barnard's Toyota MR2 - now Mr240

[barney]

Hi Barney,

What is your peak charge voltage?
That is, what is the highest voltage your charger gets to before switching off?

Hi Rod. I can answer this one...

I have noticed 2 different figures: 165ishV (in feb12) and 161V (last week, aug12).

Back in Feb I was noticed the floating charge process seemed out-of-whack (cant recall what I actually noticed, now, but I had been away for 4 weeks in January and someone else was keeping it charged up). Anyway, this particular night I did watch the main meter in the car every 30 mins or so, and at the time added notes on my 'smartphone' (aside: those phone makers have a way with words, eh?).

That evening (22feb12) charging had started around 6:15ishpm (I know, I know, but I wanted to watch it without staying up til all hours).
At 9:25pm, the pack voltage was climbing slowly towards 164.7V, and the ZIVAN charger was pushing 19.4A in. The apparent capacity was 96.9%.
By 9:35pm the Amps in had slowly dropped from 7A to 3A and by 9:35pm was 0.7A.
By 10:20pm the gauge reported 164.6V and 100% capacity. But not until 10:47pm did the Zivan beep, beep beeeeep its 'finished ok' tones and turn off.

Presumably this was some top-up charging process going on; normally the topping out is much quicker.
A few months later whilst charging I noticed it was hitting 161V and stopping. The long drawn out step from 96% to 100% seems to have gone. So 17may12 it stopped at 160V, capacity 99.6% at 10.30pm (cant recall when charging started, probably 8 or 9pm). Last week it was similar, voltage climbing to 161V and Zivan beeping finished without the long slow close of February.

By the morning the pack seems to have settles back to 150 or 151V. Does that tell us something?

hth, Barney

[roddilkes]

Maybe something amiss there I don't know for sure.

Can you use the cell loggers to measure the cells as they are approaching end of charge?
It would be useful to have this info just to check if the ones that are weak are getting fully charged.

[drgrieve]

Hi Barney,

What is your peak charge voltage?
That is, what is the highest voltage your charger gets to before switching off?

Hi Rod. I can answer this one...

I have noticed 2 different figures: 165ishV (in feb12) and 161V (last week, aug12).

Back in Feb I was noticed the floating charge process seemed out-of-whack (cant recall what I actually noticed, now, but I had been away for 4 weeks in January and someone else was keeping it charged up). Anyway, this particular night I did watch the main meter in the car every 30 mins or so, and at the time added notes on my 'smartphone' (aside: those phone makers have a way with words, eh?).

That evening (22feb12) charging had started around 6:15ishpm (I know, I know, but I wanted to watch it without staying up til all hours).
At 9:25pm, the pack voltage was climbing slowly towards 164.7V, and the ZIVAN charger was pushing 19.4A in. The apparent capacity was 96.9%.
By 9:35pm the Amps in had slowly dropped from 7A to 3A and by 9:35pm was 0.7A.
By 10:20pm the gauge reported 164.6V and 100% capacity. But not until 10:47pm did the Zivan beep, beep beeeeep its 'finished ok' tones and turn off.

Presumably this was some top-up charging process going on; normally the topping out is much quicker.
A few months later whilst charging I noticed it was hitting 161V and stopping. The long drawn out step from 96% to 100% seems to have gone. So 17may12 it stopped at 160V, capacity 99.6% at 10.30pm (cant recall when charging started, probably 8 or 9pm). Last week it was similar, voltage climbing to 161V and Zivan beeping finished without the long slow close of February.

By the morning the pack seems to have settles back to 150 or 151V. Does that tell us something?

hth, Barney

The good news is that 151v static after charge is bang on target 3.33 to 3.35 average.

This confuses me as the way you are charging is overcharging the pack.

164.6V is 3.65v per cell. Holding that voltage and terminating until current is 0 - is not good. The charge should be terminated at C /20 - which is 5 amps for you.

I would prefer you charging to 3.5v and terminating at 5 amps and on the days you need a full charge, then charge to 3.65v but still terminating at 5 amps.

LiFeP04 cells do not need a floating or topping charge like lead acid. It is damaging to them.

I hope your charger is adjustable?

[weber]

zeva,

The point is that a few cells are sagging below 1.8 V while the others are still above 2.5 V.

antiscab,

These BMUs bypass 600 mA at 3.65 V so I don't understand why you'd be concerned with a charger putting 500 mA in for another hour to ensure balancing, as jonescq suggested. Sure 3.5 V at 0.001C is full, but are you saying significant damage occurs by taking them to 3.65 V at 0.0001C for an hour?

drgrieve,

barney's system is designed for top balancing. He's trying to figure out what, if anything is wrong with his existing system. I think you're just confusing the issue with your bottom-balancing-derived statements.

Overall battery-voltage tells you almost nothing about individual cell voltages. But it's clear that 161 V is not enough to ensure top balancing of 45 LiFePO4 cells. In an extreme case you could have 35 cells fully charged at 3.65 V and the other 10 nearly flat at 3.325 V.

[zeva]

The point is that a few cells are sagging below 1.8 V while the others are still above 2.5 V.

Apologies, it is easy to get off topic and ruin a thread's signal-to-noise ratio But it is interesting to hear all the different opinions.

Although after a couple of years it's probably a good idea to do a manual balance, I don't think imbalance could explain the voltage data presented. Imbalance would only cause this kind of difference in voltage drop as cells are approaching flat, and the test data was all taken with significant SoC left (and over a significant range of SoC).

My money is on oxidised terminals (also for the sake of Barney's money). Easy to test: clean the terminals on any one of the suspect cells (e.g 4, 5, 10, 17, 25, 35, 45) with wire brush or fine sandpaper, preferably add Noalox, and see how much of a difference it makes.

If things aren't looking up after a balance and a clean, unfortunately you might have a handful of cells to replace. (High voltage sag is indicative of high internal resistance, often means a cell is sick and may have reduced capacity too.)

To wander slightly off topic again, I don't understand why people say float charging is bad for lithiums. Perhaps if you're using an incorrect charger designed for flooded PbAs it'll keep trying to push current in and end up taking cells over-voltage. But with a CCCV lithium charger, in time the cell voltage matches the charge voltage exactly, and zero current flows. It's functionally no different from wiring two cells in parallel, which we do all the time without ill effect. I would be interested to hear the justification behind float charging being bad.

[drgrieve]

I would be interested to hear the justification behind float charging being bad.

Once a cell is full and you try to push more charge into it, you start to plate. This causes loss of capacity.

3.4v static voltage is full. Charging at a higher voltage than 3.4V will cause overcharging - BUT - when charging several factors cause the charging voltage to rise above the true underlying cell voltage. Current is a main contributor. This is why you must terminate the charge at a reasonable amperage. C/20 is a widely used termination formula for current. If you can't set you charger to stop at C/20 then charge at a lower target voltage. CV @ 3.45v terminated at 0 amps should be OK.

Charging voltage is not an exact science, individual cells will react differently when charging, just because one cell is reporting a high voltage does not mean it is at a higher SOC than another cell. (I do realise that SOC is a major contributing factor to the "apparent voltage" but is it not the only factor)

Therefore it is not possible to use a BMS* to truly top balance your pack. If you want to top balance your pack then you must use static voltage as an indicator of SOC.

These would be steps to top balance your pack.

Charge the pack normally. Test each cell after 1 or 2 hours rest. Choose a static voltage where most of your cells are at. Say 3.34v.

Manually charge or discharge the other cells not at this voltage. You need to let them rest an hour or two each time.

Depending on how fast you get the knack for this it might take you a day or two. When all cells read the exact same static voltage then your pack is top balanced (and ragged bottom of course).

* I don't want to start a flame war on this topic. In the near future I hoping to finally start my build and as I live in Sydney would like to be part of a sharing community. So please don't take this the wrong way - I'm just trying to provoke some discussion.

Here is some interesting recent threads to read on going BMS free:
http://www.diyelectriccar.com/forums/sh ... 77497.html
http://www.diyelectriccar.com/forums/sh ... 63154.html
http://www.diyelectriccar.com/forums/sh ... 77331.html

[jonescg]

I think a battery monitoring system which trims any high cell down to an acceptable level during and just after charging is a good idea. This is essentially 'balancing' but not for very long. It just stops any one cell from resting at an unacceptably high voltage. Coupled with regular tune-up balancing efforts (that is, individual sub-pack balance charging, like an RC balancer) is a good combination.

I think Barney has a few dud cells which may or may not be able to be balanced back to normal. If not, replacing them is always an option.

[T1 Terry]

I think the cells that drop voltage under load are simply not being fully recharged or able to hold full charge after each charge cycle, I assume that is what most other are suggesting. Here is an example of exactly this problem with 3 week old pack under a 1C load

as you can see, the lowest cell under load always returned to the same voltage as the others when the load was removed. Once that cell was fully recharged the problem went away. These cells are in pack that has cycled 24/7 for over 12 mths now, they power my house and solar recharge.
Do you have a mini charger like one of these http://www.hobbyking.com/hobbyking/stor ... oduct=6478 They are great for topping up a wayward cell and can be adjusted back to 0.1 amp or up to 5 amps for single cell charging. I don't use a BMS system as I have had issues with friends systems that not all the BMS boards discharge to the same voltage before switching off resulting in a similar issue to the one you are describing. After the charge cycle finishes try charging the low cells with the mini charger at rate low enough for the BMS board to handle on discharge and watch the voltage the discharge starts and when it stops, if it stops below 3.45v this may be what is causing the cell to go out of balance to the rest of the pack. The boards are easy enough to change and you can give the terminals a polish up with aluminium oxide sandpaper and a paint with Alminox before reassembling.
Just suggestions, hope this helps.

T1 Terry

[weber]

So we have a number of hypotheses:
1. High R connections e.g. loose or oxidised terminals or links.
2. High-R cells (temporary) due to being out of balance.
3. High-R cells (permanent) due some kind of abuse such as over-temperature, possibly combined with a lack of cell compression.

So come on barney, test some of these hypotheses. You're keeping us all in suspense.

Are your cells compressed in any way? i.e. prevented from bulging their cases when hot?

Did you choose to have the optional temperature sensing with your EV Power BMS? If so, how often do you have a thermistor between cells? Every second cell?

By the way, the EV Power is an extremely reliable BMS and any charging problem is far more likely to lie with the choice of charger (e.g. one that is designed for Lead-acid, not Lithium-ion) or how the charger is set up, or the manner in which the charger is controlled by the BMS.

Please do what both Rod and I have suggested and datalog some cells while doing a normal charge. Up to and including the end of charge. Of course you should ensure that you include some good and some "bad" cells in the set being logged.

Here is an example of exactly this problem with 3 week old pack under a 1C load ... These cells are in pack that has cycled 24/7 for over 12 mths now, they power my house and solar recharge.
...
I don't use a BMS system as I have had issues with friends systems that not all the BMS boards discharge to the same voltage before switching off resulting in a similar issue to the one you are describing.

I've heard this kind of argument from several people as to why they don't use a BMS. But is it logical? The general form seems to be:

"A friend had a BMS that was so badly designed that <insert bad thing> happened, therefore I will use one that's so badly designed it's non-existent. That way I'll guarantee that a bad thing will eventually happen if I ever tire of being a BMS myself."

Well OK, they don't say the last bit.

[barney]

Ok, ok! No pressure then!

I'll skip my 'church' (well, uplifting music while we read papers on Sunday morning is the best I can do these days), ignore the growing grass and weedeater which wont start, the printer my wife wants to work but which grumps after installing a new server, so I can go take a photo of the battery box, and reply to y'all. I thought retirement was gonna make things easier... am I a bit off-topic here? Sorry! )

Back soon, Barney

[weber]

Once a cell is full and you try to push more charge into it, you start to plate. This causes loss of capacity.

Yes a LiFePO4 cell with a rested voltage of 3.4 V is fully charged, but that doesn't mean that you will cause significant lithium plating if you raise it to 3.6 V (or even 3.65 V) and hold it there for an hour.

How can I know this? Because of Farady's law of electroplating. The number of Li+ ions plated out as neutral Li atoms cannot be more than the number of electrons that pass through the cell. I calculate 0.26 grams of Li per amp-hour but it's probably more meaningful to think of it as a proportion of the lithium ions that are involved in a full charge or discharge.

And it will be less than the Farady's-law amount at very low charge currents because electrons are either involved in other reversible reactions or are involved in self-discharge due to the slight electron-conductivity of the electrolyte.

When you have a fully charged cell resting at 3.4 V and take it up to 3.6 V, current only flows for a very short time before falling again to next-to-no-amps. And you can hold it at 3.6 V for a long time and the current stays at next-to-no-amps. Integrate next-to-no-amps for 1 hour and you get next-to-no-amp-hours therefore next-to-no lithium plating.

It is only when you take it up past about 4.2 V that you get a significant continuous current again, and will be plating out lithium at a significant rate. I'm assuming normal temperatures here. Lithium plating occurs at much lower voltages when the cells are below 0°C. You must not charge when the cells are below 0°C.

You can see the manufacturer's specs here.
http://www.evworks.com.au/store/datashe ... ations.pdf

So one need not fear taking Sky Energy cells to 3.6 V to balance them. And there certainly won't be any balancing happening if you only ever take them to 3.4 V. But I agree the excursion to 3.6 V should be for the shortest time possible, just until all cells have reached 3.6 V and the current has fallen to bypass current plus maybe 0.05C.

Here is a great chart of failure modes of lithium-ion cells (of which LiFePO4 is one sub-type).



For the full description, see
http://www.mpoweruk.com/lithium_failures.htm

[zeva]

How can I know this? Because of Farady's law of electroplating. The number of Li+ ions plated out as neutral Li atoms cannot be more than the number of electrons that pass through the cell.

Phew, that's actually a relief to hear. I suspected this but couldn't back it up with science To add some empirical data, I have "float charged" LiFePO4s to 3.65V with a proper CCCV charger and can confirm that the charge current does drop to effectively zero - I recall my Fluke meter showing 0.03mA in one case where I'd accidentally left the charger connected for a couple of days. That works out at 0.25Ah per year (though I'm not sure I trust the accuracy of even a Fluke with sub-mA measurements; self-discharge alone should have been greater than 0.03mA, unless the 1%-per-month rule of thumb is also way off.)

[T1 Terry]

Here is an example of exactly this problem with 3 week old pack under a 1C load ... These cells are in pack that has cycled 24/7 for over 12 mths now, they power my house and solar recharge.
...
I don't use a BMS system as I have had issues with friends systems that not all the BMS boards discharge to the same voltage before switching off resulting in a similar issue to the one you are describing.

I've heard this kind of argument from several people as to why they don't use a BMS. But is it logical? The general form seems to be:

"A friend had a BMS that was so badly designed that <insert bad thing> happened, therefore I will use one that's so badly designed it's non-existent. That way I'll guarantee that a bad thing will eventually happen if I ever tire of being a BMS myself."

Well OK, they don't say the last bit.

As we only use 4 cells in series for the house power set ups a BMS is over the top but some form of cell monitoring and HCV LCV alarm is needed, we use the Junsi cell logger, does all we need. Active discharge balancing seems to be the area where problems develop, as we don't need it we don't use it.
When there are a lot more cells in series things become a lot different, still don't think I'll use active balancing but HCV and LCV will become even more important.

T1 Terry

[barney]

So we have a number of hypotheses:
1. High R connections e.g. loose or oxidised terminals or links.
2. High-R cells (temporary) due to being out of balance.
3. High-R cells (permanent) due some kind of abuse such as over-temperature, possibly combined with a lack of cell compression.

So come on barney, test some of these hypotheses. You're keeping us all in suspense.

Are your cells compressed in any way? i.e. prevented from bulging their cases when hot?

Did you choose to have the optional temperature sensing with your EV Power BMS? If so, how often do you have a thermistor between cells? Every second cell?

Please do what both Rod and I have suggested and datalog some cells while doing a normal charge. Up to and including the end of charge. Of course you should ensure that you include some good and some "bad" cells in the set being logged.

Research time: Weber, Rod, Zeva, jonesy and others want to see results of
a) reseating at least some battery connectors with sandpaper and Alminox/Noalox;
b) the logs during a charge cycle of a set of 8 batteries containing at least 1 'bad' battery and one good one;
c) individually charging up to some Li ceiling (3.7V on the charger?), at least some of the 'bad' batteries and re-insert into the system;
d) some way of measuring the state of indiv poor cells to see if they have suffered permanent damage.
e) what restraint/compression surrounds the batteries.
Also implied in the questions, and alluded to by T1Terry...
f) perhaps documenting the actual cutoff of individual BMUs on a few batteries to see if there is a correlation between lower cutoffs and the bad batteries. Swapping the batteries alone and leaving the BMUs in place would be an indirect way to test this, possibly? Im not the electrician here, so Im guessing.

Only (b), and (f) can happen without needing either Noalox, a cell charger, a reasonable multimeter, and/or removing batteries/terminals. And since there are at least 4 batteries below par in Pack A, under the car, it looks like Im taking ZEV240 off the road for a while! SO Im afraid testing will proceed slowly until I have all the tools I need.

Anyway, here's whats up so far: 1Took pics of Upper Battery box to illustrate battery restraining system: Attached; 2 Logged one set of 8 batteries during a 2hr charge cycle last night, from 70% to 100% (usu ~20%/hr, so completed in about 1.5hr or maybe 1h45: Graphs and table attached also BUT something went wrong.... The logger (set to log every 10 sec) was left running for 2h, but reports 15 min worth of data. Not sure what Ive done wrong here.

(And I looked for Noalox or Alminox paste/tube etc online. Hoping to find a tube a local auto parts store, as Jaycar online have nowt so far. Fremantle no longer has a car parts shop, so a bit of driving needed)

Battery restraining:
(part save, before trying to upload pics to site, which didnt work in th old system...)
Welded 6mm Al case, open top, tailored to each set of cells, with compression bar over and clear plastic cover bolted down on top.

[weber]

(And I looked for Noalox or Alminox paste/tube etc online. Hoping to find a tube a local auto parts store, as Jaycar online have nowt so far. Fremantle no longer has a car parts shop, so a bit of driving needed)

There are some other aluminium jointing compounds that don't get mentioned much (possibly because of their unmemorable names). These are Cabac EJC/300 (White) and Cabac EJCG125 (Grey). Your local Ideal Lighting and Electrical should have these. You can order them from their online shop if you set up an account, although this is probably not worth the trouble. http://www.idealelectrical.com

Cabac recommend the white goo for bolted aluminium and copper connections and the grey goo for crimped aluminium connections. You can read their blurb here.
http://www.cabac.com.au/shop/media/Pdf/ ... jc_300.pdf
But we've been using the grey goo for everything, including crimped copper to copper connections. Tritium show the grey goo being used for a crimped copper to tinned-copper connection in a photo on page 29 of their WaveSculptor 200 manual.
http://tritium.com.au/wp-content/upload ... Manual.pdf

Battery restraining:
(part save, before trying to upload pics to site, which didnt work in th old system...)
Welded 6mm Al case, open top, tailored to each set of cells, with compression bar over and clear plastic cover bolted down on top.

Are they wedged in tight in the narrow direction of the cells. i.e. if you turned the box upside down without its lid or downward-clamping bars, would there be enough friction to stop the cells falling out. This could be done e.g. by making the box to really precise dimensions, or a rubber pad at one end of each row and shoehorning the last cell in, or a piece of plywood or plastic sheet tapped in with a hammer at the end of each row after the cells are in.

[barney]

Ok, some pics, and an answer to weber's questions about battery restraint...
First, a view of the full pack, welded box with full-width restraining bar which holds them in as well as preventing movement in the top width (ie East-West as you look at it) dimension....



and showing that they are close to the sides, the 'West' end of the box (sorry about the dust - but it is an engine bay after all, and used to be MUCH worse)...



The big question was, were they hard to get it? Yes, and for the bottom box, there was just room to squeeze in a masonite sheet on the long edge, to provide extra firmness. Even better, would they fall out of held upside down? I dont think so, but it was a while ago. I'll ask Daniel at EV Shop and see if he remembers... I certainly remember his saying that squeezy was important, so I think they probably pass muster... The biggest question of all is whether weber likes it!

Next pics are the data from data logging during a charge cycle. I picked a set of 8 in the upper pack, pack B, so I could get to them easily. I chose batts 37-45 as they include both good ones (eg #38, #41, #44) and a very bad one, #45, the one nearest the leftmost power lead in the upper pack photo provided above.

The logging ran for c.2hrs, and the spreadsheet is long so its a screen grab in two parts. Which suits both the 150K upload limit, AND saving me hosting the pics on my site and you relying upon two servers to be running rather than one.

First the 99 datapoints (why did it stop then?).... There are some discontinuities in the data suggesting charging ran to the end (eg the drop off around 14 min) but I dont get it. Tell me if this is rubbish data and I'll just delete it and try collecting again. Is 99 points per file the CellLog recording limit? Here is hoping the graphs align next to one another...



And the data points themselves... A long file, in two parts, using Excel conditional formatting to colour the cells from Orange (lowest voltage c.3275, to Yellow (highest V, c3515). Note all the cells are 4-digit whole numbers representing 3dp voltages from 3.275V to 3.515V. Also to the right is a sum of all 8 cells giving a pack voltage for those 8 cells, ans finally a CALCULATED voltage as tho these 8 cells represented the whole pack - ie the projected voltage for 45 cells (which of course they do NOT, but it gives an _idea_ of the whole pack); to which Ive applied more conditional colouring as a bar chart, longer= higher voltage.



                               

Weber - Many thanks for the info about 'jointing compound', Cabac are supplying it tomorrow afternoon to an electrical wholesaler near me, so I can do some scraping and testing connections real soon... Wish I had a car lift in my garage, I could do more! Maybe I dont really, who would I justify THAT cost to?

And thanks everyone for your feedback; as zevaIan says, very interesting to hear the variety and use it to clarify/validate our own understandings.

Ok, now Im off to find out why the CellLog has a 99 point limit, and whether I need to separate the cells to do this individual charging they all talk about... Barney

Edit: more typos and trying to align the long spreadsheet...

[T1 Terry]

My guess with the logger is it still had old files in memory and it ran out f space. We use 30 sec logging and can record for up to a week. You need to do the control shift and control alt to spread out the graph using the mouse wheel to find the 24hr points to separate the days but the graphs give a good general idea of how things are going.
From your graph it would appear cell 45 has lost capacity or has a high resistance joint but it could be inside the cell to one or more of the pouches. We had a few cells that did strange run away charging like that but a constant voltage charge of 3.8v on those individual cells brought them back into line, I doubt the BMS card would appreciate that sort of sustained voltage though. We charged for an hr, let the cell stand for an hr, if the voltage dropped below 3.5v we charged again, some took 3 goes before they settled down. We do this with all the cells we use before we put them to work now, charge them till they will hold 3.5v rested, seems to save having to pull battery packs out and sort out the recalcitrant cells... always wanted to use that word some where

T1 Terry

[jonescg]

I learned the origins of the word recalcitrant: re - again, calcio - kick (Italian).

These cells keep on kicking? Kicking back more like it...

[weber]

Hi Barney,

I'm deliriously happy. Your cell clamping sounds fine. no need to bother Daniel.

There is a fourth hypothesis regarding the overall problem, as T1 Terry mentioned.
1. High R connections e.g. loose or oxidised terminals or links.
2. High-R cells (temporary) due to being out of balance.
3. High-R cells (permanent) due some kind of abuse.
4. Low-capacity cells (permanent) due some kind of abuse.

I suggest applying the scientific method by only changing one thing at a time. If you do all the suggested things at once and the problem goes away, you won't know which one caused the problem or what to look out for in future.

You're quite sensible to first do the thing that doesn't require you to disassemble anything -- the datalogging -- to investigate whether the charger is balancing the cells or not. And I'm still betting on that being the problem. If it is, you may never have to disassemble anything.

Unfortunately that charging data, while not exactly rubbish, doesn't cover the part of the charge we are most interested in -- the end. Although, as you say, those jumps in the curves imply gaps in the record. If we really are looking at the end of charge there, then something is very wrong since most of the cells should have got to 3.65 V before charging terminated.

What happens around 14 to 15 minutes is pretty interesting, if it is real. A couple of possible explanations as per T1 Terry.
1. A high resistance joint spontaneously reduced its resistance back to near normal.
2. A cell that was near zero SoC got enough charge into it that its internal resistance reduced to near normal.

I know which one I'd put my money on.

No need to do any great discharge before trying again. That would just risk doing more damage to the low cells. Discharging just 1% of capacity would probably be enough, and then log the charge again, hopefully after figuring out what went wrong with logging last time.

To be convinced that I was looking at the end of charge, I'd want to see cells ramping down from 3.65 V after the charger shut off.

[barney]

Hi Weber

Thanks for all this. Im getting to understand more about my batteries... should I have known more earlier? Well, isnt that always the way. But thanks for all the guidance out there, its slowly making more sense.

The hypotheses make sense, but the last two are worrisome. Not sure what causes a cell to suddenly increase resistance or lose capacity, when they are meant to be usable over thousands of re-charges; and they have never been seriously mistreated that I know of... like never reaching 12C even momentarily (the cell specs say "4C cont/ 12C pulse claimed by manufacturer"... http://www.evworks.com.au/index.php?pro ... T-SE100AHA). Still, a single cell isnt an EOW (end of the world) event, I guess...

I gather you are leaning towards 4, ie a single cell suddenly playing ball after maybe 14/15 min charging? None of them can be determined with crap data, so that's the next step. At this stage I'd like to be confident in the data Im collecting - and publishing! - before I worry about any consequences... I have the Junsi in my hand and still cannot see why it stops at 99 entries. Tempting to set it for 30s intervals like you have, t1 terry, and hope for 45 min.

I think there is a sleep problem. I did notice that when I went to check the logger, its screen had gone blank and so I pressed the MENU key and it came back on; the first time ir was showing 10m (about 1 hr into the process) and the second time 15m, about 2 hrs in. Im going to turn off 'sleep mode' which seems to be stall mode right now. Will that 'flatten' (or rather, affect the charging of) battery #1 (where the Junsi draws its tiny power needs... A small bit of soldering and it can be encouraged to draw power across all 8 cells, but it then wouldn't work on the 5 cell pack), or upset its charging cycle? Hope not. However, if its true about sleep mode, then the data at the end is indeed after turnoff.... but probably an hour later, so not enough info to decide whether the BMS was approaching the magical 3.65 V figure...

The car is at around 70%, so should be another 1.5hr to charge. I'll go charge and log this evening, but wont likely get some data up 'til tomorrow afternoon - got to do some day job stuff in the morning!

Here's hoping, Barney
PS how do you say the 'MX-ϟ' thing? Is it now an MX-zing?


edit: improved 'flatten'

[weber]

I gather you are leaning towards 4, ie a single cell suddenly playing ball after maybe 14/15 min charging?

No. That makes me lean towards number 2 (out of balance), a little more than I already was.

None of them can be determined with crap data, so that's the next step. At this stage I'd like to be confident in the data Im collecting - and publishing! - before I worry about any consequences...

Right.

I have the Junsi in my hand and still cannot see why it stops at 99 entries. Tempting to set it for 30s intervals like you have, t1 terry, and hope for 45 min.

30 s intervals would probably be OK.

Im going to turn off 'sleep mode' which seems to be stall mode right now.

Good idea.

Will that 'flatten' (or rather, affect the charging of) battery #1 (where the Junsi draws its tiny power needs...

No. It will be insignificant.

PS how do you say the 'MX-ϟ' thing? Is it now an MX-zing?

Coulomb and I have in mind "MX-lightning", but "MX-zing" is good.

[barney]

I am pleased that I seem to have collected 2h worth of data, and the logger never stopped. And some pics. The pack sat at around 164.4V, according to the main gauge, for a long time. Presumably trying to charge a recalcitrant (!) cell... guess which cell? I reckon #8 on the logged set... Or so it seemed. Nearly all the red LEDs were on, suggesting they were floating. Of all the cells to be under 3.65V, #8 was almost the only one. The charger held on for an extra 30 mins or so trying to charge it whilst the others (#1-7 in the logged set) all seemed to be floating... in fact nearly all the top pack's red LEDs were on, at least for a while... altho some grew dim towards the end, and strangely, some (#31,32) stayed red for minutes after the charger went off. I have a pic. Go figure! Oh, and when you do, tell me, please!

Barney

[barney]

Cant resist putting the graphs up now. These might be interesting.

Charging of the whole traction pack from 70% to 100% usu takes 20%/hr plus any levelling time at the end, usually quite brief. THis year this levelling period has been much more variable. Tonight it took 2.5h to charge what would normally take 1.5h.

Actual data later.
GraphAug12 PackB 38-45 0m to 1h38


GraphAug12 PackB 38-45 1h38 to 2h30...

[T1 Terry]

I can sort of explain why some cells are high and some cells are low, the charger looks at total voltage across all the cells and charges to an average, if a few cells go high during charging all the other cells must remain low for the average voltage to remain the same. I'm guessing the cells that ramp up quickly have lost capacity either though a poor connection internally or through one or more of the pouches within that cell being damaged by over voltage charge.
This one is fairly radical but if the cell is damaged anyway you aren't going to make it worse. It you charge the wayward cell you can get to easily to 4v it may be enough to clear the high resistance connection and the cell will settle down with the rest of them. If it doesn't work the cell will remain pretty much the same as it is now.

Can you tell us more about what was happening in the second graph? It shows cell cell42 went high this time but cell 45 is not fully charged, is there more graph available? Maybe leave the logger running longer after charge finishes to see where the cells settle to.
The charger from Hobby King I posted the link to earlier can balance charge 6 cells at a time through the cell logger plug but you would need an adapter plug and lead to bring the cells into groups of 4 so the pack 8 cell packs can be balanced in 2 seperate groups.
With the logger, don't run any power saving modes and don't solder the link between seven and eight as shown on the Endless Sphere site, the logger part stops working correctly if you do.

T1 Terry

[weber]

A note on terminology: When the red LEDs are on on the BMUs we'd normally say they were "bypassing" or "in bypass". "Floating" is not something we normally do with Lithium cells, and if we did, it would have to be at around 3.4 V, not 3.65 V.

That's all still rather inconclusive. We really need to know what kind of charge profile the charger is supposed to be using (plain constant-current constant-voltage (CC-CV) or something more complicated?), and whether the charger is controlled by the BMS master, and if so in what manner (merely on/off, or including a reduced-current stage for balancing).

Do we know if the charge was terminated by the charger itself, or by the BMS master turning the charger off? If by the charger itself, is this based on time, or the current falling below some threshold? If so, what threshold or what timeout?


Presumably the BMS master will only turn the charger off if any cell goes over 4.1 V. Is it wired up to do that?

Does the BMS sense cell temperatures at all? i.e. are there any thermistors connected in the BMS daisy-chain and poked down the slots between cells?

Does the charger sense the cell temperature. i.e. is there a temperature sensing cable going from the charger to the battery as well as the positive and negative charge cables. Its temperature sensor would typically be glued inside a lug bolted to a cell terminal.

Ideally, to be able to make sense of these voltage curves, we'd need to see the current logged along with the cell voltages. I don't suppose there's any way you can do that. Wristwatch, pen and paper?

Like, what the hell happened at 1:37?

Both the fact that all the red LEDs did not come on, and the measured voltage of 164.4 V for 45 LiFePO4 cells suggest that the charger voltage could do with being tweaked up a little. It needs to be definitely greater than the voltage required to put all cells into bypass at the same time. There will be some tolerance on the nominal 3.65 V bypass voltage of the BMUs so we should perhaps allow 3.7 V per cell, which would be 3.7*45 = 166.5 V. But assuming the charger doesn't have a lower constant current mode for balancing (e.g 0.5 A), if we go too high we risk having some cell go to 4.1 V and shut off charging before they all achieve bypass. So it's quite a balancing act. This method of charging LiFePO4 cells is very unsatisfactory, and becomes more so the more cells you have in series.

But there's no point in playing around with charger voltage until we eliminate the possibility of high-R connections. That possibility has not been eliminated by this charging data.

I suggest that the next step should be to unbolt, clean and rebolt the terminals of cells 44 and 45 only. And then repeat the charge cycle with the same logging (plus current logging if possible). There's no need to discharge more than about 5% before doing the charge since the first hour and 20 minutes of the previous charge was pretty boring.