks Designer - electric Trike and new eBike frame for a Montreal friend well known on the forum.
Free Personal 3D Drawings download http://grabcad.com/sg/projects
"Your tiny little human brains. How do you get around in those things?"

BYqSXt8Z   100 W

100 W
Posts: 119
Joined: Jul 03 2014 8:29pm
Location: GTA, Ontario, Canada

Re: Testing large amounts of cells quickly

Post by BYqSXt8Z » Aug 31 2019 8:07am

Bison_69 wrote:Hi guys,

Battery University came up with a quick way to analyze lithium cells using pulse frequencies...

If someone could reproduice this electronic device to adapt to many individual channels (multiplexing circuit)
then many cells could be checked in a very short time.
https://batteryuniversity.com/learn/art ... _batteries

They actually have the tool (technology) but not made for large scale testing, may be one day
someone could reverse ingineering it and build a version for the hobbyist.
https://www.cadex.com/en/technology/our ... /quicksort
I don't know if I'm misunderstanding, but it seems like they are talking about testing impedence.

Sent from my CLT-L04 using Tapatalk


User avatar
fechter   100 GW

100 GW
Posts: 14073
Joined: Dec 31 2006 3:23pm
Location: California Bay Area, USA

Re: Testing large amounts of cells quickly

Post by fechter » Aug 31 2019 9:43am

By measuring cell IR and voltage you can get a pretty good idea if a cell is bad or not, but the only way I know of to check capacity is to do a full charge/discharge test.

You might try doing the fast tests on a sample of cells and see how it correlates to the actual capacity.

If the cells were not used much, their capacity should be pretty close to spec if they are not damaged. You can tell a lot about A123 cells from their resting voltage. Anything below 2v is bad for sure. Good ones tend to stabilize around 3.3v. Bad ones also tend to have accelerated self-discharge, so that is one more fairly easy test to do. Charge a bunch of them to 3.65v and let them sit for a few days, then measure voltage.

28850 is a mountain of cells. How did you get that many?
"One test is worth a thousand opinions"

BYqSXt8Z   100 W

100 W
Posts: 119
Joined: Jul 03 2014 8:29pm
Location: GTA, Ontario, Canada

Re: Testing large amounts of cells quickly

Post by BYqSXt8Z » Aug 31 2019 10:57am

fechter wrote:By measuring cell IR and voltage you can get a pretty good idea if a cell is bad or not, but the only way I know of to check capacity is to do a full charge/discharge test.

You might try doing the fast tests on a sample of cells and see how it correlates to the actual capacity.

If the cells were not used much, their capacity should be pretty close to spec if they are not damaged. You can tell a lot about A123 cells from their resting voltage. Anything below 2v is bad for sure. Good ones tend to stabilize around 3.3v. Bad ones also tend to have accelerated self-discharge, so that is one more fairly easy test to do. Charge a bunch of them to 3.65v and let them sit for a few days, then measure voltage.

28850 is a mountain of cells. How did you get that many?
I settled on a methodology not that far off from that + a little bit. I'll stay quiet on that until I can test a few hypotheses as there I no point in spreading misinformation... Just some hunches I want to explore for now.


I got some 3500 cells XD


Sent from my CLT-L04 using Tapatalk


BYqSXt8Z   100 W

100 W
Posts: 119
Joined: Jul 03 2014 8:29pm
Location: GTA, Ontario, Canada

Re: Testing large amounts of cells quickly

Post by BYqSXt8Z » Sep 02 2019 10:29am

So to get the correlations down I'll be doing some endurance testing on some sacrificial cells (1000 cycles). The issue I'll have is it takes around a month to run a single cell. That means I need to test with many cheap testing rigs, so I've designed one from jellybean components and modules, controlled by an arduino. Purchasing in volume will be difficult financially, but would bring the cost down considerably so I can run 100+ cells at a time. If anyone is interested in a 1s cell cycler, please let me know, I'd like to sell some to help pay for the construction of the large setup. It is possible to do more than 1s, but voltage resolution goes down to 0.025v instead of 0.005v

By the way, I've attached some pictures of my current more manual setup for larger packs to show I'm serious about getting this done, and I've also already invested in a precision impedence tester.

Sent from my CLT-L04 using Tapatalk




ImageImage

User avatar
fechter   100 GW

100 GW
Posts: 14073
Joined: Dec 31 2006 3:23pm
Location: California Bay Area, USA

Re: Testing large amounts of cells quickly

Post by fechter » Sep 03 2019 7:58am

Wow, that's a setup. Bet it generates a lot of heat at times. I like the load resistors. The "big guys" use synchronous inverters as a load to dump the energy back into the grid. Not that it saves that much money but it greatly reduces the heat generation.

I've messed with cell impedance and found it doesn't really tell you much unless the cell is completely toast. Readings were all over the map depending on SOC and temp. Cell IR is a much more reliable indicator. I guess it will depend on what frequency the impedance is measured at.
"One test is worth a thousand opinions"

BYqSXt8Z   100 W

100 W
Posts: 119
Joined: Jul 03 2014 8:29pm
Location: GTA, Ontario, Canada

Re: Testing large amounts of cells quickly

Post by BYqSXt8Z » Sep 03 2019 8:08am

Thanks :) I'm not super happy with the reliability of those b6 but no surprise there. They are still very helpful for bulk processing to ready everything for the precision tests or for regular use though. V1 of the arduino prototype will be tested today when I get my high current 22650 holder in :)

Isn't cell impedence just a fancy word for cell IR? From what I read the spec provided by mfg is not just resistance but impedance. I find the best tests are at around 80% soc for the lifepo4. I'm doing 1kh tasting, from what I've found that is suppose to be the standard, but the readings fluctuate non stop is my biggest issue. I might have to build an arduino split voltage tester for proper IR measurement.

In the winter that heat will be nice with the -40 temps we get here ;) Eventually I want to use the power to charge batteries on the opposite part of the cycle, but one thing at a time!



Sent from my CLT-L04 using Tapatalk





User avatar
fechter   100 GW

100 GW
Posts: 14073
Joined: Dec 31 2006 3:23pm
Location: California Bay Area, USA

Re: Testing large amounts of cells quickly

Post by fechter » Sep 03 2019 9:47am

Impedance measured at 1kHz won't tell you the same thing as IR measured at DC, but it's easier to make the tester. DCIR measurements can be done by watching the voltage sag and recovery from a high load placed on the cell for a few seconds. There are some testers out there that do this.
"One test is worth a thousand opinions"

BYqSXt8Z   100 W

100 W
Posts: 119
Joined: Jul 03 2014 8:29pm
Location: GTA, Ontario, Canada

Re: Testing large amounts of cells quickly

Post by BYqSXt8Z » Sep 03 2019 10:02am

fechter wrote:Impedance measured at 1kHz won't tell you the same thing as IR measured at DC, but it's easier to make the tester. DCIR measurements can be done by watching the voltage sag and recovery from a high load placed on the cell for a few seconds. There are some testers out there that do this.
Ummmm I went about it back wards, I though they used 1khz impedence for the spec sheets and I was trying to compare to nominal. Making a high current tester with arduino is a piece of cake... Darn.

Sent from my CLT-L04 using Tapatalk


BYqSXt8Z   100 W

100 W
Posts: 119
Joined: Jul 03 2014 8:29pm
Location: GTA, Ontario, Canada

Re: Testing large amounts of cells quickly

Post by BYqSXt8Z » Sep 06 2019 6:48am

I just confirmed, the a123 data sheet specifies ac testing at 1khz... I'll have to compare usefulness!

Sent from my CLT-L04 using Tapatalk


BYqSXt8Z   100 W

100 W
Posts: 119
Joined: Jul 03 2014 8:29pm
Location: GTA, Ontario, Canada

Re: Testing large amounts of cells quickly

Post by BYqSXt8Z » Sep 06 2019 1:00pm

Battery cycler prototype built! Now custom pcb coming to clean it all up.

Ps since there was a comment about the resistor board, it's super simple but here I the design anyways. I use jumpers to make any combination from 1 - 15 p resistors to get a bunch of different values. It's based on putting 5w resistors in the board, all 15 identical resistance at 2/3 ohm per v of battery you plan on testing (so I rounded to 2.2 ohm for testing 1s lifepo4 for example)

https://easyeda.com/editor#id=|189505d6 ... d7c43e32f7
Image

Sent from my CLT-L04 using Tapatalk


User avatar
fechter   100 GW

100 GW
Posts: 14073
Joined: Dec 31 2006 3:23pm
Location: California Bay Area, USA

Re: Testing large amounts of cells quickly

Post by fechter » Sep 06 2019 3:35pm

What is the little display unit you're holding? Arduino, but I don't think I've seen that display...
"One test is worth a thousand opinions"

BYqSXt8Z   100 W

100 W
Posts: 119
Joined: Jul 03 2014 8:29pm
Location: GTA, Ontario, Canada

Re: Testing large amounts of cells quickly

Post by BYqSXt8Z » Sep 06 2019 4:10pm

Oled 0.91" display

Sent from my CLT-L04 using Tapatalk


BYqSXt8Z   100 W

100 W
Posts: 119
Joined: Jul 03 2014 8:29pm
Location: GTA, Ontario, Canada

Re: Testing large amounts of cells quickly

Post by BYqSXt8Z » Sep 08 2019 2:54pm

If you're going to use it, the SSD1306ASCII library is amazingly small and efficient.

Sent from my CLT-L04 using Tapatalk


Post Reply