A123 20AHr Pouch Cell Battery Build & Info Thread

[wb9k]

only 20amps on ebike in 48v pack,

OK, so just to be sure we're on the same page: the pack was 20 Ah pouch cells in 14, 15, or 16S, 1P. Max pack current was 20 Amps, or 1C. Correct?

lowest voltage was 2.4 v
that was voltage when pack was opened up

OK, so the pack was discharged to around 32 Volts (assuming 16S). I can believe that. I have a hard time believing that this particular cell didn't see some time at a very low voltage, or maybe was even driven negative at one point. Several of your photos suggest that the copper anode had begun to dissolve into the electrolyte. Flaking of the powder layer on the anodes often accompanies this, as do soft pouches. The cell has to come to rest below 300 mV to do this. Driving the cell negative is even worse. What was cell performance like after it "went soft"? Do you have any photos of the cell before it was dissected?

I think the difference between hard cells and soft cells will be when the graphite layer on
the copper starts to go off, was surprised just how good the iron phosphate layers held up.
They were pretty much unblemished ( iron phosphate layer), the graphite had starting peeling off the copper in places.

Cells can be soft for several reasons, not all of which represent a defect. The most common cause in my experience is overdischarge (which internal evidence also strongly suggests). Sometimes a cell will get a bit more electrolyte than normal, and this can make the cell less rigid, but it's not generally a problem.

When you get some electrolyte starts to be visible from the outside of the pack ( grey bubbles along seams), it looks like that
is due to breakdown of the mylar outer skin, not due to contact with copper or aluminium plates. The inner surface of the mylar pouch appears to have electrical contact with the plastic membrane, first layer under the plastic separator membrane is copper.

The pouch is made of aluminum with a thin plastic layer on both the inside and outside of the pouch. Think about what you're saying here about Mylar--an insulator--having an "electrical connection" to anything. It makes no sense. The aluminum pouch is electrically isolated from the electrode stack. The defect you describe with liver spotting and electrolyte under the outer poly layer is usually the result of breached isolation between the anode and the aluminum pouch.

The aluminum
Innerside of mylar pouch has no iron phosphate coating, but possibly a very thin plastic coating ( not sure on that)so should play no role in battery capacity, but may hold some charge or at least a conductive path, so cells with cut down tabs and mylar removed to make tabs longer might short from mylar pouch to copper tab ( a bit of silicone seams to solve that issue)

There can be some capacitive "surface charging" of the pouch by the electrode stack, but it represents a tiny amount of power. Trying to salvage cells with tabs that have been cut down below the pouch edge is a recipe for disaster. Tweak the cell just right and you've got a dead short across the pouch itself--no fuse can save you now.

The separator does not "hold" the electrolyte, BTW. The electrode stack with separator go into the pouches before the electrolyte is added. The separator is a plastic that has high enough dielectric strength to prevent the electrodes from shorting together, but is still porous enough to allow Li ions to pass through from anode to cathode and back. The lithium ions are suspended in the electrolyte.

 

[whatever]

it was 16s pack, this cell was on the positive end. This was the second cell which lost voltage on the positive end of the pack,
without any pressure on this cell, the cells were stacked so that the negative end cell was at the bottom taking most of the pressure. All other cells still good except the two cells which went soft on the positive end. The other cell that also failed previously in same position went to close to zero volts( I haven't opened that cell), the cell pictured was at 2.4v when I removed it from the pack, as far as I'm aware it never went below that. I never tried to recharge it after removal as it was obviously soft. No photos before dissection. Pack wasn't charging up correctly was when I took the cell out, but second time this had happened, it happened to another cell previously same position in the pack.
There was no breach of the electrolytic membrane, it was intact undamaged. There was no contact of copper to the mylar.
Mylar has a thin vapour deposited layer of aluminium, that layer of aluminium would be conductive, the mylar is heat sealed across the tab, I think its likely the very thin layer of aluminium would be in contact with the aluminium tab, the aluminium layer in the mylar would be extremely thin ( probably nanometer scale) and not likely to conduct are large current, but would be conductive since its aluminium. Hence the reason they have kapton tape only on the copper tabs! To stop them shorting with the aluminium layer of the mylar.
The electroytic membrane appears to me to hold the electrolyte. Just did a quick google search appears there are many types
of ion exchange membranes, they all depend on some sort of electrolyte held within the membrane, try a google search on the topic.

 

[wb9k]

it was 16s pack, this cell was on the positive end. This was the second cell which lost voltage on the positive end of the pack,
without any pressure on this cell, the cells were stacked so that the negative end cell was at the bottom taking most of the pressure. All other cells still good except the two cells which went soft on the positive end. The other cell that also failed previously in same position went to close to zero volts( I haven't opened that cell), the cell pictured was at 2.4v when I removed it from the pack, as far as I'm aware it never went below that. I never tried to recharge it after removal as it was obviously soft. No photos before dissection. Pack wasn't charging up correctly was when I took the cell out, but second time this had happened, it happened to another cell previously same position in the pack.

OK, a picture is emerging here that makes more sense. From the sound of it, there was almost no compression whatsoever on the top (+ end) cell, and slightly increasing compression for each cell underneath. 1C is about where compression starts to matter for these cells (I work at A123, you don't seem to be aware of that), so it makes sense that the top cell would fail first--twice in fact, so you now have a pattern established. This is a design flaw in your pack. With no compression, the layers of the cell develop space between them, elevating the internal impedance of the cell. I'm guessing you didn't use a BMS or balance very often or at all with this thing. Because of the elevated impedance of the end cell, a disproportionate amount of Peukert loss occurred across the cell during both charge and discharge cycles, fostering a continuously worsening imbalance (do you monitor cell voltages regularly?) The end cell eventually got far enough below the rest of the pack that it was driven negative (or very close to it) during a recent drive cycle, destroying the end cell's ability to hold a charge. This is probably when you noticed the pack "wasn't charging right". I would say with pretty high confidence at this point that this has happened to you twice now.

There was no breach of the electrolytic membrane, it was intact undamaged. There was no contact of copper to the mylar.

From your pictures, I don't think your teardown method was anywhere near informed or controlled enough to make that statement. If you had grey spotting on the external portion of the pouch, you almost certainly had a low impedance between at least one anode electrode and the aluminum pouch. Your + end cell being on top and uncompressed probably makes it more vulnerable to damage over time from not being properly affixed/protected within the pack structure.

Mylar has a thin vapour deposited layer of aluminium,

No, you're describing a bag for holding Cheetos. The cell pouches are very different--a rather thick layer of aluminum with super-thin plastic laminated onto both sides.

that layer of aluminium would be conductive, the mylar is heat sealed across the tab, I think its likely the very thin layer of aluminium would be in contact with the aluminium tab, the aluminium layer in the mylar would be extremely thin ( probably nanometer scale) and not likely to conduct are large current, but would be conductive since its aluminium. Hence the reason they have kapton tape only on the copper tabs! To stop them shorting with the aluminium layer of the mylar.

The pouches are heat sealed, but that's about the only correct statement here. The Kapton is present on both the aluminum and copper tabs. (Those are ultrasonic welds BTW.) It wouldn't matter how thin the aluminum was, if the pouch was shorting out the cell, it would fuse open and leak all over the floor or at least get hot and vent electrolyte.

The electroytic membrane appears to me to hold the electrolyte. Just did a quick google search appears there are many types
of ion exchange membranes, they all depend on some sort of electrolyte held within the membrane, try a google search on the topic.

 

[megacycle]

Hey thanks for this cell construction info wb9k.
Haven't heard this level of detail before, may go a long way in helping with pack construction, I did bring up trying an anodised aluminum case to contain a pack, to help with heat sinking, knowing the pouch foil is insulated from cell contact internally too, makes me think getting away with otherwise risky, compression of a pack directly by an aluminum box arrangement might be ok.

 

[wb9k]

Hey thanks for this cell construction info wb9k.
Haven't heard this level of detail before, may go a long way in helping with pack construction, I did bring up trying an anodised aluminum case to contain a pack, to help with heat sinking, knowing the pouch foil is insulated from cell contact internally too, makes me think getting away with otherwise risky, compression of a pack directly by an aluminum box arrangement might be ok.

It would take a quadruple fault to make your idea dangerous, but it's still possible. I would isolate the cells from the box with additional insulation. Make sure there is NO debris between or on the cells that could compromise the outer poly layer over time.

 

[megacycle]

It would take a quadruple fault to make your idea dangerous, but it's still possible. I would isolate the cells from the box with additional insulation. Make sure there is NO debris between or on the cells that could compromise the outer poly layer over time.

It's actually six layers of insulation, due to the insulating, anodised aluminum case walls, as well.
Yes understand all surfaces would need to be very clean and debris free, compromising both sides through, would be an amazing show, for all the senses

 

[whatever]

yup I wasn't aware of the importance of compression when I made the pack, very important indeed.
The cells are Korean made and definitely no kapton under the aluminium tabs on these cells, only under the copper tabs.
Oh you work at a123, they still have a plant running in usa? assuming your in usa.
I had connectors for cell log, but didn't check the pack often, my main aim in putting the pictures of the internals was just to show whats inside, not really a scientific study! interesting its ultrasound to weld the tabs, I was wondering how it was done, would be tricky to get every tab of every layer lined up and weld it. Pretty sure there was no breach of the electrolytic membrane certainly wasn't any obvious holes in it, looked to me like the very thin layer of plastic on the mylar like pouch had allowed electrolyte through, anyhow I hadn't seen any pics of the internals anywhere on the net, so thought might be useful. Compression is very important for these cells even at low amp levels.

 

[wb9k]

It would take a quadruple fault to make your idea dangerous, but it's still possible. I would isolate the cells from the box with additional insulation. Make sure there is NO debris between or on the cells that could compromise the outer poly layer over time.

It's actually six layers of insulation, due to the insulating, anodised aluminum case walls, as well.

Well, it's easy to think that, but try hipot testing the anodization layers sometime using large conductive pads, or rags damp with salty water and see how much voltage it withstands. The vast majority of anodized finished are full of pinholes and simply cannot be counted on for much in the way of electrical isolation. The cells are WAY more robust than the anodization--they can regularly pass electrode-to-pouch hipot tests of up to 6,000 Volts--which is as high as my tester goes (or any other tester Ive ever actually seen, for that matter).

Yes understand all surfaces would need to be very clean and debris free, compromising both sides through, would be an amazing show, for all the senses

Actually, even the nail penetration tests are a real bore the vast majority of the time. A123 cells are not the most energy dense cells in the world, but they fail far more benignly under extreme duress than any other equally powerful energy storage medium I've ever seen. It's one of my favorite things about them.

 

[megacycle]

Merry Christmas everyone to you and yours.
Bit of time alone before turkey

Well, it's easy to think that, but try hipot testing the anodization layers sometime using large conductive pads, or rags damp with salty water and see how much voltage it withstands. The vast majority of anodized finished are full of pinholes and simply cannot be counted on for much in the way of electrical isolation.

I'd be using a good anodised finish and if it's actual pin holes rather than raised abrasions shouldn't make much difference overall as the pouch plastic wont be able to enter the micrometer holes and testing tab to case with a low voltage source, obviously higher than pack voltage, but not doing a flash over test at 1kV, or dipping my batteries in buckets of conductive fluids, unless I was doing these tests for GM or similar, the packs are just for small, ELV, EV projects, that will never see overall pack voltage of more than about 100V or so.

The cells are WAY more robust than the anodization--they can regularly pass electrode-to-pouch hipot tests of up to 6,000 Volts--which is as high as my tester goes (or any other tester Ive ever actually seen, for that matter).

Wouldn't argue with that, unless the pack supply rails are somehow having spikes superimposed on them, i can't understand HV flash testing them, I'd test to around 2 times pack voltage.

Actually, even the nail penetration tests are a real bore the vast majority of the time. A123 cells are not the most energy dense cells in the world, but they fail far more benignly under extreme duress than any other equally powerful energy storage medium I've ever seen. It's one of my favorite things about them.

Good to here, very safe tech then.

 

[whatever]

on the korean made cells, there is no kapton under the aluminium tabs, the pouch is not insulated from the aluminium, it is in electrical contact.

 

[wb9k]

on the korean made cells, there is no kapton under the aluminium tabs, the pouch is not insulated from the aluminium, it is in electrical contact.

These must have been fairly early Korean cells. All those I've torn down (Korean included) have had the Kapton on both electrodes. A123's pouch cell mfg techniques were developed in that (now closed) Korean plant. Having the cathode tab weld exposed is not nearly as detrimental as having the anode exposed, but it's still not a good idea. It didn't stay that way long. It's doubtful it played a role in the case of your failure(s).

 

[wb9k]

yup I wasn't aware of the importance of compression when I made the pack, very important indeed.
The cells are Korean made and definitely no kapton under the aluminium tabs on these cells, only under the copper tabs.
Oh you work at a123, they still have a plant running in usa? assuming your in usa.

A123 has several plants in the US and China. There are also offices elsewhere in the world. The company never stopped producing cells, even during bankruptcy. I've been with the company since before bankruptcy and have worked there without interruption for nearly 4 years now. Not too surprising many folks don't know this--I continued to see articles in the mainstream press describing A123 as "bankrupt" over 1 year after the bankruptcy was finished and the company had totally restructured. Reporting of the bankruptcy and the circumstances around it was so poor (grossly inaccurate) that I was changed from an avowed news junkie to someone who pays almost no attention to the "news" at all. I no longer trust anything those people have to say--any of them. We no longer have reporters, just regurgitators of statements made by vested parties.

I had connectors for cell log, but didn't check the pack often, my main aim in putting the pictures of the internals was just to show whats inside, not really a scientific study!

No problem. There's been so much misinformation out there about A123 over the years, I always try to clarify what is said in threads about the cells. It's difficult enough to try to find a defect inside a cell that hasn't had the electrodes begin to dissolve. A cell in a state like this one is even more difficult to successfully troubleshoot. I don't want to discourage experimentation and learning, but it's also vitally important that blind guesswork not become conventional wisdom via the internet--so I chime in on these threads quite a bit.

interesting its ultrasound to weld the tabs, I was wondering how it was done, would be tricky to get every tab of every layer lined up and weld it. Pretty sure there was no breach of the electrolytic membrane certainly wasn't any obvious holes in it, looked to me like the very thin layer of plastic on the mylar like pouch had allowed electrolyte through, anyhow I hadn't seen any pics of the internals anywhere on the net, so thought might be useful. Compression is very important for these cells even at low amp levels.

Problems with the separator are usually limited to small creases on the outermost layer. The inner plastic layer of the pouch is practically invisible. We never look for holes in the plastic, we look for burrs and projections that could have pierced it. The outer layer is not really visible either unless it starts to peel. This usually happens because the aluminum has become brittle and cracked, allowing the electrolyte to then attack the outer plastic layer from the inside. Plastics cannot contain the electrolyte for long periods of time...they are too porous. This is why solid aluminum MUST be used for the pouches. It is the primary container, not the plastic. The plastic layers are present only for electrical isolation and corrosion resistance.

So it sounds like my theory for your failure seems plausible to you? No BMS and no regular checking of cell voltages/balance? Big time gotchas for the hobbyist community. Thanks for sharing your experience.

 

[arkmundi]

So it sounds like my theory for your failure seems plausible to you? No BMS and no regular checking of cell voltages/balance? Big time gotchas for the hobbyist community. Thanks for sharing your experience.

Yea, important lessons - either a BMS the other BMS - brain monitor system, aka active monitoring. My latest A123 AMP20 battery build - for the MAC 10T/Trek Shift 3 build - has 5 sets of wires:

• the main +/- leads to the controller
• the 2S*8S1P series lead; I've mounted the pack pannier style with 24V-8S on a side
• the +/- leads to the charger
• two 9-pin JST pigtails for the two celllogs I use for active monitoring
• a 4*4S-12V set of wires so I can plug in a simple 12V DC to AC adapter, for my laptop or other chargers

I've encased the two 8S bricks in aluminium plate cut to size and duct taped so as to provide a rigid box and compression. Part of active monitoring is a periodic inspection of the cells to see any visible signs of de-foiling or electrolyte leakage, proof that the compression is sufficient. While riding, I stay in the safe 80%<SOC<20% range. Hence my pack is never fully charged or discharged. I have yet to see a single cell fall out of range of 3.5<V<3.3, and is most often at 3.34V nominal. All to extend longevity of the cells. Hoping for 5 years.

 

[999zip999]

I have been balance charging and used a 1420 hyperionto it went south (In Dnmum hands for now ), then two thunder 1220, both are lighting up, but fade out. So I been charging my 24s with two 36v2a from pedago, not for lifepo4 but stops at 42v. Been using it for three months. So 83.8v. Charged, only 0.035% different.
Discharged with 30amp controller to 76.72v 0.035% or 3.220~3.170v 16.50ah
So for you guys took it back out.
73.1v and 17.86ah 0% 0.150% off
3.165v ~ 2.973v
Recharged with the two 2amp. to 83.8v or 95% at 0.040 , 3.519 ~ 3.487v
Just came back in balance. Wow. Bms what bms. O.k. most times just discharge to 15ah at the very most. 573 cycles 7653 miles.

 

[whatever]

@wb9k
certainly if I had actually used the cell logs I would have recognised there was a problem, it seems no compression was the culprit, I've been off esforum for a long time, so I forget usernames and the who the people are, I was aware you worked at a123 but long time off forum didn't recognise your handle.
Using kapton tape on both tabs would make alot of sense, quite possibly these are early cells. Do the serial numbers on the cells include a date, I might be able to find what year they were made

 

[wb9k]

@wb9k
certainly if I had actually used the cell logs I would have recognised there was a problem, it seems no compression was the culprit, I've been off esforum for a long time, so I forget usernames and the who the people are, I was aware you worked at a123 but long time off forum didn't recognise your handle.
Using kapton tape on both tabs would make alot of sense, quite possibly these are early cells. Do the serial numbers on the cells include a date, I might be able to find what year they were made

No worries, just wanted to make it clear I had some experience looking at these things. I also like to know if what I'm thinking makes sense or if there's some other scenario I haven't seen or thought of before (this is my job, after all), so I appreciate you "manning up" and letting us know that I was correct in my assumptions. Hopefully now you can revise your build so that the majority that remains will last a long time for you yet and we all got to learn something in the process. Life is good!

Dates are indeed embedded in the serial numbers, but decoding SN's is one bit of info I can't share publicly, sorry.

 

[wb9k]

So it sounds like my theory for your failure seems plausible to you? No BMS and no regular checking of cell voltages/balance? Big time gotchas for the hobbyist community. Thanks for sharing your experience.

Yea, important lessons - either a BMS the other BMS - brain monitor system, aka active monitoring. My latest A123 AMP20 battery build - for the MAC 10T/Trek Shift 3 build - has 5 sets of wires:

• the main +/- leads to the controller
• the 2S*8S1P series lead; I've mounted the pack pannier style with 24V-8S on a side
• the +/- leads to the charger
• two 9-pin JST pigtails for the two celllogs I use for active monitoring
• a 4*4S-12V set of wires so I can plug in a simple 12V DC to AC adapter, for my laptop or other chargers

I've encased the two 8S bricks in aluminium plate cut to size and duct taped so as to provide a rigid box and compression. Part of active monitoring is a periodic inspection of the cells to see any visible signs of de-foiling or electrolyte leakage, proof that the compression is sufficient. While riding, I stay in the safe 80%<SOC<20% range. Hence my pack is never fully charged or discharged. I have yet to see a single cell fall out of range of 3.5<V<3.3, and is most often at 3.34V nominal. All to extend longevity of the cells. Hoping for 5 years.

Sounds like this should last a good long time...hopefully well over 5 years. How old is it now? Have you been using the pack, or is this still "in progress"?

 

[whatever]

@wb9k
heres a serial number
app72161227,vgj5045b0149
made in korea
300396-001

can you extract a date of manufacture? without sharing your 'secret'

 

[wb9k]

@wb9k
heres a serial number
app72161227,vgj5045b0149
made in korea
300396-001

can you extract a date of manufacture? without sharing your 'secret'

That's an unusual format---indecipherable to me. That model number is for an older version of the cell....It would appear this predates any production I'm familiar with. May have even been from an experimental build.

 

[whatever]

intersting comments, I googled a123 20ah usa, and the serial numbers appear to be in the same format
as the korean cells I have,
for example this picture
http://shop.lipopower.de/bilder/produkte/gross/1438.jpg
it has a usa made cell and same format
So I'm not sure how you could say its an older cell, and an experimental cell based on what info?
What format are the cells you deal with in? can you post a pic of an example?

 

[whatever]

a bit more google image searching and I can only find cells with same format serial numbers as on my cells, either korean or usa made.
it would be a bit much to assume that all a123 cells pictured on the internet showing serial numbers either usa or korean are experimental cells and old cells?? dont understand how you can come to that conclusion?It might be better if you back up your statements with some evidence rather than speculation not based on anything apparent. Some examples of actual cells showing the serial number format your talking about would be useful, as it appears the serial number format is the same no matter the source of the cells.
It would be very useful to be able to identify year of manufacture.

 

[whatever]

plenty of pics in this thread showing usa made cells with same format for serial number
http://endless-sphere.com/forums/viewtopic.php?f=14&t=33939&start=100

 

[wb9k]

I see the problem now....you didn't give the number that contains the date code information. IOW, that's not the serial number. The model number of the cell given is now obsolete...I don't recognize the model number off the top of my head, but assumed, perhaps erroneously, that it was an obscure early model number. The cells have undergone several minor revisions since the Korean plant was closed, so nothing at all from Korea is truly representative of what is made in Livonia today, or even a year ago.

The lack of Kapton on the cathode weld is not what caused your failure, if that's your concern. Only an isolation breach on the anode side will lead to corrosion of the pouch.

 

[wb9k]

I see the problem now....you didn't give the number that contains the date code information. IOW, that's not the serial number. The model number of the cell given is now obsolete...I don't recognize the model number off the top of my head, but assumed, combined with the lack of Kapton on the cathode, that it was an obscure early model number. The cells have undergone several minor revisions since the Korean plant was closed, so nothing at all from Korea is truly representative of what is made in Livonia today, or even a year ago.

The lack of Kapton on the cathode weld is not what caused your failure, if that's your concern. Only an isolation breach on the anode side will lead to corrosion of the pouch.

 

[whatever]

wbk9 wrote:
I see the problem now....you didn't give the number that contains the date code information

that doesn't make any sense, as mentioned all the a123 20ahr cells that I can find on the net from usa/korea etc have same format, same format as I've given from the cells I have. All numbers are included, you seemed confused.
Have you got any pictures showing kapton tape under the aluminium tabs?