The aftermarket cell replacement saga

[nico]

(Kiev, thank you for the picture of the battery management system.)

If your replacement cell has more capacity than the old ones, all the other cells will charge faster that the new one. Correct? If the charging rate difference is higher than what the battery management system can compensate, your new cell will only be partially charged.
I understand that the battery pack's performance can be improved only to the level of the worst cell in the pack. I would recommend using a cell of the same type as the original when inserting a replacement. Secondlife-evbatteries.com batteries are a safe and good idea, but they will just get you back to the degraded performance of an old battery. They are definitely a good value if your pack was close to its original capacity but lost a few cells (or if your car is not usable anymore).

In my case, I am looking at replacing all the cells, hopefully under US$10k, which is cheaper than any new EV. The limit will then be imposed by the 42 Ah register value. Are there any hints that the value can be changed? If it can, the cell balancing routine should allow to benefit from higher capacity cells. If the value can't be changed (or the current sensor fouled), the extra cell capacity should allow to maintain the pack's estimated capacity value at the maximum possible for longer. If the register value can be changed, it could be interesting to pay extra for 100Ah cells, instead of 70Ah.

DBMandrake, you mentioned that the battery pack needed to deliver a 150 amp current. The LEV50 specification calls for a 300 amp continuous discharge current. Cells with 300 amp continuous discharge are difficult to find. Is that number overkill? Where did you find the 150 amp figure?

Any info on the temperature sensors? Are they glued between cells? Do they have a part number?

Does anyone know how and what info the battery management system communicates with to rest of the vehicle? Can a different battery management system transmit the same messages? In that case, we could install a new pack (including its own BMS) that would report with the same messages as the original pack...

 

[DBMandrake]

If your replacement cell has more capacity than the old ones, all the other cells will charge faster that the new one. Correct? If the charging rate difference is higher than what the battery management system can compensate, your new cell will only be partially charged.

What should happen is that after balancing all cells will be 100% charged at the same time then as the pack discharges the replacement cells will simply not be discharged as much as the others.

If they are only a few Ah better than the others it should be a problem - after all I already have a 7Ah spread in cell capacity and the BMS is still mostly working fine. Only if the difference in cell capacities is huge should there be a problem.

In any case the cells I have to fit have a capacity that is only slightly higher than the current best cell(s) in the pack so they will be a good match.

I understand that the battery pack's performance can be improved only to the level of the worst cell in the pack. I would recommend using a cell of the same type as the original when inserting a replacement. Secondlife-evbatteries.com batteries are a safe and good idea, but they will just get you back to the degraded performance of an old battery. They are definitely a good value if your pack was close to its original capacity but lost a few cells (or if your car is not usable anymore).

In my case I have some defective cells that have a lot lower capacity than the majority of cells, and more importantly they are continuing to degrade at about 4x the rate of other cells. At this rate the car will be unusable to me within a year. Replacing them not only restores some of the lost capacity (I'm hoping to get back up to around 38Ah) but should slow the degradation rate back down to what it should be.

In my case, I am looking at replacing all the cells, hopefully under US$10k, which is cheaper than any new EV. The limit will then be imposed by the 42 Ah register value. Are there any hints that the value can be changed? If it can, the cell balancing routine should allow to benefit from higher capacity cells. If the value can't be changed (or the current sensor fouled), the extra cell capacity should allow to maintain the pack's estimated capacity value at the maximum possible for longer. If the register value can be changed, it could be interesting to pay extra for 100Ah cells, instead of 70Ah.

The way to change the Ah capacity value is to perform the battery calibration routine with the diagnostic tool. As mentioned I've performed this process twice now so I have a good idea of how it works.

You start the calibration and run the battery down to about 3.75 volts per cell using the heater. Then it asks you to plug the car in and charge to 100%. The BMS does a capacity measurement of every cell in the car individually during this charging step. After that you complete one final step and the new Ah value is written into memory and becomes immediately active - I actually monitored this using Canion using an OBD-II Y-cable so I saw at exactly what point the Ah figure was updated at the end of the process.

So if you put a larger capacity than original battery in you would have to run this calibration routine for the BMS to "learn" about the larger battery. The only question is how much larger could you go before the BMS would reject the capacity as an unrealistic or false reading ?

For example it might accept a figure up to 50-60Ah, but would it accept 100Ah ? Unfortunately without reverse engineering the firmware of the BMS we simply have no way to know.

Someone is going to have to be the first to try retrofitting a much larger capacity battery and then performing the battery calibration - but if the BMS will not accept the higher value it will be a very expensive and time consuming failure! :shock:

In the case that someone did go to all the trouble to fit a much larger battery and the BMS would not accept the higher value, your options would be to get an ECU unlock specialist to try to modify the firmware in the BMS to accept higher values (this seems feasible, if you can find someone willing to do it) or replace the BMS with a programmable after market unit - which would be a major project in itself requiring substantial reverse engineering work.

DBMandrake, you mentioned that the battery pack needed to deliver a 150 amp current. The LEV50 specification calls for a 300 amp continuous discharge current. Cells with 300 amp continuous discharge are difficult to find. Is that number overkill? Where did you find the 150 amp figure?

150 amps is not a precise figure it's approximate. Maximum output from the battery under full acceleration is 54.5kW according to the BMS. 54500/360 = 151 amps. If the figure is taken at a slightly lower voltage it will be a little bit higher, but it will be under about 170 amps.

Does anyone know the size of the main fuse in the pack ?

A 300 amp continuous rating from the cells is to give a bit of headroom. For something that you want to last for 10+ years you don't want to be running it close to it's limits all the time.

Any info on the temperature sensors? Are they glued between cells? Do they have a part number?

I haven't taken a pack apart (yet...) but I believe the temperature sensors are on the under side of the PCB's and press down to make thermal contact with the flat part of the battery terminal next to the threaded stud. So they are measuring the temperature of that flat part of the terminal which is bonded internally to the electrodes.

Does anyone know how and what info the battery management system communicates with to rest of the vehicle? Can a different battery management system transmit the same messages? In that case, we could install a new pack (including its own BMS) that would report with the same messages as the original pack...

That's a big project you're describing there.... :lol: but ultimately what needs to be figured out for people to retrofit newer technology batteries to old cars.

The BMS in this car is located under the rear passenger seat and is quite easy to get to. It has overall control of the battery. The CMU (cell monitoring units) are on the boards in the cell modules inside the pack.

The CMU's mainly just send individual cell voltage and temperature readings to the BMU and perform cell balancing when requested. The BMU has to be able to talk to the CMU's, but also to the MCU (motor control unit) the onboard charger, and the heating system. (To request cooling during charging based on cell temperatures)

Any replacement BMS system would have to be able to talk to all these different ECU systems. It's not impossible but it's certainly a very challenging task that will need a lot of reverse engineering if nobody has already been there and done this before. For the small number of i-Miev's that were sold I'm not sure that we'll ever see anyone take the trouble to reverse engineer the car enough to fit a fully custom BMS. Patching the firmware in the existing BMS seems like a much easier way to go.

 

[nico]

I found this article which brings some information about the current sensor in the Miev:

Battery Management System Hardware Concepts: An Overview
by Markus Lelie, Thomas Braun, Marcus Knips, Hannes Nordmann, Florian Ringbeck, Hendrik Zappen and Dirk Uwe Sauer
Appl. Sci. 2018, 8, 534; doi:10.3390/app8040534
http://www.mdpi.com/journal/applsci

Here is an extract:
"The first example is the traction battery of a Mitsubishi i-MiEV (initial registration: February 2014),
shown in Figure 4a. It contains 10 Modules of eight cells and two modules with four cells, which leads
to a total amount of 88 prismatic cells, all connected serially using screwed contacts. On top of each of
the modules, a PCB is mounted, which—among other things—contains an LTC6802G-2. This IC is
designed to monitor up to 12 lithium-ion cells, which are connected in series. The same PCB design
is used for the module versions with four and eight cells. When used with four cells, the PCB is not
fully populated, as four of eight available channels are not needed. The eight-cell modules use a
second PCB to connect the second half of the module to the four remaining channels. The PCB on
top of the modules is called the Cell Management Unit (CMU) in the official service manual for the
car [30]. In addition to voltage measurement, each PCB contains three temperature sensors, which are
connected to a controller located next to the Linear Technology BMS IC.
Apart from the cell modules, the battery housing contains contactors (separate ones for the
connection to the inverter and for DC charging), fuses, a service plug, an LEM (Freiburg im Üechtland,
Switzerland) current transducer, an insulation monitor and a fan to extract cooling air. The service
plug splits up the pack into two sections when it is removed. In the car, it can be found below the
left hand seat. The main fuse also splits the pack in the middle. The cooling air originates from the
vehicle’s air conditioning system and is partially guided by short plastic pipes and forwarded to the
outside by the mentioned fan.
Not contained in the battery housing is the BMS master, or Battery Management Unit (BMU) [30],
which communicates to the rest of the vehicle. It is located below the rear bench seat of the vehicle.
The contactors, the current transducer and the insulation monitor are connected directly to the BMU.
The CMUs on top of the cell modules are connected to each other and to the BMU via an internal CAN
bus. Some of the signals on the battery internal CAN bus can also be found on the car’s main CAN
bus in lower resolution. Compared to the smart’s battery, which is described below, there is a lot of
free space in the iMiEV’s battery housing, which may be a side effect of the air cooling."

Further in the document, we find that the sensor is a LEM CAB300 which has these specifications:

The CAB family is for battery monitoring applications where high accuracy and very low offset are required.
Features
• Transducer using Fluxgate technology
• Zero offset
• Unlimited over-current capability
• Panel mounting
• Unipolar + 12 V battery power supply
• Ta= - 40 °C ... + 105 °C
• Accuracy (over temperature range)
• eG= Sensitivity error < 0.5 %
• et= Linearity error < 0.1 %
• Output signal: CAN.C (500 kbps)
• Optional internal digital low-pass frequency filter.
Special feature
• Connector type Tyco AMP 1473672-1.
Advantages
• No offset error
• High accuracy
• Low linearity error
• Full galvanic separation.
Automotive applications
• Hybrid and electric vehicle battery pack
• Accurate current measurement for battery management applications (SOC. SOH, SOF etc.).

 

[DBMandrake]

Nice find, with some interesting information.

Your link only goes to the front page of the website though, here is a direct link to the article, which I was able to download in PDF format:

https://www.mdpi.com/2076-3417/8/4/534

 

[DonDakin]

Dbmandrake,

Thanks for your super informative post. You seem to have all the details of a battery retrofit thought thru.

I was wondering how many Kilometers your car is down to with the 4 weak cells. Do you think you could manage to get a warranty replacement before the 2022 deadline for the 2012 models ?

Don.....

 

[DBMandrake]

I was wondering how many Kilometers your car is down to with the 4 weak cells. Do you think you could manage to get a warranty replacement before the 2022 deadline for the 2012 models ?

Don.....

I'm going to assume you're based in the US. There is no such 10 year battery warranty in the UK.

Peugeot offered a 60K mile 8 year warranty for the battery, but only if you serviced the car yearly with them. (Otherwise I think it was 3 years, or 5 years, can't remember which) My 2011 car is now outside this 8 year period and it is very close to hitting 60K miles now as well.

Also there is no minimum capacity clause in the "warranty". In other words they will not replace the battery due to reduced range. Only if the battery fails completely. Which as we all know, makes the "warranty" completely worthless, as EV batteries rarely fail completely, they just drop to an unusable range.

In short, forget about any kind of battery warranty on an Ion/C-Zero in the UK because without a minimum capacity clause it is worthless.

So I am stuck replacing cells myself.

 

[DonDakin]

Hi DBMandrake,

I just updated my location. I've been a member since 2012 but never updated that. I'm in Montreal so close to the US but not quite....

Anyways did not connect the dots that you are in Scotland. We are fortunate on this side of the pond to have 10 years on the battery and recently the charger also. Your right that that warranty only covers a failed cell and does not really cover "normal" degradation (whatever that is) I guess with enough time someone will replace the battery completely with larger capacity cells and we will all be enlightened.....

I have 2x2012 imievs one with 95,000 KM (the more used one) on it and recently before it got cold I got 70 KM on the highway (at 90-100 kph) down to the turtle as I pulled into the driveway. I wonder what range you are seeing. If I baby the car I can still get 100 KM plus.

I have Canion but I never see anything in the AH row. I have a "batman" STN1170 OBDII-DEFD dongle that I got when CANION first came out. I don't know if it's filtering out the pid with the AH information or not. It would be nice to know my AH capacity as computed by the car.

Anyways very interesting the cell replacement saga I will follow it closely.

Regards

Don.....

 

[DBMandrake]

My battery is currently down to 32.1Ah and I would say that my summer no heater range is now down to about 55 miles and winter with heater range about 37 miles...

So that's a loss of about 10 summer miles and 6 winter miles compared to the 39.9Ah that the car had at the start of 2017. I'm not expecting to get all of that back however judging by the voltage differences between the cells I'll be replacing and the rest I'm hoping that it will go back up to about 37-38Ah, and more importantly go back to a "normal" degradation rate as the degradation rate is currently about 4x what I would consider normal. (It's dropping 0.4Ah per month)

 

[nico]

Cell dimensions:

With the LEV50 beeing 111 x 171 x 44 (in mm) how much available space do we have to fit a higher/larger/thicker cell?

The thickness (44mm) is limited by the spacing between the sensor holes on the CMU board. Does anyone have the spacing measurement?

Height (111mm) will be limited by the frame that encloses the cells and the available space above it in the battery case. Do we have some room? Can the frame be modified to house higher or larger cells? Smaller dimensions can always be fitted with spacing blocs so they are high enough to connect to the CMU board... We would nee to find a suitable non flammable, possibly non conductive, material for the spacers...

Cells are often available in 173mm width (instead of 171mm); would they fit?

Some batteries available from China are (in mm): 85 x 173 x 42
https://ru.aliexpress.com/item/4000322949420.html?spm=a2g0o.productlist.0.0.4d9d3498TVxT5y&algo_pvid=a880ac8d-c9b1-4e42-b6f7-035ef58f30c2&algo_expid=a880ac8d-c9b1-4e42-b6f7-035ef58f30c2-16&btsid=f40e9ae9-3b16-48fb-8b2e-4b0d8cf9fdf3&ws_ab_test=searchweb0_0,searchweb201602_9,searchweb201603_52

Battery transportation:

I am trying to purchase a used 70000km traction battery to replace my aging 160000km one. One issue I just came across is that traction batteries are considered dangerous goods and they require the documents showing conformity with UN Manual of Tests and Criteria part III subsection 38.3. Also, packaging must be labeled according to UN3480 Class 9 Hazardous material. Unless they are carried within the car, I need to fill the paperwork to move the traction battery to my home. Does anyone have a copy of a filled dangerous goods transportation form for a i-Miev traction battery?

The closest I found is this document that applies to LEV50 and LIM50E, which seems to be the industrial version of the LEV50.
http://www.gsyuasa-lp.com/SpecSheets/LIM50E-MSDS.pd

It does not mention conformity with subsection 38.3.

The issue will also arise if we try to import cells. The manufacturer will need to have performed the UN tests in order to be able to fill the paperwork required to ship the cells.

A lot of information is available on this Canadian government site:
https://www.tc.gc.ca/eng/tdg/transporting-batteries.html
It looks like the primary class placard may not be required for a gross mass under 500kg (which is more that the weight of an i-Miev traction battery), but the paperwork is still required in the transporting vehicle...

 

[dracekvo]

Interesting

https://elektroauto-forum.de/kleinanzeigen/peugeot-ion-lion-umbauset-f%C3%BCr-traktionsbatterie-unvollst%C3%A4ndig-muss-erg%C3%A4nzt-werden.93/

62,5Ah on 25P means one 18650 cells have 2500mAh.

On DC charging (130A) one cell must eat 5A.
Braking return similar current.

How long can work this battery in iMiev?

 

[coulomb]

I found this article which brings some information about the current sensor in the Miev:

Battery Management System Hardware Concepts: An Overview
by Markus Lelie, Thomas Braun, Marcus Knips, Hannes Nordmann, Florian Ringbeck, Hendrik Zappen and Dirk Uwe Sauer
Appl. Sci. 2018, 8, 534; doi:10.3390/app8040534
http://www.mdpi.com/journal/applsci

I note that this article refers to several example EVs, not just the iMiEV, e.g. the Smart 4-2, Tesla.

Further in the document, we find that the sensor is a LEM CAB300 which has these specifications: ...
• Unipolar + 12 V battery power supply ...
• Output signal: CAN.C (500 kbps)

This must be the LEM current sensor for another vehicle. The iMiEV sensor (yes, a LEM, but they make a lot of sensors) operates from 5 V and the output is analog, not CAN.

Information on the iMiEV current sensor can be found here: http://myimiev.com/forum/viewtopic.php?f=23&t=4718

 

[dajsinjo]

Interesting

https://elektroauto-forum.de/kleina...atterie-unvollständig-muss-ergänzt-werden.93/

62,5Ah on 25P means one 18650 cells have 2500mAh.

On DC charging (130A) one cell must eat 5A.
Braking return similar current.

How long can work this battery in iMiev?

Cells with low internal resistance is a must, as high charge and discharge rates is needed.
Lev50n is used from 2.75V to 4.1V range, so you can't use full capacity from usual cells that is usually from 2.5 to 4.2V
Also turtle mode is suppose to kick in before, so useful capacity is even lower.
If they chose cells with higher internal resistance it will also lower range because of losses,
it could be dangerous when overheating from fast charging and regenerative braking.

 

[Rational]

So I'd read that NMC cells are a bit more volatile compared to LFP, is that right? But have a bit more range and are perhaps a tad lighter?

What about if I went with LFP instead eventually when I do this?

 

[kiev]

What about if I went with LFP instead eventually when I do this?

It might be made to work if it fits into the volume available and can connect to the CMU boards.

But not sure if the energy density of LFP would provide any increase in range, and the usable voltage range is outside of that programmed into the BMU. e.g. Max charge voltage

 

[piev]

What about if I went with LFP instead eventually when I do this?

What Kiev said is true but the bigger issue with LFP is the voltage. The NMC cells are charged to 4.15v or 4.1v I can’t remember but LFP cells can only be charged to 3.65V. I believe you would need to completely replace the BMU or reprogram it and perhaps the CMUs. Their may be others here that know more details but it would be a considerable effort.

I will say I have thought of doing this but could find no documentation on the BMU processor to reprogram it. YUASA does make a LFP cell in the same form factor.
Yuasa LFP Cell LEV60

 

[MickeyS70]

What Kiev said is true but the bigger issue with LFP is the voltage. The NMC cells are charged to 4.15v or 4.1v I can’t remember but LFP cells can only be charged to 3.65V. I believe you would need to completely replace the BMU or reprogram it and perhaps the CMUs. There may be others here that know more details but it would be a considerable effort.

Not necessarily, in theory one could use a second can bridge on BAT-CAN and add a .45V offset to any voltage reported from the CMU. What may be trickier is to determine a reliable voltage based SoC algorithm due to almost flat discharge curves?

YUASA does make a LFP cell in the same form factor.
Yuasa LFP Cell LEV60

Great find..

 

[Jiminy]

Besides chemistry, the 55mm thickness on LEV60F would also be an issue...

 

[MickeyS70]

According to link below there is such a thing as OEM LEV61 cell, given that it seems to match the dimensions of the LEV50 exactly and the BMU is capable of dealing with 60Ah cells these would be literally plug and play, no hacks required, just a bit of patience.

https://www.gs-yuasa.com/en/products/vehicle/phev.php

Question is though, are they available to order and at what price?

 

[Rational]

but the bigger issue with LFP is the voltage. The NMC cells are charged to 4.15v or 4.1v I can’t remember but LFP cells can only be charged to 3.65V.

Thanks for that. I was forgetting that difference between the different chemistry voltages. Sounds like a good reason to just stick with the NMC chemistry it was designed for originally.

How do you like your CATL cells? Did they turn out to be genuine upon arrival? If so I might consider these.

But what's interesting is that both links posted above show the cells as being 3.65v not 4.15v cells. Is that right?

https://batteryhookup.com/products/new-case-with-6x-3-2v-74ah-lifepo4-prismatic-cells

https://www.alibaba.com/product-det...8.html?spm=a2700.wholesale.0.0.256e30b1jevR7c

Or is that nominal voltage and they actually charge to the full 4.15?

It looks like the LEV60 original type are only 74ah vs 93ah for the CATL cells.

Oh, I may have found my own answer to the voltage question: CATL sales page says, "Maximum charging voltage 4.2V".

But the LEV60 says, "Full Charge Voltage: 3.65V"... hmmm. What am I missing here.

 

[MickeyS70]

Thanks for that. I was forgetting that difference between the different chemistry voltages. Sounds like a good reason to just stick with the NMC chemistry it was designed for originally.

LEV50s are most likely LMO chemistry

What am I missing here.

LEV60F cells simply aren’t suitable as a direct replacement; while the voltage difference could be managed, they won’t fit physically (see @Jiminy ’s post #247)

Regardless of what the chemistry’s upper/lower limit is, the native BMU will stop charging once a cell hits around 4.1V and will limit power below 3V.

There is (on paper) another option: Yuasa’s LEV61 cell has identical dimensions and a 20% higher capacity compared to LEV50.

https://myimiev.com/threads/the-aftermarket-cell-replacement-saga.4403/page-2