 Notice. New forum software under development. It's going to miss a few functions and look a bit ugly for a while, but I'm working on it full time now as the old forum was too unstable. Couple days, all good. If you notice any issues, please contact me. |
Forum Index : Electronics : "balancing" lithium batteries
| Author | Message | ||||
| isochronic Guru  Joined: 21/01/2012 Location: AustraliaPosts: 689 |
I am wondering, is "balancing" lithium batteries sensible ? The logic goes, that slightly discharging some cells in a pack to match voltage to the lowest one or so, is beneficial. (?) That will reduce the pack voltage, and so the recharger will kick in, and attempt to bring the pack up to 100% recharged voltage by pushing current through. So the worst cell will now be getting over 100% charge - which is likely to cause more damage to that cell !! I can see the common sense with aqueous electrolytes - where the side water split/recombine reactions allow overcharge to be used to balance cells - but lithium/nonaqueous electrolyte is way different. Have I missed some logic there ? |
||||
| davef Guru Joined: 14/05/2006 Location: New ZealandPosts: 499 |
If all the cells are reduced to the lower one and then recharged how would the lowest cell be charged any higher than the others? The lowest cell is possibly "low" because it doesn't charge up as quickly (or as efficiently) as the other cells. |
||||
| isochronic Guru Joined: 21/01/2012 Location: AustraliaPosts: 689 |
Say the pack is 100 amphour capacity and has 6 cells abcdef in serial and it has been recharged. So all the cells have 100 amphours on tap. Cell d has a slightly lower voltage, so all the other cells are slightly discharged until the voltages match. So cell d still has 100 amphours charge, the others are at say 98 amphours. The pack is then used and discharges 98 amphours. The charger then kicks in and charges 100 amphours. Most cells are fine but cell d is now 102 amphours charge. (Edit - that is, 100 amphours on top of the remaining 2 amphours ) The problem is the cell voltage is being used as a charge indicator. A slightly off-spec cell may have the rated charge capacity but have a lower voltage. In fact a damaged cell may not charge to the rated voltage at all, quite a dangerous situation. |
||||
Downwind Guru Joined: 09/09/2009 Location: AustraliaPosts: 2333 |
Confuse to weather you are asking a question or stating a result, Your question is valid. If its a statement than its not valid. Pete. Sometimes it just works |
||||
| davef Guru Joined: 14/05/2006 Location: New ZealandPosts: 499 |
I see I was looking at this from a charge voltage point of view rather than a capacity point of view. How does cell voltage relate to the rated or actual charge capacity? What type of charger measures the number of A/H it is putting into the string of batteries? And takes into account the charge efficiency of each cell? |
||||
| isochronic Guru Joined: 21/01/2012 Location: AustraliaPosts: 689 |
The point is, a balancer may be counterproductive with lithium batteries. I think a charge-measuring (coulometric) charger would be a good idea, I am fairly sure some large-capacity charging like electric vehicles etc already does it, but it relies on an assumption of ca 100% charge efficiency which is only really viable with lithium. |
||||
yahoo2 Guru Joined: 05/04/2011 Location: AustraliaPosts: 1166 |
that is the question. Also, why is the lowest capacity cell the last to come up to voltage? or If a cell is so damaged that it starts to heat at a lower voltage, can it still be used in a functional pack? if so , HOW? I'm confused, no wait... maybe I'm not... |
||||
| isochronic Guru Joined: 21/01/2012 Location: AustraliaPosts: 689 |
This is good : http://batteryuniversity.com/learn/article/charging_lithium_ion_batteries |
||||
powerednut Senior Member  Joined: 09/12/2009 Location: AustraliaPosts: 221 |
umm.. no, never. because physics. |
||||
| yahoo2 Guru Joined: 05/04/2011 Location: AustraliaPosts: 1166 |
so that is no saturation charge, no trickle charge, no equalisation charge, in fact almost no time at all at full capacity, no temperature/voltage compensation. And the majority of degradation coming from holding full charge at elevated temperatures (like the older tech nissan leaf with its very average cooling system) Dalhousie uni testing has shown huge increases in lifespan of cells with 4 and 5 electrolyte additives because it keeps a nice stable open structure to the solid electrolyte interface. A LiNiCo cell in a Tesla for example will have an additive blend like this in it, they have been smart by supporting this research from the start. So over time our LiFePo cells without fancy additives will increase resistance and lose capacity. None of that gives us the scenario that you are proposing. I can think of two 1 cells were not evenly charged from day one and were then charged with no BMS and now cell (D) has a plated anode. 2 cells were held at full charge for extended periods and cell (D) got hot. 3 poor quality control at the factory. 4 faulty BMS draining a cell 5 poor connections 6 charger that never switches to float/off because of phantom loads or poor settings. OK  I thought of a couple more, I have to learn to type faster!
the super low resistance and the voltage spike at the almost full stage make it quite easy to get a feel for what voltages we need to stay below 99.9% of the time. My way of thinking about this, is that if we are fitting cell sensors for high and low voltage cutout then we may as well incorporate a balancing mechanism in the build from the get go BUT I would like the option of just letting the battery cycle on their own. wadda-ya reckon? I'm confused, no wait... maybe I'm not... |
||||
| isochronic Guru Joined: 21/01/2012 Location: AustraliaPosts: 689 |
It is a bit ambiguous, instead of "charge efficiency" (I was thinking electric-charge instead of battery-charge) I should have said "amphour efficiency" which is clearer. I am surprised to to see the cells will keep accepting a small charging current, at the highest voltage, this must become overcharge at some point. In any case, a group of cells will always have small voltage and amphour differences, which of course is minimised by matching as far as possible. A simple approach would be to recharge/discharge the pack, using the weakest cell as a guide, and stop charging/discharging when that cell was recharged/discharged, say at arbitrary 95%/5% levels. The pack will have practically all its amp-hour capacity and should last well as the cells will be out of the high-voltage damage zone. Slightly less capacity, sure. In a perfect world each cell would have a cell manager IC that measured amp-hours in and out, and regulated the schedule accordingly...you could then add indiviual cells in a round-robin replacement cycle , kind of expensive to set up of course. Having said all that, I agree, seeing as the cells are slowly degrading over a few years anyway as you say, you may as well get the most actual grunt out of them by pushing them a bit ! I guess it depends on the intended lifespan and use, eg day-to-day power, or as a longer term reserve energy bank |
||||
| yahoo2 Guru Joined: 05/04/2011 Location: AustraliaPosts: 1166 |
Hi Chronic, you really need to watch a big bank in action to appreciate what low resistance means, charging a 400Ah bank at 1C it will still be taking 400 amps at 99% full and 50-250 amps at 99.5% full if you let it and the cell variation is usually TIGHT more like 0.05% to 0.2%. The big dogs (1000Ah cells) are a factor of 10 better again. I think I mentioned that on the LFP demo unit I set up a couple of years ago we have now balanced the bank a grand total of three times. I checked the voltages last month while they were floating @ 3.3375 and there is a variation of 0.009 volts, the variation would be a lot higher @ 3.6 volts but we never get there as we charge to 3.4875 volts and cut to float after 15 seconds?? (i cant remember the exact time setting). Whichever way I slice it, that is going to be less than 1-3 watt variation per cell on a 10 Kw bank after 9 months of cycling with no balancing. its not much 
if I had built a 38 x 80Ah cell bank on a budget instead of 8 x 400Ah I would expect a lot more variation. In my dream scenario I would let it run and datalog right at the end of the charge cycle, when the highest cell is say 0.03 volts above the pack average @ 3.4850 volts then I would perform a manual balance. There may come a point when the tactics need to change because one cell is reaching empty (low voltage disconnect) before the others, perhaps we could use the same idea and datalog and manually balance at the bottom and just float it at the top and squeeze a few more years from a bank. All my small stuff is charged using a PL8 Cellpro Powerlab workstation, in that case every cell is individually charged. I struggle to see a reason why I would make it any more complicated that. P.S. I call "float" with lithium's when no charge is flowing in or out of the bank and 3.4875 volts/cell is around 99.7% full with this bank. I'm confused, no wait... maybe I'm not... |
||||
| isochronic Guru Joined: 21/01/2012 Location: AustraliaPosts: 689 |
Just the thing for a spot of welding  Not !!!
I was thinking more of the smaller scale stuff, where there is much more variation between the cells - and a whole heap of chargers and projects that try to compensate by balancing. IMHO I think it is a mistake. Buying groups of cells at a discount - eg for a small pack - necessarily means pot luck to an extent. I don't think the vendors even sell deliberate sets at the lower end of the market (maybe some do ?) - and the bought set is used regardless unless there is a conspicuous fault. For smaller scales it is probably not cost-effective to do anything else I guess. |
||||
| yahoo2 Guru Joined: 05/04/2011 Location: AustraliaPosts: 1166 |
A bank made from 1000Ah cells wont get off the couch for less than 5C discharge
Just use some Panasonic 18650's when someone asks you say, dont tell anyone, these are actually Tesla cells SSSHHHH! if they mention the endcaps are not the right shape, you need a completely oulandish story ready to bluff your way out. Basil Faulty style. Or you could do some extensive testing on some lithium titanate oxide (LTO) cells and post it here, for the good of the community. 
if you dont, I may be forced to buy my own Christmas present this year....and it wont be socks! I'm confused, no wait... maybe I'm not... |
||||
| isochronic Guru Joined: 21/01/2012 Location: AustraliaPosts: 689 |
Is this a fair impression, of the various types: Li-CobaltOxide - highest power density Li-LiFePhosphate - stability, long life Li-Polymer - light but fragile and expands/contracts, limited life ? What is the advantage of the titanate cells ? they seem to have lower voltage.. Or graphene - when it is something practical (!) |
||||
| yahoo2 Guru Joined: 05/04/2011 Location: AustraliaPosts: 1166 |
titanate has very high discharge and lots of cycles I am thinking of power for a very compact mobile workshop. An upscaled version of this with a small inverter generator. Swap the 12Kva genset for 2Kva and fit my own modular version of Oztules inverter/transformer build. 
LFP cells are probably not long life compared to the better LiCo cells. All this stuff is subjective to how they are used and $$$. I'm confused, no wait... maybe I'm not... |
||||
KarlJ Guru Joined: 19/05/2008 Location: AustraliaPosts: 1178 |
nice! I've built 3 x 1kWh packs with (new but 3 years old) laptop cells at $1.1 each. theory is we build 1.1million packs for 1million laptops and the 100K packs are spares/replacements only problem is laptop goes in the bin when the batts die and they buy a new one = loads of 18650 cells un-used but 3 years old. test every one - this is time consuming but can be sped up by bulk charging ie 40 cells 20A 4.2V (connect via some books and aluminium foil connect chage wires via a couple of neo magnets to foil) gives 0.5C charge or thereabouts. Result! some time + $1.1/cell at ~7-8Wh/cell = rougly 142 cells/kWh and this translates to $160/kwh + time + cell holders + nickel strips + BMS units all in i call it less than $200/ kwh. Problems - no more than 0.5C charge / discharge for unknown cells dont charge more than 4.1V per cell (rarely if ever) as this halves the cycles from 4.15V which halves again to 4.1V What does this mean? rated `400 cycles at 100% DOD and 4.2V full OK full discharge and 90% charge to 4.10V =double cycle life Full discharge and 85% charge to 4.05V =4x rated cycle = 1600 cycles NOT full discharge to 90% DOD cycles double to 80% dod cycles quadrouple so conclusion cheap cells <0.5C charge/discharge = good. not full charge or discharge (85% =4x rated max discharge 80%= 4x rated) Gives rated at 400cycles x 4lower max charge x 4higher min DOD = 6400 cycles in these conditions. TESLA powerwall 2x units of 7S 60P 18650 format batts waranted at 60% original cap remaining at 5000 cycles. Hence A) my maths is horribly wrong or B) powerall aint all its cracked up to be given its ~$8K vs my $1500..... i'll roll the dice and build my own thanks Luck favours the well prepared |
||||
 Print this page |
