Vectrix 102 cell NiMH Battery Analysis

docnjoj said:
Looks pretty good to me also! I wish my 10 ah cells were in that good shape! Thats only 5% difference at 20 amps! those are really good quality cells!
otherDoc
Most of them, yes.

But have another look at cell 103-48; it has severe problems!

Once I have capacity tested all cells I'll run more tests on the low capacity cells.
 
At least U are narrowing down what could be only 1-2 bad cells! Great testing! I may have yo do something similar for my 42 volt 35 cell pack! It really sags alot!
Wont pull over 800 watts and individual cells are 10 AH nominal!
otherDoc
 
Here are the graphic results of Block 13-7.

This is the overlay of the capacity testing results for the 8 cells in Block 13-7:


Not just one cell, but four cells have markedly reduced capacity in this block!


If you want to see any more detail click on the individual thumbnails below:








 
Mr Mik! I'm afraid U have motivated me to do something rash! I have ordered a 10 pack of 10ah Nimh from (gulp!) Batteryspace and will integrate this in my present NiMh pack. If all goes well I will end up with a 48 volt 10ah pack with five of the worst cells replaced. If I find more............? Then it may not have been such a great idea. My reasoning is that the present market for LiFepo4 is wild and wooly, and it seems like the batteries, even Pings are going up in leaps and bounds! If my Ni cells can last another year, there may be some market stabilization and even lower pricing. Certainly there will be better quality control! Now, if I just get that data logger...............! BAM! There goes the budget! :oops:
otherDoc
 
Yes, the risks with LiPO batteries are much higher than with NiMH.

Although I have put a marble plate on the workbench and removed curtains and other easily flammable materials from around the battery testing area, I 'd still be worried much more if I was trying do muck around with LiPO's of any description at this stage.
Seems to me one needs advanced electronics skills to safely experiment with Lithium battery technology; NiMH appears much more forgiving.

I also believe that maybe people are moving to LiPO to quickly - there might be a lot of so far untapped potential in NiMH.

The long lived RAV EV packs, NiMH in Prius and Vectrix etc tells me there are too many problems with Lithium battery packs at this stage.

Once I have cut my teeth on NiMH for a while I might of course move on. But by then I will not be a beginner any more!

It might also be possible to "repair" NiMH cells by reconditioning and maybe even matching parallel additions to raise capacity and lower resistance of weakened cells.

That is the next question I'll need to tackle once I have taken this "snapshot" of what I have: How do I fix it ??????

With LiPO it would be very stupid for me to jerry-rig things together; but I might get away with it with NiMH!
 
Here is the next lot of capacity testing screen shots.
Block 13-8; it appears to be well balanced without bad cells.

The overlay of the 8 cells:


And the individual cells:








 
This next Block, 13-9, has a low capacity cell in it:

The overlay shows cell 103-75 way below the others:


The individual graphs for cells 103-68 to 103-75:







 
The screenshots from Block 13-10 (Cells 103-76 to 103-84) show big capacity differences between the 9 cells.

The overlay:



The individual cells;








 
Here are the screenshots of Block 13-11 (Cells 103-85 to 103-93).

Cell 93 has a reduced capacity compare to the rest.


The overlay of all 9 cells:




The individual cells:









 
Here are the results of the capacity testing for Block 13-12 (Cells 103-94 to 103-102).

Not particularly well balanced, but no extreme outliers, either.

The overlay of the 9 cells:




The individual cells:













That leaves the blocks 13-1, 13-2 and 13-3 to be reported; I initially measured them with too much resistance due to cables and connectors and the cutoff of 1.1V was reached long before the cells were empty.
 
I capacity tested the first two blocks of cells with cables and connectors that introduced too much additional resistance. The first few cells of the third block were affected by this, too.

Continuation of capacity testing at 20A after fitting thicker cables and after completing capacity testing of the other cells, produced results for the capacities of cells 103-1 to 103-27 which appear to be very comparable to the rest of the cells.

So, 'll treat the results as if they were obtained by using the same cables used for cells 103-28 to 103-102.

But, and that's the reason for the long introduction, the graphical depiction of the tested capacities is more complex than for the straight foreward capacity tests done after I fitted the thicker cables.

The graphs look like this:





So instead of producing more confusing graphs (imagine the overlay for these double graphs), here is the Vectux battery capacity summary at 20A as text for the Blocks 103-1 to 103-3:

Pack 13-1:

1 20.25Ah
2 22.47Ah
3 23.58Ah
4 23.91Ah
5 23.65Ah
6 23.35Ah
7 23.36Ah
8 22.40Ah
9 16.40Ah


Pack 13-2:

10 19.92Ah
11 21.21Ah
12 23.30Ah
13 22.95Ah
14 21.58Ah
15 23.18Ah
16 23.19Ah
17 23.04Ah
18 24.33Ah

Pack 13-3:

19 20.75Ah
20 18.50Ah
21 18.15Ah
22 20.66Ah
23 25.85Ah
24 16.49Ah
25 25.71Ah
26 22.47Ah
27 17.79Ah

Pack 13-3 is the most unbalanced of them all. It contains the cells with the highest capacity and several with very low capacity.
 
This overlay shows the lowest capacity cell (at 20A) and the second highest capacity cell in the string.
(The highest capacity cell is very similar but I have no graph of a continuous discharge for it).

Cells103-48and103-25-1.jpg
 
Finally some very encouraging results to keep me going with this mammoth task.....

I have reconditioned Cell 103-001 using the CBA2 and the capacity at 20A has increased massively!

The three graphs below show the reconditioned Cell 103-001 (black, on top), then the second best cell in the series without reconditioning (red, middle), and a comparison cell (green, bottom) which is very similar to the initial performance of Cell 103-001. (I have no continuous discharge graph for Cell 103-001 or the "best" cell in the series.)

Note the much flatter curve with higher voltage throughout with a very sharp drop at the end.

Vectux103-001afterreconditioning-1.jpg


Of course this is only a step of many, but at least I know something can be done about the low capacity of some cells.


Problems remaining are:

1) The reconditioning algorithm I used takes a lot of time and needs repeated operator input.
It would take 3 months to get all cells done at the rate I am going now! By then the first ones will have self discharged and they are unbalanced again.....

2) Not all cells might have the same problem and just one cell that does not respond to the reconditioning will severely limit the packs performance.

3) The effect of the reconditioning might be short lived. It might take 100 times as long to recondition the cells compared to the duration of the improvement.

4) The conditioning might be dependent on a slow, individual cell recharge at the end of the process. In that case it will take 102 days to recondition all cells with a single charger! I am in the process of conditioning the other cells in Block 13-01 and then I will trickle charge the whole block in series and test if the cells improve with that approach, too.

Not a practical solution to the low battery blues yet, but I think I am on the right track!
 
Some very "mixed" sort of news: I do not know if this is good or bad, but probably bad for the life expectancy of the Vectux pack....

I can now identify the bad cells much quicker - that's the good news.

The bad news is that I can identify them because they have bulged, probably due to excessive internal pressure.

It is not very obvious, but once you know about it then it can be felt (more than seen) on the narrow sides of the cells. There is a very high correlation between cells which tested initially as having poor capacity at 20A, and the bulging. A few cells have a little bit of bulging although they do not have a very low capacity. I have not checked it obsessively enough to be sure, but I feel very certain about it. The correlation is close to 100%.
I managed to run my fingers along the sides of the 9- and 8-packs of cells and could identify the cells that tested low for capacity at 20A! I can literally find them with my eyes closed.
And find (a very few) more that the CBA2 missed....

Part of the good news is that a Vectrix owner with just a few cells gone bad can rapidly (in a few hours) identify them, rather than spending weeks testing day and night, and then replace them.

That only helps if you can get replacement cells, of course.

I have also deciphered the battery labels on the side of each cell:

C3008061221000125 means, for example: Capacity 30Ah Date of manufacture: 2006-12-21, number made that day: 125

A superficial look for a correlation between manufacturing date and capacity/bulging showed that there is no strong correlation, if any. In fact, some of the best cells were made on the same day as some of the worst. But I really need to look at that issue more closely to be certain.

There is also an additional sticker on the most positive cell in each pack of 9 or 8 cells, which probably specifies the date they were assembled and/or tested as a pack. It is usually located on the lower half of the cell and starts with a different letter, "M".
Example: M3008 YYMMDD 000xxx

And some more bad news: The "bad" cells do not respond as well to the reconditioning as the fairly good cell 103-001 shown above. There is some improvement, but the capacity gets nowhere near back to full level.

Very, very annoying!!!
 
I have found no significant weight reduction between low capacity and high capacity cells.

In fact, some of the low capacity cells are a little heavier than some of the high capacity cells, but not significantly, either.

I am not sure how many grams of oxygen and hydrogen could theoretically be produced and vented.

Similarly, internal resistance is a poor predictor of capacity (at 20A to 1.1V cutoff) for these cells.

The open cell voltage does not help, either, as expected.

The only tests I could find to distinguish low capacity from high capacity cells are visual + tactile inspection for swelling, and full capacity testing.

The inspection as well as the capacity testing require dismantling of the battery packs.

Inspection of the cells is faster by several orders of magnitude compared to capacity testing, but capacity testing is the first step to reconditioning of the low capacity cells.
 
The attached PDF file shows the original configuration of cells in the Vectux.

I have added the serial numbers to analyze if there is a relationship between cell capacity, swelling and serial number / manufacturing date.
I could not see any pattern there, it seems that the uneven distribution of low-capacity cells is unrelated to the production date of the cells.

The low capacity cells and swollen cells were almost exclusively located in the bottom layer of three.

I believe it is radiant heat exposure from hot road surfaces that caused this, but have not tested it yet.

I plan to locate a temperature sensor below the cells (the original sensors all sit on top of the cells and miss heating up from underneath until it is too late).

If the temperature in the bottom of the battery compartment is indeed too hot on future testing, then I will paint the underside with heat reflective paint to reduce the problem.

I have marked the cells with <19Ah capacity on initial testing at 20A to 1.1V cutoff, and those with between 19Ah and 20Ah, as well as those with swelling. All those with <19Ah capacity have also swollen.

The Serial numbers have this structure: C3008 YYMMDD 00xxxx , with YYMMDD being the date of manufacture.
There is also a date of assembly for the Modules of 8 or 9 cells: M3008 YYMMDD 00xxxx .

Can anyone come up with an alternative explanation for the uneven distribution of low capacity + swollen cells?
 

Attachments

  • Vectux Cell Serial Numbers (Original configuration).pdf
    88.8 KB · Views: 775
This post was originally posted at: http://www.endless-sphere.com/forums/viewtopic.php?f=14&t=7512&st=0&sk=t&sd=a&start=15#p116416

I partially duplicate it here because of it's relevance to the subject.

;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
After many trials and errors I found a way to measure internal resistance of the cells with great repeatability, and hopefully accuracy.
The thumbnails below show all that is needed (click to enlarge):




A CBA2 with upgraded cables, a DMM with min/max hold function, and a timer.

It's basically a 4-cable measurement setup for the CBA2. I believe it would work just as well with the original cables, but have not tested it.

The most repeatable and most plausible results for IR were obtained with this procedure:

Connect CBA2 and DMM, set DMM to minimum voltage recording.

Set CBA2 to 11A, hit start, hit timer to start countdown from 35s.

When timer sounds the alarm, stop the CBA2, save the file, reset to 1A, record minimum DMM voltage during first part of test, reset the DMM minimum voltage memory, then hit start on the CBA2 again and start the timer again. All as fast as possible.
Stop after another 30s (10s to 15s would probably be enough for the 1A testing, but not for the 11A part.)

Then note the second minimum voltage value from the DMM.

Subtract the two voltages from each other and divide by ten, there is your resistance result, with a resolution of 0.1mΩ .

Ignore whatever voltage the CBA2 displays, it is not important at all.

I did a large number of tests on many of the 102 cells in that way, and got repeatable results (when SOC, time since last charge, cell temp etc. were similar).

I did not find the results very useful, though.

I did not test all the cells because the results were almost identical between the worst and the best cells, and the testing is time-consuming.

At full SOC the IR is about 1.2mΩ.


The only difference between higher capacity cells and damaged cells (in regards to IR) seems to be at the 20A / 1.1V cut-off level:

The damaged cells showed higher resistance than the better cells at that (practically empty) SOC, i.e. about 4mΩ to 6mΩ . The better cells were about 3mΩ at the 1.1V cut-off level. I am not sure yet if this is cause or result of the reduced capacity of these cells, or both. I have not tested IR at lower SOC than 1.1V cutoff at 20A.

Otherwise the IR values were identical for all tested cells (at 10Ah SOC, 20Ah SOC and fully charged), always about 1.2mΩ to 1.3mΩ.
 
Mr. Mik said:
2) The battery recall was indeed necessary - it appears to me that the bolts holding the "inter-level-connectors" have been torqued to 5Nm (the lowest one was only 3Nm) instead of 10Nm. These bolts are the same size as all the others, but have different letters and numbers on the top.
My guess is that the batteries were assembled into packs of 8 and 9 cells at the battery factory, using correct torque settings; and then the assembly into complete packs of 102 cells was buggered up somewhere else by using 5Nm instead of 10Nm to tighten the bolts holding the battery temp sensors and the connectors between the three layers of cells.
None of the connectors in the rear battery had actually rattled loose - guess that says something about the tolerance level built into that part of the scooter.

I do now think that this was only partially correct:

The inter-module-connector-bolts (IMCB's) were most likely tightened to 10Nm by Vectrix, but only once.

But these bolts do become loose again and need to be re-tightened several times before they hold. I learned that during the slow and careful reassembly of the rear battery pack of the Vectux, and then confirmed it by "reworking" the previously reworked front battery pack (because the fuses for the M-BMS needed to be replaced): The inter-module connectors could be undone with about 5Nm torque, slightly better than when I dismantled it for the first time (in original, pre-rework state). And I know that I had religiously tightened them to 10Nm......

The IMC's get squished flat by the bolt because their end plates are soldered to the braided cable. After a few re-tightenings they do hold, at least when the batteries are just being cycled on the workbench. I believe it will be fine in a running scooter, too, but will check the top cell level after a while once the Vectux is running again.
 
Hi Mr. Mik,

I think that you will find that the individual cell 'failures' that you are finding are likely NOT to be the result of a manufacturing defect. The most likely explanation is that they have been damaged by reverse charging when the battery voltage has been allowed to reach too low a level on an otherwise normal discharge cycle.

Each cell will have a 'nominal capacity of (in the case of the Vectrix) 30Ampere hr. There will always be some measurable variation in a mass produced product. If a battery of cells (series connected) starts fully charged and is progressively discharged, then the cells with a 'lower' capacity will empty first and their voltage will drop first. This is a simple and locical effect.

If the discharge is not stopped at the point when the first (lower capacity) cells are 'empty' then those lower capacity cells can be 'reverse charged' by the current passing through them from the rest of the battery. This reverse charging will almost certainly result in cell damage of the type you have described. i.e. Cell bulging through excess internal pressure and a serious lowering of capacity due to the mechanical and Chemical damage done by the reverse charge.

There is no 'cure' for this problem. The best solution is to match the cells during the initial battery assembly. Then to monitor the cell voltages on discharge and STOPPING the discharge at the point when any cell reaches the predetermined minimum allowable voltage.

The whole object of the 'Equalising Charge' is to ensure that all cells are 'full' to capacity prior to discharge. Unfortunately an 'Equalising Charge' entails overcharging the 'Better' cells and in turn, as we know, will stress those better cells and will reduce there life expectancy.

Prius and others have got it right. Don't discharge the battery to the point of allowing any cell voltage to reduce below 1.0V
this would probably mean leaving the last 10% of the charge unused. Don't charge the battery past 80% to 90% of it's capacity, this would eliminate the problems of overheating which are associated with the final stages of a 'full' charge.

I don't know whether any of this helps or hinders your cause but I offer you my best wishes in your endeavors to sort out your Vectrix.
Keep up the good work.
Sandy
 
Sandy said:
Hi Mr. Mik,

I think that you will find that the individual cell 'failures' that you are finding are likely NOT to be the result of a manufacturing defect. The most likely explanation is that they have been damaged by reverse charging when the battery voltage has been allowed to reach too low a level on an otherwise normal discharge cycle.

Each cell will have a 'nominal capacity of (in the case of the Vectrix) 30Ampere hr. There will always be some measurable variation in a mass produced product. If a battery of cells (series connected) starts fully charged and is progressively discharged, then the cells with a 'lower' capacity will empty first and their voltage will drop first. This is a simple and locical effect.

I think you are right with most of what you said.

In addition: Once there is reduced capacity in some cells it will be those cells that get over-charged in an equalization charge.


The keys to the Vectrix battery problems are these:

A) The cell voltages are poorly monitored: only three sub-strings are monitored, 27 cells, 48 cells, 27 cells.

B) The cells are exposed to very different temperatures depending on their location within the pack. This causes different self-discharge rates, followed by unbalanced cells, followed by reverse charging of some cells.

C) The cells that are over heated (and maybe also over cooled) are distributed evenly in all three monitored sub-strings, causing the erroneous measurement results of a balanced pack, when in fact the bottom layer of cells (34 cells) are much deeper discharged than the rest.

D) The advice given by Vectrix to deeply discharge the batteries five times, without clear instructions on how to do it and when to avoid it. These deep discharges are poison for the pack whenever it is very hot or very cold due to the above three points.


The solutions:

A) Monitor at least each Module of cells (8 and 9 cell modules).

B) Thermal insulation to keep all cells at the same temperature.
 
By measuring the temperature at multiple points in the battery and around it, whilst parking and/or charging it in full sunlight, I found that my theory about the mechanism damaging the bottom layer cells was INCORRECT.

But the overall principle, and the suggested solutions, remain valid.

It turns out that the heat transfer from the top of the battery downwards dwarfs the heat transfer upwards from the hot road and the aluminium bottom of the battery.
This leads to increased self-discharge of the two top layers of cells, and in turn the best cells in the bottom layer get over-charged during the next full charge!
This damages the best cells n the bottom layer first, because they had the lowest self-discharge rate and could therefore not absorb as much charge (before overheating and venting) as the poorer cells could.

The cure: Heat insulation mostly focusing on the top of the battery!
 
TylerDurden said:
Forced air cooling?

Forced air cooling is already integrated into the stock Vectrix. It has two 12V 2.5A impellers sucking air through the batteries during charging and when the batteries are very warm during riding. The impellers can also run after riding or after charging if the battery temperature is still high.
See: http://www.endless-sphere.com/forums/viewtopic.php?p=133439#p133439

I have attached an external 12V 5A power supply to the impellers so that I can run them at will in addition to the "spontaneous" impeller operation by the stock electronics.

But the problem is that the top layers, mainly the upper layers of the front battery, heat up when the scooter is parked in the sun, without impellers running.
The black plastic cover gets very hot and transmits the heat downwards into the battery.

If the top layer of cells is 40dC warm, and the bottom layer 25dC, then the pack becomes imbalanced due to the massively increased self-discharge rate of heated NiMH cells. The fairly unsophisticated stock BMS cannot see this imbalance, because each monitored string of cells has 1/3 top, 1/3 middle and 1/3 bottom cells in it! The stock BMS assumes therefore that all cells have the same SOC, but because half of them are significantly discharged, the more charged cells get overcharged, bulge and possibly vent through the safety valve.

That's my theory, anyways!
 
They make solar powered fans for car interiors: when the car is parked in the sun, the hot air gets vented.

Perhaps a similar setup could help your battery enclosure.
 

Attachments

  • solar-car-fan.jpg
    solar-car-fan.jpg
    20.8 KB · Views: 4,151
Hello!
I have two batteries from two Vectrix.
BMS one of them has heat sensors and wires monitors battery voltage.
BMS from second scooter has only thermal sensors.
Does anyone know why?
 

Attachments

  • BMS-1.JPG
    BMS-1.JPG
    44 KB · Views: 3,845
Back
Top