PreTest New Cells (Delta OCV) B4 Spot Welding

[eMark]

It's really your responsibilty (not the manufacturer) to Pre-TEST your Brand New 18650 cells and decide whether you'd rate them as Grade A or Grade B before building your pack. Hopefully, saving you some future frustration. You'll need to devise your own benchmarks for grading new cells, cells from a used pack or bargain basement salvaged cells. Then sort the cells according to the best use: Raw Performance, Casual Crusing or another application.

It's your call how precise to be (e.g. 0.001V, 0.005v, 0.010v -- A, B, C) depending on the application. Depending on your criteria you may disqualify more cells that you were led to believe were Grade A cells, because 10% of the cells fail according to your pre-test benchmark as not being Grade A According to your demarcation line between A & B.

The following diagrams are from an article on how the Delta OCV (self-discharge) of NEW cells is measured. The unknown is the percent of a manufacturing run that's tested with the objective being zero defect tolerance. Thus how manufactures arrive at what they consider Grade A cells and Grade B cells. Every manufacturer determines what they consider the allowable difference between a Grade A cell and a Grade B cell. Manufactures use the more expensive Potentiostatic Method for grading cells because of its immediacy (within 1-2 hrs) ...

Two methods for determining Delta OCV self-discharge whether new cells, useful pack cells or unknown salvaged bargain basement 18650 cells

This is basically the method that goatman and i are using to test capacity and self-discharge of his 30Q 17s4p and my 30Q 10s3p. Because our cells are not new (many c/d cycles) we decided to rest our packs (equalize cell voltage) for 5 days instead of 10 days because only -0.001mV after 5 days is unlikely with our used (problematic 30Q) cells. Moreso the scenario with my pack then goatman's **regenerated** pack.

We started our tests on December 16th and have been PMing each other to establish testing criteria, resulting data of our two used 30Q packs and what comparative results to share (KIS). Hopefully easy enuf for others to follow along and participate with their own suggestions, testing criteria, test data, etc. We'll start sharing our comparative results in a couple days.

Both 30Q packs (Plasti-Dipped 17s4p & Vruzend V2.1 10s3p) were equalized with p-groups (68 cells & 30 cells) at 3.228v avg & 3.226v (17s4p & 10s3p) before charging to 90% at 2.0A rate instead of FULL at 4.20v. Charge time could've been shortened by only charging p-groups to 3.90v (70% charge) as shown in above chart, but we both wanted to see what p-group capacity was with a 90% charge. Temperature 75°F during 17s4p charge at 2.0A rate. Temperature 73°F during 10s3p 2.0A charge.

While we decide how best to go about sharing our comparative Delta OCV info (storage capacity & self-discharge) the following article gives an idea how manufaturers go about testing 18650 cells ... https://www.batterytechonline.com/testing-and-safety/measuring-lithium-ion-cell-self-discharge

 

[eMark]

Only a few have the interest, time and equipment to do justice to PreTesting 68 brand new cells (17s4p) or even just 30 brand new cells (10s3p) to determine they're really “Grade A” before building their DIY spot-welded battery pack. Most just take it for granted that the expensive brand name cells they purchase are "Grade A". What is passed on to a consumer as "Grade A" may be more a matter of the manufacture's opinion than fact.

If only Elon Musk could hold all 18650 cell manufactures to adhere to his benchmark so we can be confidient the "Grade A" cell we buy meets Elon's benchmark for a "Grade A" 18650 cell. Do we actually believe 18650 cell manufactures hold to that 18650 standard of the above article when classifying what cells they pass on to us as “Grade A” ?

Thanks in part to what some consider the infamos 30Q this ES Battery Forum sees its share of homemade designs for ease of disassembly to replace what was supposed to be a "Grade A" cell. The latest thread being APL’s Pressure Contact Battery box ingenious design … https://endless-sphere.com/forums/viewtopic.php?f=14&t=113512 ... The NESE kit is a popular aternative to soldering or spot-welding for ease of replacing a bad cell.

Another way to determine if your cells are actually "Grade A" is to test the individual cells after at least 100 c/d cycles. That's assuming your pack design is so inclined. In other words is your packs' design such as to easily test the OCV self-discharge of the individual cells (not just the p-groups). The above “Delta” OCV self-discharge article test gives one the impression that the Delta OCV of all 18650 cells sold as "Grade A" have no more OCV self-discharge than 1-2mV .. Yeh Right!

My benchmark for determining which of my thirty 30Q "141" cells were "Grade A" in my experimental Vruzend 10s3p pack couldn't be determined until after 125 c/d cycles in 2020 ...

These thirty 30Q "141" cells were from 5 different manufacuturing runs (removed label shrink wrap). The following grading of the thirty 30Q "141" cells (after 125 c/d cycles in 2020) was with cells at 3.75-74v and after resting for 14 days from 1/1/21 to 1/15/21 (see above chart) ...

Grade A … 20 out of 30 (no more than -0.01v OCV self-discharge after 14 day rest)
Grade A- to B+ ... 5 out of 30 between -0.02v to -0.05v (3.73v cell to 3.70v cell)
Grade B to B- … 4 out of 30 between (3.68v cell to 3.57v cell)
Grade Fail … 1 out of 30 (3.43v cell)

After 125 additional c/d cycles in 2021 (total of 250 c/d cycles) 6 more of the "141" Grade A cells need to be replaced because of unacceptable OCV self-discharge. Samsung 30Q cells are rated for cycle life of 250 c/d cycles. What to do and how to it with this pack going forward into 2022 is still undecided

NOTE: My understanding of the use of "Delta" OCV self-discharge is only for PreTesting of NEW Lion 18605 cells to determine if they are good or bad. Good enuf to pass as a "Grade A" cell or bad enuf to pass as a "Grade B" or "C". The unknown is whether there is any industry standard that 18650 cell manufacturers adhere to when it comes to determining what qualifies as a "Grade A" cell.

 

[99t4]

These thirty 30Q "141" cells were from 5 different manufacuturing runs (removed label shrink wrap).

Do any of the outlier cells correlate to a particular manufacturing run in any meaningful way?

 

[eMark]

Do any of the outlier cells correlate to a particular manufacturing run in any meaningful way?

The problem is that 18650 manufactures don't test every cell. At best maybe 10% (MIL-STD-105E) of what is thought to be a Grade A line ... until QC testing at least 10% of the run to verify if they're "Grade A".

Even if we were a QC plant technician what's passed as "Grade A" on a 1st Shift run may have a minor hickup toward end of 1st Shift and continued on 2nd Shift with the next Production Supervisor/Mgr deciding to pass them as Grade A even though they may be closer to Grade A- or B+. Just being realistic having been around the block more than a few times. Now at 77 yrs young having worked in different fields i've seen human nature for what it is in all it's brilliance and frailties. Thus the push for Artificial Intelligence to eliminate human error whether unintentional or on purpose.

It's really the responsibility of the DIY builder to PreTest his expensive 18650 cells and/or use a non-spot welding design so as to periodically check OCV self-discharge of "individual" cells with more and more c/d cycles. Even those expensive "Grade A" 18650 cells may include a few A- or B+ cells. You really won't know the actual cell condition until at least 100 c/d cycles ... if not sooner (JMO).

 

[eMark]

Do any of the outlier cells correlate to a particular manufacturing run in any meaningful way?

The answer is definitely NO (as best explained in above reply).

In a way your question goes to show how we've all been taken in (me included) by what marketing wants us to believe. Samsung's 30Q has been bad-mouthed (problematic) for years because of higher self-discharge than other high energy dense cells. Did Samsung do anything to correct the problem by improving Quality Control? Samsung's latest Made In China 30Q6-T supposedly has a noticeable "improvement" in power.

What users really want more than improved power is less self-discharge, more "Grade A" cells and fewer Grade B cells included in their order. Out of the thirty 30Q cells in my 10s3p rated only 20 as Grade A, 9 as Grade B, 1 as Grade C ... based on **unacceptale** self-discharge after 125 "Real Life" c/d cycles in 2020 (see chart ... https://endless-sphere.com/forums/download/file.php?id=288919 )

How many want to take the time (i didn't) to pretest 30 individual cells (10s3p) or 68 new cells (17s4p) according to this test ... we're all in a hurry to build our pack and get on the road.

The Dilemma: So you do the above test and conclude that 1/3 of your new cells are "Grade B" according to the above test as was the case with my 30Q cells (a best my hindsite) after 12I5 c/d "Real-Life" casual discharge cycles ... not raw performance.

Seems like what we all need is a pack design that's easy to disassemble so you can periodically check your cells' condition ... OR ... use only cells that are "Grade A" and even then it wouldn't hurt to pretest them.

 

[eMark]

Here's a less sophiscated test, but good enuf whether new 30Q6-T or older 30Q "136" cells. You really won't know how many are good enuf for your application use until you grade them according to your benchmark of what's **acceptable** to you and what's not.

FIRST: Check the resting voltages of all your new or not-so-new 30Q cells. For the sake of this example lets say all are at3.40v. If not them group them as best as possible. You could have as many as 5-10 groups with old cells even though they've never been out of the box.

SECOND: Connect all your 30Q cells in parallel or as many as possible at one time and charge the string/group of paralleled cells to 4.10v (90% capacity) and let them equalize for 5-10 days. The higher the charge the more likely the increase in Delta OCV self-discharge. In the above test the cells were charged 70% (3.90V). This is better from the standpoint that the Delta OCV self-discharge shouldn't be as great. Plus it takes less time to charge :thumb:

THIRD: Separate to individual cells and let them rest (72-75 F) for 10 days.

FOURTH: Measure individual cell voltages for degree of Delta OCV self-discharge. Hopefully most will be "Good Enuf" ... congratulations! "Good Enuf" depends on your own application grading system for what you consider is **acceptable** Delta OCV self-discharge after 10 days. For some (depending on application) their benchmark decline after 10 days shouldn't be more than -2mV for others depending on application -5mV may be acceptable after 10 days (from 4.100v to 4.095v). The benhmark of an ES member who makes custom packs doesn't use any cell that declines more than 7mv. From his experience that usually means at least 3 out of 100 30Qs don't make the grade. So one can only imagine how many wouldn't make the grade using the professional industry benchmark in the above chart.

FIFTH: Establish your own grading benchmark that you consider **acceptable** for your 18650 application.

As a guessimate grading benchmark for old (never used) 30Qs that can be purchased for $1 each (quantity of 100) is first to see how many don't decline more than 7mV after 10 day rest (4.100v to 4.093v) having been parallel charged together to at least 3.900v and equalized together for at least 5 days if not 10 days. Then separate and let individual cells rest for 10 days (72-75 F) before grading.

You'll need to get a DMM that reads three decimal places. This one was suggested to me by the ES member that uses 7mV as his benchmark. The actual voltage may not be as accurate as a $100 DMM, but all your using it for is making dc voltage comparisons in millivolts ... https://www.amazon.com/gp/product/B07TWQ4ZSJ/ref=ppx_yo_dt_b_asin_title_o06_s00?ie=UTF8&psc=1

 

[99t4]

What users really want more than improved power is less self-discharge, more "Grade A" cells and fewer Grade B cells included in their order.

Or maybe, at least for all the cells to exhibit similar characteristics (capacity, current delivery, self-discharge, etc.) so as not to cause future problems over the assembled battery pack's lifetime.

 

[docw009]

I began my battery building a few months ago. I've built two packs, 13S-3P and 10S-2P. using new cells. Didn't pretest, other than spot checking one cell for mah. Turned out OK and it got cold before I could really see how they worked. Using a 24 buck aliexpress welder, but it works with .20 mm nickel.

I recently bought 100 new, but "old" Samsung 30Q's for a dollar each from Battery CLearing House.I sampled five at 3100MAH. Have four more in the Littokala tester. The cells were shipped at 3.43 volts. I plan to charge them all to 4.2V, and check the self discharge for a few week. I am using 4.2V only because that's what my charger does. I don't think it will hurt for a few weeks.

 

[MorbidlyObeseKoala]

Thanks for putting together this procedure. I think as more and more used cells hit the market at affordable prices, DIYers need to have some kind of reliable and repeatable method to weed out weak cells so as not to risk their own or neighbors' safety. Ideally, running basic tests will get the lowest hanging fruit and weakest cells out of the mix so one can rely on parallel group measurements to monitor as the pack ages.

A data point that a number of individual cell chargers offer now is an IR measurement. Is that not worth using as part of this initial test phase? It seems like another data point to group similar cells together. Anything that had really high IR you could mark with a red pen to keep an eye out for on the self discharge test.

Thus the push for Artificial Intelligence to eliminate human error whether unintentional or on purpose.

This is an entire topic of its own but AI is really another man made tool and should be treated as such. I.e.- It has inherent flaws and biases built in from its maker that must be reviewed with a critical eye. Our machines and measuring devices have greater precision and speed than ever before but they still need calibration to make sure they are pointing to the true measurement/solution.

 

[eMark]

A data point that a number of individual cell chargers offer now is an IR measurement. Is that not worth using as part of this initial test phase? It seems like another data point to group similar cells together. Anything that had really high IR you could mark with a red pen to keep an eye out for on the self discharge test.

From my preliminary tests a year ago there wasn't enuf IR difference between a bad cell at 3.43v with OCV self-discharge of -0.031mV after 14 restful days and a good cell at 3.74v with Delta OCV self-discharge no more than 0.010mV(if that much) after 44-45 restful days. The difference in IR, if my memory is correct was only 1-2 Ω difference between these two cells.

One needs a more accurate IR test than that of the 4-point AideTek SM8124A in order for IR cell tests to be a better test for distinquishing a good cell from a bad cell than that of Delta OCV test graph shown above.

 

[eMark]

OPPSY ... double post after an edit clarification ... thought i was editing this post ... so will delete this post in favor of the following ...

 

[eMark]

I plan to charge them all to 4.2V, and check the self discharge for a few week. I am using 4.2V only because that's what my charger does. I don't think it will hurt for a few weeks.

You definitely don't want to charge FULL to 4.2V and then let them rest for "a few week". The higher the charge the more likely the higher the OCV self-discharge. Also, only charging 70% (3.90V-3.91V) takes less time. A 70% charge is what you want as you will initially use this chart as your guide. Then you'll establish your own benchmark for what's a good enuf outdated 30Q 136 cell for your application. Hopefully more easy going applcation than raw performance ...

Periodically check the charging voltage and when the paralleled cells reach 3.90v-3.91v stop the charge. Then let the cells rest for 10 days before recording the Delta OCV self-discharge of each individual cell. Then it's up to you to decide your benchmark if you want at least 80 out of 100 to be good enuf for your application. Hopefully no more than an OCV self-discharge than -0.007mV after resting for 10 days.

Need to clarify that you'd want to let the cells relax for nine days after charging before recording individual cell voltages to 3 decimal places. Then rest another 10 days and again record individual cell voltages. This voltage(mV) decline from the 9th day to the 19th day (according to the chart) will be your timeline for determining what cells are good enuf and what cells don't get a pass. Hopefully, no more than 20 will have a Delta OCV self-discharge decline more than -0.007mV.

With these outdated (but unused) 30Q cells you may decide to give any cell that doesn't decline more than 8mV of self-discharge a passing grade. It really depends on your self-discharge benchmark as to what self-discharge you deem allowable if your objective is say to give at least 84 out of 100 a passing grade. Maybe an OCV self-discharge that's no more than -0.008mV will be your benchmark for grading at least 84 out of 100 (12s7p or 14s6p) as good enuf for more of an Easy Going Cruising application than that of Raw Performance.

A 6p pack will/should provide more cycle life than say a 3p or 4p pack. Something to consider when deciding whether to build a 12s7p pack or 14s6p pack with 84 cells good enuf to get a passing grade. If only 65 get a passing grade and all you really need is a 13s5p pack than 65 cells out of 100 is your benchmark :thumb:

Hopefully you'll use a pack design that's easier to assemble, disassemble and reassemble than a DIY spot-welded pack; especially when using older(unused) outdated cells. If a few of the unused 35 (13s5p) just barely missed getting a passing grade you could try using them to replace an underperforming cell in your pack. You might want to use 5mV as your benchmark. If that means only 50 out of 100 get a passing grade then a 10s5p pack may be just what you need if your DIY plan is a spot-welded pack.

 

[eMark]

Thanks for putting together this procedure. I think as more and more used cells hit the market at affordable prices, DIYers need to have some kind of reliable and repeatable method to weed out weak cells so as not to risk their own or neighbors' safety. Ideally, running basic tests will get the lowest hanging fruit and weakest cells out of the mix so one can rely on parallel group measurements to monitor as the pack ages.

Currently testing an easier method than the above (new cell chart) when pre-testing unused outdated 30Q cells as well as used salvaged 30Q cells.

The problem is that it's still a matter of opinion (based on application) as to which used/salvaged/outdated cells (30Q) are still good enuf to provide say at least 50 to 100 c/d cycles. It would take several field tests and by that time a relatively inexpensive DIY pack design could be available that's easy to assemble, disassemble and reassemble when a p-group cell needs to be replaced.

If a DIYer were to pre-test 100 used/salv. 30Q cells to hopefully find 68 cells good enuf (17s4p) he may realize that what passes as good enuf REALLY isn't good enuf. But what else is he going to do, but come to the conclusion that he needs a pack design that is easier to assemble, disassemble and reassemble than spot-welding.

A data point that a number of individual cell chargers offer now is an IR measurement. Is that not worth using as part of this initial test phase? It seems like another data point to group similar cells together. Anything that had really high IR you could mark with a red pen to keep an eye out for on the self discharge test.

To measure the IR of new 30Q cells would probably require the likes of a +$20,000 piece of equipment like the Keysight BT2152A Self-Discharge Analyzer ... https://www.keysight.com/us/en/product/BT2152A/selfdischarge-analyzer.html?jmpid=zzfindbt2152a . As far as testing the IR of ALL cells (not just p-groups) at say 125 cycles, 250 cycles, etc that option is already possible with at least two balance chargers costing less than $100. The problem, if it is a problem, is for a DIYer to originally use a pack design that is relatively easy to assemble, disassemble and reassemble than is a spot-welded pack.

Hopefully Samsung's latest 30Q6 150 ___T undergoes improved QC testing and imrpoved grading of cells, such that the 30Q6T cells that are passed as Grade A are all legit Grade A (not prone to high unacceptable self-discharge).

If Samsung QC had set their bar higher (with 136, 138 and 141 30Q cells) for weeding out 30Q cells prone for unacceptable self-discharge there wouldn't be the need for pre-testing new 30Q cells. Until then a DIYer should make some attempt to determine the condition of 30Q cells (whether new or used) before spot-welding a DIY pack only later to find out to determine which ones are good (Grade A) and which ones are bad (Grade B).