Aftermarket iMiEV battery upgrade project

[nuggetgalore]

So my deduction is that if the drive is on,it bleeds a small amount from the cell,allowing the lower cells to catch up.

Good, that's the way that makes most sense.

Perhaps you need to discharge the battery to a point where roughly 50% of the cells are bypassing. All bypassing does no good (just wastes stored energy), and no or few bypassing does little good.

When I started to monitor the cell balance drivers, I did it with my old degraded pack first and then with the new 60Ah cells. They seemed to behave similar with a degraded pack and an "unknown" one, with very few drives off at any time. So last Tuesday evening I had the opportunity to check charging a 2012 iMiEV ( Odo 104,913km). This car was fully charged during a very warm/hot night, displaying 110 km on the GoM. The owner drove normally,never babying, for 89 km arriving home with 4 bars on the fuel gauge and 24 km on the GoM.Being a warm/hot day certainly made difference, but this and many observations before "proves" that his cells are most likely LEV50Ns,and much improved from my 2010 LEV50s.
Surprisingly the number of cell balance drives "OFF" were no more than what I observed on my packs (as a matter of fact it started probably with one only (73 low cell,out of range to monitor) then low cell(s) alternated between CMU 03 cells C and D (19,20),their drives being on either one or both until charging to about seven bars on the fuel gauge when we ended charging. I found no other drive to be on but cannot rule out the low probability of one in CMUs 10 to 12.Maybe one day I can capture the drive status' at full charge.

[richi]

The model year 2013 changeover, when the N cells were introduced, was roughly July/Aug 2012. So if that car has N cells, it's a MY2013, presumably first sold in late 2012.

[nuggetgalore]

The model year 2013 changeover, when the N cells were introduced, was roughly July/Aug 2012. So if that car has N cells, it's a MY2013, presumably first sold in late 2012.

I just confirmed with the owner that he bought it in 2013, it was basically new (but like all iMiEVs at a price of 55K AUD did not sell for a long time). He was able to get it for half the price, the dealer wanted it out of the yard!
You must be right,it would be a MY2013.
I checked the degradation of three MY2013 cars during the last 3 years or so and found it to be a lot less than on the two cars the same age as mine (I assume 2010 but maybe 2011).
By the way, the two older cars both have only about half or fewer kilometers on the odometer than the three younger ones. Age versus charging cycles a big question mark.
Edit:typos

[richi]

Compare with my car, a MY2012 with 72,000 km: In summer, I'll get a genuine 110 km in normal driving. BMU reports 39 Ah health—or 82% (of 47.5 Ah).

So, all else being equal (which it rarely is) an N-cell car with 50% more on the odo gets the same range as a late-model with the original cells. Figures.

[dracekvo]

I have same car year. Buy with 56000 km. Now 85 kkm.
I don´t have knowledge and diagnostics when buy this car. If i know, that battery have only 34Ah, i was newer buy it.

About one year this car work in dealer. I mean, workers use this car only with full gass in city and charge only DC 50kW.

Real range is 60-70km. (i need 45 to my work). I use parking heater. With integrated heater i have real range about 30-35 km.

There is my Ah graph history.
B: km
C: AC kWh, some time reset if internal battery die.
D: Ah

http://elektroforum.cz/viewtopic.php?f= ... 100#p10352

Sorry for my bad english.

[nuggetgalore]

I don't quite get this. Do you mean just leaving the battery at rest without being charged? All the balance drives are OFF as soon as the charging stops, irrespective whether the EVSE cable is plugged in or not. As far as if understand, no balancing happens except if charging.

Oops. I was thinking continuous bypassing like with the Nissan Leaf.

How can individual cells be manually balanced without taking the pack out?

I believe it can be done with 0x3C3 CAN messages (on the bus between the BMU and the CMUs, I don't know if that can be accessed from the OBD II port). I've seen posts on DiyElectriccar indicating that it can be done, but the poster didn't give details. I might take another look at the CMU firmware to see if I can figure out what the flags in bytes 3 and 4 actually do. Edit: bytes 1 and 2 are known to contain the threshold voltage, above which the bypass FETs turn or pulse on.

Could the special function "PROCEDURES FOR SMOOTHING CELL VOLTAGE" be the tool to balance the cells? The manual does not stipulate any state of charge resp. cell voltage for this procedure, so if it does what I think it does, the cells could be both top and /or bottom balanced?
I got hold of a supposedly MUT3 clone, this special function is there but does not execute, so no joy there yet. Another special function is "MAIN BATTERY CAPACITY MEASUREMENT". That function is missing on this "clone".
It is not all bad news though, by accident, not by design, a certain value got reset to zero without me knowing. I was browsing through the functions of above mentioned clone (that can only be done with the tool connected to the car), and the counter of the months since the battery was installed got wiped.
Now this under normal circumstances would be a big problem, but in my case a blessing in disguise.
It in turn reset the Ah capacity to 48 Ah (the factory default?). What this means is that I can now discharge said Ah rather than the 30 odd that the original pack allowed. It appears that cell voltage plays a minor role in determining calculated range remaining,tortoise and complete shutdown. It is all the fault of coulomb counting,lol!
Charging still restricted to about 359 VDC (4.08 average) because I think the bms cannot overcome the cell volt differences. Balanced cells charging to 362 VDC (4.11V) would add some extra energy into the battery I guess, but beggars can't be choosers.
Tested range so far 87km (fuel gauge 3 bars and RR 21) so a possible 100km in real life.

[richi]

Didn't I suggest you did that a few weeks ago?

When replacing the traction battery, Mitsubishi tells techs to "initialise" the BMU. The procedure is two separate MUT-III commands, both of which include the word "reset" in their names:
1. Batt. capa. estimated info. reset
2. Control information reset

[nuggetgalore]

Perhaps you need to discharge the battery to a point where roughly 50% of the cells are bypassing. All bypassing does no good (just wastes stored energy), and no or few bypassing does little good. Then just leaving the battery at about that state of charge (it might be necessary to top up or bleed down now and then) might allow better balancing while you do other things, even while you sleep. Of course, it would take at least weeks to balance it, but you might as well leave it at the ≈50% balancing point as much as possible.

One more twist in the cell balancer drive saga.
Following the 87 km test drive, and charging, I connected the MUT3 clone again and that is when I noticed the reset Ah figure of 47.9 instead of the previous 31.70 Ah.
With this clone it is a bit easier(not much*) to check on cell voltages and balance drive status.
At the start of charging all balance drives were OFF! I kept double checking the low volt cell(s) against the drive status,but must have missed the time when the first drives turned on.
At 14 bars on the fuel gauge (charge state (control)65.50% ;charge state (display) 86.50% {yes,no typos}) one or two drives were ON (steady state) and about a dozen were changing state on an irregular frequency. This scenario continued until battery was full, never reaching a stage where
just about all drives were on previous to the Ah capacity reset.

*all 88 cell balance drives are accessible, but one (CMU10 cell A) is probably faulty as instead of
displaying a status, it shows a dash -.Edit: voltage display on this cell is ok.

[coulomb]

Following the 87 km test drive, and charging, I connected the MUT3 clone again and that is when I noticed the reset Ah figure of 47.9 instead of the previous 31.70 Ah.

I predict that what will happen is that the estimated Ah capacity will gradually (perhaps not so gradually) fall from the reset figure until it reaches a figure somewhere in the low thirties (it might even go further than that, we didn't let it continue past that). This is based on a few days of observations, so there could easily have been confounding factors.

You could continue to reset the estimated capacity, to get a fixed capacity of around 48 Ah. But that's still rather short of the nominal capacity (60 Ah I believe).

That's why I tried current spoofing. Unfortunately, it's now been shown that current spoofing on its own leads to unreliable range estimates, and frequent towing. So that's been stopped for now. There will be a test car converted soon, where we can watch the results more closely than an actual customer's vehicle.

It seems that your estimated capacity has already fallen from 48.0 to 47.9 (but that's such a small drop that it may not be significant; it could be something like a rounding error). But please keep an eye on that figure and record its value over time if possible.

[ Edit: I no longer believe the below text in red; I now believe that the Ah estimate will be reset to around 48.0 Ah. See a few posts down. ]
My notes say that the reset capacity is 45.8 h, because of the restricted voltage range the car uses for longevity of the cells (4.1 to 3.6 VPC per my notes). The cells are nominally 50 Ah. It's interesting that you're seeing 48.0 initially. It may be that the clone actually gets it wrong, setting the estimated figure to the wrong value. But if so, that would mean that the capacity can be set to any figure, possibly including 60.0 Ah. I think I read somewhere that the highest estimated capacity that the BMU can store is a little over 60 Ah. That's why I tried current spoofing for the 93 Ah cells; since it can't learn a high enough Ah estimate, it had to look like the car was amazingly efficient, i.e. making normal driving appear to draw less current than it actually does. The reason for the unreliable and wildly varying Range Remaining estimates that apparently result from current spoofing are completely unknown to me.

[wovenrovings]

Was just wondering which part does the coulomb counting and the GOM calculations? I would guess it is the VCU not the BMS master?

[coulomb]

I believe it can be done with 0x3C3 CAN messages (on the bus between the BMU and the CMUs, I don't know if that can be accessed from the OBD II port). I've seen posts on DiyElectriccar indicating that it can be done, but the poster didn't give details. I might take another look at the CMU firmware to see if I can figure out what the flags in bytes 3 and 4 actually do. Edit: bytes 1 and 2 are known to contain the threshold voltage, above which the bypass FETs turn or pulse on.

I've been spending some time this last week re-reading the CMU firmware. I've posted very brief notes on it here. It's very hard going, but I've had some minor breakthroughs. I'm still working on the fourth byte flags, however. I still need another breakthrough: how tasks get awoken.

Could the special function "PROCEDURES FOR SMOOTHING CELL VOLTAGE" be the tool to balance the cells? The manual does not stipulate any state of charge resp. cell voltage for this procedure, so if it does what I think it does, the cells could be both top and /or bottom balanced?

What that procedure does it mysterious to me. It could well be what you want. But the clones (even expensive clones like the Gscan) don't seem to get it right; it has the function, but when activated, it just says "Testing" and nothing seems to happen.

If you had a tool that did this function properly, it will be assuming original cells (LEV50s, possibly LEV50N). At least the LEV50s, which I believe are LMO chemistry, have a flatter voltage versus SOC characteristic than NMCs. I don't know how this would affect balancing of NMC cells; it may be that NMCs could be balanced at pretty much any SOC / cell voltage, but the tool may impose restrictions that only make sense with LMO chemistry.

Even though you could smoothe the cells at nearly any SOC, it's probably worth considering at what SOC you want to smoothe them. If you balance them at the top, you might be able to charge them to a higher average cell voltage, increasing range slightly. But that would unbalance them at the low end, causing less range because any cell instantaneously under 3.0 V will produce a temporary (?) turtle. Plus, it may be that the BMU does most of its capacity estimation updates at lower SOCs. Perhaps it depends on where your biggest outlier cell is compared to most of the others. For example, if your biggest outlier is higher SOC than all the rest, it might be better to smoothe at the top, so it doesn't prevent most cells from charging to 4.1 V. But if the biggest outlier is at a lower SOC than the rest, it might be better to smoothe at the low end, so that it doesn't cause the turtle prematurely. It might also depend on the relative capaciity of the outlier to the rest, rather than the present SOC compared to the rest.

[coulomb]

Was just wondering which part does the coulomb counting and the GOM calculations? I would guess it is the VCU not the BMS master?

My guess is that the BMU (BMS master) would almost certainly do the coulomb counting. It connects to the current sensor in the battery.

I would think that it would also calculate Range Remaining, but as you say, that could be the VCU. Sadly, I can't get hold of the BMU or VCU firmware for analysis (at least so far).

[nuggetgalore]

icle.

It may be that the clone actually gets it wrong, setting the estimated figure to the wrong value. But if so, that would mean that the capacity can be set to any figure, possibly including 60.0 Ah. I think I read somewhere that the highest estimated capacity that the BMU can store is a little over 60 Ah.

re the "clone". This rather poor excuse for a professional device is actually a scanner rather than a clone, but the distributor claims it is made specifically for Mitsubishi and Nissan to do all the checks,resets and adjustments. This is obviously not true.
As far as the "clone" setting the estimated figure wrong, I doubt that very much, as there are the very fields missing that would allow this and other values to set.
I think all that happened was the value for months since the battery was installed was wiped and reset to zero. That in turn reset the Ah value to a factory default but by the BMU, not by the "clone".
Edit:
Yes setting the Ah to the correct value may be possible with a MUT3 or a good quality clone.

[nuggetgalore]

An appeal to any body who has had some success with a multi user tool 3 (MUTIII) or a clone to share that information.
Thanks.
Reasons I am reluctant to get a clone from any of the internet sellers are that they work on windows xp only, reviews are not encouraging and the instructions are vague.

[coulomb]

As far as the "clone" setting the estimated figure wrong, I doubt that very much, as there are the very fields missing that would allow this and other values to set.
I think all that happened was the value for months since the battery was installed was wiped and reset to zero. That in turn reset the Ah value to a factory default but by the BMU, not by the "clone".

Yes, I think you are probably right. I was misled by this post (about half way). Looking back through my notes, I see several readings of over 45.8 Ah, one at 47.8 Ah, which could well have been fairly soon after a battery capacity reset (using a Gscan tool). So I now believe that the BMS resets to 48.0 Ah or close to it.

Edit:

As for BMU data list items 23 and 24, as you probably know, 23 is the estimated capacity that gets reset to 48.0 Ah and comes down rapidly when the BMU detects "anomolous" behaviour (actually due to the changed chemistry), and 24 is just the present estimated Ah remaining on which SOC is based. It presumably starts at the same value as item 23 when the battery is deemed to be full, and reduces (or increases) as integrated Ah are measured going out (or into) the battery. That's obviously where I saw the 60 Ah maximum.

Thanks for posting that; it has only just dawned on me that this is the reason that all the various scanner tools agree so closely on their descriptions of the data list items. It comes from Mitsubishi themselves. I was getting an idea at one point that these somehow are readable from the BMS itself.

[nuggetgalore]

Hello everybody,
best wishes for a happy, healthy and prosperous 2021.

I have now a bit more data to play with.

As for BMU data list items 23 and 24, as you probably know, 23 is the estimated capacity that gets reset to 48.0 Ah and comes down rapidly when the BMU detects "anomolous" behaviour (actually due to the changed chemistry)

this value has already steadily dropped to 46.90 Ah, no idea if because of an inbuilt reduction value for new cells or if the BMS was just confused.

and 24 is just the present estimated Ah remaining on which SOC is based. It presumably starts at the same value as item 23 when the battery is deemed to be full, and reduces (or increases) as integrated Ah are measured going out (or into) the battery.

BMU data list item 27 is "Charge current integrated value". I am puzzled with this as it has not changed from 25.70 Ah even though the Ah charged and /or discharged from full to empty varied widely.So really I have no idea what it means.

That's obviously where I saw the 60 Ah maximum.

Thanks for posting that; it has only just dawned on me that this is the reason that all the various scanner tools agree so closely on their descriptions of the data list items. It comes from Mitsubishi themselves. I was getting an idea at one point that these somehow are readable from the BMS itself.

For an unknown reason the new scanner refuses now to read the EV_ECU , that is where by fluke or accident I reset the months since battery install to zero, I did see this increasing to about .5 or so after a couple of weeks (in line with the decrease of the estimated capacity value item 23 ).
So far I traveled 748 km with the new cells.
A couple of times (both during warm weather, I was able to get above 70 resp 80 km range.
All the other times the real life range was between 50 and 60.
I am sure the BMS is not able to overcome the big discrepancy in cell capacity or variation of cell charge behavior. In an earlier post I wrote that the cell balance behavior of the LEV50Ns is the same as the 60Ah NMC-LIYUANs at the start of charging at low SoC, I can confirm that this is also the case at near full. The difference is that the LEV50 high volt cell(s) are within 0.01 V or better all the way from bottom to top where as the 60Ah NMC-LIYUAN cells are 0.04 V with short lived moments of 0.03V. Discharging is even more pronounced, at full tank and modestly accelerating (~50 A) low volt cell drops up to 0.2 V below top V cell,increasing to 0.4V when tank is half empty. At rest they bounce back to 0.04 or 0.05 difference.
I am contemplating to pay Mitsubishi to do the PROCEDURE FOR SMOOTHING CELL VOLTAGE.
Correct me if I'm wrong,I assume this means balancing the cells. But where? Top ,middle or bottom? Since I had the cells balanced ( 3.985 / 3.992 V) before charging them as a pack the first time, I hold no great hope that it will fix the problem.
One more anomaly is getting more pronounced.
Yesterday I drove until two bars. Then I turned on the heater full blast until the fuel gauge was empty and GoM 2 km. Let it rest for 40 minutes. SoC 10.50% (control),11.5%(display),
3.65 V(44),3.61 V(2)
Charged to "full" ,13.18 kWh of the wall in 356 minutes, disconnect cord.
46.90 Ah (item 23) 43.60Ah (24),78.5% (c),86%(d), total 359.7 V . RR (GoM) 108km, fuel gauge14 bars.
Let it rest for a while and kept checking to find SoC increased a bit, total 359.4 V ,GoM now 117km ,fuel gauge 15 bar.
This morning I checked again to find State of charge 93% (C),96%(D), max and min cell V 4.09 resp 4.06, total 359.4 V ,GoM now 127 km and fuel gauge 16 bar!
One mystery after another!

[francisco.shi]

@coulomb Do you know who decides the battery is 49Ah? Is it the BMS in the battery or the car computer?
Do you know where I can get info on what wires the CAN bus runs on?
I am trying to get this problem sorted too but I just started.

[richi]

The BMS is implemented in the BMU, which I think is one of the boxes under the back seat, rather than in the pack. The electronics in the pack are the CMUs (cell monitoring units), which just report Voltages and temperatures and do balancing as micromanaged by the BMU.

[francisco.shi]

Do you know where I can get the pinout for the plug?
I want to see what messages are being sent in the CAN bus.
I am thinking that it may be that the reason the current spoofing didn't work is because the BMU may also be getting current information from the charger and the motor.

[brendon_m]

This is for a 2011

HET15E14BC00ENG (1).pdf

(67.53 KiB) Downloaded 8 times

HET15E14AC00ENG (1).pdf

(116.87 KiB) Downloaded 7 times

Screenshot_20210125-054302.png (340.67 KiB) Viewed 286 times

Screenshot_20210125-053227.png (226.36 KiB) Viewed 286 times

[brendon_m]

And the BMU diagram as a whole

HET04E03AC00ENG.pdf

(107.59 KiB) Downloaded 9 times

HET04E03BC00ENG.pdf

(151.13 KiB) Downloaded 8 times

HET04E03CC00ENG.pdf

(32.98 KiB) Downloaded 8 times

Screenshot_20210125-055305.png (408.09 KiB) Viewed 286 times

Screenshot_20210125-055235.png (333.82 KiB) Viewed 286 times

Screenshot_20210125-055204.png (114.54 KiB) Viewed 286 times

[brendon_m]

The BMS is implemented in the BMU, which I think is one of the boxes under the back seat, rather than in the pack.

Yes it is

Screenshot_20210125-060721.png (392.58 KiB) Viewed 286 times

[coulomb]

@coulomb Do you know who decides the battery is 49Ah? Is it the BMS in the battery or the car computer?

It pretty much has to be the BMU, since that 49Ah (initially) figure is reported as a value of the BMU by the scanners. It's possible that the EV-ECU has some influence over the result.

As far as I know, no-one has analysed or reverse engineered the BMU firmware. It's possible that that may change over the next few months. A person in Russia claims to have software that with suitable hardware will be able to read and write the firmware from/to vehicles, even if they are in a "bricked" state. That sounds extraordinarily to me, but time will tell. This software has never been used on EVs as far as I know, though other software by the same author has been used on Mitsubishi Outlander PHEVs.

Do you know where I can get info on what wires the CAN bus runs on?

@Richi has provided some great information above. The iMiEV 2011, 2012, and 2013 service manuals are available for free from this site. Sadly, the minivan service manuals are much harder to find, and I don't know of any sites that host them.

[nuggetgalore]

@coulomb Do you know who decides the battery is 49Ah? Is it the BMS in the battery or the car computer?
Do you know where I can get info on what wires the CAN bus runs on?
I am trying to get this problem sorted too but I just started.

The 49 Ah (Edit: I think it may be closer to 48 Ah) seems to be the factory default for the pack of LEV50 cells.
On my pack (that obviously is nowhere near the original LEV50 specification), this value has come down to 46.2 Ah @1.75 months since resetting. I wonder if this is a preprogrammed reduction,using the elapsed time and several other factor assuming the drop in capacity is strongest with new cells and then the decrease flattens out.
Edit :see above

[francisco.shi]

What I was thinking was that if we intercept the cell voltage values and any message going to the BMU that makes it think the battery is different and replace it with what the BMU expects to see and then send out what we want the car to see then it may fix the problem. So it could be fixed by a man in the middle approach.