PDA

View Full Version : Cell balancing - top, bottom, static, and dynamic



podolefsky
05 June 2011, 1656
I've been thinking about this all day and realized that there is a lot of confusion and misunderstanding about cell balancing. I'm starting this thread to mostly help me understand all this stuff.

The first thing I want to talk about is static vs dynamic voltage and how that impacts balancing. Whether you're top or bottom balancing, it is based on voltage when the cells are at rest (or very nearly at rest), not at large current draw. This seems to obscure the fact that cells at the same voltage with an open circuit (zero current) will not be at the same voltage under load, because they have different internal resistances.

So let's pick on bottom balancing because it illustrates the point (not because I'm taking a side). Suppose you bottom balance all your cells to 2.8V. Then you put 60Ah into all of them. They all sit at different voltages, but that's fine as long as none go over HVC.

Now you discharge again. If you're pulling, say, 0.1C they will all hit 2.8V at the same time and you're all good. But what if you're pulling 1C, or 5C? The cells with higher Ri will dip below 2.8V before the others because voltage is a dynamic thing. The higher Ri cells will also heat up faster and get damaged in spite of being bottom balanced, again because balancing isn't done with large amp draw.

If you top balance, then weaker cells will run out sooner and you get something even worse. I'm just thinking about bottom balancing and how it might not actually work as people say.

Now, maybe the differences in Ri aren't large enough to have much effect. I don't really know, I'm just thinking out loud.

BaldBruce
05 June 2011, 1923
To put the problem into a form that might make more sense for you, I present a simple battery model. Note that many others have modified this model to include more complicated parts of the problem such as the effect of SOC or temperature on Ri.

Shepherd Generic Battery Model Equation:

Et = Eo - Ri*I - Ki*(1/(1 - f))
Where

Et = battery terminal voltage [volts]
Eo = open circuit voltage of a battery cell when fully charged [volts]
Ri = internal (ohmic) resistance of the battery [ohms]
Ki = polarization resistance [ohms]
Q = battery capacity [ampere-hour]
I = instantaneous current [amps]
f = integral of I*dTIME/Qo = accumulated ampere-hours divided by full battery capacity.
So with no current draw I, the internal resistance is no longer detectable or measurablle. Clearly shows the difference between static and dynamic measurements.

With LiFePO4 batterries, Ki is very small. This is both a blessing and a curse. Means we can draw lots of power and the battery voltage will only change by how much current we draw and not the SOC. Of course that is also the negative in that the SOC cannot be easilty determined becasue of that very flat voltage responce.

podolefsky
05 June 2011, 2112
Thanks Bruce. This model works very nicely, but I think it leaves out a major feature of lithium battery discharge curves. Namely, loss in capacity at high discharge rates.

In the graph below, note how the discharge curve for higher C-rates falls off at smaller capacity. This is actually a diffusion effect - when lithium ions get depleted, they can't diffuse fast enough to maintain the current you are trying to pull, and voltage drops.

I found that if I added an exponent to the f term I could approximate this effect. As in:

f = I^alpha * dTime / Qo

Still playing around with alpha.

[edit] that doesn't work. Hmmm....

http://kilowatt-age.com/yahoo_site_admin/assets/images/image007.323213728_std.png

magicsmoke
06 June 2011, 0232
Hi Noah, no need to reinvent the wheel, or in this case the mathematical model of the power source that drives it :)
If you google 'lithium cell model' you'll get a ream of useful links, some better than others.

As it happens, the first is one of the better ones ...
http://vtb.engr.sc.edu/vtbwebsite/downloads/publications/LithiunIonBattery_Lius.pdf

Rob

billmi
06 June 2011, 1032
Quoting from Dave's 1 bad cell thread that headed in the direction of this new thread:

Top-balanced packs tend to fall out of balance more as the BMS/charger end up forcing more charge into full cells.
That's exactly opposite of what most BMS' do. Most BMS monitor cell level voltage to shut the charger off as any one cell nears its capacity. Moreover, BMS' which have a shunting balance feature reduce or eliminate the charge going into a cell approaching capacity so that it doesn't get more charge forced into it while the other cells continue to charge. That's how they top-balance the pack. If they forced more charge into full cells they would not be doing any balancing.

Forcing more charge into full cells is not a problem of top balancing and BMS systems, it can be a problem though with charging a pack that has been bottom balanced.

Whether a pack is balanced to the top, or to the bottom, the other side of the pack's charge state will be unbalanced. That's the nature of the beast (unless you can afford to buy hundreds, or even thousands of cells, test them and group them in perfectly or near perfectly matched sets like they do for satellites and space probes where the cost of perfect matching outweighs the cost of launching the additional weight of a BMS). Either you top balance, and risk over-discharging the weakest cell during use, or you can bottom balance, and you risk over-charging the weakest cell during the charge cycle for exactly the same reason.


So let's pick on bottom balancing because it illustrates the point (not because I'm taking a side). Suppose you bottom balance all your cells to 2.8V. Then you put 60Ah into all of them. They all sit at different voltages, but that's fine as long as none go over HVC.

Here's what I don't get about the BMS-less bottom-balancing approach... Suppose I do the above, then I go out for a ride, and come back. I never hit pack LVC, so I know everything is honkey-dory. How do I know how much to charge? I can't charge by amp-hous unless I counted how many amp-hours I pulled out - and even if I did I'm now creating a risk situation if there's inaccuracy in my data. I can't charge to full pack-level HVC, because my cells are balanced on the bottom, not the top. I could under-charge the pack, reducing the HVC value, but even then, I'm assuming I'm reducing it enough to clear the weakest cell - which is a serious assumption as cell capacity will degrade over time.

DaveAK
06 June 2011, 1044
Either you top balance, and risk over-discharging the weakest cell during use, or you can bottom balance, and you risk over-charging the weakest cell during the charge cycle for exactly the same reason.
That's what I said!!


Here's what I don't get about the BMS-less bottom-balancing approach... Suppose I do the above, then I go out for a ride, and come back. I never hit pack LVC, so I know everything is honkey-dory. How do I know how much to charge? I can't charge by amp-hous unless I counted how many amp-hours I pulled out - and even if I did I'm now creating a risk situation if there's inaccuracy in my data. I can't charge to full pack-level HVC, because my cells are balanced on the bottom, not the top. I could under-charge the pack, reducing the HVC value, but even then, I'm assuming I'm reducing it enough to clear the weakest cell - which is a serious assumption as cell capacity will degrade over time.
I agree. Even if you deiced to charge to 3.5V/cell sooner or later you're going to have a cell that charges higher than 3.65V and without warning you're going to toast it.

I just don't see any benefit to bottom balancing.

podolefsky
06 June 2011, 1047
Yeah, I thought that sounded fishy - BMS's pull charge off of the full cells. But, it depends on how much charge the BMS takes out compared to the charge rate, and what algorithm is used. My BMS balances at 3W, about 1A for a 3.2V cell, but it charges at about 10A. So it has to shut down the charger and balance, then turn the charger back on. It goes through this cycle until the cells are all charged and balanced. (And it makes a pretty light show while it's doing it...I just wish it went "bing".)

If the BMS didn't have control over the charger, I could see it adding Ah faster than it could shunt around the full cells. Not sure, but seems plausible.

I think that with BMS-less bottom balancing, you can just monitor pack voltage and charge to an average of, say, 3.65V per cell. Most lithium can actually take up to 4.0V or more, so as long as all the cells are within 10% of each other, you should be OK.

DaveAK
06 June 2011, 1053
I think that with BMS-less bottom balancing, you can just monitor pack voltage and charge to an average of, say, 3.65V per cell. Most lithium can actually take up to 4.0V or more, so as long as all the cells are within 10% of each other, you should be OK.
But this is the same argument as top-balancing and only discharging to say an average of 2.8V/cell.

podolefsky
06 June 2011, 1100
So...back to my original question about bottom balancing. Take a look at the discharge chart up above. Suppose you're cruising along very happy at 1C. Your at 90% discharge, and pack voltage is about 76V (24 cell pack). Yeah, I know, don't go below 80% dischage...fine. But suppose you are.

Now you go to accelerate, pull about 3C. The knee at 3C will be a little different for each cell, and it's really sharp so some of them will be OK and others will fall off a cliff. So your statically bottom balanced pack all of a sudden becomes extremely unbalanced in a dynamic situation. Y'all with me?

podolefsky
06 June 2011, 1106
But this is the same argument as top-balancing and only discharging to say an average of 2.8V/cell.

Sort of - but this is the static (or "quasi-static") vs dynamic thing. You usually charge at like 1/5C or something, but you can discharge at 3C. It's not being at 2.8V or 4.0V per se, it's the temperature rise due to high current combined with high internal resistance.

DaveAK
06 June 2011, 1115
So...back to my original question about bottom balancing. Take a look at the discharge chart up above. Suppose you're cruising along very happy at 1C. Your at 90% discharge, and pack voltage is about 76V (24 cell pack). Yeah, I know, don't go below 80% dischage...fine. But suppose you are.

Now you go to accelerate, pull about 3C. The knee at 3C will be a little different for each cell, and it's really sharp so some of them will be OK and others will fall off a cliff. So your statically bottom balanced pack all of a sudden becomes extremely unbalanced in a dynamic situation. Y'all with me?


Sort of - but this is the static (or "quasi-static") vs dynamic thing. You usually charge at like 1/5C or something, but you can discharge at 3C. It's not being at 2.8V or 4.0V per se, it's the temperature rise due to high current combined with high internal resistance.
I'm with you.

frodus
06 June 2011, 1116
so the long and short, no matter how you balance, you still need monitoring to keep you from falling off the cliff on both charge and discharge..... and a means of disconnecting (or reducing) either throttle or charging.....

Because, yeah, you're right, if you're sitting there at 2.8V a cell at 1C and then do a blip to 3C, those cells will fall off the cliff quickly, and some will faster than others.....

but I guess you could always IR match ;)

DRZ400
06 June 2011, 1138
When I got my 24 Calb cells they had litte Alpha stickers on them.......... 20 C's, 3 B's and 1 D. I'm thinking they were sorted, but by what I don't know.....guess I also expected 24 C's then too. I sent them an email asking but they never answered.

DaveAK
06 June 2011, 1141
Mine were all F's. I trust this is just a sorting system and not a grading one!! :D

podolefsky
06 June 2011, 1146
But they all get an A for effort (especially that little one that gave his life to save the rest of the pack) :)

jsantala
22 February 2014, 0622
I'm rocking a 25 series pack of CALB SE40AHA's in my motorcycle that I've bottom balanced in the summer of 2012. I charge at 20 amps and discharge at up to 400 amps (as much as the cells can put out). I charge the pack to 87.6 V which is 3.5 volts per cell on average. Some cells don't reach 3.5 and some go slightly over it, but it has made no difference.

I've twice driven the pack down to 50 volts (2 volts per cell), at which point they didn't have enough juice to move the bike anymore. Which is why they are unable to kill their buddies either, even if those buddies were a little bit more empty. In a top balanced scenario strongest cells would destroy the weakest at this point.

If cell drift exists, I cannot prove it any more than I can prove the existence of God. Yes, I'm an atheist.

In my experience, variations in cell temperature, charge/discharge speed or internal resistance have no measurable impact in reality. QED

podolefsky
22 February 2014, 1104
If cell drift exists, I cannot prove it any more than I can prove the existence of God. Yes, I'm an atheist.

Drift definitely exists, but with lithium cells it's very small. Enough that a balanced pack (top, bottom, mid, whatever) could go for months without becoming noticeably unbalanced.



In my experience, variations in cell temperature, charge/discharge speed or internal resistance have no measurable impact in reality. QED

Yup, that's exactly right. The only thing that unbalances a pack is a variation in self-discharge rate between cells.

podolefsky
22 February 2014, 1122
Just to add to the thread, this is a really good tutorial on top vs bottom balancing. Davide obviously favors top, but I think the reasons he gives make a lot of sense. He does go to some length to make the case for bottom as well, with caveats.

http://elithion.com/education.php#Video_tutorials


And just to be fair (and balanced?), here is Jack Rickard's classic post on bottom balancing.

http://jackrickard.blogspot.com/2009/11/get-rid-of-those-shunt-balancing.html


Should go without saying, but I'm just putting these up for reference. Arguments for either side are well hashed out in the blog comments, other places on the internet, etc.

chipper6
22 February 2014, 1212
To compliment the theories, some empirical data from John Hardy.

Summary: He has performed cycle tests with an arduino/cell log8 set up with Headway cells and Calb Grey cells 40Ah.

Charts and data:
http://tovey-books.co.uk/testing.php

Youtube channel : http://www.youtube.com/user/jlghardy/videos

chipper6
22 February 2014, 1252
Just interesting info here:

http://www.youtube.com/watch?v=pxP0Cu00sZs

1 hour seminar on the effects of heat on Li cells from Professor Jeff Dahn of Dalhousie University. He has all kinds of data on different chemistry, charge rates, etc.

Question for people who top balance and actively balance. Each time you charge, how long are your cells held above the 100% SOC static resting voltage? (for example. My GBS cells rest at ~3.35V when charged. How long are your cells held above that voltage while your system is waiting to balance all the cells?)

podolefsky
22 February 2014, 1305
Interesting data, thanks for posting.

He did about 500 cycles, each taking about 1.5 hours. If my math is right, that's roughly 1 month. Over that time, a li-ion cell would be expected to self discharge 1-2%. All of the cells will be about that, which means the variance in self-discharge will be a lot smaller. So, not surprising that he didn't see any difference at the end. (A max of about 0.01V, which is usually the allowed delta in a "balanced" pack.)

The theory is that it's just self-discharge rate, not cycling, that creates unbalance. Would be interesting to see the same test, but no cycling. Just leave them on a shelf, then check them again a month later.

Incidentally, the small variance is why most BMSs have very low balance currents, usually 0.2-0.5A. Even with that tiny current, you can keep a 1000Ah+ pack balanced.

podolefsky
22 February 2014, 1314
Question for people who top balance and actively balance. Each time you charge, how long are your cells held above the 100% SOC static resting voltage? (for example. My GBS cells rest at ~3.35V when charged. How long are your cells held above that voltage while your system is waiting to balance all the cells?)

With my 60Ah GBS pack, less than an hour. There's the CV phase, which is probably around 20-30 min. Balancing is usually done in 15-20 min - a lot of that time is spent waiting for the cells to rest to get a good measurement. The time spent actually adding / removing charge is maybe 1/2 of that (rough guess).

jsantala
22 February 2014, 1358
Drift definitely exists, but with lithium cells it's very small. Enough that a balanced pack (top, bottom, mid, whatever) could go for months without becoming noticeably unbalanced.

Yup, that's exactly right. The only thing that unbalances a pack is a variation in self-discharge rate between cells.

I have not observed any measurable self-discharge (or cell drift). If it exists in the cells I'm using it doesn't make any difference whatsoever.

Let me reiterate. I have seen no reason to do anything about alleged self-discharge or cell drift. If I don't see them I don't need to do anything about them.

If I were to install a cell level BMS, I'm certain I would see discharge and drift, but it would be caused by the BMS. The cells themselves exhibit none.

chipper6
22 February 2014, 1415
Also last summer Jack pulled some old thundersky batteries he recieved in 2008ish from their original shipping crate that were still at the same SOC as the ones he originally used in 2008. So for the case of cells in a fixed enviroment, not connected to anything there is evidence of no self discharge.

I have not seen a case where a pack if cells connected in series and left in storage with no connected devices.

But like Jake's example I have talked to a dozen other owners of cars that have bottom balanced and charge to 3.4-3.6v and have driven 10k+ miles over the span of one to two years without observing cell drift or self discharge. Until data is presented to show evidence, it is hard for me to justify the cost and complication (and usually lack of quality) of a BMS.

Sent from my HTC One using Tapatalk

podolefsky
22 February 2014, 1454
Yeah, like I said, self discharge rate is so small you wouldn't notice unless you let the cells sit for months without charging them. Since most people drive their vehicles at least once a month, they won't see it. If your pack stays healthy, then sure, soldier on.

Just clarification - drift isn't the same as self discharge - it is the *difference* in self discharge between cells. If the self discharge rates are all the same, you won't have any drift.

If you bottom balance, then the cells are always at different states of charge (except at the bottom, which you almost never hit). You'll almost never be in a place where you could take voltage measurements and confirm that the cells haven't drifted. I'm curious how you can determine that you have no drift if the cells are never at a point where they should be equal.

I won't argue the merits of using a BMS, their cost or quality. Most of the claims that BMSs cause problems are due to misunderstanding of how they work. Since the thread is about balancing, I'll just say that balancing is only one function of a BMS, and actually not the most important.

podolefsky
22 February 2014, 1506
Here are 3 sources of evidence for self discharge (not drift) in lithium cells.

http://www.mpoweruk.com/performance.htm

http://nordicgroup.us/battery/#Advantages_of_Li-Ion_Batteries/Disadvantages_of_NiMH_batteries_(AA/AAA)_

http://powerelectronics.com/site-files/powerelectronics.com/files/archive/powerelectronics.com/images/0606batterypacks-Figure04.jpg

(from here (http://powerelectronics.com/mobile/designing-battery-packs-thermal-extremes))

One thing you'll notice is that self discharge is reduced at lower temperatures. Since people tend to store their EVs during the winter, that could be another explanation for why they don't see much self discharge. But I think it's mostly because they're always charging them.

furyphoto
22 February 2014, 1822
Here's an interesting video from June 2010. If you haven't seen it, it's worth a watch.

At about 1:05:00 he talks briefly about balancing, and fields a question about BMS, with some interesting insights, but no particular conclusion. The first 45 min of the video is about how lithium batteries work. It's like a basic intro to advanced info on batteries.

http://www.ri.cmu.edu/video_view.html?menu_id=387&video_id=60

Jay Whitacre, a Carnegie Mellon professor with a joint appointment in Materials Science and Engineering and Public Policy, gives a quick overview of Lithium-ion batteries. Starting at the chemical level, he explains the properties and mechanics of the battery which give rise to macroscopic behavior, especially focusing on issues relevant to electric vehicles and other high-power systems. He then fields questions about the batteries themselves, ongoing research at Carnegie Mellon, and tricks for improving long-term performance, such as pulse-charging and pairing batteries with similar defects.

podolefsky
22 February 2014, 1900
Great video. I've watched it about 1000 times. Note some things he says on BMSs:

- The best way to manage is individual cell level (but it's expensive).
- It's cheaper, if you're Toyota, to get perfectly matched cells from a top tier supplier (Sony, Panasonic). If you're not Toyota (i.e., you're one us DIYers), then you should have a cell level BMS.
- Tesla doesn't monitor all 7000 cells in it's battery. What he doesn't say is that you don't need to do that. I don't know exactly, but that battery is something like 100s70p. You only need to monitor parallel banks, so you only need 70 sensors (one for each parallel set), not 7000.
- They still have cell level monitoring (which is cheap). Just no way to "take a cell out of the system", by which I think he means electrically disconnect a bad cell.

Another thing he doesn't say, those big automotive packs aren't pushed to their limits. Mostly to get longer cycle life but also to stay in the safe zone. If you never get close to the high or low voltage limit, then you don't need to worry as much about slight imbalances.

[edit - nevermind - see next post] All that said, I'm now dealing with yet another bad cell board - common problem with distributed BMSs - which means buying more, about 4 hours of work to replace it, then hoping one doesn't go bad again. If I had room I'd switch to a centralized BMS, but there's nothing out there for the DIY crowd compact enough for my bike. So, I'm considering just using CellLogs for high and low voltage alarms, and a Clean Power state of charge display. Then balance with an RC charger every once in a while as needed, which isn't very often (as others have pointed out).

podolefsky
22 February 2014, 1925
I just figured out I didn't have a bad board, it was that my pack had been sitting since November. I forgot to disconnect my DC/DC converter (doh!). The pack drained and one cell dipped so low that the cell board stopped messaging the BMS. I forced it to charge, and now I see that cell is coming up about 0.2V lower than the rest.

I've pushed my pack pretty hard, so I wouldn't be surprised if that one cell had developed problems. Could be dendrite growth (crystals around the electrodes). These can increase self discharge (http://www.mpoweruk.com/failure_modes.htm). Could also be that it popped its valve at some point and lost electrolyte (has happened to me before). Whatever it is, I'll have to keep an eye on it now.

podolefsky
22 February 2014, 2205
OK, this is my last post here for a while, I just got to reading about Tesla's battery and had to share.

They have a ton of safety features built in, down to the cell level, including double fusing on each cell (which is a way of "taking cells out of the system").

http://large.stanford.edu/publications/coal/references/docs/tesla.pdf


I kept trying to find out how / whether Tesla does balancing - hard to find, since they keep everything secret. I finally found it. It does indeed, and it does it after charging, which means top balancing. They apparently have a "storage mode", which cycles the cells at low SOC if the car is sitting for more than 2 weeks. They recommend switching to "standard mode" after the car has been stored - in order to rebalance the pack.

http://www.teslamotorsclub.com/showthread.php/3848-Tesla-Roadster-Battery-Care

Hugues
23 February 2014, 0011
OK, this is my last post here for a while, I just got to reading about Tesla's battery and had to share.

They have a ton of safety features built in, down to the cell level, including double fusing on each cell (which is a way of "taking cells out of the system").

http://large.stanford.edu/publications/coal/references/docs/tesla.pdf


I kept trying to find out how / whether Tesla does balancing - hard to find, since they keep everything secret. I finally found it. It does indeed, and it does it after charging, which means top balancing. They apparently have a "storage mode", which cycles the cells at low SOC if the car is sitting for more than 2 weeks. They recommend switching to "standard mode" after the car has been stored - in order to rebalance the pack.

http://www.teslamotorsclub.com/showthread.php/3848-Tesla-Roadster-Battery-Care

Interesting read, thanks for posting.

Important point though is both of these articles refer to the Roadster model, not the more recent model S, and date from a few years back. Would be good to know if they have used similar techniques for the model S.

jsantala
23 February 2014, 0011
After charging and letting sit for a day all the (bottom balanced) cells are always in the same voltage to at least two decimal places. Maybe more. Nevermind their theoretical SOC. My total voltage always shows 83.4 volts on a full pack. Every time. No variation whatsoever. I'd know immediately if there was a problem, if my voltage wouldn't be 83.4 volts. Has not happened.

I've left the cells sitting over the winter, for months, in the cold, at -20'C and when I've checked the voltages afterwards, they've still been identical. No cell drift.

On 40Ah cells even 1% per month should have resulted in several amp hours lost during months of storage. I have not seen such loss. No self discharge.

Really, checking cell voltages on a bottom balanced pack gets really boring, really fast. They're always the same. To me that proves it just works.

If the cell voltages vary, there is a problem. The problem is a broken cell, which needs replacing, or a BMS, which is harming the cells.

podolefsky
23 February 2014, 1108
After charging and letting sit for a day all the (bottom balanced) cells are always in the same voltage to at least two decimal places. Maybe more. Nevermind their theoretical SOC. My total voltage always shows 83.4 volts on a full pack. Every time. No variation whatsoever. I'd know immediately if there was a problem, if my voltage wouldn't be 83.4 volts. Has not happened.

I've left the cells sitting over the winter, for months, in the cold, at -20'C and when I've checked the voltages afterwards, they've still been identical. No cell drift.

On 40Ah cells even 1% per month should have resulted in several amp hours lost during months of storage. I have not seen such loss. No self discharge.

Really, checking cell voltages on a bottom balanced pack gets really boring, really fast. They're always the same. To me that proves it just works.

If the cell voltages vary, there is a problem. The problem is a broken cell, which needs replacing, or a BMS, which is harming the cells.


I get where your coming from - if you have a system that seems to be working, by all means go with it. There's no one right way (although there are definitely some wrong ways).

I'm not trying to be contrary here, or convince you to change what you're doing. Just explaining what I think is going on - you're saying what you measured proves that it works, and I'm just saying that you could get those measurements even if it wasn't working.

LiFePO4 cells will tend to sit at the same rest voltage no matter how they are balanced, and, really, even if they're not balanced. Rest voltage tells you almost nothing about their state of charge, so measuring after they've sat for a day won't tell you anything about the balance of the pack.

The same goes for pack voltage. 83.4V means 3.475V per cell. Again, they might all sit there at rest, but that doesn't mean they're at the same state of charge. (It also doesn't mean that they're not, it just means you don't know.)

Like I said, if it's cold self discharge will be much less. At -20'C, it will be nearly zero. Not to go too deep into it, but the reason it happens is that you have a chemical reaction waiting to happen in the cell (it's like holding a stretched rubber band, discharging is like letting the band contract). That reaction is going to happen, it's just a matter of how fast. The lower the temperature, the less the molecules jiggle, and the slower the chemical reaction. Again, imagine freezing a stretched rubber band - it would contract more slowly.

frodus
23 February 2014, 1222
Voltage is in no way an indicator of SOC. You can have two cells at basically the same resting voltage, yet completely different SOCs. Rest voltage means nothing. It will not tell you if a cell has self discharged, or if a cell capacity has drifted to a lower value (i.e. 10Ah and now its 9.8Ah capacity). You need to load a cell to even remotely look at SOC.

In order to even test for capacity drift or self discharge, you need to actually discharge it all the way and measure the Wh of each cell individually. This is something no one wants to do nor has the equipment (most times) to do. Unless you've got a CBA or PowerLab (or other load) and routinely take your pack apart and discharge your cells to measure this, then you won't know what SOC, drift or self discharge is.



Now, back to the actual topic. Balancing = Voltage balancing. All we're doing with balancing on the top or bottom, is getting the cells within a fairly close voltage cluster so that they're predicable. We want them to hit LVC and HVC at the same time. Thats it.

The thing about bottom balancing, is that we're loading the cells, so they show ACTUAL LOADED VOLTAGE, not resting voltage. You're loading the cell and letting it discharge to the same point on the curve. Every cell will hopefully discharge to that same resting voltage. Then when you charge the pack, you charge them all up to some total pack voltage. When you do this, I can almost promise you your cells will be at slightly different voltage levels, because each one has a slightly different IR and capacity. I've tested many batteries and all cells, even high quality cells, showed a slightly different capacity and IR. Low quality cells were worse, however.

Top balancing is kind of like loaded voltage. Its not a resting voltage that you see. Cells will get to the top of the curve, hopefully at the same time. Each cell has a different capacity, so when they do discharge to the bottom, because of different IR and capacities, they will show slightly different bottom voltages.

Its only "balanced" at either top or bottom (whichever you chose to balance). Resting voltage could be anything, but it doesn't tell you where each cell is individually in its SOC.

The entire idea of balancing, is that you pick one or the other (top or bottom), and monitor on a cell-level basis all of the voltages. When one cell gets too low, cut off discharge. When one cell gets too high, cut off charge. In the middle, you have no idea of the cells SOC. A BMS is there to protect your cells. I could care less if people balance their batteries. I honestly don't care what you do, but I WILL say that I think its super important to monitor them, because in either top or bottom balancing, if you aren't monitoring a cell, and it hits LVC and you don't know it, you're permanently damaging that battery. If you don't monitor, you'll never know because you're only looking at pack voltage, or resting voltage.

Ted actually has a simple type of BMS, it doesn't balance but it does monitor. He uses to let him know when a cell goes low or high. Its technically there to protect the batteries from him overdischarging. Its by far the simplest and cheapest way that I've seen to do it, and I fully recommend one to those that want to run without BMS. What we're trying to protect from here is not balance, drift, IR..... the whole point of a BMS is to protect from going under the low voltage cutoff, over the high voltage cutoff and over the max current.... But on a per cell basis.

EV_Scoot
23 February 2014, 1741
There's a saying that most of you probably already know.

You can't manage what you don't measure.

BMSs fall under this, in my opinion. I'll be using a BMS to give me an overview of what's going on, not do the dishes and make coffee.

podolefsky
23 February 2014, 1850
Exactly. What I was saying, and Travis explained really well, is that measuring voltage is not the same as measuring state of charge. It's only a reasonable measure of SOC at certain points, under certain conditions. You have to measure, and you have to meaure the right thing.

I was just rereading JR's blog (http://jackrickard.blogspot.com/2009/11/get-rid-of-those-shunt-balancing.html)on bottom balancing, that, as far as I know, started the whole thing. I get his point, but there's one thing in there that gets me every time - he says "But I don't care about the top. I don't lose cells at the top..."

Wut? Overcharging can totally kill cells, even start the fires that he's so worried about. In fact, I'd argue it's worse at the top, since that's when the cells have the most energy in them. Worse, that's when your EV is in the garage, charging unattended. Talk about a bad idea. Especially if you do what JR says, which is charge to the nominal pack voltage, letting some cells go over their high voltage limit. At the very least, you should know the pack voltage where the first cell hits high voltage and never charge beyond that.

chipper6
23 February 2014, 2212
JR relies heavily on the repeatability of the chargers he selects to terminate at a specific voltage. I believe his saying is "bottom balance and undercharge". But he has come around to at least a monitoring system of some sort, but not cell level monitoring.

One thing that they have shown is that it may be better for cells that are bottom balance to not have a CV stage at the end of charge. That is to have the charger drive at a constant current and terminate at a voltage equal to ~3.5v per charge without tapering the current holding constant voltage.

The other point he tries to make is that the current while charging is significantly less than the current pulled from the motor under load. If the pack is top balanced, the user relies on the BMS to stop them from killing a cell at 200-500 amps. And they rely on the BMS to stop the charger. He relies on the charger terminating at the same voltage every time. Most controller he deals with also have a pack level LVC.

Less components to fail, less things to buy and install, less complication, that's why it has caught people's attention as a possible better solution.


Sent from my HTC One using Tapatalk

frodus
23 February 2014, 2230
But..... He has also mentioned that he'd like to have his gevcu act as a modular monitoring system. He's also used cell-log8's and reap systems was on his show. I think he's OK with monitoring. It's the balancing that he's against.

If you are the bms and you can design a system that can, for the most part, keep cells in their operating range.... Then do it.

Most people, however, won't pay that close attention to their batteries. I've seen many people without a bms damage an entire pack of cells from either overcharging or over discharging. Some of then are high profile EV members.

Running without monitoring is far from a "better" system. It's just easier and cheaper. It's still a gamble.

chipper6
23 February 2014, 2318
I hate to say it, but some sort of standard (ISO, CE, SAE, etc.) would assist in bringing the quality up. There might even be something but its probably costly to follow.

podolefsky
23 February 2014, 2339
Chipper, I appreciate that you said people are interested because it's cheaper and less complicated. That's a perfectly legitimate reason. I'll even admit that it does seem like an elegant solution to the balancing problem, and with all the BMS parts I've replaced I can see the appeal. I just think that when you dig into it, you find as many problems as solutions. The biggest problem I see isn't even with the method, it's that it goes hand in hand with people not doing adequate monitoring. People seem to think that if you bottom balance, all you need is pack level voltage and you're safe. That's only if everything goes according to plan - and you know Murphy's law.

Toni
24 February 2014, 0026
I've got an indentical setup as jsantala has in his bike. From what I've witnessed so far, these blue CALB cells seem pretty quality stuff and have really minimal (if any) cell drift. In my use I typically use about 2+kWh from the 3.2kWh total so I very rarely drive them anywhere near empty. I had balanced them at 2.75V and when fully charged they settled within 0.01V from each other. And the couple times I've monitored them so far they've stayed that way, both fully charged and after about 2/3 used.

But as a lot of you has pointed out, with bottom-balancing the user should be the BMS. And I will be periodically checking my own pack cell balance.

jsantala
24 February 2014, 0211
I have 25 cells so 83.4 volts is 3.336 per cell.

Rest voltage (after like a day or two) is actually the only way to truly measure voltage. Voltage under load or charge is unreliable and cannot be trusted. It's simple enough to prove. Connect load, watch voltage change, disconnect load, watch voltage change. Let the cell rest and after a day or two and after that the voltage will not change at all, if you keep the cell disconnected and at the same temperature.

I can assure you that after some power drain and rest my pack voltage will not be the same 83.4 volts as it's charged at. So you can look at total pack rest voltage and tell it's not full, therefore seeing SOC in voltage. I could make a plot to see which voltage levels represent which SOC, but it's quite unnecessary and cumbersome, because you'd need to wait a day to see where you are. Easier to just count Wh or Ah.

Measuring SOC from rest voltage when the cell is quite full is hard, because it changes so little. However there is no need to do so. Voltage when near empty, like around 2.75 volts on a LiFePO4, changes very rapidly based on SOC. Therefore it is very easy and reliable to bottom balance the cells using rest voltage near empty. You just need time to let the cells settle after discharge. After the balancing you can just count Coulombs.

Empirical evidence is what proves stuff. Theories and formulas try to explain and represent observations, not the other way around.

jsantala
24 February 2014, 0400
I would also like anyone to point me in the direction of a well documented, failed bottom balancing experience, where a pack or cells have been destroyed.

Honestly. I am not aware of a single one. I'd really like to know what could go wrong. Not in theory, but in real life. Theories I've seen plenty.

jsantala
24 February 2014, 0448
Great discussion, and thanks all for adding the resources, experience and links. My take away is this, from page two ^^^. I'm not sure I see any disagreement here, right? Top or bottom balance, you need to monitor, and monitor at rest and under load? Or am I missing something?

I certainly can vouch for it in my own experience. The addition of the cell logs to my top-balance "lolsystem" made the difference between it working well for me and not.

Ted, in my bottom balancing scenario, which I've be veen using without problems, I "monitor" as follows:

- On discharge: I set my controller low limit at 2 volts times cell count, so on 25 cells = 50 volts. A bottom balanced pack can be pushed to 2 volts on average, because all cells become empty at the same time and they don't have energy left to harm their buddies. They can't even move the vehicle in the very end and you'll feel the pack going empty in your throttle as the controller starts hitting that limit. The cells will bounce back from that 2 volts closer to 3 volts when the load is disconnected.

- On charge: I just charge to 3.5 volts times cell count, so about 87.6 volts (when I purchased the charger I had 24 cells which I charged to 3.65, but I added 25th cell). Some cells don't reach 3.5 and some go above, but it makes no difference. The charge is the monitor then and stops the charge very reliably, which not all BMS do.

- On rest: I check that my pack reads 83.4 volts when fully charged and rested. If not, there's a problem.

These figures are of course for CALB LiFePO4. If I get a Renault 24kWh pack from Jack & co, I will put 11 modules (22 cells) in series and charge them to 87.6 volts so 3.98 volts per cell. Hopefully that will be enough to let me use most of the capacity. Otherwise I'll need to get a new charger.

Your "lolsystem" works because it's not doing anything bad to the cells. You're just monitoring them. With a little uneven load maybe, if you're using all 8 channels, but as long you're aware of it you know how to deal with it. KISS usually works better than overengineering. :)

Warren
24 February 2014, 0738
The Nissan/Renault modules are actually 4 cells, 2p2s.

I have lots of concerns about JR personally, but he does lots of projects, and he is very open about problems he has had. It will be interesting to see how well bottom balancing works with chemistries other than LiFePO4. I have groaned through all his shows, and it seems clear that for prismatic Chinese cells, at least, bottom balancing works great.

He, and several others that I am aware of have done unintentional "experiments" with their packs. Basically, leaving some load turned on that drains the entire pack to less than a volt per cell at rest, in one case down to only several hundredths of a volt per cell. They started off with a trickle charge, and once the cells were back up to 2 volts per cell, they charged them up normally...no venting or swelling, no apparent capacity loss. On his 76 kWh Escalade, he has done it twice with no ill effects.

He is now offering two packs for RV's and golf carts, a "12 volt", and a "48 volt", which are bottom balanced, and sealed up so you can't discharge the individual cells. He dares you to ruin these packs, in the applications they are designed for.

jsantala
24 February 2014, 0820
So you're doing no cell-level voltage readings, either under load or at rest?

That's correct.

I'd recommend checking the pack once a year or so: Drive it as empty as you can, check that the cell voltages are in line when empty and that no bolts have come loose. Just as a routine maintenance. Every vehicle needs a yearly check anyway, so you can just add this on the list.

Obviously you can do this as often as you like. Depends on how paranoid you are. I find it rather boring.

Daily check is that the pack voltage after charge and rest is the same every time. Right after charge it may be higher, of course.

Cell level monitoring will not hurt bottom balancing, but only if you can be absolutely certain that cells are not loaded unevenly.

And everyone knows each and every monitoring wire needs to have a fuse at the battery end, right? First and last wire will have full or near full pack voltage between them.

jsantala
24 February 2014, 0822
The Nissan/Renault modules are actually 4 cells, 2p2s.

You are indeed correct. I will parallel 2-3 modules and then put them in series. So I suppose it will be 2p2s2p11s or 2p2s3p11s... :D Bottom balanced, of course.

Alternatively I could put 11 modules in series and parallel 2-3 of them, so, umm, 2p22s2p or 2p22s3p, but probably won't.

Oh, and I won't ask the cells what they think about all this. I'm also quite sure they won't tell me.

podolefsky
24 February 2014, 1035
Empirical evidence is what proves stuff. Theories and formulas try to explain and represent observations, not the other way around.

With due respect, I take some issue with this statement. You never prove anything in science, you only disprove things and eliminate wrong ideas, leaving the better ones standing.

As far as theory, it goes both ways. Data gives you clues about how close theory is to reality. But data is not the same as reality. You need a model of how data relates to the unobservable thing you want to know. That's the issue with using voltage as a measure of SOC - you can measure voltage, you can't measure SOC. So any voltage measurement is only as good as your model of how it relates to SOC. In other words, your data is only as good as your theory.

It sounds like you do have a model of voltage vs SOC. The question is: how good is that model? Does it predict SOC in a repeatable way?

Many times, theory will tell you that you are misinterpreting the data - or even that what you think you saw didn't actually occur. This happens all the time in physics. People thought they had found faster-then-light neutrinos, theory said that shouldn't be possible. They spent months going back over their measurements, low and behold their data was wrong. Theory trumped data, in that case.

I'm not saying your measurements are wrong, or even your model - just taking issue specifically with this statement on data vs theory.

frodus
24 February 2014, 1108
In the end, why a debate? Everyone has a solution to their own design.

I mean we all agree that starting with top or bottom balancing is important. Also we agree that you should have some way of ensuring that you stay between the upper and lower limits of a cell voltage.

Both can be be done via BMS/top balancing or by undercharging/bottom balancing. Both can be done with or without a monitoring system. With no bms, you are the monitoring system. With a bms, you're letting the equipment do it for you.

There is no wrong here..... There are just more than one way to keep cells in their happy place. Whatever works for you, and whatever your acceptable risk is (cost of bms or cost of replacing a cell), is irrelevant. The fact that we're all keeping out packs safe is the important thing.

But I will disagree with one thing, there is measurable drift and there is measurable self discharge. It can be high, and it can be not very much and often it's so small you won't see much difference between cells that have similar IR values. This is fact and I've seen it time and time again in my lab while testing batteries for myself and other clients.

I had a whole battery pack of headway cells that self discharged over the course of 2 years while in storage they started charged and 2 years later they were at a lower SOC, tested with a CBA. I also noticed differences in their ah capacity from when I got them to when they were retested. All of these cells had never been used, only charged and stored.

jsantala
24 February 2014, 1343
Ok, I'll take your word for it. Headways may have these problems. It may be a feature of their design. What it doesn't mean is that all LiFePO4 cells would be the same. Luckily. :) I haven't used Headways and the more I hear of them the luckier I feel that I didn't go with them in the first place. CALB cells seem to have none of these issues.

frodus
24 February 2014, 1501
Ok, I'll take your word for it. Headways may have these problems. It may be a feature of their design. What it doesn't mean is that all LiFePO4 cells would be the same. Luckily. :) I haven't used Headways and the more I hear of them the luckier I feel that I didn't go with them in the first place. CALB cells seem to have none of these issues.

Reread what I said...... I didn't say they self discharged all the way, only that I measured a difference. It actually wasn't that much.... But it was measurable. I've tested CALB as well and found that they also self discharge similarly... So yes they have these same issues.

All batteries I've measured had some amount of self discharge and all had varying IR. This is present in all battery types and technology to some measurable degree. Period. Higher quality like Kokam, Enerdel and A123 were much closer in IR and Ah capacity, but still showed slight variances.

Look at this:
Http://liionbms.com/php/wp_short_discharge_time.php

CALB and headway are right next to each other.... Near the bottom of the list. Note: I helped with a couple of the measurements on that page.

Now.... Go back and reread my first post in this thread about SOC.


Voltage is in no way an indicator of SOC. You can have two cells at basically the same resting voltage, yet completely different SOCs. Rest voltage means nothing. It will not tell you if a cell has self discharged, or if a cell capacity has drifted to a lower value (i.e. 10Ah and now its 9.8Ah capacity). You need to load a cell to even remotely look at SOC.

In order to even test for capacity drift or self discharge, you need to actually discharge it all the way and measure the Wh of each cell individually. This is something no one wants to do nor has the equipment (most times) to do. Unless you've got a CBA or PowerLab (or other load) and routinely take your pack apart and discharge your cells to measure this, then you won't know what SOC, drift or self discharge is.

You have not done any testing on your cells to support your claim of no self discharge or drift. You have not measured SOC of a cell, only the resting voltage of a cell. You need to discharge a cell while measuring amps and volts and time, and compare that to what you started with in order to judge self discharge or drift. You've done neither before or after. Your so-called "fact" is just a theory because it is completely unsupported with empirical data.

Few people purchase equipment necessary to measure the wh into and out of a cell. Nor do they leave them for 6+ months then come back to test how much was left. No one except for a few, including myself, have done that. Mostly because of lack of the proper equipment or motivation ($$) to do so.

Athlon
24 February 2014, 1517
Http://liionbms.com.com/php/wp_short_discharge_time.php
.

correct url http://liionbms.com/php/wp_short_discharge_time.php , there is a .com typo

BTW , nice page , im reading it and I will comment later

frodus
24 February 2014, 1526
Updated.... Sorry. Doing all this from my phone and had to manually type in.

The short discharge time doesn't really tell you much about drift or self discharge.... That's not my point.... More of a tangent.... But it does illustrate IR and overall quality of cells being similar between calb and headway.

jsantala
25 February 2014, 0032
It's just that if what frodus is telling is true, how can bottom balancing just work, charge after charge, year after year? If the cells truly exhibit these issues I don't see how it could just work again and again. It really shouldn't. But it does.

Btw, frodus, do you or have you sold BMS, or are you planning to sell BMS?

podolefsky
25 February 2014, 0051
Everyone has their knowledge and their experience, and what I've found with these arguments over and over is that everyone is usually right in some respect, but people are talking across each other.

Since I started this thread, I'm calling a time out. Everyone chill for a day, read back over what people have said and try to take it in. Then we can start up again with clear heads.

Sound OK?

jsantala
25 February 2014, 0420
I'd like to hear frodus' answer to jsantala's question, please, because there's something there I'm missing. I'd also like to know from jsantala how you bottom balance if you don't monitor each individual cell for it's voltage... balancing in my understanding means monitoring voltage under load or charge so it's the same as it's buddies.


I balance once by draining each cell individually to a voltage below the knee of it's discharge curve, where the voltage drop is the fastest. On CALB LiFePO4 a suitable value is 2.75 volts. Note that once you disconnect the draining load the cell will bounce back up, so you must repeat the procedure until it no longer bounces back. Best to wait until the next day to see if they're still where you want them. And don't worry, the cells will not be harmed by this, unless they were DOA anyway.

You may program a device to do this for you, like a Revolectrix PowerLab 8, but if you do it by hand you'll get a good feel of the cells and will definitely notice if they're not all ok. To make things faster you should start with a rather empty pack. I used the lights etc. on my motorcycle to drain the cells first four at a time and then used the motor to drain the rest.

After this you may assemble your pack and as long as you always charge and discharge it only as a whole and undercharge slightly (average 3.5 volts on CALBs), it should keep its balance. You may check your pack occasionally by driving it empty, so that the cells are all below their knee, and check voltages. In the long term the undercharging will also pay off since your cells will last longer.



I'd like to know more details about any method of balancing when it fails. jsantala brought that up before, and though I've heard stories, I'd like to see the examples of bottom and top balancing failing, damaging packs, and learn more about why.


I am not aware of any bottom balacing failures. Top balancing failures I only have hearsay of, but since most people top balance and use a BMS to protect their cells, yet still apparently lose cells left and right, meaning the BMS was not able to protect the cells after all, there must be a problem somewhere. Otherwise nobody would ever lose any cells, right?

Toni
25 February 2014, 0615
It was actually only after following jsantala building his pack and using bottom-balancing, I had the guts to get into the whole "build your own electric motorcycle" business. So I also bought 25pcs of CALB 40Ah cells and a charger that had cutoff voltage of 87.6 volts.

I used Graupner Ultramat 16 computerized charger that I had (German-built RC lipo-charger with balancing boards for LiPos and also programs for LiFePo4, NiMH and lead cells) to bottom-balance the cells. All the cells were initially in about 50% charge when I got them and then I drained them with the Ultramat, one by one, aiming for the 2.75V. The charger confirmed that during the discharge about 20Ah were drained from each cell before voltage dropped below 3 volts. I had to repeat the discharge a couple times for each cell, since the voltages always bounced back somewhat (first to 2.9V or so, then less and less). Finally I had to actually set the charger target voltage to about 2.7V, since they would always bounce back above 2.75V. When I was close enough, after a couple days of resting all had crept up to 2.76V and the pack total was 69V when connected in series. I called that "good enough".

That's also pretty safe way to do it as the charger won't allow you to discharge them too much and stops at certain voltage (which is still way above dangerous). But at that point they have very little energy in them. After that I connected the pack in series and charged as a whole. After the charge was complete for the first time, all the cells were within +/-0.01V deviation.

chipper6
25 February 2014, 0928
You can also do a CC/CV discharge with the power lab8. And select to terminate at C/10 or C/20. I did that with some a123 20Ahr cells to 2.65v 5 months ago and they are all sitting at 2.81-2.83 currently. Suppose I should charge them since your not suppose to store them empty... :/

Sent from my HTC One using Tapatalk

frodus
25 February 2014, 1018
It's just that if what frodus is telling is true, how can bottom balancing just work, charge after charge, year after year? If the cells truly exhibit these issues I don't see how it could just work again and again. It really shouldn't. But it does.

Btw, frodus, do you or have you sold BMS, or are you planning to sell BMS?

BMS VS NO BMS is not my debate...

I'm staying that You have no idea if your cells are drifting or self discharging because you haven't tested it, And neither have most people. So to say that your cells do not exhibit self discharge or drift is unsubstantiated. Your think it works one way.... But real life testing shows otherwise.

I could care less if people want to use a bms or not. As long as you keep an eye on them, who cares. Someone or something is taking care of that pack. I think we agree there.

It doesn't negate the fact that a cell will have a certain measurable self discharge of a drift, regardless of whether you use a bms or not!

And I'll be honest.... Yes I do sell one bms only as a package with GBS batteries because they'll warranty the entire package that way, and it's integrated with charger. It's best for my customers to have a warranty on cells IMHO. I don't push bms. I've sold many cells without bms at all! It's up to the user and I support them either way. It's all on them to take care of the cells. Most of my business however, is motors and controllers, chargers and Dc-dc converters. I think a good chunk of people on this forum have bought from me, and not once did I push a bms on them, if that's what your getting at.....

But since you don't know me.... I also spent 2+ years doing tech support... in my spare time... for Elithion and Manzanita Micro. I've built my own bms for lead and lithium, active, passive, power shuttling, capacitive and monitoring only. I've owned an Elithion Pro and helped Davide with some of the features.... and now I own an Empulse R. I spent $3k on battery testing equipment.... Because no one else would do these tests aside from a manufacturer and I thought it was fun! I built an 800A discharger to monitor a single cell and logged in on the computer.... I've since sold it all... But I've seen what cells do in many weird situations and measured discharge, capacity and IR on cells from all over the place.

podolefsky
25 February 2014, 1124
To try and tease apart the issue of balancing from monitoring, I started a new thread here (http://www.elmoto.net/showthread.php?3455-Why-you-need-cell-level-monitoring).

picaroon
25 February 2014, 1442
I've often heard it’s a good idea to parallel all the cells and charge them to ~3.6v (or whatever you fancy for whatever your chemistry) before someone builds a battery pack which they will keep top balanced. Now I can understand this as all the voltages are equal if they’re connected in parallel.
Whenever I read about bottom balancing, the cells have all been discharged to the same voltage individually, to me it always sounds easier and more accurate to parallel them all and discharge them in one hit but I've never read of this being done.
So what’s up with my way of thinking? Why isn't this how you would bottom balance cells?

chipper6
25 February 2014, 1506
There are people doing car sized packs who paralleled 50 calbs. NewElectric of Amsterdam and a couple guys on YouTube. But not much for good publish detail on it.

Sent from my HTC One using Tapatalk

podolefsky
25 February 2014, 1610
I'm doing my own little bottom balance experiment, and I'm doing bulk parallel discharge. First at C/10 to 2.8V, then at C/30. That C/30 lets them stay pretty close to open circuit, left them a hair over 2.8V.

jsantala
26 February 2014, 0005
I'd have to think that actually using the cells in your vehicle is closer to real life than testing in a lab.

podolefsky
26 February 2014, 0016
I'd have to think that actually using the cells in your vehicle is closer to real life than testing in a lab.

You can learn a lot from both.

chipper6
26 February 2014, 0039
I follow this guy on youtube (mostly for the build aspect) but you can learn from everyone. He actively top balances and developed his own BMS. http://www.youtube.com/user/tgamber?feature=watch

He just started a series talking about his top balancing scheme : http://www.youtube.com/watch?v=AdLVk-KVcL0

He's probably on the forum but thought I would ad this to the pot.

frodus
26 February 2014, 0956
I'd have to think that actually using the cells in your vehicle is closer to real life than testing in a lab.

I'm confused why you you try to discount me because you may think I'm trying to sell bms... in reality I could care less if I sell them.

Then you think because I've tested in the lab that the results are somehow less valuable than your unmeasured estimates.

I Respectfully disagree.

I've been doing this EV thing since the time this forum was created in July 2007 and was one of the first 5 members. My evfr blog has been online since August 2007. I started a business not to sell things, but to help people get parts for cheap. It morphed into consulting and building prototypes. I've worked with several companies on battery systems (synkromotive, Brooklyn motorized, motoczysz , ev components, Elithion, Manzanita micro, pure watercraft, am racing), I've built my own EV motorcycle, helped with an ev race bike and consulted with scores of people. I've seen real life pack characteristics.

But after seeing some of these characteristics... I did the same sorts of things in the lab. High discharge rates, cycling, drive cycles of varying currents, DC IR testing, fast charge, slow charge.... And I've seen these characteristics in a lab settng.

Self discharge and drift do exist on all cells to varying degrees. That is my direct experience with many different cells of different chemistries.

Anyway..... I'm going to agree to disagree with you.

Your pack should be fine because you're keeping an eye on it. Bottom or top balancing is fine... Just make sure you're managing cells... Either yourself or with a monitoring system.

Go have fun, ride and create.

podolefsky
26 February 2014, 1807
I thought it would be useful to have a picture of what a bottom balanced pack looks like charging and discharging. I bottom balanced 6 LiFePO4 cells down to 2.8V. I paralleled them, discharged at C/5, then C/30 to 2.78V. When the charge stopped they popped up to 2.81V each.

I then connected them in series with taps monitoring each cell. I charged at C/5 until the first one hit 3.65V. Then let them sit about 30 min, then discharged at C/5 back down to 16.8V (6 x 2.8).

Graph below shows the results (PDF link is a higher res version). You can see they all start together and stay within about 0.01V through most of the charge. Then near the end, cell 1 is close to full and it shoots up very steeply (almost a vertical line). The pack is balanced, but clearly the cells are not very well matched.

On discharge, they all start very close, about 3.35V. Note that from the charge curve, we know that cell 1 is at a higher SOC than the others, but at rest all the voltages are nearly identical (within 0.01V). In fact, since cell 1 filled up first, we also know that it was at a higher SOC for most of the charge and discharge, even though all the voltages differed by less than 0.01V nearly the entire time. The voltages are also flat through big parts of the curves, again telling you that voltage isn't changing even though SOC is.

They discharge, again staying pretty well together, down to about 2.8V each at the bottom. Actually, not exactly 2.8V each. The table below shows the starting, charged, and ending voltages on each cell (during charge/discharge, not resting). The end voltages are close, but the max delta is 0.047V. That's not insignificant, but it's under load (although only C/5).

**BIG DISCLAIMER** These are somewhat old cells that I had sitting around. It illustrates the charge/discharge process. It doesn't mean that new, good condition cells will be this mismatched. What you're seeing here would happen with any pack, it's just magnified, which helps show the main effects. I'm also not saying in any way that bottom balancing is bad - I just want to show how it works. It does appear to work, even with fairly unmatched cells.

http://www.colorado.edu/physics/EducationIssues/podolefsky/EV_project/bottom_balance.png

PDF version (http://www.colorado.edu/physics/EducationIssues/podolefsky/EV_project/bottom_balance.pdf)




Cell 1
Cell 2
Cell 3
Cell 4
Cell 5
Cell 6
Delta
Total V


Start V
2.811
2.815
2.811
2.810
2.810
2.808
0.007
16.865


Charged V
3.651
3.417
3.469
3.440
3.419
3.428
0.234
20.824


End V
2.821
2.825
2.816
2.828
2.800
2.781
0.047
16.871

robo
26 February 2014, 2057
Nice data Noah. Your graph also clearly demonstrates that voltage has almost no perceptible change while draining through over ~30% of SOC. And, this is even with mismatched cells of differing capacity.

So, measuring cell (or pack) voltage anywhere in that flat spot would tell us nothing about SOC of any cell. The voltage would be the same ('cause the voltage line is, uh, flat), but SOC could be anywhere along that flat part of the curve(s). This is why measuring/monitoring present cell capacity (not voltage) is the only way to know cell SOC.

Different cell capacities will only show significant voltage differences at the ends of the curve, not in the middle. But, those changes at the ends occur pretty quickly and without much warning, depending on the c-rate.

Bottom-balanced cells will show the cell differences at the top of the charge curve. Top balanced cells will show the differences at the bottom of the discharge curve.

Either balancing technique just moves the risk from one side of the curve to the other. Both techniques still require some sort of keeping an eye out or some technique- bms or not.

I know a bunch of you all know this stuff, but sometimes it's useful for me to think things out loud and to write them down.

podolefsky
26 February 2014, 2124
Sounds like a good summary to me, Rob.

A couple caveats - this is for LiFePO4. Other chemistries have different curves. LiCo (like hobby lipo) and NMC (like Kokam) are more "slopey", as they say. You can get some idea that SOC is going down with voltage, but it's non-linear so it's not a direct relationship. You'd need something like a lookup table, so it would be a pain, and probably not very reliable as the cells vary with load, like in a vehicle. If you know your pack, you might be able to get some intuition about where you're at (basically you have a lookup table in your head).

The two places where you can get SOC with LiFePO4 are at the ends. You know when the cell is full and empty at the voltage extremes, say 3.65 and 2.5, but only during charge or discharge. When you stop driving current either direction, they come back to rest voltage and cells of different SOC will all be very close.

robo
26 February 2014, 2333
Good point(s) Noah. Explicitly saying LiFePo4 is a good thing, although your data also indicates that. I probably wasn't clear there.

I appreciate the stuff I get to learn here. I still thank BaldBruce for saving my cells from my own ignorance when I did my first parallel balancing charge.

Also, I should have explicitly said that if a cell is really jacked up with severely reduced capacity, its "end of the curve" may show up in the "middle of the curve" for the healthy, higher-capacity cells. All looks good voltage-wise and it seems like plenty of SOC (capacity) to go...and then that cell commits suicide. That's another case where capacity measurement is useful.

Fun stuff!

podolefsky
05 March 2014, 2202
This has been rattling around in my head. It's a technical question - if you're bottom balancing and not charging to a standard voltage, what do you use for a charger?

Reading back over, I see jsantala and Toni are using a 24s charger with 25s pack, so they get about 3.5V/cell. That makes sense. But what if you want to use standard numbers, 24s, 32s, etc? Or if you don't want exactly 3.5V/cell for some reason? What do you use? I suppose a variable power supply, or maybe a specially programmed charger? Or do you monitor and just cut off the charge?

In my little experiment, I watched the cells and stopped when one hit 3.65V, which was at pack voltage ~20.8V. I'm not sure how I would repeat this without manually stopping the charge. I'm genuinely intrigued by bottom balancing so this question has been nagging at me.

jsantala
05 March 2014, 2224
podolefsky, you can match your charger to your pack or your pack to your charger, which ever is more convenient. You can get a charger with any charge voltage you like, for example from http://www.evassemble.com - or if you have a CAN- or otherwise controlled charger you can obviously use one of those too.

Or, you can indeed do as you suggested and monitor pack voltage while charging with something that has relays, like this JLD404:

http://store.evtv.me/proddetail.php?prod=JLD404AH (with shunt)
http://www.lightobject.com/Programmable-Digital-AH-meter-red-led-Ideal-for-battery-monitoring-P870.aspx (cheaper, no shunt)

That model can also count Ah if you have a shunt. Jack has written a manual for it: http://media3.evtv.me/JLD404AH.pdf

I would however recommend using a charger that does stop charging at some reasonable point by itself, so that if your programmable meter breaks you won't overcharge the cells too bad. Also note that if you use a meter to cut off the charge then you lose the tapering or constant voltage (CV) phase of the charge, so the cells won't get as full.

However, the most important factor in the longevity of your cells (besides not charging while frozen) seems to be the time spent charging and how full you charge them. So if you just quick charge as fast as you can and stop when you reach 3.5 on average, you're putting the least amount of strain on your cells. You may not get them quite full, but you may increase cycle count by an order of magnitude, so in the long run you may actually get more out of your cells than if you had always charged to 100% SOC.

Edit: Here's a lecture on Lithium battery life http://youtu.be/pxP0Cu00sZs

Hugues
05 March 2014, 2322
This has been rattling around in my head. It's a technical question - if you're bottom balancing and not charging to a standard voltage, what do you use for a charger?

Reading back over, I see jsantala and Toni are using a 24s charger with 25s pack, so they get about 3.5V/cell. That makes sense. But what if you want to use standard numbers, 24s, 32s, etc? Or if you don't want exactly 3.5V/cell for some reason? What do you use? I suppose a variable power supply, or maybe a specially programmed charger? Or do you monitor and just cut off the charge?

In my little experiment, I watched the cells and stopped when one hit 3.65V, which was at pack voltage ~20.8V. I'm not sure how I would repeat this without manually stopping the charge. I'm genuinely intrigued by bottom balancing so this question has been nagging at me.

on my electric bike charger i opened it up and found a little pot to tune down the max voltage. I put it down to 4.0 v / cell (LiCo)

podolefsky
05 March 2014, 2349
Thanks, that makes sense. I'm definitely aware of chargers like Elcon that can be programmed or controlled with CAN (using a PFC5000 with CAN in a project, actually). I was mostly wondering what most people did - seems like whatever works. Cool.

I agree it's good to have a redundant system.

I watched that Jeff Dahn lecture before, very interesting. At first I was confused, because when he said slow charging killed cells that seemed counter to what I've read elsewhere (like here (http://www.mpoweruk.com/life.htm), under "charging rate"). You generally read that high charge / discharge rates reduce cycle life. I just watched the beginning again and I think I get it now - Dahn seems to be saying something very specific, that slow charging at high temperature kills cells. Reason being that at high temperature, you get parasitic reactions that you don't get at lower temps. I believe that at lower temperatures the opposite is true - lower charge rates give you longer cycle life. At least, at the extreme it's true - if you charge at a C-rate of 100 you will kill most cells, at any temperature.

I could be wrong, but that's how I reconcile the two seemingly opposite claims about how charge rate affects cycle life.

jsantala
06 March 2014, 0156
It's just that prolonged charging causes heat within the cell and thus is not a good idea. If I recall correctly, what they did is measure how much energy was lost during charge-discharge cycle and more was lost in longer charging so more more of it turned into heat and heat is bad. High temperature of course makes it worse the hotter you get.

I think this new research pretty much debunks the old myths regarding slow vs fast charge.

Fast charging doesn't seem to have any of the problems they traditionally claim it has. It must be either lead legacy or applies to some of the old cells, but personally I'd charge as fast as I can. In my opinion, producing more current becomes a problem much sooner than it's a problem for the cells to absorb it. It's in any case much slower than discharge, hundreds of amps versus tens of amps.

What those Leaf people with dimished capacity were apparently doing was slow charging overnight at home and then slow charging at work all day, all the while in hot climate. That's just about the worst it could get for the cells.

podolefsky
06 March 2014, 0918
Do you know of research that debunks it at normal temperatures, like aroud 30C? Seemed like Dahn was just talking about 60C temps, but I might have missed something.

jsantala
06 March 2014, 1040
Do you know of research that debunks it at normal temperatures, like aroud 30C? Seemed like Dahn was just talking about 60C temps, but I might have missed something.

You're right, he was talking about slow vs fast charge at high temps (55-60'C), but the point is that even then, faster is better, so there's probably no harm in it at lower temperatures either. And who wants to charge slow? I know I always want to charge as fast as I can, even if I don't have to.

chipper6
06 March 2014, 1044
He was very much speaking about an isolated situation of high temps. At the end he said something along the lines of "you want to keep the batteries a cold as possible" without putting a limit on his statment. As I understand more severe damage can occure if you charge at too low a temperature.

Sent from my HTC One using Tapatalk

jsantala
06 March 2014, 1050
Yeah, there's no doubt that charging below freezing is a bad idea and does hurt your cells.

podolefsky
06 March 2014, 1105
My understanding is that you want to keep cells in the Goldilocks zone - not too hot, not too cold. Right where the reactions that drive charge/discharge work most efficiently without overheating or plating.

Most lithium cells should not be charged below freezing, due to plating. But some have additives that allow them to be charged below freezing, some I know of as low as -20 deg C (at low C-rates).

podolefsky
08 March 2014, 1536
Here's a question for a sunny Saturday afternoon (here in Colorado anyway).

If slow charging is bad, how fast is fast enough? And is it practical?

In that video, Dahn compared charging at 1.5C, which was good for cells, to C/56, which was bad (and also REALLY slow - 2.5 days...)

Most cells can take 1.5C charge. The question is whether you can actually do it. My bike has roughly 4.6 kWh. 1.5C is 7000W. That's a 220V circuit pulling over 30A. Most people would need to install a special circuit breaker for it. If I just want to use a standard 110V, 15A outlet I'm stuck with 1500W at most.

The Leaf battery he talks about is 24 kWh. 1.5C on that is 36 kW. It's very unlikely anyone would be able to pull 36 kW in their house. Even most shops don't have that kind of power. The Leaf charger is 6.6 kW (which happens to be 220V at 30A). That's roughly C/4. Actually up from their older 3.6 kW charger, about C/6. He said Leaf owners were killing their packs in the heat, so can I assume C/6 isn't fast enough? Maybe not even C/4?

Now think about the Model S, with 84 kWh. Tesla's standard charger is 10 kW, which is a little under C/8. Their twin charger is 20 kW, C/4. To do 1.5C, you would need 126 kW. That's just never going to happen outside of a supercharging station.

So what I'm left with is hoping that for the most part, charging at C/8 to C/4 is pretty OK. Otherwise, the EV industry has a serious problem.

My ray of hope is from a recent article in Charged, where they said that the Tesla Roadster batteries were exceeding their expected cycle life. Since no Roadster owner has a 75 kW charger, or likely even a 25 kW charger, that's seems to bode well.

chipper6
08 March 2014, 1541
Another note to add: Tesla has active temperature control on their pack. I think the early leafs don't. I'm on my ice moto now or else I would verify.

Sent from my HTC One using Tapatalk

podolefsky
08 March 2014, 1602
Posting from your bike? I hope you pulled over ;)

Yeah, Tesla has active thermal management. Liquid cooling (but the Roadster has an air cooled motor - go figure). Probably it keeps the cells around room temperature charging, but that's just a guess.

Early Leaf's had air cooled packs. I think the new ones too.

http://www.plugincars.com/no-active-thermal-management-did-nissan-make-right-call.html

Severe battery degradation only seems to be a southwest thing, evidently Leafs (Leaves?) in cooler climates do just fine.

http://insideevs.com/tesla-roadster-battery-study-proves-that-nissan-leaf-battery-is-long-lasting-too/

chipper6
08 March 2014, 1643
Leavi... Ya, I think Dahn was saying that people in the SW (especially Arizona) who charge in 100+F weather and with 110V are doing the most damage without knowing it.

Sent from my HTC One using Tapatalk