Biggest bottleneck in an 18650 pack

Hi Guys,
So we seem to have the buss and fuse system worked out pretty well.
I'm just getting started and only have a few hundred 18650s ready to go, so I'm at the avid research phase. However, I'll be starting out with smaller packs, like 7s5p to run 24v systems. So since I don't have multiple 100+ cell packs like some of you, I'm more likely to pull a heavier intermittent load of 2-3C (not like Pete who jumps up to 0.1C when his air conditioner kicks in
There fore I'm going to need fuses that can withstand at least 8 amps with a 25% margin on top of that, so say 10 amps per cell. Also our biggest current bottleneck with the present state of design is the fuses themselves. By my calculations, even a 2cm fuse of say 30ga tinned wire is going to provide enough resistance at 2C draw to provide over 4 watts of heat per fuse (assuming a single fuse per cell), so that is over 200 watts of wasted power per pack on a 100+ cell pack if you aren't just sipping at it like Pete. Not to mention the massive voltage drop just from the fuses alone.
So what happens when a cell shorts out? Every other cell trys to dump short circuit through it right? So we are talking about way over 100 amps.
So I'd like to start a discussion about making the fuses about 5X as big as what is currently being implemented by y'all in the community. Are there any drawbacks I have overlooked? Or are we safe to open the bottleneck?

Oh and btw, I got some laptop batteries that had 26650's. They are only rated 3000ma storage but 50 amps sustained delivery! So they will need a different architecture yet.

Thanx for the bboard guys! :idea:

This is why i dont use fuses, unnecessary and wasteful and a false sense of safety.
Providing you dont over charge or under discharge, your batteries will be fine.

This is exactly why laptops cells dont have fuses, it adds a level of unacceptable risk.

Large scale installations have fuses for indemnity reasons, not safety reasons.
Tesla do this to ensure there is no gap left for liability, but its a safety feature not requirement.

Instead you should ensure your inverter and charging setup treats the cells properly, and have global fusing to ensure short circuit situations dont kill your wiring. Make sure your ceiling and LVD are set to be kind to the cells.

Fuse the whole pack and the global if you really feel you have to, have global isolation.

I currently have two installations with nearly 2000 cells, no fuses and have drawn over a megawatt of power safely from the combined 12kWh of storage.

I have also induced cell failure and run a 2kw rig with over 30% dead 0v cells to test the reaction from said dead cells, no issue at all.

kind regards

Yeah the only way to figure out fuses for higher currents is to do the testing yourself, for now anyway.
I want to do some testing to find what fuses blow at 10, 20, 20A etc,

You can get plastic holders for 26650 cells from the usual places.

3nergE said:

This is why i dont use fuses, unnecessary and wasteful and a false sense of safety.
Providing you dont over charge or under discharge, your batteries will be fine.

This is exactly why laptops cells dont have fuses, it adds a level of unacceptable risk.

Large scale installations have fuses for indemnity reasons, not safety reasons.
Tesla do this to ensure there is no gap left for liability, but its a safety feature not requirement.

Instead you should ensure your inverter and charging setup treats the cells properly, and have global fusing to ensure short circuit situations dont kill your wiring. Make sure your ceiling and LVD are set to be kind to the cells.

Fuse the whole pack and the global if you really feel you have to, have global isolation.

I currently have two installations with nearly 2000 cells, no fuses and have drawn over a megawatt of power safely from the combined 12kWh of storage.

I have also induced cell failure and run a 2kw rig with over 30% dead 0v cells to test the reaction from said dead cells, no issue at all.

kind regards

Click to expand...

My main install is well protected, I have a 250 amp dc breaker on the ground for everything, and I run Overcurrent/undervoltage protection on the positive line and it shuts down the 440 amp solenoid relay much faster than the actual breaker can ever trip.
But I think I'd still like to have pack protection in case a rogue cell decides to go postal nuclear on all his neighbors. Currently I'm considering something like 18-14ga pure copper wire with the ends smashed to provide for better soldering or welding, haven't decided yet. Those should blow at 30-40 amps per cell sustained and pretty fast at 60 amps+. The pure copper will provide minimal resistance for the amps they will melt at. Using things like speaker wire that I see some using, you may as well be using nichrome (heater wire).
~cheers~

Aspendell said:

My main install is well protected, I have a 250 amp dc breaker

Click to expand...

You really need to be using HRC fuses. If there is a dead short in your system, your breaker will not be able to interrupt the current.

3nergE said:

This is exactly why laptops cells dont have fuses, it adds a level of unacceptable risk.

Click to expand...

Laptop batteries don't have fuses you say? Then what's that thermal fuse hanging out from this laptop controller board?




The problem is that the powerwall community just needs to do some testing on which fuse is best, because we just haven't found one yet.

Beware of that many of the used laptop batteries will have to struggle damn hard to blow a 10A fuse-wire.... You generally need 13-15A to blow that one fast and most of them cannot deliver this amount of power.

Laptop batteries do not like to go far above 1C if you dont want to stress them to hard so my recomendation is to not choose fuses that blow on to high current.
I have done several tests here and yes they can trip 15A some of them but what if they are in their lower SOC? They have very hard time to above 10.
Im using 5A fuse wire and this one will trip pretty fast at 7-8A and thats whats needed. In my case my system is designed to have max 1A per cell and thats quite a difference compared to yours though.

Thats my 5 cents to the whole. But beware that you never use fuses that cant be blown by the cells you have.. And also dont forget to test them with uncharged cells before use!

Fuses that wont blow is of no use

daromer said:

Beware of that many of the used laptop batteries will have to struggle damn hard to blow a 10A fuse-wire.... You generally need 13-15A to blow that one fast and most of them cannot deliver this amount of power.

Laptop batteries do not like to go far above 1C if you dont want to stress them to hard so my recomendation is to not choose fuses that blow on to high current.
I have done several tests here and yes they can trip 15A some of them but what if they are in their lower SOC?

Fuses that wont blow is of no use

Click to expand...

True, for single cells. But I'm more concerned about the other ~99 cells trying to dump their individual currents thru a shorted cell in a parallel pack, in which case hundreds of amps would be expected. I think this is where most catastrophic failures occur (from what I've seen). ie, a +buss bar coming in contact with a bare spot on the cell casing even.

Which actually makes a good case for putting our fuses on the negative side, as 95% of the potential contact surface area of an 18650 is negative. I can't justify putting fuses on both sides and potentially doubling the circuit resistance of fuses and hence lowering voltage and increasing pack heat for no real advantage that I can make out.

It's kinda like that old addage that, when in bear country, your best protection is to have a friend with you that is a much slower runner. Is the bear really gonna choose you if you have 2 slow friends instead of 1?

CUDAcores89 said:

Laptop batteries don't have fuses you say? Then what's that thermal fuse hanging out from this laptop controller board?


The problem is that the powerwall community just needs to do some testing on which fuse is best, because we just haven't found one yet.

Click to expand...

That brown thing is a NTC thermistor, i (re)used it on my early FCDS prototype.
But you are right every laptop battery has fuses, called Self Control Protector. It means it can blown by faulty cells and it can blown by the AFE (analog front end ic).

Ohh and usually this SCPs rated between 5-10A.

CUDAcores89 said:

3nergE said:

This is exactly why laptops cells dont have fuses, it adds a level of unacceptable risk.

Click to expand...

Laptop batteries don't have fuses you say? Then what's that thermal fuse hanging out from this laptop controller board?


The problem is that the powerwall community just needs to do some testing on which fuse is best, because we just haven't found one yet.

Click to expand...

No fuses here either

I think Tesla cells probably do require indivicual fuses, more so thanregular cellsbecause they have removed the integrated safety features (PTC, CID and the outer insulator wrap)from their cells. The PTC effectively works as a resettible thermal fuseand the outerinsulator wrap of a battery is pretty good for stopping shorts

I was thinking about this, and I came to this conclusion (correct me if I'm wrong or plz adjust my thinking):

The larger are battery pack is, the more cells that it will have. Assuming that all the cells are in parallel, each cell added will drop the required amperage pulled per cell. So, if you have 12 cells, and pull 12 amps, each cell is supplying 1 amp. However, if you have 120 cells, each cell is supplying 100ma.

So here's what I'm thinking. If you have a large pack, then you could use smaller gauge wire without the threat of heating up while in normal use. In the example given, you could use a 1 amp load wire/fuse since each cell would never even get close to a full amp discharge (assuming the pack never goes above 12 amps of current draw).

If you know the pack will never go above a certain amperage draw, you could then use that amperage and divide it into the number of cells and multiple by 10 to get a safe margin for the fuses. Or something close to that.

Am I thinking correctly here, or way off base

Korishan said:

I was thinking about this, and I came to this conclusion (correct me if I'm wrong or plz adjust my thinking):

The larger are battery pack is, the more cells that it will have. Assuming that all the cells are in parallel, each cell added will drop the required amperage pulled per cell. So, if you have 12 cells, and pull 12 amps, each cell is supplying 1 amp. However, if you have 120 cells, each cell is supplying 100ma.

So here's what I'm thinking. If you have a large pack, then you could use smaller gauge wire without the threat of heating up while in normal use. In the example given, you could use a 1 amp load wire/fuse since each cell would never even get close to a full amp discharge (assuming the pack never goes above 12 amps of current draw).

If you know the pack will never go above a certain amperage draw, you could then use that amperage and divide it into the number of cells and multiple by 10 to get a safe margin for the fuses. Or something close to that.

Am I thinking correctly here, or way off base

Click to expand...

Yes you are correct Kor if you are creating strictly parallel packs.
But I'm operating from the perspective that everyone is actually misconstruing the biggest issue that we are protecting against. See everyone seems to be considering only protecting a single cell (essentially from itself). I propose that the biggest threat by far is Not making sure that a single cell doesn't pass too much current. But rather that if there is too much current flowing, the rest of the pack will be protected. So as I see it the 2 most probable overcurrent scenarios in a pack are:
1) a single cell shorting out, which would cause all the other cells to try to dump their current through a short circuit path. So the other 99 cells trying to pass current through a single cells fuse. or
2) The positive buss coming in contact with any of the 95% of a cell that isn't positive (as we have seen very recently)

So in both of these 2 scenarios you have the entire pack trying to pass current through a single fuse, that once blown will save the whole pack. So we aren't talking about a couple amps but more like 60-300 amps.
Sure the fuses are only maybe 2 or 3 ohms total per pack for fuse losses passing the current. But that is like 4X the combined other losses in busses, connectors, cables and all else in a properly designed pack.

Yeah, I agree with what you were saying there. I guess I could of further elaborated on it...

What you would be able to do then, would be able to put a larger wire that would burn out at say, 10 amps. Since the single cell will only ever go to a max of a hundreds of mah, the fuse is safe, it stays cool, and u could have little ohms of resistance.

Altho, I will admit, I'm not fully aware of how many ohms are in a piece of wire X size in diameter and Y in length and made out of Z material. I do know those are the key factors; however, the larger the size, the less resistance(?)

What I was kind of getting at is instead of using 30 awg (~.24 mm) wire, one could go with a heavier wire of say, 14 awg (~1.63 mm), therefore lowering resistance, thermal buildup, and not to mention easier to solder on to those connections. 30 awg would blow at about 5-6 amps, whereas 14 would blow around 15 amps. Any more than 10 cells in a pack, there's well more than 15 amps available.
Are there other safety issues here, or viewpoint I may be missing? Learn as much as I can before I work on mine

Let me give an analogy for any that may not be well versed in electrical theory.
Most of the packs I have seen in the community are single cell, parallel packs. So lets say 3.8v.
It would be like if you had 1 meter of 12 gauge wire going from each pack to your inverter/charger or whatever your load is. We would never do this because it would generate too much heat.
And every time you double the voltage you can pass 4x the power (voltage x current) through the same conductor while observing the same % voltage drop. But we are talking about 3.8v so conductor size is even more critical.

But the above example is essentially what we are doing by using minimal size fuses, up to 200 per pack x 1-3cm. So at say 0.2C charge and draw you may still be generating 10-20 watts of heat per pack. Unnecessary losses going in and out. And the fuses are connected directly to the cell so guess where the heat goes?

I'm not for no fuses. Just trying to get us to think about this so we don't all end up building a Yugo

Yeah, I definitely like not having it look like a nuke went off in my house, or shed, or whatever. I wasn't thinking of the lower voltage being a factor. Guess ya need to pay attention to both sides of the equation

Oohh, we could use those light up blown fuses that are designed for cars. When the fuse blows, the led comes on to show you

Fully protecting a system, any system, is implemented simply by protecting the individual components - each cell is a component, hence the cell level fusing - each pack of cells is a component, so that is protected by the sum of the individual cell fuses - when all packs are combined into a battery a much larger capacity suitably specified protection device is used - when the battery is attached to a charger/inverter again protection and isolation should be implemented - the output of inverters should again be protected by over current devices, along with earth leakage .......

All of these protective measures will result in losses, you can minimise these losses (to some extent) by using quality, fit for purpose protective devices, but seeking to negate entirely these losses by the removal of protection is a recipie for disaster.

EDIT: Below calculation was a bit wrong. I did not use 5mm but actually 8mm+. At 5mm my wire did not blow that fast when tested to see distance. So my avg loss is atleast 25% more.

------------------------------------


I have 0.5cm per cell and the measured resistance is roughly 1ohm per 1meter! Im using 35awg as mentioned earlier. And yes there are losses here but for instance lets calculate some based on my setup and my max that is 10kW!

I have roughly 4480 cells in total for this purpose. Just to make sure we all are at same here we will calculate this based on that all cells have equal IR. it gets our calculations a bit better but no matter the loss will be the kind of the same.

We will then have 4 strings. And that's 2500 watt per string.
Every string have packs of 80 cells. Per cell we then get 2500w/51.8v = 48A. That is at max 0.6A per cell (I base it on 3.7v per cell so we have an avg number where we most of the time will be)

0.6A over a 0.005Ohm resistor gives us a drop of. 0.003v. And that would result in 0.003*0.6 = 0,0018w in heat. If we take the total here we will then loose 4480*0,0018 = 8,064W

Those 8 watt is spread over a huge set of cells and i would consider them irrelevant.
BUT
This is for me that is running very low C as you also said

So lets say we have 1/4th of the pack i got/plan for then we will have 4 times the heat on the cells. Ie 32watt among 14 packs. Still thats not much but of course its alot more. Also worth noting that the internal heat in the cells will most likely be bigger due to that internal resistance of the cells are alot higher than 5mOhm


And im on Seans track fully. Removing cell fuses because of above losses is NOT something i would do. In my case they are close to none and when compared to the loss you already have in the batteries or in the inverter or anywhere else in the chain....

I'm only fusing one side. The other side have a 4 times thicker wire at least so i have left them out of the calculation above but can easily be included if you wanted to.

That is awesome Dar.
It sounds like you fully thought this out. Which is what I was aiming for. So you got your fuses down to 0.5cm and only primary fusing one side (hopefully the negative side). I would suggest that you have probably 1/5th of the innate(losses) from fuse resistance of the average of packs I've seen so far.
What type of bussbars did you use that you could get the fuses nice and short, or sis you just go with large solder blobs? 5mm is an excellent length. And if your resistance per meter is that low, did you use pure copper wire?

*I haven't seen anyone suggest removing fuses as was eluded to above..

I am still working the last details of mine out. Based on my past experience, mostly with lead acid systems, I'm concerned about future corrosion from going from tinned-copper to copper to steel and back again. But at this point I'll probably end up going with what is readily available within my budget. But of course steel wire has about 3X the resistance per cross-section area.

Thanx for sharing yours!

I have my fuses on the positive side im afraid Why? Because that's how i started it and thats the side that will blow if something happens and so I want the fuses not to hold it back.
Actually i have no good answer and i generally always have fuses on the negative otherwise but not here. Im up for changing in future but have not seen anything that could change me so far. One more reason to have on positive is that the fuses become slightly shorter. You might have seen the videos i made and you see how the fuses look like as well.

Im using 3*2.5mm2 copper wires twinned. (per row, so i have 6*2.5mm2 per pack) And this is between the cells so the distance is really short to the actual cells. Especially the positive side since it sits a bit further up.

Im using thinned copper wire that is made for fusing and is said to be 5A. Blows rather quick at 7-8A and even my really bad <1500mAh and >10 year old cells seem to blow the wire easily.

Data for the wire picked here:
http://www.daycounter.com/Calculators/AWG.phtml
And this is the wire i have used:
http://www.ebay.com/itm/121664920466 (Note that the image on ebay is wrong)

I have not measured the wire myself since i cannot measure this small resistance with decent accuracy. I could put out 10meter of this wire and try to check but im afraid there is to many other factors that could cause issues so i "blindly" followed the general table above.

So my numbers could be off quite a lot if any of above assumptions i made isnt correct