Replace bad LEV50 cell with LEV40

[MickeyS70]

Hello I’m new to this forum and while it has been a great source of information I haven’t found an answer to my query:

I’m hoping to buy a 2012 IMIEV with about 20k miles on the clock. It’s going cheap as it won’t charge over 3 bars and has a very limited driving range. From posts on this forum I suspect that at least one cell is faulty or has degraded rapidly. I haven’t managed to connect to any apps as my dongles didn’t work but I got hold of an OBDLink LX and will try again.

Assuming that I need to replace at least one cell, the next problem is what to use instead. I gather the original LEV50 are impossible to come by, however I can easily get my hands on LEV40s from a 2015 PHEV Outlander.

I think that the reduced capacity of the old LEV50 cells are close enough to the newer LEV40 replacement(s). A battery calibration should therefore be possible. As long as I don’t fast charge the lower max current should not really matter?

It seems the obvious solution to get this car back on the road but the fact that I haven’t found anyone mention this positively leaves me wondering?? Is this because it’s essentially a downgrade or am I missing something?

Any comments would be much appreciated.

Mickey

 

[kiev]

What is the build date on the label in the door jamb--that plus the actual sales or lease date and it may still be under the pack 10-year warranty. Not charging to Full 16 Bars is the key.

 

[MickeyS70]

Thanks for your reply, unfortunately this car wasn’t sold in the US, it was initially bought in UK and therefore only had an 8 year battery warranty, at least that’s what I have been told?

 

[kiev]

There have been a few guys that have swapped out cells, but they have used old LEV50 cells. But your idea sounds feasible if the LEV40 is newer and has higher capacity and will fit in the same volume. Let us know how it turns out.

 

[stoatwblr]

LEV40s are significantly smaller than LEV50s (PEV2 format vs BEV2 format), in all three dimensions and bolt spacings so they'd be a ***** to make fit

8/4 pack modules have the same issue - smaller than the older units - which means mounting issues

I'm not saying it can't be done. It's just not a "drop in" replacement and whilst there are a LOT of LEV40s floating around the UK most of them are of questionable SOH (they were mostly replaced with LEV46) for traction pack use. You absolutely need to check their AC impedance before installing and avoid using anything reporting higher than ~1.25 milliohms @ 1kHz (1.8 is definitely faulty)

Lev40 8 packs are selling for £250-300 at the moment, which is more than I'd like to pay for them, all things considered (At that price it's worth considering the CATL repowered option if it becomes available outside Australia)

 

[MickeyS70]

I’m aware of the different dimensions, however I consider length as the most important and there is only 1mm in the difference. I would have thought that bolt spacing would also be similar enough to fit without major modifications, you seem to suggest otherwise, do you have the actual measurements?

Even if that’s the case, the fact that the cell is not as tall should leave enough room to fit some sort of an adapter. The difference in width can be taken care of with suitable filler materials.

I can get single LEV40s for about £35 a piece or a complete outlander battery (80 LEV40s) for about £850. LEV46 are going for £50/cell. Do I take it that LEV46 would be best suited for the task?

At the moment my plan is just that as I have yet to verify the status of the battery, if only one cell is affected I may give it a go..

 

[MickeyS70]

You absolutely need to check their AC impedance before installing and avoid using anything reporting higher than ~1.25 milliohms @ 1kHz (1.8 is definitely faulty)

I’ve seen some battery impedance meters on Amazon for around £50, will they do or do you need to get more sophisticated tools?

 

[stoatwblr]

You're better off going to Aliexpress

My meter is a RC3563 - the best way to buy is to find a vendor whose prices seem reasonable and has shipped a lot of units

YR1030 family devices don't really measure low enough to characterise high capacity traction cells (they only go to 0.01 milliohms and you really do need the extra resolution at these impedances)

NB: Even after zeroing, your measurements are pretty much unique to your meter. "absolute accuracy" and resolution are two different things and at these impedances even the individual lead quality matters (along with the quality of contact made with the battery terminals. Make sure they're spotlessly clean).

What this means is that you need to test replacement cells to assess their relative impedence vs your existing ones.

Yes, there are better meters - but they cost a lot more. The RC3563 is good enough to provide useful results without breaking the bank

When testing: What you're looking for is anything that's markedly different to the others. A lithium cell's impedance shouldn't change much over its lifespan (or with charge state) but when it does change, it's an early warning of impending failure. For the cells in the car you'll be able to crosscheck impedance with voltage sag under load when heavily discharged as reported by canion or hobdrive (and voltage whilst charging, particularly if using Chademo: the ones reaching 4.11V first are likely to be troublemakers as they're relatively lower Ah compared to their brethren. They're usually (but not always) the ones reaching 2.8V under load first too.

What you don't want to do is put in cells which have a wildly different impedance to the "good" cells (in particular: higher impedance).

You can test these kinds of differences with ordinary 18650 cells. Compare some high discharge 2100mAh units with high capacity 3600mAh(or higher) ones. The higher capacity ones will have higher impedance - and if you check their spec sheets you'll see they're rated for much lower peak/continuous currents. (This is why you have to use the right cells for the right task. Puting laptop cells into a power tool risks causing a fire whilst power tool cells in a laptop won't let it run very long)

If you have older cells - particularly ones which have been deep discharged - you can see how the impedance changes over time. This shows the probable maximum current (sag, etc) which may be be quite markedly different between cells which test as notionally the same Ah capacity

The impedance is related to the thickness of the SEI (Solid Electrolyte Interphase) - this is a really complex subject. The main takeaway for testing is that you want to weed out anything that's "odd" (high) in the car pack and ensure you're not introducing weak cells as replacements

 

[stoatwblr]

Hm

Taking a closer look at the spec sheets, I think the only documented difference is that the LEV40s are 10mm narrower than the LEV50 (35mm vs 45mm)

Assuming no other changes(*) you'll just have to figure out how to safely pad the cells so they're spaced and secure in the carrier. This might give you better thermals if you can ensure some airflow in the spaces

(*) so far - undocumented by Yuasa - I've seen that purple LEV50s are slightly taller than yellow/blue ones (ie - can't be mixed), yellow ones have 2 locating terminal bumps whilst blue ones only have one (meaning you can fit blues into a yellow pack, but not vice-versa) - and I still have no technical data on the internal differences between LEV50/LEV50N

edit: scrub all that - they're electrically incompatible.

LEV40: 3.65V nominal voltage; 40Ah; 170x25x100; 1.65kg; 0% SoC 2.65V - 100% 4.1V; M8 terminal 9-13 NM.
LEV50: 3.75V nominal voltage; 50Ah; 171x44x115; 1.65kg; 0% SoC 2.75V - 100% 4.2V; M8 terminal 9-13 NM.

The imiev's charging system will probably kill LEV40s unless completely reprogrammed. Worst case you may end up with the pack burning up

The dimensions here are different to the source I was previously using, which was quoting 170x45x115 vs 170x35x115 - which may mean that sizing has altered over time or that Yuasa has been cirulating different (wrong) data at different times)

LEV46 fail the electrical compatibility test too (curiously they're quoted as 170x35x100)

 

[MickeyS70]

Hm

Taking a closer look at the spec sheets, I think the only documented difference is that the LEV40s are 10mm narrower than the LEV50 (35mm vs 45mm)

Assuming no other changes(*) you'll just have to figure out how to safely pad the cells so they're spaced and secure in the carrier. This might give you better thermals if you can ensure some airflow in the spaces

(*) so far - undocumented by Yuasa - I've seen that purple LEV50s are slightly taller than yellow/blue ones (ie - can't be mixed), yellow ones have 2 locating terminal bumps whilst blue ones only have one (meaning you can fit blues into a yellow pack, but not vice-versa) - and I still have no technical data on the internal differences between LEV50/LEV50N

Thanks a million for the invaluable information regarding battery impedance and I will closely follow your advice.

The faulty I-Miev I’m looking to buy is advertised as a 2012 model, however it was manufactured in August 2011 and is therefore unlikely to have the LEV50N cells fitted. On the plus side it has only about 21k miles on the clock so I’m optimistic that the pack in general has good SOH, but that remains to be verified…

 

[stoatwblr]

Please check what I just edited into my previous post. You CANNOT use LEV4x cells - they're electrically incompatible and the BMU would need to be reprogrammed (0.1V terminal voltage difference - potential fire hazard)

==================================================================

Unusally low milage is _not_ a good indicator of battery health - if anything it's a warning to be extra careful

==================================================================
If the pack has been let discharge and sat for a while that will cause damage
If the pack has been continually being left on charge, _that will cause damage

There are a batch of 2011-2012 ex-council vehicles which were due to be let onto the market about now which spent most of their lifespan not being driven (mostly attached to charge points in council garages) and according to the maintenance techs have packs in rotten shape...

My Ion only has 53k miles on it and the pack is more or less knackered. It's much the same as yours - 2012 model but appears to be made in late 2011 - it has yellow cells

The LEV50s I put in it came from a 2010 imiev and are blue. Obviously neither will be LEV50N, but there may be internal differences other than the "two mounting bumps on terminals" of the yellows vs "one bump" of the blues.

I can see 40mV difference between the two cell types after charging/levelling, which is out of spec - that's a big warning sign

Yuasa are not responsive to requests for information

 

[MickeyS70]

I don’t give up easily:

LEV40: 3.65V nominal voltage; 40Ah; 170x25x100; 1.65kg; 0% SoC 2.65V - 100% 4.1V; M8 terminal 9-13 NM.

Although capable of charging LEV50s to 4.2V the BMS sets 100% SOC @ 4.1V, which is within the LEV4x spec. Assuming that a newer LEV4x has a higher capacity than the original 2011 LEV50, it should never get close to 4.1V anyway as the lowest capacity cells reach 100% first and charging stops?

I would be a bit concerned about fast DC charging due to the lower max current rating, should not be an issue using an AC home charger.

Bear in mind that my only goal at this stage is to get the car to charge over it’s current 3 bar limit, preferably without blowing it up in the process.

 

[stoatwblr]

"Although capable of charging LEV50s to 4.2V the BMS sets 100% SOC @ 4.1V, which is within the LEV4x spec"

yes.....but......

that last 0.1V is really important to battery longevity and more important the charging system lets cells go _higher_ than 4.1V, particularly if you're rapid charging (I've noticed this whilst monitoring my weak cells). (Going from 4.1 to 4.2V can take as long again as going from dead flat to 4.1V and only gets you 5%. You're now putting a cell in the chain which will do that from 4.0-4.1 instead, It won't end well)

At the other end of the scale the BMS will shut down the car whilst the cells still have 25-30% charge in them.

Remember that the BMU regards 40mV differences between cells as "out of spec", will throw error codes if it sees anything much higher than this and will shut the car down if it regards the differences as dangerous

When you hit go under 2.8V, that's it, as Australian DIYers who directly swapped in NMC cells found out - putting in 104Ah cells had the car behaving as if they were 30Ah thanks to the different charge/discharge curves and voltage endpoints - and that was despite the NMCs being 2.6-4.2V (LiMn cells fall off a cliff at the low end, whilst NMCs fall more gradually)

These kinds of differences really are that critical - and it's why you shouldn't mix chemistry, capacity etc etc

I'd be far less concerned about the small capacity difference if the endpoint voltages were the same but at this point I believe the only safe solution would be to replace the entire pack and reprogram the BMU. At best a single cell changeout is likely to work horribly (even worse than your existing bad cell) and at worst you could have a fire on your hands. Proceed with extreme caution and a lot of instrumentation

 

[MickeyS70]

Really appreciate your input on this topic, lots of food for thought…

Enough speculation and high time to to connect my dongle and run some diagnostics; fingers crossed, it may not even be a cell problem after all

I’ll keep ye posted..

 

[stoatwblr]

Some more input for you

I've managed to overcome the link rot and grab the original Yuasa tech documents, then translate them from Japanese.

They all talk about LEV50s being 4.1V cells, not 4.2

It may be possible without major firmware modification after all. I'd still be extremely careful - ideally running cycles with the pack open and on a test bench rather than buttoned up inside the car


other notes:

1: As suspected, mixing LEV50 and LEV50N is a bad idea. Their discharge/charge characteristics vs temperature are significantly different

I'll stick my neck out and say that in all liklihood the post-2012 BMU firmware tuning of LEV50N packs is such that putting LEV50s into them is likely to be dangerous (LEV50N has a wider set of tolerances at both high and low temperatures), but LEV50N in a LEV50 pack is unlikely to result in damage to the cell or pack (but will give suboptimal performance)

2: There are LEV40N/40N too, but LEV46 are all post-2012 and therefore "N" formulation, so be careful

3: LIM cells are the ancestor of LEV cells, but internally different. I was hoping that LIM50 (stationary standby power) packs might be able to be used as a substitute but this looks unlikely. That's a pity as they're still available from Yuasa

4: Yuasa offer up zero clues as to how the barcodes on the cells translate to model numbers


I think it would be extremely useful to obtain firmware dumps from various years and see how the BMU settings vary

Do you know how to run a battery capacity recalibration procedure?

 

[MickeyS70]

Just as I was about to throw in the towel, some good news finally, I take sub-optimal over dangerous any day….

Bench testing as in extending the HV and signal cables and run the open pack outside the car?

I have read about the battery recalibration procedure but as far as I’m aware it can only be done with a MUT-3 terminal?

 

[stoatwblr]

Just as I was about to throw in the towel, some good news finally, I take sub-optimal over dangerous any day….

Sub optimal ranges between "barely works" and "not very well".

I think it'll be okish if the LEV40s are in reasonable condition but in my opinion the figures being asked in the UK for LEV40/46 are far too high, especially when compared to the cost of an australian 104Ah conversion - in brief, the cost of a set of used cells of unknown provenance - which will at give you at best a 70km (45 miles) range - is half the cost of a complete rebuild including labour and a warranty which gives a 240km range (150 miles)

That's why I'm pursuing this option.

Incidentally, one of the more interesting things I discovered is that when LEV50Ns were released, the price was HALVED and Yuasa was shipping the things to Mitsubishi for a few months before officially announcing them

Bench testing as in extending the HV and signal cables and run the open pack outside the car?

Yes. You can use the heater as a dummy load. Beware of nasty bitey DC. If it grabs you it does not let go, unlike AC. This is "wear gloves. work with one hand in your pocket and DO NOT work alone" territory (I'm a service tech by training with a background in HVDC radio stuff, There are some nasty stories there when people got careless)

I have read about the battery recalibration procedure but as far as I’m aware it can only be done with a MUT-3 terminal?

The latest Hobdrive beta has it too. I've run it a few times on my own car but they need more testers with different model years to verify the produecure works ok

 

[MickeyS70]

Ok, the guessing game is over, managed to connect caniOn: I haven’t figured out how to post images so will describe what I can observe:

Average 3.874v all very close 0.02V diff except two outliers #85: 4.075V # 86 3.675V SoC 18%. Batt temp 12.6 deg C RR 6km

Questions:
- #85 reduced capacity and therefore fully charged first?
- #86 bad or stopped charging because of #85?
- is it significant that problem cells are next to each other, BMU problem?
- any potential quick fixes without changing modules?
- if bad module which one or both?


By the way I’m now the proud owner of this cute I-Miev and I even have a lead on getting my hands on LEV50 cells.

 

[coulomb]

- any potential quick fixes without changing modules?

I think it might be worth hand balancing those two cells. But the question would be at what SoC? Perhaps 50% display [ edit: was real ] SoC would be best, for minimum difference at the top and at the bottom, both of which can reduce range.

Edit: I changed my mind to 50% "display" SoC, since that's effectively the usable SoC. It doesn't matter what would theoretically happen in the buffer zones, since the BMS won't let the cells go there.

 

[MickeyS70]

By hand balancing you mean removing the cells and charge/discharge them externally to a voltage that the rest of them have?