DIY HI POWERED PACK: First time, General questions

[Jonathan1981]

Cant you increase the voltage any higher?

Nope. No room, won't work with chosen motor or controller.. 21700's are the future. I'm sold. I'm just prepping for this project anyway, no rush. By the time I'm ready, they'll probably come down in $$..

 

[Jonathan1981]

...you guys never mentioned the magical 21700 form factor! ...

My next pack (2+ years from now) will likely be 21700 cells for the very reasons you cite.

M

No doubt, they're magic! better in every way except price.... I'll build this pack and by the time I'm ready for another pack, solid state should be on the market ...

 

[SamRich]

...you guys never mentioned the magical 21700 form factor! ...

My next pack (2+ years from now) will likely be 21700 cells for the very reasons you cite.

M

No doubt, they're magic! better in every way except price.... I'll build this pack and by the time I'm ready for another pack, solid state should be on the market ...

I'm on a similar boat looking to build a 20S 15P pack @ 100A peak for commuting on moped so at least 3000 miles a year for 5 years.
I looked at the LG50 21700, but they seem to have the same energy and power density as MJ1s/M36s but at a higher cost per Wh. (Search on ES for test results) I've looked at cost/weight/power/cycles and MJ1 has come up on top for me.

See table below:

 

[flippy]

Try making that list with the 29E

 

[SamRich]

Try making that list with the 29E

I included it with a best guest of IR.
It the cheapest pack but comparable to the MJ1 in terms of Wh/$ so I'd rather buy that extra capacity.

 

[flippy]

22E IR is 22.7 from a random cell i pulled fresh from a new box (i got 3000 or so on the shelf here). most are between 19 and 25 depending on how you measure.
and because the price is lower you can buy MORE cells increasing the current capacity, lifespan, range and so on.

also the lifecycle is way wrong according to my data. i got a 4.2kWh pack that was charged to 4.05v and discharged to 3.5 or so and it lost less then 5% capacity over 1200 cycles.

i already posted lifecycle wear from my testbench on other topics.

 

[docware]

I included it with a best guest of IR.

The IR you included seems to be out of reality. Real values should be around these values ( ± 1 miliohm ) :

 

[SamRich]

I included it with a best guest of IR.

The IR you included seems to be out of reality. Real values should be around these values ( ± 1 miliohm ) :



ACIR v DCIR.jpg

The values you share are for news cells. I was estimating average IR over the life-cycle based on shared results on ES. For example 35E ages faster overtime than Mj1 even if both have similar IR when new.
Thanks for sharing the data.

 

[docware]

Well, I didn´t catch you mean aged cells estimation. I understand that better DCIR doesn´t ensure better lifecycle.

 

[Jonathan1981]

...you guys never mentioned the magical 21700 form factor! ...

My next pack (2+ years from now) will likely be 21700 cells for the very reasons you cite.

M

No doubt, they're magic! better in every way except price.... I'll build this pack and by the time I'm ready for another pack, solid state should be on the market ...

I'm on a similar boat looking to build a 20S 15P pack @ 100A peak for commuting on moped so at least 3000 miles a year for 5 years.
I looked at the LG50 21700, but they seem to have the same energy and power density as MJ1s/M36s but at a higher cost per Wh. (Search on ES for test results) I've looked at cost/weight/power/cycles and MJ1 has come up on top for me.

See table below:

Mj31's are 3.5ah, LG M50's are 5ah with just the slightest increase in cell size.... so M50's win hands down except for price...
am I missing something?

 

[docware]

Yes, you are missing the fact that your expectation regarding both 18650 and 21700 is unrealistic.

 

[SamRich]

Mj31's are 3.5ah, LG M50's are 5ah with just the slightest increase in cell size.... so M50's win hands down except for price...
am I missing something?

I don't have the numbers here for the m50s but when I calculated how many cells I could fit in the same volume, I got pretty much the same Ah in mJ1 as m50s. Your box might take better advantage of the 21700 form factor but the increase in size isn't as "slight" as you think.

Volume of 18650 = 16532mm^3 -> 0.212 mAh/mm^3 for 3500mAh
Volume of 21700 = 24233mm^3 -> 0.206 mAh/mm^3 for 5000mAh

So Mj1 is actually slightly higher energy per volume + 18650 will also have a higher packing fraction (less air between cells).
21700 are cheaper to produce than 18650 (per mAh) and cheaper top assemble in batteries (fewer cells per Ah). So that's likely why we are seeing push towards largers cells.
But for DIYer 18650s MJ1 are better options at current cost.

 

[flippy]

considering its a moped (usually constructed for SLA) there is pletly of room for a bigger pack. i put 4+kwh packs in them all the time. using cheaper cells and more of them increases the capacity, lowers IR, increases lifespan by several orders and makes for a better experience in general.

 

[Jonathan1981]

Yes, you are missing the fact that your expectation regarding both 18650 and 21700 is unrealistic.

care to explain?

 

[Jonathan1981]

Mj31's are 3.5ah, LG M50's are 5ah with just the slightest increase in cell size.... so M50's win hands down except for price...
am I missing something?

I don't have the numbers here for the m50s but when I calculated how many cells I could fit in the same volume, I got pretty much the same Ah in mJ1 as m50s. Your box might take better advantage of the 21700 form factor but the increase in size isn't as "slight" as you think.

Volume of 18650 = 16532mm^3 -> 0.212 mAh/mm^3 for 3500mAh
Volume of 21700 = 24233mm^3 -> 0.206 mAh/mm^3 for 5000mAh

So Mj1 is actually slightly higher energy per volume + 18650 will also have a higher packing fraction (less air between cells).
21700 are cheaper to produce than 18650 (per mAh) and cheaper top assemble in batteries (fewer cells per Ah). So that's likely why we are seeing push towards largers cells.
But for DIYer 18650s MJ1 are better options at current cost.

Sam, you don't need to go as far a pack size, just do a simple ratio comparing dims vs. capacity.

71/65= ~1.09
700/650 = ~ 1.09 so let's just say .09% larger...

capacity figures: 5/3.5 = ~ .43 % more capacity. .43-.09 = .034... so if my math and logic are correct, per unit size, the 21700 is 34% more energy dense... or in other words, if you had an 18650 (Mj1 etc.) and a 21700 of equal size, the 21700 would be 34% more capacity... so it wins.

This is obviously a rough estimate and is probably less in real life..... oh wait. Here you go.

"According to Tesla, Power density of 21700 goes up to 300 Wh/kg, which is about 20% more than current 8650 batteries that rates at 250Wh/kg."

But I don't think this negates 18650's, they'll be around for a long time still, especially as 21700's become the norm, 18650's are going to get really cheap making them a nice balance of kwh/ah/cost....

I'm going to design a pack for both cells...

Anyone know of a good place to get cell holders CNC machined out of sheets of ABS or acrylic???

cheers

 

[danlim]

Not particular on ebike application, more on powerwall application...
You may want to take a look: http://bit.ly/416OFPPK

 

[docware]

Yes, you are missing the fact that your expectation regarding both 18650 and 21700 is unrealistic.

care to explain?

I am afraid that your project is unrealistic. Warming of LG HG2 and Samsung 30Q in the closed battery pack even at 5A discharge is too high.

View attachment 4

At 5A discharge you can expect 30Q warming at the surface of the cell cca 28 °C. At ambient temperature 23 °C it means that metal can of the cell more than 50 °C warm. Temperature within the jelly roll is higher.

https://endless-sphere.com/forums/viewtopic.php?f=14&t=96360&start=25

Moreover, what range do you expect at 2C, 3C discharge ? What range do you expect from 30Q battery pulling 5A/cell ? Is 20 - 25 minutes ride enough for you ?

With 21700 the warming is only slightly better. Here is measurement of Samsung 21700-50E. DCIR of the Samsung 50E is cca 29 miliohm.





Warming of the cell was measured in expanded polystyrene isolation by thermocouples K, wires diameter 0,127 mm fixed to cell by Kapton tape 0,05 mm.


View attachment 2

View attachment 1



Rate of the cell cooling down after 5A switch-off indicates that expanded polystyrene isolation is not perfect lossless vacuum, that isolation used still have some heat dissipation.





Simplified, the warming is linear function of DCIR and quadratic of current.
LG M50 is similar high energy cell, you can expect similar results.

Moreover, your math and logic on 21700 is not correct.


My measurement is illustrative only. Heat development at 5A continuous current inside the real battery is much more complex. The cells in contact with the battery shell may have lower temp. The cells in the core of battery surounded by other cells may have higher temperatures that indicated here.


Flippy here sufficiently described problems of the heat development in such big packet.
He mentioned also need for cooling. So that some complication for you.

Maybe pouch cells would be appropriate solution.

 

[flippy]

pouches run into the same problem of heating, and expansion prevention that requires massive steel support sturctures.

2 options remain: more cells to spread the load or cells with lower capacity and much lower IR to prevent temperature melting the pack.
my option is just to make the pack bigger. it makes a battery better in every regard.
a properly sized pack runs well under 1C and preferably under 0.5C.

 

[Ohbse]

your personal experience is not valid for this use case. this is a motorbike, not a scooter or ebike. 17km is done in less then 9 minutes with a motorbike on a highway.

the >average< load for a motorbike is around 5kW. with your battery that would mean 70A continous (5.8A per cell). well into the range of self heating. this pack is also considerably bigger and the center of the pack would get WAY too hot with such currents and a high IR cell like the 30Q. you are hardly using those cells over their full range. a motorbike usually runs its pack almost empty.
500 cycles on a pack that will cost 2300USD in cells alone is economically "unviable" as you would wear out the pack in just tens of thousands of miles, for a daily driver that would mean 3 years at best so you would write off 1000+ bucks a year just in pack wear and that ignores the case the cells will get incredibly hot also reduding lifespan and possibly start a "pack ending event" due to overheating. especially during the summer on a highway.

no, in order to be economically viable you need longer lasting cells that are higher current, preferably cheaper and more of them. how much more depends on the physical space that is available. in order to increase efficencly and reducing currents i would crank up the voltage as high as possible. 144V or so if the controller can take that.

I disagree my experience is not relevant - realistically my bike is a featherweight motorcycle, averaging 48 wh/mile in real use. The OP described a pack that's substantially larger than mine - 42p rather than 12p. Average load will of course be higher for highway use, but in discharge C terms would be lower with a much larger pack. The 30q is ideal for my use, his use is *easier* on the cell than I am, so how is this going to melt to slag exactly?

Proposing a 29E cell to avoid performance and heat troubles over a 30q is ridiculous, it has a higher IR and less capacity, how exactly does that translate to superior performance? You are probably correct that the 29e performs better in extended cycle testing, however my point is that constant high load discharges are not representative of the real world load pattern of a motorcycle ridden in a city - pulling 300amps for more than a few seconds is not possible outside of top speed runs - you'll be crashing into traffic in no time. Testing projected pack temperature is an admirable goal, however testing temperature rise at 3c is irrelevant when per cell discharges will be closer to 0.6c at highway cruise. At this level internal heating will be entirely manageable, even in a densely packed, weather proof pack.

I absolutely agree with you regarding incorporating as much active material as you can in your given physical constraints - a very large 29e pack is very good, however a very large 30q pack is better in all respects other than price and perhaps theoretical longevity, again irrelevant if the projected mileage when that becomes an issue is 100,000+ miles.

To OP - 21700 cells are great, however they are the same energy/capacity volumetrically and by mass, despite being (last time I looked) substantially more $/wh and with availability concerns. You do have some advantage in pack construction, by merit of less connection points, however the packaging density factor is the same irrespective of the diameter of the cylinders. If your specific dimension constraints are better filled with 70mm length cylinders then you may be ahead of the game, but other wise 18650's are (for right now) a better bet in bang for buck terms.

 

[SamRich]

Mj31's are 3.5ah, LG M50's are 5ah with just the slightest increase in cell size.... so M50's win hands down except for price...
am I missing something?

I don't have the numbers here for the m50s but when I calculated how many cells I could fit in the same volume, I got pretty much the same Ah in mJ1 as m50s. Your box might take better advantage of the 21700 form factor but the increase in size isn't as "slight" as you think.

Volume of 18650 = 16532mm^3 -> 0.212 mAh/mm^3 for 3500mAh
Volume of 21700 = 24233mm^3 -> 0.206 mAh/mm^3 for 5000mAh

So Mj1 is actually slightly higher energy per volume + 18650 will also have a higher packing fraction (less air between cells).
21700 are cheaper to produce than 18650 (per mAh) and cheaper top assemble in batteries (fewer cells per Ah). So that's likely why we are seeing push towards largers cells.
But for DIYer 18650s MJ1 are better options at current cost.

Sam, you don't need to go as far a pack size, just do a simple ratio comparing dims vs. capacity.

71/65= ~1.09
700/650 = ~ 1.09 so let's just say .09% larger...

capacity figures: 5/3.5 = ~ .43 % more capacity. .43-.09 = .034... so if my math and logic are correct, per unit size, the 21700 is 34% more energy dense... or in other words, if you had an 18650 (Mj1 etc.) and a 21700 of equal size, the 21700 would be 34% more capacity... so it wins.

This is obviously a rough estimate and is probably less in real life..... oh wait. Here you go.

"According to Tesla, Power density of 21700 goes up to 300 Wh/kg, which is about 20% more than current 8650 batteries that rates at 250Wh/kg."

But I don't think this negates 18650's, they'll be around for a long time still, especially as 21700's become the norm, 18650's are going to get really cheap making them a nice balance of kwh/ah/cost....

I'm going to design a pack for both cells...

Anyone know of a good place to get cell holders CNC machined out of sheets of ABS or acrylic???

cheers

The numbers from Tesla are just general numbers for no specific cells and also you forgot to include the increased radius.

So let's try this again in more detail....
Compare energy per volume of 18650 vs 21700 specifically for LG MJ1 and M50s and note volume of cylinder is Pi*Radius^2*Height.


Volume of ONE 18650 is 16532mm^3 -> 3500mah/16532mm^3= 0.212 mAh/mm^3 for LG MJ1
Volume of ONE 21700 is 24233mm^3 -> 5000mah/24233mm^3 = 0.206 mAh/mm^3 for LG M50


Energy per volume is the same for both.

 

[Jonathan1981]

Yeah, that math is better.... like I said, was just a rough calc...but Tesla makes it seems like it's a more energy dense battery, it's misleading.

 

[SamRich]

Yeah, that math is better.... like I said, was just a rough calc...but Tesla makes it seems like it's a more energy dense battery, it's misleading.

Yeah it's just basic physics, unless you dramatically change the chemistry you can't get more energy out of a unit of li-ion battery.
Now that cells performance (energy and power density) has plateaued most effort is towards reducing cost. Going to larger volumes is one solution as explained above.

I personally would prefer effort going towards increased cycle life.

 

[Jonathan1981]

Yeah totally realizing that now.... it's just a different "can" size lol . Just crossing fingers that goodenoughs' glass substrate research comes through.

Has anyone here experimented with materials to help passively cool the batteries, like some sort of gel or putty that can be molded around the batteries that will absorb excess heat under high load? I'm a huge fan of EGO products, I've bought literally every one of their tools, and they're AMAZING. I understand they use a proprietary polymer around the cells in the battery that absorbs heat under high load....but I'm sure their proprietary stuff is expensive...

it would be awesome to find a cheap mold-able material the could do that. I was thinking something like mixing aluminum powder into a soft two part epoxy and molding a thin layer around each cell?..... thoughts, links etc???? just trying to think of a cost effective DIY cooling solution for such a large pack..

 

[SamRich]

Yeah totally realizing that now.... it's just a different "can" size lol . Just crossing fingers that goodenoughs' glass substrate research comes through.

Has anyone here experimented with materials to help passively cool the batteries, like some sort of gel or putty that can be molded around the batteries that will absorb excess heat under high load? I'm a huge fan of EGO products, I've bought literally every one of their tools, and they're AMAZING. I understand they use a proprietary polymer around the cells in the battery that absorbs heat under high load....but I'm sure their proprietary stuff is expensive...

it would be awesome to find a cheap mold-able material the could do that. I was thinking something like mixing aluminum powder into a soft two part epoxy and molding a thin layer around each cell?..... thoughts, links etc???? just trying to think of a cost effective DIY cooling solution for such a large pack..

I was hopeful too...until I did the math. 21700 are just a more expansive version of 18650s. The only upside I see is fewer cells to assemble.

 

[Ohbse]

it would be awesome to find a cheap mold-able material the could do that. I was thinking something like mixing aluminum powder into a soft two part epoxy and molding a thin layer around each cell?..... thoughts, links etc???? just trying to think of a cost effective DIY cooling solution for such a large pack..

Wax. http://www.allcelltech.com/index.php/technology/pcc-thermal-management

Phase change temperature can be varied with composition to target the sweet spot for optimal cell performance/lifespan.

The downsides are primarily mass and complexity. In many circumstances you would be better off taking the mass dedicated to the cell cooling and adding additional active material in the form of more cells. More of the same cell with the same power demands lowers per cell heating OR more cells allows for a more 'power' focused chemistry while hitting your energy requirements, again reducing per cell heating due to lower IR.

Essentially to maximise performance and longevity, you should use the largest possible pack you can package. The higher % of mass of your vehicle is active material (cells OR motor) the higher system efficiency will be in general.

On my latest skate prototype, over 75% of the total weight is in cells or motors. Next one will be even higher. On my bike the proportion is about 60%.