Home made LiIo "pipe" battery possible voltage sag issue

[Jez]

Hi Experts! Got a question regarding a battery built for an electric bike. The battery is a pair of 7s4p batteries used in series. I elected to use this pattern as I am not using a BMS and have purchased a super-duper iCharger duo battery charger which will charge both batteries simultaneously but each channel can do a max 8s battery with balancing.

Each cell is around 3.2Ah which makes this battery a 50.4V 12.8Ah battery. Anyone have a rough idea of expected eBike range with a batttery of this size?

So, the battery behaves nicely for the first 20 miles running it as hard as I can. I don't see much of a drop. Then it abruptly drops away. Okay, relatively normal behaviour for LiIo I guess. What my inexperience with the real world of these batteries is not telling me is what I should be seeing from the cells. At about the 20 mile mark my battery monitors start to beep (set at 2.7v) but when I get home, all the cells are well into the 3v range. Doing a short blast will see the (very simplistic) battery monitor on my handlebar display drop down and up again. So with the battery in this state I can then put the bike on the stand and put the system under load by running the bike on the hand throttle with the brakes on.

I can see that some of the cells drop markedly - down to 1.2V briefly. I'm told (is this true?) that it is the resting voltage you need to watch so that it doesn't drop below 2.5V. You can dip below this under load.

So in this load testing scenario, as soon as I stop load testing, the cells come back up to over 3V.

So does all this sound normal to you guys and gals?

So just to summarise:

What would be the expected range of the bike with this battery size?
Is it true to say that a LiIo cell can drop below its 2.5V threshold under load? (Resting voltage is the key measure here)
Is the behaviour described above a normal cycle for a battery pack?

One of the reasons I'm asking these questions is that the design I adopted for this battery makes it very safe because individual celll is joined to its buddies via short piece of 5A fuse wire before hooking up to the main pipe bus. Obviously this wire is very thin but the couple of cm of wire could potentially be a resistance which causes a voltage drop under load. My theory is that it's okay and 20 miles is about right for this battery and these indicators indicate I need to recharge, but I'm not sure so just checkin'!!

 

[pwd]

First and foremost, it sounds like you are over-discharging your cells below the recommended voltage. When I was using LiPo, I stopped discharged as soon as any cell hit around 3.4v. You are going below 2.5V under load? Way too low. At that point you are hitting the "lipo cliff" and are doing excess damage to the cells.

I would be careful using these cells again, especially while charging....

What cells are you using and what kind of load are you pulling from them?

I highly recommend reading through this stickied thread on Lipo: it may save you a lot of trouble
https://endless-sphere.com/forums/viewtopic.php?t=9170

 

[Jez]

Hi. Thanks for your response. No worries on the voltage. I'm using LiIo Lithium Ion cells. The spec on the cells say my low cutoff voltage is 2.5V. My key concern on this issue is that is it okay to see a load voltage below this threshold as I've seen? The resting voltage in this situation is around 3.1V but this "sag" I'm seeing is what I'm not sure about and am trying to find out if it's normal or a result of the fine fuse wire linking my cells which is potentially producing resistance in the battery as a whole thus bringing down performance.

Having said all that I did a charge today. All went well. Total charge time for the cells at the rated current is 4 hours again according to the spec and the charge time was a tad under that so the implications of that little bit of data is that the batteries are indeed discharged and this behavious is normal. Just hoping someone can confirm this!

Would also be nice to know if this dip-under 2.5V under load is also acceptable for Lithium Ion. Research consensus says it is but some of the data is hearsay so hoping there is someone who either has empirical evidence or is a Lithium Ion superhero!

 

[john61ct]

The spec on the cells say my low cutoff voltage is 2.5V

Acting in accordance with that is not conducive to longevity.

Yes large voltage drops are normal, more so at high C rates and especially with lesser quality cells or those not formulated for such usage.

No one can say if your "fuse wire" connections are adding to those factors without testing with known good instruments. I suspect so, and personally would not use such a design.

IMO all bank wiring infrastructure should be capable of handling - without dropping voltage -far more current than X, the maximum they will see in use.

The exception being quality fuses or breakers, rated for 15-20% below X.


> Would also be nice to know if this dip-under 2.5V under load is also acceptable for Lithium Ion.

Lithium Ion is a general umbrella term, not a specific chemistry. Same with LiPo.

But yes, at the low end all LI chemistries are pretty similar as far as the care specs.

IMO opinion cutoff should be well above 3V even under heavy load, but that is a tradeoff with longevity, some are happy to replace at under 50 cycles, others strive for 500 or more.

> resting voltage in this situation is around 3.1V

Not immediately "damaging" but again, drastically sacrificing life cycles, so your call.

 

[john61ct]

Link to your actual cells would help.

Each cell is around 3.2Ah which makes this battery a 50.4V 12.8Ah battery.

Do you mean 3.2Vpv at rest "fully" charged?

50.4V at 14S comes to 3.6Vpc, is that a stop-charge setpoint, or actually measured?

If the latter, with the bank in what state?

e.g., after a full charge but before resting, maybe some surface charge remaining?

Do you know your charger's termination algorithm, have you verified it with an ammeter? Link to the charger?

 

[amberwolf]

The spec on the cells say my low cutoff voltage is 2.5V

That's likely the absolute max "never go below this ever no matter what" limit.

3.3-3.5v is MUCH safer as an LVC, and will give you much better cell life, and will require a lot less balancing of the pack every time you use it.

Using 2.5v you might get several dozen cycles out of the pack before it's seriously degrading, if the cells were very good quality to start with. If they weren't, probably less.


You won't see much difference in capacity between the two points, maybe 10% or a bit more, so it is worth the bit of sacrificed range per charge for the longer pack life and lesser problems balancing every charge.

 

[Hillhater]

How can anyone comment usefully without knowing what cells/chemistry, are being discussed ?
“Lilo” is not a cell i have ever heard of ?
Could be LiPo,, or LiCo, or even Lipo4 if the 3.6 v charged is correct ....big differences !
Likewise, useful range is impossible to prdict without knowing what motor/controller system is used, what gearing, how much pedaling, what speeds, what type of roads/hills, etc etc.
EG..if you have a 250W hub motor, and pedal enough,..you can have a 100 mile range !
So..more info needed :thumb:

 

[john61ct]

How can anyone comment usefully without knowing what cells/chemistry, are being discussed ?

Hence this, but crickets

Link to your actual cells would help.

Each cell is around 3.2Ah which makes this battery a 50.4V 12.8Ah battery.

Do you mean 3.2Vpv at rest "fully" charged?

But then,

> Would also be nice to know if this dip-under 2.5V under load is also acceptable for Lithium Ion.

Lithium Ion is a general umbrella term, not a specific chemistry. Same with LiPo.

But yes, at the low end all LI chemistries are pretty similar as far as the care specs

The top end, charge termination is where things get more critically chemistry-specific.

 

[Jez]

Hi all. Thanks so much for showing an interest. You are all correct in that I was not clear in my earlier post. So here's more info. I have also berated myself somewhat as there are some obvious things I can do myself to check out any resistive losses in the design.

So first of all. The batteries are Panasonic NCR18650B Lithium Ion (LiIo) which according to my spec have a capacity of 3350 mAh and a max discharge current of 4.875A. I think I right (please correct me if I'm wrong) in saying that the C rating for this battery is just over 1C given those values.

John you are of course correct about the final resting voltage. I was bunging figures out into the forum to elicit a response and of course garbage in = garbage out. The final charge voltage is of course 4.11V (average) per cell block giving a voltage on each battery block of around 28.7V. This is what each battery shows as a final charge voltage. All good too. Apologies again John. I built this battery a year ago and rather than check to remind myself of details I just chucked out the nominal voltage from the spec.

What is of much greater interest is this business of the discharge voltage and voltage under load. You guys are focussing in on this which is great. So my understanding is that given an internal resistance for any given cell the sag under load will be greater for a higher internal resistance. The spec for these batteries quote a max internal resistance of <48mOhms.

Given that I'm running a 500W motor then maximum current drain at that nominal 3.6V is going to be just shy of 10A which results in a cell current of 2.5A which is nicely within the cell capacity.

So if a cell is maxed at 2.5A with a 48mOhm (must work out how to do an ohm sign) internal resistance, the voltage sag under load would be 0.12V

I was seeing a load across the fuse wire of 0.048V. Each end of the battery is connected by a fuse wire so would need to double that so (2*0.048)+0.12=0.216V which is the theoretical sag of each cell under full load given a worse case battery internal resistance.

Hmmmm..... it's interesting when I do those sums that the loss across the fuse wire adds up doesn't it. I wanted it very safe. Got scared seeing all the youtube videos of bikes going bang so I think I'll keep the design but there is a loss. Scale that up to 14 cells and that's around 3V. Without the fuse wire it would be 1.68V So that's nearly double the voltage drop because of the fuse wire.

Let's see if the empirical evidence matches up...
Gone off to measure. Resting voltage is 50.6. I've been for a few runs so that's fine. Works out at roughly 3.6V per cell.
Running at full throttle with the brake on drops the voltage to 45.9V. Call it 46V for luck. I'm seeing over 4V sag under load which is close enough to my 3V theoretical drop for everything to make sense.

So I know that my voltage drop measurement over the fuse wire was quite rough and ready. Consider losses in the connectors (banana plugs) as well and that would account for the difference.

Have I just answered my own question? Am posting this along with apologies again for wasting your time if I have this right. It may serve to help others. I wouldn't have headed this way without your input John so thanks again. I think my battery design could use a bit of work to alleviate the losses over the fuse wire. The battery joints are soldered (I've been soldering since I was 4 years old so was comfy with this approach) and am aware of the debates about soldering v welding. But to join a much thicker wire to one battery terminal via a solder joint would mean more heat which is not a good thing.

On the business of this acceptable level of voltage before recharge issue I'm not so clear.

What I'm taking away is that the closer to 2.5V I get on a regular basis the shorter my battery life will be. Generally allowing the voltage to sag "low" is also not conducive to longevity. Accordingly I must balance range with battery life. Does that about sum it all up? Conscious that Lithium Polymer cells behave differently but the rules are the same for Lithium ion too.

Hopefully I've done a better job with info this time. This stuff is such fun isn't it!!

 

[john61ct]

NCR18650B are NMC, which is part of the LI family. Good cells for high discharge use like this, rather than being optimised for longevity, but still decent # of cycles compared to others if not abused "too bad".

Letting them get below 3.0V under heavy load will murder them pretty quickly, likely ready to be replaced in weeks rather than months.

If you want a year or more, follow amberwolf's advice above.

Generally allowing the voltage to sag "low" is also not conducive to longevity. Accordingly I must balance range with battery life.

You really are not gaining much if any useful range at all between 3.05Vpc and 2.50

Stopping at say 3.20V maybe 5-10% ? With enormous longevity gains, at least stopped murdering them.


And yes your fuse wiring to every cell is IMO wasting a lot of energy to resistance.

Difficult to measure that directly enough to guide design decisions, just work to reduce V sag as much as possible, easy to measure.

 

[john61ct]

Conscious that Lithium Polymer cells behave differently but the rules are the same for Lithium ion too.

LiIo is not used much, really LI is the broadest possible umbrella, includes LFP and LTO.

Best if the specific chemistry is unknown, just specify the actual cells by mfg / model.

18650s can be LFP, LCO, LMN, NCA or NMC.

LiPo as a label is really just about physical packaging, not in itself indicative of the specific chemistry, which is what dictates performance, balancing between discharge energy vs power density, thermal stability vs longevity vs low-temperature performance etc.

That said, I guess most LiPo can be assumed to be LCO.

 

[Jez]

Hi John! Thanks for that quick response. I've been busy meanwhile and prepared a post which I'll drop in below. Am interested to know how much further I can get by dumping my fuse wire. So here goes...

Going to take this to the next level because basically what this is all about is how much further can I get if I don't use fuse wire.

So let's assume from my previous post that there is a combined voltage drop of 3V across the cells with the fuse wire but only a 1.5V drop without the fuse wire.

Let's further assume that if I were to build the battery with thick coppper wire (my initial thinking is UK spec 30A mains solid copper wire (don't have an AWG rating to hand but it's about 3mm diameter roughly)) the losses across that would be negligible. Might change that idea later but just for the sake of this thought experiment.

Let's further assume that the bike does 20 miles before the cell monitors start to bleat about low voltage (which is set to 2.7V) and so because of that I'm going to assume an end of ride resting voltage of 2.7V under load.

We know from previous post that the sag under full load across each cell is (2*0.048)+0.12=0.216V which is made up of the two bits of fuse wire at each end of the battery plus the internal cell resistance. But now I'm taking away my fuse wire so the actual sag should only be 0.216V. This is based upon the maximum internal resistance of the cell which should be less but I'm not allowing for any losses over the copper connections so I'll stick with this for now.

Also assuming that the nice design of a Lithium Ion battery provides a constant rate of discharge.

So in 20 miles, the voltage drops from 4.11V no load to 2.7V under load which means that the no load voltage should be 2.7+0.216=2.916V Empirically it's higher. Closer to 3.1V which suggests my losses over the fuse wire is higher than measured but am going to pick the difference to make it easy and go with 3V.

So 4.11-3=1.11V in 20 miles which is 0.055V per mile

So if each cell is connected by a fuse wire dropping 2*0.048V (there is one at each end of the battery) then that's 2*0.048=0.096 so that by removing all the fuse wire I get 0.096/0.055=1.7455 extra miles from the battery.

Appreciate there is lots of assumptions in this calc but wanted to get a ball-park idea of if it's worth rebuilding the battery.

One assertion I know is wrong is that the assumptions in the calculations are that I'm riding at max speed which I won't be. The 20V for the given voltage drop is fairly accurate but there are question marks regarding the voltage losses over the fuse wire which I think are worse than measured. In theory I could go further than the 1.7 miles if I didn't push it as the current drain would be lower and the consequent cutoff point wouldn't be hit so soon. So could be worth maybe 2-4 miles extra range?

The battery is a 14S4P with each cell having 3.2Ah so battery pack is 12Ah at 57.5V (sorry about that again John giving me a range of 20 miles. Watt hours is 3.2*57.5=184Wh

Can you guys poke holes in that? Have I gone wrong anywhere?

Given a trade-off between cell safety and performance I'm happy to stick with the fuse wire to join stuff together (but mostly because I hate the idea of taking the battery to bits and building it again!) I think the Mk2 would only have fuse wire at one end though. I would take the risk of soldering a thicker wire to the positive end perhaps (as this end I believe is more tolerant to heat) and fuse the negative end.


Just for info on the empirical stuff. I tend to ride everywhere in the Tong sheng "speed" or "turbo" mode which for those without a Tongsheng machine would be level 3 and 4 on a scale of 1-4.