Which cell loses capacity fastest? Which retains it best?

Now that we have the capacity to read all cell-pair voltages, I think it would be good to start looking at the performance of our packs to try to determine if there are consistent trends between cars as to which cell(s) lose capacity fastest and which cell(s) lose capacity slowest. This information will be useful in determining if module rotation is warranted and, if so, it should also aid in determining which rotation patterns might be best.

For reference, page EVB-20 of the November 2010 Edition, April 2011 Revision of the Nissan LEAF Model ZE0 Series service manual contains a pictorial view of the layout of the 48 battery modules in the 2011/2012 LEAF. Each module contains two pairs of cells, so there are a total of 96 cell voltages which are measured and reported by the LEAF. The service manual numbers the modules from MD1 to MD48 and the cell voltages from Cell-1 to Cell-96. Cell-1 is at the highest-voltage end of the stack with its positive terminal connected to the positive terminal of the battery. Cell-96 is at the lowest-voltage end of the stack with its negative terminal connected to the negative terminal of the battery. To determine which cells are contained in a given module, simply multiply the module number by two for the higher cell number and subtract one from that product for the lower cell number. For example, module MD23 contains the following two cell voltages: Cell-45 and Cell-46.

Modules MD1 through MD24 are contained in a stack under the rear seat with MD1 on the far passenger's side and MD24 on the far driver's side. Modules MD25 through MD28 are located under the rear driver's side footwell. Modules MD29 through MD36 are located under the front driver's seat. Modules MD37 through MD44 are located under the front passenger's seat. Modules MD45 through MD48 are located under the rear passenger's side footwell.

The exact location of each module can be seen in the MODULE LAYOUT diagram from page EVB-20 of the service manual, shown below:



In order to determine the relative capacity of the cells in your LEAF, please follow these steps:

1) You must first achieve a high degree of cell balancing, which is occurs at the top of the charge range. So you will need to perform at least one 100% charge to balance your pack. Sometimes, you will need to charge to 100% several days in a row to achieve a good balance. I find that charging to 100% using 120V gives the best amount of cell balancing. You can also let your LEAF sit, plugged in, following a 100% charge for a few hours to let it attempt to rebalance once the pack fully cools.
2) Once your pack is fully balanced, measure and record your cell voltages using the one of the methods shown in the second post of the this thread, which I will reserve and update as new techniques for reading cell voltages are developed. This is your record of how well balanced you pack is.
3) Now, discharge the battery pack to a point where the lowest cell voltage is 3712 mV or lower. This voltage is from Nissan's CELL VOLTAGE LOSS INSPECTION procedure found on page EVB-65 of the service manual. In order to do this, Nissan recommends putting the vehicle in READY and setting the following:
- A/C set temperature: Full hot
- A/C fan speed: Maximum speed
- A/C air outlet: Defroster
- Headlamp: High beam ON
- Door glass: Full open
Monitor the available charge gauge and watch for it to go down to two bars or less. This should take about four hours if the battery was fully charged. It may take less time if your pack is degraded.
4) Once your pack is at or below 2 charge bars, measure and record your cell voltages. Ensure that the lowest voltage recorded is 3712 mV or lower. If it is above this voltage, repeat step 3) above. This is your record of how much voltage difference there is between all of the cells in your LEAF at a low SOC. If you pack was very well balanced in step 2) above, then the cells with the highest voltages here should have the LEAST capacity degradation and the cells with the lowest voltages should have the MOST capacity degradation. (Note: this test does NOT tell how MUCH capacity degradation your battery pack has, only how much difference there is between different cells in the pack.)

Once you have performed the test, please post images of your results from both step 2) and step 4) here (preferred) or simply type in the top six and bottom six voltages from both steps along with their cell-pair numbers. I will collect the data, perhaps in a Google spreadsheet, and will try to find capacity loss trends by plotting histograms of results which we achieve.

Please post if you find any errors in the test procedure or would like to point out new ways we can measure cell-pair voltages. I can then update either this post or the second one. Thanks in advance!

Links to posts describing equipment/techniques for reading your cell-pair voltages:

- Using ELM327 Bluetooth OBDII Adapter and Turbo3's LEAF Battery App - Thanks, Turbo3!
- Using ELM327 Bluetooth OBDII Adapter and Hyperterminal - Thanks again, Turbo3!

I posted a thread last night in the CAN section of the forum to post screenshots from the Android app. Guess great minds think alike!

JeremyW said:

I posted a thread last night in the CAN section of the forum to post screenshots from the Android app. Guess great minds think alike!

Click to expand...

Ha! Very funny! :ugeek:

I thought about posting there, but that area is not widely read. I wanted to gather together the procedures and layouts in one place to try to get a consistent view of data.

Do you know if anyone has done the full test including good balancing followed by testing at low SOC? I don't have the equipment to do it just yet, but I will.

First of all, let me add my thanks to Turbo and the others who reversed engineered the CAN bus messages and have made this app that uses the ELM327.
I bought my ELM327 on Ebay from romorus for $10.70 with free shipping. It works with my LG Nexus 4 running Android 4.2.2 without problems.

I did the test of running the battery down from full to nearly empty. As I normally only charge to 80%, two nights in a row I charged to 100% and after it was done, told it to start charging again. This presumably balanced the battery cells. I was surprised to find the SOC starts at only 92.08%. I then drove. I got LBW at 21.6% and VLBW at 10.72%. I then drove slowly home.

The screen shots for fully charged and nearly empty are below.

RegGuheert said:

... try to determine if there are consistent trends between cars as to which cell(s) lose capacity fastest and which cell(s) lose capacity slowest. ...

Click to expand...

You'll likely find that cells in the middle of the pack which are hottest tend to lose capacity the soonest.

thimel said:

I was surprised to find the SOC starts at only 92.08%. I then drove. I got LBW at 21.6% and VLBW at 10.72%. I then drove slowly home.

Click to expand...

Great data, thanks for sharing! Based on the CAP value, I would hazard a guess that you lived in South Bay. Is that correct? Could you perhaps help narrow it down a little? I have some thoughts on the low maximum SOC, but it's likely premature to talk about it. How often do you charge to 100%? Is the LEAF parked outside or inside at night? Do you do much freeway driving, and if so, how fast would you typically go? Any chance that your car was represented in the PIA battery survey? Hope you don't mind all the questions

In my spare time at the end of the month I rotate my cells for longer life. No sense in not dropping the pack once in while and spending a full day rotating modules. I worked on an spreadsheet to track the rotations and a graph on the time spend VS percent capacity gained. I divided that by the number of days in the year multiplied by how I would rather do something else and came up with time spent doing other things would be a better choice. Next up I'm going to plot the wear of my wiper blades from end to end and measure the wear deviations. I may put this on Google Docs.

South Bay is correct. Sunnyvale to be exact. I got the car in June 2011 and it has 19k miles on it. It is parked in a garage at night and out in the sun on weekdays. I charge to 100% about once every two months. I have quick charged once just to see if it worked. I drive about 25 freeway miles every weekday. Speed is around 65 except when there is traffic.

thimel said:

South Bay is correct. Sunnyvale to be exact. I got the car in June 2011 and it has 19k miles on it. It is parked in a garage at night and out in the sun on weekdays. I charge to 100% about once every two months. I have quick charged once just to see if it worked. I drive about 25 freeway miles every weekday. Speed is around 65 except when there is traffic.

Click to expand...

Great, thanks! Now that you have more hard data about the health of your pack, please consider contributing it to Tom Saxton's battery survey. This appears to be the only way we can get some information about longevity and real-world performance of these batteries.

EVDRIVER said:

In my spare time at the end of the month I rotate my cells for longer life.

Click to expand...

Why not purchase a fan and blow smoke at the pack every night? A reliable source told me that this will add one Gid per month, regardless of ambient temperature. Plus smoked Gids are better than raw ones. Or so I've heard. With little luck, the doors will rust through long before the battery gives up its ghost. Look forward to studying the Google spreadsheet!

thimel said:

I did the test of running the battery down from full to nearly empty. As I normally only charge to 80%, two nights in a row I charged to 100% and after it was done, told it to start charging again. This presumably balanced the battery cells. I was surprised to find the SOC starts at only 92.08%. I then drove. I got LBW at 21.6% and VLBW at 10.72%. I then drove slowly home.

Click to expand...

Very cool!

Here's my analysis at the data.

1. The overall shape of the 92% and 8.75% (or 100% and slightly below VLBW) appear to be very similar, leading me to believe that the relative health of your cells are fairly close to the same.
2. A SOC of 92% is low - that should read about 95% on a 100% charge.
3. At 8.75% you have a very wide spread in cell-pair voltages at 115 mV.
4. The 4 lowest voltage cell-pairs at 8.75% are also some of the lowest at your 100% charge - in particular #s 20, 26 and 35.

It seems fairly evident that more balancing would get you more usable capacity of your pack - and that if the BMS had the capability to shunt energy to lower voltage cells it might be able to do a

I think that when grabbing this data, also grabbing a screenshot when LBW and VLBW might be useful. For people that normally charge to 80%, a screenshot at 80% and a screen shot right after a 100% charge and after sitting at 100% for a couple hours would be interesting.

Can't wait for my BT adapter to get here!

thimel said:

I did the test of running the battery down from full to nearly empty. As I normally only charge to 80%, two nights in a row I charged to 100% and after it was done, told it to start charging again. This presumably balanced the battery cells. I was surprised to find the SOC starts at only 92.08%. I then drove. I got LBW at 21.6% and VLBW at 10.72%. I then drove slowly home.

Click to expand...

Thanks! Very interesting!

Can you tell us how many miles are on your car? (Perhaps I need to add a short questionnaire to the instructions for some vehicle details? MY, VIN, Purchase Date, location, miles, charging habits, driving habits, etc. OTOH, I don't want this to be too difficult!)

It seems clear that the stack of modules which stand on edge under the rear seat appear to be degrading faster than the ones lying flat under the rear floor and front seats. Even Cell-48 in the Module 24 on the far end of the rear stack is pretty degraded. Perhaps any module rotation scheme might see a swap between the modules up front and those in the back, among other possible changes.

I will admit that I had imagined that the cells in modules MD35 through MD38 (Cell-69 through Cell-76) at the tops of the stacks under the front seats might get the hottest due to long path down to the heat sink. In fact, these tend to do better than the cells in the rear stack and Cell-69 seems to be doing particularly well.

I suspect the vertical cells in the back develop a temperature gradient across the cell which might also result in a current gradient across the cell. If so, such a current gradient could cause the vertical cells to appear to have lower capacity than an isothermal cell, even before any degradation occurs. It could also accelerate degradation. It will be interesting to see similar data for a brand-new LEAF with all cells nearly undegraded. Alternatively, it might be interesting to see results comparing a car which is driven versus one that is discharged using the heater approach. With the heater, the cells should have a lower temperature gradient than driving and the results might be different.

Reg, I don't see any evidence that the first 24 modules are significantly more degraded that the others, only that for some reason they all have a lower SOC than the other half of pack (they aren't charged as much!).

Weak cells will show up as having high voltage when the pack is charged and low voltage when the pack is discharged.

This pack simply looks out of balance.

Comparing to other cars will be interesting. It's way too early to draw conclusions from a single data point.

drees said:

Weak cells will show up as having high voltage when the pack is charged and low voltage when the pack is discharged.

Click to expand...

I guess I don't believe this since the pack is top balanced. They should all come up to the same SOC if the balancing is done perfectly (which it clearly isn't here). However, a degraded cell not only has lower capacity, but it also has a higher resistance. As a result, even if the top balancing were done perfectly, one discharge and charge cycle should result in the degraded cells being at a lower SOC at the end of the charge cycle.

Agreed that more data points will be interesting, particularly if we can get some new battery results.

Yes, the equalizing looks suspect.

Also, there is an unusual difference between the cells
on the left half and the right half of the graph.

To start off, no I did not follow the instructions properly, but when I found myself at VLB last night, and I finally had the ELM327 working, I decided to snap a photo or two. I hope they are of use.

On this day, I used a charge timer set for 100% and then commuted to work. Commuted home and charged for 10 min while I got ready for a movie, then drove to/from the movie. Plugged in and let my timer charge overnight again (it finishes about 1 hour before I leave). Typically, I get home with LBW about 25% of the time in winter and rarely in summer. I use a 100% timer every day of the week, with midday charging about 4 times a week, including weekends. I drive about 60-65 on the highway often, but only sometimes get into spirited driving. I have 40,000 miles in Seattle since I purchased my 2011 Leaf in late June 2011. My car is one of the “pollen cars”, produced on March 8th and languished at the Long Beach port for three months before delivery to me.

It seems to me that my pack could use more opportunity for cell balancing, so this morning I have disabled my charging timers. I’ll try to take some readings again at the end of the week and see how it goes.

VLB after driving:




Full after 100% timer charging:

FairwoodRed said:

To start off, no I did not follow the instructions properly, but when I found myself at VLB last night, and I finally had the ELM327 working, I decided to snap a photo or two. I hope they are of use.

Click to expand...

Thanks! Yes, very useful!

I will note that there is not a big difference between the rear stack (first 48 cells) and the front stacks like we saw with the other data.

BTW, I recommend using L1 to try to achieve the best balance possible in the LEAF, at least for the last couple of bars of charge.

Here's my pic - started out at 27 gid parked in the garage, then ran it down to 24 gid (VLBW) with the heater/defrost on max. Will get 80% charge tomorrow and if I have time, 100% charge, too.

Drive to get to this point - started at 80% charge, mix of freeway/surface streets, LBW around 48 miles, pulled into garage at 55.4 miles with 27 gids at 4.5 mi/kWh. Interestingly, my Ah reading is basically the same as thimels earlier - perhaps not surprisingly given our near identical age of vehicle and driving habits despite being situated on opposite ends of California.

Looks like modules in the mid-20s like to read significantly lower than the rest of the pack when low like other posts - the animation shows how quickly max-min climbs - in 4 gits it went from 40-87mV. And cell-pairs 49-96 are significantly higher on average than cell-pairs 1-48 like others.

Other notes of interest: VLBW went on at 13.30% SOC compared to thimel's 10.72%. Despite my battery temp readings of 74-78F, I only showed 5 temp bars on the dash. Wow - it was pretty hot in Sunnyvale on 4/29! Pack temps a few degrees over ambient correlate well with my pack temp today and today's max around 75F.

It may be a bit early to draw conclusions, but looking at drees' neat pic as well as thimel's. It sure looks like the middle of the back stack (modules 1-24, cells 1-48) is where the weakest cells are. Not suprisingly, these are the modules that are getting the least exposure to the outside world and are probably running pretty hot compared to the rest of the pack.

To add annother confusing data point, I've noticed when doing heavy quick charging, the first temp sensor is often the highest by a few degrees C. As this is located in the front of the pack, it contradicts the above paragraph. :roll:

JeremyW said:

To add annother confusing data point, I've noticed when doing heavy quick charging, the first temp sensor is often the highest by a few degrees C. As this is located in the front of the pack, it contradicts the above paragraph. :roll:

Click to expand...

I have to wonder if the BMS is reducing voltage on areas of the pack that tend to be hotter on purpose...

It seems pretty consistent that when reading the 4 temp sensors the first two are higher than the latter two by a couple degrees.

Anyway, here's what my pack looks like this morning after charging to 80%. Just eyeballing it, relative balance appears to be similar to the pics from VLBW. Not going to charge to 100% since it's supposed to be a scorcher today will be interesting to see how hot the pack gets. The pack didn't seem to cool off much from last night. Garage temps a steady ~70F with an overnight low of 59F.