Battery assembly kit for 18650 cells - first prototype

[trevc2]

Great project. Just curious, what do the measurements end up at for the assembled pack?

Thanks

 

[circuit]

For 72 cells, roughly: 114x227x69 mm.

 

[circuit]

Continued discussion about cell-level fuses from this topic:
http://endless-sphere.com/forums/viewtopic.php?f=14&t=57291&p=856358#p856358

""
On every EV race contest. This is mandatory for all participants, written black on white in racing rules.
Also such fusing is available in my 18650 kit.
""

Well, I'm working as a consultant on a formula student car building the battery pack and I can tell you with 100% confidence that only one pack fuse is needed. So no, cell level fusing is not mandatory in every EV race contest.

Other rules include thermal monitoring of 2 thirds of cells and cell voltage monitoring on every cell.

This is strange, because many customers were talking about cell-level fuses. Also I recall this being discussed here on forums.
I will ask to show the document, when I get a chance.

Did a quick google search and found a nice english document. It is about snowmobiles, but probably not that much different. From page 51:

EV6.1.5 If more than one battery cell is used to form a set of single cells in parallel such that groups of parallel cells are then combined in series, then either each cell must be appropriately fused or the cell manufacturer must certify that it is acceptable to use this number of single cells in parallel. Any certification must be included in the ESF.

EDIT: the same text is in "Formula hybrid 2014 rules". Document attached.

The rules for formula student are avatible online, they are extremely boring. The pack I'm helping on is using single pouch cells of the required capacity, so there may be a special clause for 18650 cells. The whole pack comes in at 16Ah 540V nominal and a tad under 55 kg (of this, 45kg is cells). It should be able to deliver 85kW continously and ~120kW peak. There are high demands to cable isolation, pack relays, isolation barriers, no charging in car and, above all, documentation.

Did I mention this stuff is redicoulusly expencive? The final price of the pack is close to 3 USD pr Wh.

As for cell level fusing. You have a nice thing going with your 18650 kit Circuit. The downside is the cost of laser cutting or stamping these specialized sheets. As I've understood it, Tesla uses ultrasonically welded single conductor to each cell. Basically this lowers production cost and improves safety. I can see the need for fuses in a ~100p big car pack, but does the same apply to a normal ~4p ebike pack?

My opinion is that a properly designed BMS system, with thermal monitoring on all submodules, negates the need for cell-level fusing. You can find the same discussion about fusing batteries "the traditional way". In many cases it is simply a liabillity since it introduces two extra points of failure. Compared to a properly designed overcurrent shutdown system fuses are stone-age-tech.

We're abit off topic here, should we create a new thread? (Very interesting topic)

PS: I wrote this before you updated with Formula rules.

The most expensive part of my kit is plastic, by far... Milling adds up to a huge bill. If big enough order came, it would be reasonable to make a mold, but we would be looking at quantities of at least 2+k kits to at least cover the cost of mold and at least 10+k kits to be economically viable to offer a consumer-level price.
I also had a huge problem with no one being able to laser-cut the tabs, because it is too precise for most melatworking companies and too big for most engravers. Luckily, I managed to solve this issue. Stamping could also be a solution, but requires huge tooling investment as well.

I have reviewed many patents of tesla's and think that my design is unique... Probably worth a shot.

 

[trevc2]

I think this sort of cell level protection is very clever. It's never going to be really easy to swap a dead cell in any design, but at least this could save adjacent cells (or the pack) if something goes horribly wrong.

I sure hope you can find a way to manufacture these somewhat economically. I'd be interested in buying some sets ASAP

 

[spinningmagnets]

I think this is a wonderful project, and thanks for posting. I think there are a couple of things that are holding back more enthusiasms for this:

1) A cheap spot-welder has been seen on youtube made from a transformer from a free trashed microwave oven, but no data was published on the requirements for which transformer would be the best in order to get consistent spot-welds (whether new or used). Publishing a "how to" for a cheap spot-welder made from a transformer and foot-switch would be a big help.

There has been a thread on assembling a DIY $100 capacitive spot-welder, but even at only $100, it is a road-block to a lot of potential customers.

2) I am persuaded by you and LFP that individual cell-fusing is best, and likely will become the future gold-standard for factory and DIY packs. If a single cell shorts its fuse, describe how someone with your kit would "re-fuse" the new cell. Dropping-in a new cell and spot-welding it into the bus-strip is easy to understand, but should pack builders stock a few spare fused cell-tabs? (is the "fuse" simply the thin section of the strip between the central button that's over the cell and the rest of the strip?) pics would help.

 

[circuit]

I think this is a wonderful project, and thanks for posting. I think there are a couple of things that are holding back more enthusiasms for this:

1) A cheap spot-welder has been seen on youtube made from a transformer from a free trashed microwave oven, but no data was published on the requirements for which transformer would be the best in order to get consistent spot-welds (whether new or used). Publishing a "how to" for a cheap spot-welder made from a transformer and foot-switch would be a big help.

There has been a thread on assembling a DIY $100 capacitive spot-welder, but even at only $100, it is a road-block to a lot of potential customers.

you are very lucky, because I have built a spot welder from microwave transformer 4 years ago. And I have published it publicly, PCB layout, MCU program, etc. Here it is:
http://translate.google.com/translate?hl=en&sl=lt&tl=en&u=http%3A%2F%2Fcircuit.lt%2F%3Fsection%3Dprojektai%26page%3Dwelder
It is written in my native language (Lithuanian), but I hope automatic translation will be better than nothing.

2) I am persuaded by you and LFP that individual cell-fusing is best, and likely will become the future gold-standard for factory and DIY packs. If a single cell shorts its fuse, describe how someone with your kit would "re-fuse" the new cell. Dropping-in a new cell and spot-welding it into the bus-strip is easy to understand, but should pack builders stock a few spare fused cell-tabs? (is the "fuse" simply the thin section of the strip between the central button that's over the cell and the rest of the strip?) pics would help.

This is how one "tab" piece looks for 6p configuration:

So if one fuse goes.. Whole tab plate should be replaced. Or one could cut out damaged part, leaving healthy part intact and weld a new plate on top.

 

[mistercrash]

Can I assume here that if you take the already available nickel strips out there and punch half holes in it between the positive side of every cell, then let's say 1.2 to 1.5mm left from that half hole becomes a fuse? Or am I just desperately trying to simplify something that needs to be complicated.

 

[circuit]

Can I assume here that if you take the already available nickel strips out there and punch half holes in it between the positive side of every cell, then let's say 1.2 to 1.5mm left from that half hole becomes a fuse? Or am I just desperately trying to simplify something that needs to be complicated.

You assume correctly. That will decrease tab price, but assembly time will increase a lot. For DIY, probably it is not important.

 

[agniusm]

So how to rate a fuse link per material/material thickness chosen? Is there a formula or error/trial method? I would like to implement fuses into my future large pack and not sure what fuse rating should be chosen for individual cell, between series connection of paralleled cells and battery pack fuse? I have tested .15mm nickel strip trimmed to 1.5mm and it would burn at roughly 100A on a single fully charged 2000mAh 4.2V cell. Still need to calculate the time it takes to do so.
Another question i have is what length should be the fuse to avoid arcing?

 

[Honk]

That have been one of my thoughts as well...the fuse rate of the 4 wire cell connection.
These wires look very stury and low ohmic (which you don't want in a fuse system) and it
felt like it would take several hundreds of amps to burn of the connections but then it's
already to late to protect the cell as it already burn as hell....

If your rating of 16amps per fuse string is correct then it comes to 64amps going into a shorted cell.
This feels very high as it could easily start a fire at that current level.

Long ago I tested thin 0.05mm dia enamelled copper wire being 3mm long as a fuse.
The wire resistance goes up as it heats up.

The results:
Amp Load = V loss = Resistance
0.5A Load = 14mV = 28mR
1.0A Load = 33mV = 33mR
1.5A Load = 53mV = 35mR
2.0A Load = 81mV = 40mR
2.5A Load = 122mV = 49mR
3.0A Load = 190mV = 63mR
3.5A Load = Fuse wire breaks

And this was just a single strand 3mm x 0.05mm dia.
Imagine the capacity of the 4 heavy duty looking connections to each cell.

My point is that any fuse system should be matched to the current level needed.
If each individual cell in a pack shall deliver e'g a peak of 10 amps then the fuse
break rating should be set 50% higher at approx 15amps but no more than 20 amps.

 

[circuit]

I can assure you that I have done some calculations and real-life tests with various bridge widths. It works good enough

 

[Honk]

But have you tested a semi broken cell situation when it's not fully shorted yet consumes
lots of current below the 64amp fusing level?
Let's say the cell shorts down to 3V and draws 50 amps without fusing....that equals to
150W heating inside the cell...feels like a high risk for fire as the electrolyte quickly heats
up to fire level....

 

[circuit]

But have you tested a semi broken cell situation when it's not fully shorted yet consumes
lots of current below the 64amp fusing level?
Let's say the cell shorts down to 3V and draws 50 amps without fusing....that equals to
150W heating inside the cell...feels like a high risk for fire as the electrolyte quickly heats
up to fire level....

I have tested and tuned my individual fuses to 5amp/bridge. So this gives you four options (by cutting off 3, 2, 1 or none bridges): 5/10/15/20 amps or 1.7/3.3/5/6.7 C per cell, if you are using 3Ah cells, for example.

I don't know where do you get those numbers of 50 amps or 100A quoted by agniusm. Even for copper that should not be that high. My fuse material is not copper.

 

[agniusm]

Well not 100A but close. Here is a shorted 18650 cell:
[youtube]0kmL8e2-20Q[/youtube]
The 1,64mm fuse clears at 89A (1:21) I think it takes about 40ms to blow.
Another thing was getting the cell shorted without a fuse. Here is LG 2600mAh fully charged cell:
[youtube]DTTMs3GeAb0[/youtube]
no venting or fire. Am i missing the point here?

 

[Honk]

I have tested and tuned my individual fuses to 5amp/bridge. So this gives you four options (by cutting off 3, 2, 1 or none bridges): 5/10/15/20 amps or 1.7/3.3/5/6.7 C per cell, if you are using 3Ah cells, for example.

Well that's great, and it answers my questions and calms my concerns.
If you had posted these numbers within post #1 it would all be clear and no worries would emerge!

Great invention and I hope you succed getting it into the market for easy access and purchase.

 

[agniusm]

Here is another opinion on individual cell fuses by Davide from Elithion:

With N cells in parallel, when 1 to N-1 switches are open, the voltage on the switches is, indeed, cell voltage.

But when the LAST switch opens, all the open switches see pack voltage, negated, across them: fireworks.

Please see this video:
[youtube]9RVYLvn-jL8[/youtube]
that explains why the entire pack voltage appears across an open connection mid pack.

When just a few switches open, the rest have to carry higher loads, and they too get closer to opening. So they open faster and faster, until the last switch opens, and: fireworks.

Say you have 4 cells in parallel, each with a protecting switch built in.
One switch opens: no big deal (the capacity is down to 3 /4) {= first cell}
Second switch opens: we're in trouble, because now all the current is going through the remaining 2 cells
Third cell can't stand the extra current (double normal) and opens; still no voltage across the switches
Fourth cell is now asked to carry ALL the current! Its switch opens and full pack voltage appears across that switch {= last cell}

Any views on that?

 

[circuit]

Here is another opinion on individual cell fuses by Davide from Elithion:

With N cells in parallel, when 1 to N-1 switches are open, the voltage on the switches is, indeed, cell voltage.

But when the LAST switch opens, all the open switches see pack voltage, negated, across them: fireworks.

Any views on that?

Sorry I did not watch the whole video, as it is too dull for my taste.

I have already explained this issue somewhere before. Yes, individual cell fuses probably are not capable to withstand a blow with above 100V on them, this *could* result in arcing. These fuses are meant to protect against internal cell short or short between adjacent cell shells (in case of mechanical damage, overheating and melting of tray, etc).
However, the situation being explained is when the whole pack is shorted. I should remind you that at least one 'main' fuse must be used on whole pack. And two fuses is a better practice, protecting you in case of mechanical damage and several shorts to chassis. These 'main' fuses are the ones that must go when pack is overloaded, and individual fuses should remain intact in such case. It all requires adequate calculation and tests to work perfectly.

 

[agniusm]

I thought it was rather clever of him to make it simple for dull person. So how often an internal short occurs in practice? Might this be a fuss over nothing and series fuse and main fuse would suffice?

 

[circuit]

So how often an internal short occurs in practice?

Not very often. LiPO fires are also not very often. Many of us use LiPO a lot and most never had our houses burned down... But some do.

 

[agniusm]

Yes, some do, but i was referring to systems that have proper circuitry behind battery pack, e.g. management system. No point in talking about macaroni lipo packs and RC chargers where human factor does most, which is the case of almost all fires.

 

[circuit]

I was tracking battery fires for some time. Surprisingly I saw many fires with BMS and LiFePO4. Cause unknown. Not as many as our DIYers messing with "bulk charging" and "celllogs" on LiPO, but still, numbers not low enough to ignore.
Cell-level fusing comes at not high cost in my design, so it is wise to use it as additional level of protection.

 

[agniusm]

...and I like the idea as long as all the dots are placed and it does not make it worse. Sure your design will avoid arcing as it is low voltage and internal shorts occur probably never in most cases. Another thing, Tesla uses it on large packs so it must be viable. I'd like to see Davides point in action thou and I would like to hear explanation why my fully charged 2.6ah cell did not burst into flabes, exploded or smoked? Is it a matter of time?

 

[circuit]

I'd like to see Davides point in action thou

Maybe I will do such test when some spare parts will be lying around. But again, this has no big importance, because pack should be protected by at least one main fuse anyway.

and I would like to hear explanation why my fully charged 2.6ah cell did not burst into flabes, exploded or smoked? Is it a matter of time?

Probably too high internal resistance, as normal power cells overheat and vent in such situation.
Single cells are designed to withstand external short or overcharge by venting, which avoids temperature climbing further and eventually flaming up.
A real problem rises when cell shorts out internally. If it does, Ri drops way below rated Ri, so even "low power" cell with high Ri can become very conductive. In such case, if there is no fuse on it, all parallel cells will supply maximum current to this cell, whole group will start overheating and venting, tab on shorted cell heats up and melts, probably causing sparks and igniting electrolyte, already venting from other cells.

As to how common cell shorts are... Common enough. My colleague had this happen on his car. 100Ah ThunderSky cell:

Reads 0.001 Ω. He was lucky, because there were no cells in parallel. If there were... Hard to guess how it could end.
I have an article on this: http://e-motion.lt/2012/01/16/thundersky-kas-viduje/

 

[spinningmagnets]

I am not an electrician, but this sounds very good to have the individual cell fusing.

Concerning the issue of the fusing of the entire pack, I was very impressed by ES member teklektik's set-up on his Yuba Mundo. The hot-rod E-bikes get more clicks at the electricbike.com website, but I was able to sneak in an article on his Mundo because he is a professional electrician, and he provided a good explanation for every component he chose.

He used a breaker, and in spite of his caution, he ended up tripping it a couple times. The good news is that it worked quite well. He corrected the electrical problem, and then simply reset the breaker switch.

Alan used a 50A magnetic marine breaker for the main power disconnect switch (Blue Sea model #7230). These types of breakers are very rugged and are intended for high-humidity/salt-air use. They are designed for up to 65V-DC and can handle interruption currents up to 7500A making them virtually indestructible in this application. “They start to trip at 62A and with a couple of controller shorts, the CycleAnalyst never recorded a spike greater than 180A before the breaker switched off”...The 50A breaker has worked flawlessly and tripped so quickly that the 50A Maxi fuses in the battery packs never popped. “For around $15, I cannot recommend this breaker more highly”

http://www.electricbike.com/custom-build-gallery-teklektiks-dual-motor-yuba-mundo/

 

[circuit]

Yup, circuit breaker is a good idea, especially in prototyping stage. And we know how long prototyping stages DIY guys encounter... "There is nothing more permanent than temporary"... I've seen those on some builds.
And probably it is cheaper and faster than a good, safe and quick fuse.