Custom battery pack with 18650s

[JakeTheDog420]

Hi,

I'm about ready to start purchasing some new cells from a seemingly legitimate source on ebay offering Panasonic cells. (Note I am aware of the inherent dangerous of purchasing cells off eBay but the source does seem trustworthy and I'll be using paypal to cover myself with their buyer protection programme.)

My question to you is this. I can purchase both High Capacity and/or High Discharge cells. Am I able to buy both and make a mix of the two for my back or will it produce negative effects? I know you cant parralel two sets of batteries with different voltages or capacity but can the same be said about discharge rate/volts etc??

And my idea was this to avoid issues;
13s4p 52 cells.
the top two sets of series connections could be with the higher discharge and the bottom two halves with higher capacity. (Note as well that the capacity difference wouldn't be massive in the two different types I think the main difference would be the discharge rate and maybe a couple hundred mah diference max between the two cells?)

I'm out now but can provide data sheets for both cells later!

Thanks

 

[redilast]

You never want to mix and match cells in the same pack. You want to use all the same new A-grade cells preferably from the same manufacturing batch.

If you are undecided between high capacity and high discharge rate, the best choice is probably to go with something like the Sanyo NCR18650GA's (Sanyo is owned by Panasonic). Good balance between high capacity (actually it is the highest capacity in a readily available 18650 cell) and mid level discharge current, rated at 10A continuous current.

What kind of motor do you plan on running so we can recommend what discharge rate cells you go with? I'd say generally speaking you want to keep your max discharge current to <1/2 of what the cells rated max is in order to get good pack cycle life and reduce heat.

 

[JakeTheDog420]

You never want to mix and match cells in the same pack. You want to use all the same new A-grade cells preferably from the same manufacturing batch.

If you are undecided between high capacity and high discharge rate, the best choice is probably to go with something like the Sanyo NCR18650GA's (Sanyo is owned by Panasonic). Good balance between high capacity (actually it is the highest capacity in a readily available 18650 cell) and mid level discharge current, rated at 10A continuous current.

What kind of motor do you plan on running so we can recommend what discharge rate cells you go with? I'd say generally speaking you want to keep your max discharge current to <1/2 of what the cells rated max is in order to get good pack cycle life and reduce heat.

Thanks for your quick reply, the motor I'm using is a full rear kit I purchased from cyclotricity link is below:
http://cyclotricity.com/kit-configurator/?xt=5
Configuration was as follows:
Custom bike
/ My Own Bike / Rear 48V 1000W Kit / 26inch / PAS&LCD / 7sp Freewheel
£364.97

Unfortunately it doesn't seem like they publish exactly what the max discharge current is etc...

And thank you for your suggestion of Sanyo cells. I did not know they were owned by panasonic! As I've already found a reputable vendor of Panasonic cells after much much MUCH effort I feel it would be slightly silly of me to go looking now for a vendor of Sanyo cells however if after further research it does seem in my best interest I may do so...

I knew mixing and matching cells isn't best practice but I did think if they were both A-Grade cells same manufacturer etc it would be okay but I asked the question cause I'm certainly no professional lol!
I'll post the datasheets and maybe I'll be steered in the right direction for which cell to go for? I'm thinking the higher capacity panasonics... Does anybody know what the specifications are of the 18650s in the tesla? Ideally I want my ebike to have a good range and be able to get up hills without much struggle. Not after too much speed but I did get the 48v 1000w version so it has that option open to me.

Regards

 

[redilast]

If you are going with a 1000W motor, many of these motors can actually draw 25A and surge even higher. But I guess it largely depends on also the controller.

If we assume 25A then with 4 cells in parallel that would mean each cell needs to put out 6.25A. In that case I’d probably recommend if you do go with the 10A cells that you go with more than 4 cells in parallel if they will fit. Maybe 5 or 6.

The Tesla cells are custom Panasonics, but the rumor has it that they are pretty close to the NCR18650BE cell, which is a 10A 3200mAh cell. It’s actually also a very good choice, lower cost than the GA’s per watt hour but slightly less energy density. I believe the internal resistance is also slightly lower meaning they should hold up to higher continuous loads better. I used BE's in my last scooter pack for a 24v SLA replacement.

 

[JakeTheDog420]

If you are going with a 1000W motor, many of these motors can actually draw 25A and surge even higher. But I guess it largely depends on also the controller.

If we assume 25A then with 4 cells in parallel that would mean each cell needs to put out 6.25A. In that case I’d probably recommend if you do go with the 10A cells that you go with more than 4 cells in parallel if they will fit. Maybe 5 or 6.

The Tesla cells are custom Panasonics, but the rumor has it that they are pretty close to the NCR18650BE cell, which is a 10A 3200mAh cell. It’s actually also a very good choice, lower cost than the GA’s per watt hour but slightly less energy density. I believe the internal resistance is also slightly lower meaning they should hold up to higher continuous loads better. I used BE's in my last scooter pack for a 24v SLA replacement.

Cheers for the reply, well there are two specs of batteries I'm currently looking at. NCR18650B which is the higher capacity cells and also the NCR18650PF which is the higher drain lower capacity cells. This is because they are the two cells that are being sold by seller currently at a decent price.

The specsheet for the 18650Bs I believe was sent to me so let me upload them now.
With regards to your PM/Offer to sell me cells I will reply to that in a little bit but ideally you'd be able to provide the cells in a timely manner, of all genuine format (of course I do test each cell before it's used in project) and you take paypal as a payment method...

Regards,

 

[999zip999]

5p or 6p for longer lasting pack.

 

[Overclocker]

Cheers for the reply, well there are two specs of batteries I'm currently looking at. NCR18650B which is the higher capacity cells and also the NCR18650PF which is the higher drain lower capacity cells. This is because they are the two cells that are being sold by seller currently at a decent price.

i've always wondered what kind of "current sharing" would happen when you use 2 different cells. i guess it depends on internal resistance, so the PF would supply the majority

 

[999zip999]

I think a 84v 24s A123 20ah and a 84v 20s 18650 20ah pack will have differen discharge curves. The A123 more of a flater higher voltage thru discharge ? Wouldn't the 18650 have a lower voltage aafter 17ah. Or would it try and drain voltage for a higher voltage. A123 pack. How would it do with the voltage irregularities, becase of discharge curve.

 

[redilast]

Cheers for the reply, well there are two specs of batteries I'm currently looking at. NCR18650B which is the higher capacity cells and also the NCR18650PF which is the higher drain lower capacity cells. This is because they are the two cells that are being sold by seller currently at a decent price.

i've always wondered what kind of "current sharing" would happen when you use 2 different cells. i guess it depends on internal resistance, so the PF would supply the majority

There are a few inaccuracies in the above posts.
Your hybrid battery is a good idea with the caveat that the cells must be matched for LVC compatability.

You CAN mix different capacity cells.

OC is thinking correctly. The performance of your battery will sway according to the ratio of cells and their IR.

You can but it is never recommended by ANY battery manufacturer on the planet and I don’t think there is a single commercially used application for mixing high and low current batteries. There are many reasons why it isn't a good idea.

High current batteries will hold more voltage under load causing them to charge the lower current batteries at the same time as they are being discharged. One battery will provide better cycle life so you will eventually have basically dead cells mixed with decently functioning cells. You will also have the high current cells depleting their capacity faster than the higher capacity cells because they are both lower capacity and because they are trying to charge the other higher capacity cells because of a lower voltage under load. And lastly there is absolutely no reason to even consider mixing cells, given the fact that as of the last few years we now have very high capacity “mid current” 10A continuous capable cells which when used with multiple cells in parallel are an excellent choice for EV packs.

Cheers for the reply, well there are two specs of batteries I'm currently looking at. NCR18650B which is the higher capacity cells and also the NCR18650PF which is the higher drain lower capacity cells. This is because they are the two cells that are being sold by seller currently at a decent price.
,

For your application I wouldn't recommend the NCR18650B's unless you do at least 6-7+ in parallel. PF's are good cells but they are a bit outdated now and lower capacity than the BE's. 2900mAh vs 3200mAh. The funny thing is they are about the same price, which is why I don't go with the PF's. I'm not even sure if PF's are being made anymore.

 

[Overclocker]

High current batteries will hold more voltage under load causing them to charge the lower current batteries at the same time as they are being discharged. One battery will provide better cycle life so you will eventually have basically dead cells mixed with decently functioning cells. You will also have the high current cells depleting their capacity faster than the higher capacity cells because they are both lower capacity and because they are trying to charge the other higher capacity cells because of a lower voltage under load. And lastly there is absolutely no reason to even consider mixing cells, given the fact that as of the last few years we now have very high capacity “mid current” 10A continuous capable cells which when used with multiple cells in parallel are an excellent choice for EV packs..

hi are you the same redilast@candlepowerforums? you make protected batteries right?

interesting thread. have you verified this empirically? the way i see it the NCR-PF has lower resistance so it ends up putting out more current and sagging by X amount. meanwhile the NCR-B having higher resistance puts out less current but sags the same amount because they're in parallel.

now if you remove the load, since the PF has lower capacity and having put out more energy, would have a lower resting voltage compared the the NCR-B. so the B would start charging the PF since they're in parallel

is this correct? it does seem like a bad idea to mix cells

 

[redilast]

hi are you the same redilast@candlepowerforums? you make protected batteries right?

interesting thread. have you verified this empirically? the way i see it the NCR-PF has lower resistance so it ends up putting out more current and sagging by X amount. meanwhile the NCR-B having higher resistance puts out less current but sags the same amount because they're in parallel.

now if you remove the load, since the PF has lower capacity and having put out more energy, would have a lower resting voltage compared the the NCR-B. so the B would start charging the PF since they're in parallel

is this correct? it does seem like a bad idea to mix cells

My above post was mostly my own logic based assumptions. I do have the test equipment to probably test what actually happens but that would take a lot of work, and I believe if it was a viable solution that the big battery companies would put in the R&D and put forth their recommendations. I also know that based on years of experience with 18650 flashlight batteries that mixing two cells of different performance and or greatly different SOC's can in some cases lead to disastrous results. Yes I have posted on CPF back in the days, but not lately. But my original point that no manufacturer recommends this and I don't know of a single commercial solution which paralleled high current batteries with low discharge batteries exists on the planet. If you have any examples, please let me know.

@redilast I appreciate your thoughtful input. While your concerns are not unwarranted I feel that some clarity could be helpful as I've done a large amount of research on this topic.
For anyone casually browsing I've highlighted below.

There are several merits to hybridization: improved power, capacity, and longevity.

Hybridizing results in a setup that lies somewhere between the two components used, such that you gain the proportional benefits of both (high power AND high capacity, for example):



The second link has a LOT more info, but the first link has links to detailed reports with graphs from local users.
https://endless-sphere.com/forums/viewtopic.php?f=14&t=75705
READ ME:
https://endless-sphere.com/forums/viewtopic.php?f=14&t=78434

Your above picture and example is quite different from paralleling High current 18650’s and low current 18650’s in the same pack. The above picture you are referring to is using ultra capacitors in conjunction with a LIB to increase short term high discharge rate capabilities. That doesn’t seem like anything new or groundbreaking to me, I don’t see that as being too different from using a starting capacitor on an AC or fridge compressor to reduce the voltage spike on the mains line in a home.

Source to your original image file is here: http://cas.umkc.edu/physics/caruso/research/advanced_energy_recovery.html

If you have any examples of mixing high and low current Li-ion cells together in the same paralleled pack from any companies, research groups or universities please post those. The OP is asking about mixing power tool 18650 cells with laptop grade cells in the same pack, instead of 18650's and ultra capacitors.

I do believe its entirely possible to make an EV run off of high current batteries and low current higher capacity batteries. However believe if this were to be implemented it would be done by means other than paralleling cells of different types and capacities in the same pack. It would be done by having 2 different packs with intelligent circuitry that can sense high current draws and then by using relays or other solid state switching change from the low current pack to the higher current pack for shorter term surges.

 

[Overclocker]

ok so this got me thinking because i'm planning on building a hybrid pack as well

i guess i just have rig up 2 ammeters and 2 test cells and see them in parallel action. but i think this is what's gonna happen:

initially the PF would give out more current but as soon as it starts depleting (i.e. voltage going down) its current share starts going down. while the B picks up the slack and contributes a bit more current.

this balancing act then continues in a way that depletes BOTH cells at a rate proportional to their capacities (i.e. capacity at the particular current draw) with variations induced by differences in the discharge curves of the two cells

at no load the charging between cells off of each other would probably be minimal if the discharge curves of the two cells are similar

but that's just my theory...

 

[JakeTheDog420]

Thanks for everyones replies.

Based on what's been said I'm going to assume me producing a hybrid pack is too risky of a job for myself when I am not doing this for tests but to build a functioning and mid-decent battery pack..

I've spoken to the manufacturers who have told me the Max Continuous Current of the controller is 30A. Based on this what battery recommendations do people have and what are the reasons so I will be able to know for the future?

My pack design is a 13S4P and while if necessary I can increase this slightly my budget is not unlimited... I do want to get decent range on the bike so probably wont push it anywhere near its max unless I'm travelling short distances but I want the battery pack to be able to give me the max power of the bike when I want to use it.

Thanks for all the suggestions so far, really informative posts! Also, I'm in the UK so if anyone has battery supplier suggestions if I'm not going for either of the 2I've posted feel free to suggest suppliers (with a proven track-record).

Cheers

 

[JakeTheDog420]

In terms of a practical standpoint, perhaps it is not for the OP but please don't be dissuaded by underinformed FUD-based opinions. There are at least two examples from local ES users and if you follow my links you can learn a LOT more.

@redilast Please read, especially the READ ME link. You'll find answers and examples you hoped to see, including a large EV with relays as you propose(lead-acid/lithium hybrid). The linked picture clearly demonstrates the resulting performance of a hybrid concept, apologies if I wasn't clear.

@Overclocker If you're curious how it would work, all of the answers have been linked above. I'm happy to illustrate further via PM as well.

Cheers to all...

Checked your links but it's all very confusing! If my max continuous current is 30a then what battery configuration / type would you recommend? I do want to keep cost down to a minimum while achieving good range and performance overall.. Initally was thinking a 13s4p pack would do good for this...
If each cell is 3000mah and I was using a 13s4p then I would need 52cells total which adds up to 156,000mah overall capacity between all cells. Now if I'm not mistaken all series connections are ignored in terms of capacity so the true capacity of the cells is actually a lot lower at only 12,000mah (3000 * 4 instead of 3000 * 52) is that correct?

A 12aH pack does not sound like a particularly large pack so I'm not sure of its total range either...

Are there any good cell suppliers (preferably fastest shipping to UK) who I should be going to instead? I initially thought 13s4p when I was going to purchase panasonic cells due to their price tag but I have heard 18650 tech has come a long way and now there are many other cell suppliers who can offer similar if not better cell specification at a more competitive price. If this is true then I can afford more cells for the same original budget meaning a potential to increase the overall paralleled connections?

Lastly if I am to increase the P configuration perhaps there would be a bog-standard way of configuring the overall cell layout so it can be upgraded at a later stage? Or would one suggest building a pack with whatever money I can afford and taking the entire pack apart and re-building once I have more funds and am able to upgrade.?

Thanks for your support everyone !

 

[redilast]

@redilast Please read, especially the READ ME link. You'll find answers and examples you hoped to see, including a large EV with relays as you propose(lead-acid/lithium hybrid). The linked picture clearly demonstrates the resulting performance of a hybrid concept, apologies if I wasn't clear.

I did look at your links. But there is no information of real world applications regarding mixing cells of different discharge types in the same pack. The link with all the PDF’s are all regarding ultracapacitors which is quite different. I’m not sure exactly why you are recommending that people build hybrid packs with mixing 18650’s when I’m not seeing where it has ever been done on a commercial scale of any kind. Seems like kind of a dangerous experiment to me, unless you rig up some circuitry to control it.

One of the articles even warns against paralleling LiFePo4’s with Ultracapaictors without special control circuitry. Page 6, 7, 8 in the A123 + Ioxus. It even describes something that I basically mentioned before an intelligent switching circuit for connecting and disconnecting which power source provides the load, (capacitor or li-ion).

My pack design is a 13S4P and while if necessary I can increase this slightly my budget is not unlimited... I do want to get decent range on the bike so probably wont push it anywhere near its max unless I'm travelling short distances but I want the battery pack to be able to give me the max power of the bike when I want to use it.

With just 4P, the NCR18650B is out of the question. The PF or other 10A cells like BE or GA, would be pushing them pretty hard. If it draws up to 30A, then you are pulling up to 7.5A per cell. In this situation I'd suggest either expanding to at least 6P, or you are going to want to go with a higher current cell. The Samsung 30Q is 15A rated and 3000mAh and LG HG2 is 20A rated and 3000mAh. But they are both a little expensive. If you step down to 2500/2600mAh you can get 20A+ capable cells like the Samsung 25R, LG HE4 or Sanyo NSX 2600mAh. And in this case these cells would be fine for 4P.

 

[Overclocker]

https://endless-sphere.com/forums/viewtopic.php?f=14&t=43127

thanks eTrike!

i didn't read every word from those links you gave but i don't think they discussed inter-cell charging. but i think it would be negligible

 

[redilast]

Hybrid pack with random used laptop cells That makes my head hurt. That link although shows that someone has done it, quite honestly provides no actual useful data on how the cells behave when a paralleled, and how they will behave in the following situations:

1) Current sharing
2) One cell charging the other
3) What happens when the higher capacity cell still has energy and the high current cell has been depleted
4) Monitoring the health of a mixed pack
5) The results after quite a few cycles when one cell type is nearing end of life and the other cell type is still healthy.

And then the question remains if you are building a brand new pack, what could you accomplish with a hybrid pack that you couldn't with a pack of all the same model cells?

For example mixing NCR18650B with Samsung 25R, might theoretically allow the pack to discharge at higher rates, but why would you do that instead of just buying some 30Q's or HG2's.

It's funny now that I think about it all this talk about "hybrid packs" the manufacturers are making hybrid cells for us already! It's perfect. No longer are high current cells limited to pure LiMn / IMR with low 1500mAh capacity, but they are now hybrid cells which can feature a good balance between high discharge and high capacity.

 

[Overclocker]

Hybrid pack with random used laptop cells That makes my head hurt. That link although shows that someone has done it, quite honestly provides no actual useful data on how the cells behave when a paralleled, and how they will behave in the following situations:

1) Current sharing
2) One cell charging the other
3) What happens when the higher capacity cell still has energy and the high current cell has been depleted
4) Monitoring the health of a mixed pack
5) The results after quite a few cycles when one cell type is nearing end of life and the other cell type is still healthy.

And then the question remains if you are building a brand new pack, what could you accomplish with a hybrid pack that you couldn't with a pack of all the same model cells?

For example mixing NCR18650B with Samsung 25R, might theoretically allow the pack to discharge at higher rates, but why would you do that instead of just buying some 30Q's or HG2's.

It's funny now that I think about it all this talk about "hybrid packs" the manufacturers are making hybrid cells for us already! It's perfect. No longer are high current cells limited to pure LiMn / IMR with low 1500mAh capacity, but they are now hybrid cells which can feature a good balance between high discharge and high capacity.

yes there are questions that need to be answered. i just ordered two 10A digital ammeters to help shed light on the finer details

anyway here's my scenario. i have lying around 96pcs Samsung 26F 2600mah, and 96pcs Amperex ATL INR18650 2200mah. all cells are very healthy. so i'm planning to build two separate 16s6p packs then wire their terminals together. their discharge curves have a very similar shape i.e have the "knee" around 3.3v.

but i agree that if you're buying new cells you should probably spare yourself the trouble of going hybrid

 

[redilast]

@redilast, thanks for reading some of the links. The a123+ioxus is sort of my favorite on that list since they both use nanotech. Did you see where they reported greater results from the combination than either of the two added together? That's where we're going with this. The benefits are the same as illustrated in the power vs energy image, except adjusted according to the two components (batteries) used. The best info in there is the reduced load on the weaker power member, allowing this weaker member to provide more energy overall. You might have a look at the pdf done by the University of Chile in Santiago for more interesting circuitry, as well as the battery-battery hybrid of lead/Life for a relay example.

I have no doubt that a supercapacitor hybrid system has its benefits. I’ve been looking at some of those insane maxwell supercaps lately. They have caps the size of “D” cells that can do 250A. So if implemented properly this would be great for high inrush current situations of quick but short acceleration bursts as well as hard regenerative braking.

Allow me some liberty with numbers here as I explain further. Above details must be heeded for building but as they irrelevant for this example will be omitted.
Let's take an 1Ah A123 cell @ .06 ohm resistance and combine it with a 3.4Ah NCR18650B cell with 0.11 ohm resistance(again, you would NOT want to parallel these two cells at the cell level due to voltage difference). The resulting IR is ~.039 ohm. Thus the resulting cell can deliver more power than either individually and has a total of 4.4Ah of capacity. Now you might reiterate why not use 2.2Ah cells in the first place. The answer results in improved optimization of "power, capacity, cost, etc." when you consider different pairing ratios of cells.

I’m not sure what numbers you are using as they are not numbers I am familiar with. Are we talking DC ohms, rather than the industry standard of 1Khz AC ohms?

I have a high end extech 1Khz AC internal resistance tester which I use to measure cells and its usually spot on with what data sheets say.

The A123 18650 is about 12 miliohms, and NCR18650B around 38-40 mili ohms. So if we say 12 and 40, then the paralleled resistance (not counting conductor resistance) would be ~9.23 mili ohms. And the total capacity would be 4.4Ah.

But if we take 2x Samsung 25R’s which are around 14 mili ohms and parallel them we get 7 mili ohms for 5Ah.

Can you think of a situation with mixing 18650’s that would actually lead to a benefit that can not be achieved by just a single cell, which is tried and tested. Several years ago before we started getting hybrid cell chemistry's it might have made sense (if done properly) to mix say a 1.5Ah IMR with say a NCR18650 to get a good balance between capacity and discharge rate, but with whats available today I still don’t see any reason to hybrid 18650’s. Hybrid LIB’s with capacitors though is another story and is worthwhile in some cases if done properly.

yes there are questions that need to be answered. i just ordered two 10A digital ammeters to help shed light on the finer details

Cool, good luck with your testing. Btw you might consider using a current shunt resistor instead as it will have a lot less resistance than the meter with its long leads and will probably be more accurate.

 

[Overclocker]

Thanks OC-- that thread by spuzzete is one I linked to in the thread my first link. Good to know someone is reading in depth Any hybrid pack with similar voltage cells should be paralleled at the cell level, but in the case of spuzzete the data you seek can be gleaned from his log charts. He also confirmed what should be intuitive about paralleling at the cell level in his last post in that thread.

so you think i should do cell-level parallel on my samsung + amperex hybrid pack? the reason i wanted to keep them separate was so that i could monitor them individually, because the Amperex cells are quite a bit older than the samsungs...

what do you think would be the issue if i do pack-level parallel?