Charge Shuffling for Li cells

How do you store and manage your electricity?
Nevilleh
Senior Member
Posts: 773
Joined: Thu, 15 Jan 2009, 18:09
Real Name: Neville Harlick
Location: Tauranga NZ

Charge Shuffling for Li cells

Post by Nevilleh »

Now that my Low Cost BMS is doing its thing, I've been thinking more about the cell balancing side of things and I had a look at Lee Hart's Battery Balancer, intended for 12V PbA batteries. I'm thinking that a similar principle might work for my Li cells too.
Basically, his scheme uses a 12V charger powered either by the ac supply or by the whole battery pack and able to be relay switched to any of the batteries in the stack. A micro measures the voltages and switches the charger to the lowest for a time proportional to the difference in voltage between it and the average value. Needs a relay per battery and he used 30A automotive ones.
With a Li battery such as my 45 cell one, that would need 45 such relays and I'd need a 3.6V constant current charger able to be driven by the full battery voltage. If I worked on say a 15A charge, that would require something like 54 watts of output, so its not a very big charger at all, say about 65 watts input which would draw less than 1/2A from the 144V battery stack. So I'd be taking 1/2 A from all the cells and putting 15A back into the weakest one.
My present BMS could provide the voltage measurement needed, but I would need to run a separate wire from each cell, capable of carrying 15A, back to the "switch" board. That switching would probably be best done with relays and I haven't thought too much about what it might look like, but I suspect DPST might be needed. The thought of semiconductors is a bit daunting.
This sort of balancing could go on almost all the time.
Any thoughts, comments, criticisms, helpful ideas?

Last edited by Nevilleh on Wed, 07 Dec 2011, 08:40, edited 1 time in total.

User avatar
woody
Senior Member
Posts: 1715
Joined: Sat, 21 Jun 2008, 02:03
Real Name: Anthony Wood
Location: Mt Colah

Charge Shuffling for Li cells

Post by woody »

My thought is: How badly matched are the cells that you need to balance even 1 amp let alone 15 amps while driving? Isn't it better to figure out the problem cell and fix it?
Planned EV: '63 Cortina using AC and LiFePO4 Battery Pack

Tritium_James
Senior Member
Posts: 683
Joined: Wed, 04 Mar 2009, 17:15
Real Name: James Kennedy
Contact:

Charge Shuffling for Li cells

Post by Tritium_James »

Those relays need to be rated to your pack voltage, with a DC rating. They will not be small or low cost. Since I doubt any of this would fit inside the pack, and you're bringing reasonably chunky 15A cables all back to one central point (where they can easily get shorted together/to the chassis/etc), all those wires also need to be fused (with pack voltage DC rated fuses) at the cell ends. Also not cheap.

Do what Woody says and fix the cell with the problem...

Nevilleh
Senior Member
Posts: 773
Joined: Thu, 15 Jan 2009, 18:09
Real Name: Neville Harlick
Location: Tauranga NZ

Charge Shuffling for Li cells

Post by Nevilleh »

Fair comment from both woody and Tritium_James and perhaps I didn't choose a very appropriate name for this topic, should've been "Charge Shuffling".
If I do nothing apart from balancing my cells and all that does is ensure that all cells have at least as much charge stored in them as the weakest one, then my battery pack can deliver only the amount of charge per cell that is in the weakest one. So if I have a stack of nominal 130 AH cells, but one of them is capable of only 120 AH, it doesn't matter whether I top balance, bottom balance or whatever, all I can draw from that battery is 120 AH. Now the whole point of charge shuffling - which both the above seem to have missed - is that I can "shore up" my weak cell by moving charge into it from my stronger ones and so extend the range obtainable from my battery. In a perfect system the available AHs would be the average of all my cells, so in my 45 cell system with 44 at 130AH and 1 at 120 AH, I would get 129.8 AH instead of 120 AH. That would be worthwhile, would it not?

Further, T-J says I would need relays and fuses rated for the full battery voltage, but I can't see why. Any relay in the system is only connected across a single cell, ie no more than 3.6V. Sure, its isolation insulation would have to withstand the full battery voltage, but it never has to switch more than a single cell voltage. Same for the fuses, they would only ever see a single cell voltage between their ends although the holder would need full battery voltage isolation.

But relays are still expensive, the best I can find are about $8 each, so I'm up for $360 just for those. Something to ponder.

User avatar
Johny
Senior Member
Posts: 3729
Joined: Mon, 23 Jun 2008, 16:26
Real Name: John Wright
Location: Melbourne
Contact:

Charge Shuffling for Li cells

Post by Johny »

Nevilleh wrote:Any relay in the system is only connected across a single cell, ie no more than 3.6V.
I like the concept but implementation would be the problem. The fuses have to be there to handle a software runaway (closing 2 relays or more at once) or relay contact "stuck" (they are after all only $8 relays).

User avatar
weber
Site Admin
Posts: 2641
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Charge Shuffling for Li cells

Post by weber »

Nevilleh wrote: Fair comment from both woody and Tritium_James and perhaps I didn't choose a very appropriate name for this topic, should've been "Charge Shuffling".
You can change the topic name by editing the first message.
In a perfect system the available AHs would be the average of all my cells, so in my 45 cell system with 44 at 130AH and 1 at 120 AH, I would get 129.8 AH instead of 120 AH. That would be worthwhile, would it not?
Yes it would.

Way back when Coulomb & I started thinking about this stuff we briefly considered a different method. Although its purpose was primarily cell protection it would also prevent a low-capacity cell from limiting the effective amp hours of the whole battery.

The idea was to use a MOSFET half-bridge on every cell as a changover switch, so a cell could remove itself from the string and completely bypass itself. This was to be controlled locally from a cell-top board and powered only from the cell itself so there were no issues with fusing or isolation. It would be the ultimate in autonomous cell self-protection. A cell would remove itself from the string if it was being distressed in any way (overtemperature, overvoltage, undervoltage). Some inductance was to be provided between the battery and the rest of the world, to deal with the sudden voltage jumps.

But it simply proved too expensive.
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).

User avatar
woody
Senior Member
Posts: 1715
Joined: Sat, 21 Jun 2008, 02:03
Real Name: Anthony Wood
Location: Mt Colah

Charge Shuffling for Li cells

Post by woody »

What about a cell can opt out with a "Stop the bus I want to get off" message.
Next time the car is stopped with the brakes on, main contactor opens, the cell bypasses itself and then the main contactor closes, then you're on your way...

This would need a cheaper per cell switch right?

Bush mechanic fix would be put a 10AH headway in parallel with your dud cell...
Planned EV: '63 Cortina using AC and LiFePO4 Battery Pack

Tritium_James
Senior Member
Posts: 683
Joined: Wed, 04 Mar 2009, 17:15
Real Name: James Kennedy
Contact:

Charge Shuffling for Li cells

Post by Tritium_James »

Woody/Weber - the system with two FETs on each cell (one to disconnect the cell, the other to shunt the connection past where the cell used to be in the string) is certainly possible, but like Weber pointed out, it's not cheap.

But even if you don't care about the cost, you've now gone and stuck #CELLS worth of MOSFETs in series with your pack, one per cell. Since the FETs have to be rated for the full pack voltage, they're not going to be fantastic on-resistance, and you've probably just multiplied your pack's impedance by 10 or 20x. Not any good in a car where you need high peak power.

Nevilleh
Senior Member
Posts: 773
Joined: Thu, 15 Jan 2009, 18:09
Real Name: Neville Harlick
Location: Tauranga NZ

Charge Shuffling for Li cells

Post by Nevilleh »

The problem with switching a cell out of the chain is that the switches have to carry the full motor current and withstand the whole battery voltage which makes them expensive plus resistive losses become prohibitive.
The idea of charge shuffling means you can simply bring a low cell up and keep on doing it until all cells are exhausted. In principle its a great idea but the implementation is not so easy. Relays provide a fairly straightforward way to do it, but they are electromechanical devices and as such they wear out, can stick, and so on, so making it fail safe becomes pretty important. One could use MOSFETs as switches and you would need high voltage ones in that case, but they'd only need to carry the "shuffle" current of say 15A or so. I'm not sure how you'd do the gate drive, probably need to opt-couple it and then you'd need devices that can turn on with only the cell voltage as drive.
Another thought is a small, high frequency transformer and rectifier across each cell. The primary could be driven by a suitable source when one wanted to put some charge into a cell. That would certainly provide all the isolation needed! I guess the "charger" would be something powered by the full battery voltage and producing an output at say 50 kHz or so and maybe 36V so the current in the primary would only be 1.5A or so. Each cell's transformer would be a step down and it could be mounted right at the cell minimising the wire that carries 15A.

User avatar
weber
Site Admin
Posts: 2641
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Charge Shuffling for Li cells

Post by weber »

Tritium_James wrote:But even if you don't care about the cost, you've now gone and stuck #CELLS worth of MOSFETs in series with your pack, one per cell. Since the FETs have to be rated for the full pack voltage, they're not going to be fantastic on-resistance, and you've probably just multiplied your pack's impedance by 10 or 20x. Not any good in a car where you need high peak power.

Why do they need to be rated for the full pack voltage? Because I certainly agree there is no way they can have low enough on resistance unless they are 60 V devices or less.
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).

Tritium_James
Senior Member
Posts: 683
Joined: Wed, 04 Mar 2009, 17:15
Real Name: James Kennedy
Contact:

Charge Shuffling for Li cells

Post by Tritium_James »

If you can guarantee that there's never any current flowing when you do the switch then they can be lower voltage. But this means you've pretty much got to open the main contactors any time you want to chop out a cell.

User avatar
weber
Site Admin
Posts: 2641
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Charge Shuffling for Li cells

Post by weber »

Sorry if this is a bit off-topic Neville. And BTW, I'm sure there's another thread on this forum where we've discussed charge shuffling before.
Tritium_James wrote: If you can guarantee that there's never any current flowing when you do the switch then they can be lower voltage. But this means you've pretty much got to open the main contactors any time you want to chop out a cell.

It would be best to delay switching until the current is low, but it doesn't need to be zero. Perhaps you're not aware that the MOSFETs can be arranged so that the body diodes will carry the current during deadtime (and thereby limit the voltage across the open devices) irrespective of whether we're in drive or regen (as in the following diagram), or perhaps I'm missing something else.

I mentioned previously that the two MOSFETs form a standard half-bridge. The cell is across its DC bus.

Image
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).

User avatar
weber
Site Admin
Posts: 2641
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Charge Shuffling for Li cells

Post by weber »

Here's where we discussed charge shuffling before.
viewtopic.php?t=900&p=25588#p25588
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).

Nevilleh
Senior Member
Posts: 773
Joined: Thu, 15 Jan 2009, 18:09
Real Name: Neville Harlick
Location: Tauranga NZ

Charge Shuffling for Li cells

Post by Nevilleh »

Sorry, I had forgotten about the previous topic/discussion. But nothing came of it and it is something that would repay deeper investigation I think. Particularly the inductive thing - I dug it out of my "things waiting for further attention" bin (jumbo!) after reading those posts and I might put a bit of effort into looking at it again.
Last edited by Nevilleh on Thu, 08 Dec 2011, 00:43, edited 1 time in total.

User avatar
PlanB
Senior Member
Posts: 465
Joined: Sat, 16 Jan 2010, 15:24
Real Name: Kris McLean
Location: Freemans reach
Contact:

Charge Shuffling for Li cells

Post by PlanB »

Does paralleling batteries reduce cell variation? Say for example I had 100 cells with a mean variation of 5% in capacity wired is series, then the likely variation from one cell to the next would be 5%. But if I wired pairs of cells in parallel then connected those parallel combinations in series would the percentage variation amongst the parallel pairs be any less?

antiscab
Senior Member
Posts: 2499
Joined: Mon, 26 Nov 2007, 05:39
Real Name: Matthew Lacey
Location: Perth, WA

Charge Shuffling for Li cells

Post by antiscab »

if you don't measure the capacity of the cells and match them, parralleling keeps the variation the same (2 x cells both 5% lower parralleled = 5% lower total)

if you do match them, than yes you can reduce the variation

Matt
Matt
2017 Renault zoe - 25'000km
2007 vectrix - 156'000km
1998 prius - needs Batt
1999 Prius - needs batt
2000 prius - has 200 x headway 38120 cells

Nevilleh
Senior Member
Posts: 773
Joined: Thu, 15 Jan 2009, 18:09
Real Name: Neville Harlick
Location: Tauranga NZ

Charge Shuffling for Li cells

Post by Nevilleh »

I'm pretty sure that I have convinced myself that charge shuffling in any shape or form is not worth the expense.
TI's "PowerPump" technique is all right, but there is the cost of a control chip, an inductor, some caps and R's and a couple of mosfets pretty much for each cell. On top of that, the current transfer isn't vey big.
Lee Hart's scheme of using relays to connect a decent charger to each cell as needed is pretty good, but costly to implement. You could do it with mosfets instead of relays, but you need fets capable of withstanding the entire battery voltage and the on res of those is relatively high and so lossy. Probably end up costing about the same as a relay.
Using a high frequency transformer for each cell is not so cheap either. But then I thought that perhaps you could make a single, 3.6V charger for each cell, capable of 15A or so and either powered by the battery pack for balancing or by 230V mains for charging. That would mean saving the cost of a charger which might pay for the lots of little chargers and achieve balancing - charge shuffling actually - as a bonus. That's probably the only method that might make economic sense. I wonder how small and for how much I could make a single cell charger?

User avatar
PlanB
Senior Member
Posts: 465
Joined: Sat, 16 Jan 2010, 15:24
Real Name: Kris McLean
Location: Freemans reach
Contact:

Charge Shuffling for Li cells

Post by PlanB »

That's a sweet idea in some respects. I got some questions....
1)Would it be possible to get enough isolation between mini chargers to allow for 650v packs?
2)Does it make sense to do sets of cells rather than individuals? A 650v pack has 200 cells in series which is a lot of chargers. If you did pairs of cells that would cut the number in half. I've noticed that laptop battery packs run up to 3 cells in series without a balancer.
3) Is it possible to both run the chargers from both the pack & the mains at different times?

Nevilleh
Senior Member
Posts: 773
Joined: Thu, 15 Jan 2009, 18:09
Real Name: Neville Harlick
Location: Tauranga NZ

Charge Shuffling for Li cells

Post by Nevilleh »

Yes, isolation should not be a problem, its something that is built into the transformer.
Hadn't thought of doing cells in pairs, but maybe that's a possibility especially if you match the cells at the start. My thinking is that these chargers would be controlled by my bms anyway and hat should be reporting on all of the cells. Mind you, I think my present micro runs out of ram at about 150 cells.
I've had a brief look at a suitable topology and a simple forward converter would be best - and cheapest - I think. I have done an initial design that will work from 120v dc right up to 400v dc and supply 3.60V at 15A. I'll cost it out before doing anything else and I hope it could be made small enough to fit right on top of a battery cell. The idea being hat you would run a pair of wires around the place capable of carrying either the dc battery voltage or rectified ac - user selects which one depending on whether you are in balancing or charging modes. This means that running from 650V is not on, you would need a double system for that. Also, running 45 cells as I do at present, each one needs about 54 watts when charging and that means about 70 watts of input for a total of 3150 watts which is a bit much for a 230v single phase supply. That means the bus supplying the chargers would have to carry something like 10 amps which is also a bit much.
It all gets very complicated!
I suppose that if one starts at the 230V end and says 10A is the limit, so 2300 watts is divided among the cells for 51 watts each (for me anyway). Multiply that by 80% efficiency and that means my max charging current is 11.3 A.
How much is a charger that will do that current at 160-odd volts, because that's what we would be competing with although you might pay a bit more to do such terrific balancing! I think my present charger, which does 15A at 162V off a single-phase supply (its very efficient, about 92%) sells for about $1500, so I have to be able to make a single-cell charger for about $35 - which is quite possible.
Interesting thoughts.
Last edited by Nevilleh on Sat, 10 Dec 2011, 06:20, edited 1 time in total.

User avatar
weber
Site Admin
Posts: 2641
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Charge Shuffling for Li cells

Post by weber »

Neville, your cells are in "buddy triples". And they are Sky Energy which (in my experience) come with a list of the factory-measured capacity of every (serial numbered) cell. When you tripled them up, did you make some attempt to find the optimum triples to maximise the available capacity of the whole battery?

As well as getting the list of capacities and internal impedances on paper with the cells, they were emailed to us as a spreadsheet. Did you get that too?

A simple heuristic might be to calculate the average capacity, then arrange the cell records in order of capacity and combine the highest with the lowest and then combine those with the other one that will bring their total capacity closest to 3 times the average. Remove those 3 from the list and repeat.
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).

Nevilleh
Senior Member
Posts: 773
Joined: Thu, 15 Jan 2009, 18:09
Real Name: Neville Harlick
Location: Tauranga NZ

Charge Shuffling for Li cells

Post by Nevilleh »

weber wrote:

As well as getting the list of capacities and internal impedances on paper with the cells, they were emailed to us as a spreadsheet. Did you get that too?


They weren't that generous with mine, all I received were printed sheets of the cell impedances, but no capacities. Nothing at all came by email. Not knowing any better at the time, I matched up the triples by impedance.

Their impedance measurement was done by some particular instrument, I forget exactly what, but they did describe it in detail. However, that measurement was quite different from the actual cell internal resistance measured by me dc-wise, that simply put a 100A load current on the cell and measured the voltage drop. Probably because my measurement would vary a bit depending on the state of charge and theirs is an ac one.

I couldn't be bothered doing that for more than one cell though and I also measured the capacity of that cell with my 100A load as being 46.6 AHrs, discharging to 2.5V. This is going back nearly 2 years and I didn't have much appreciation of the "care and handling" of Li cells.

What I'm finding with the bms is that one triple (number 9 for what its worth) is consistently lower than the others and shows a significantly larger voltage drop under load. I will check its strap connections before doing anything else though. It is, of course, one of the least accessible cells in the whole battery pack and I have to get motivated to unbolt some retaining brackets and slide the box rearwards to get at it, plus undo about a dozen screws holding the plastic cover in place.

I tried making a video of the bms display while driving but was defeated by the big increase in light level, once I left the garage, causing the camera to stop down to the point where the bms lcd was too dark to read! I didn't discover this until I re-played the video which thus turned out to be quite useless.

User avatar
Johny
Senior Member
Posts: 3729
Joined: Mon, 23 Jun 2008, 16:26
Real Name: John Wright
Location: Melbourne
Contact:

Charge Shuffling for Li cells

Post by Johny »

The charge threshold would have to changed if running in DC BMW mode otherwise you would be trying to bring every cell up to 3.6v while driving under load. Kind of a waste. Otherwise a great idea. The voltage set point would have to.based on total pack voltage I think. Lots to think about.

Edit: typo - changed - not chanel
Last edited by Johny on Mon, 12 Dec 2011, 06:49, edited 1 time in total.

Nevilleh
Senior Member
Posts: 773
Joined: Thu, 15 Jan 2009, 18:09
Real Name: Neville Harlick
Location: Tauranga NZ

Charge Shuffling for Li cells

Post by Nevilleh »

Wellforces (Meanwell) make a LPS-75-3.3 which is a 75 watt, 3.3V @ 15A converter that sells here for $NZ33 in 100+ qty. The output is adjustable +- 10%, so I can set it to 3.6V. Its current limiting is "hiccup" mode which might not be the most suitable for use as a charger, but I think it might be useful for playing around with the idea a bit more, so I've ordered one. A lot simpler, quicker and cheaper than designing and building one myself!
Last edited by Nevilleh on Tue, 13 Dec 2011, 03:50, edited 1 time in total.

User avatar
Richo
Senior Member
Posts: 3737
Joined: Mon, 16 Jun 2008, 00:19
Real Name: Richard
Location: Perth, WA

Charge Shuffling for Li cells

Post by Richo »

Wouldn't a 25W PSU be better for running from mains?
Still hard to justfy the cost for 100 chargers vs 100 BMS + 1 charger.
About the same efficiency either way too.

I looked at it for charging 400+ headway cells with 5W module.
Also looked at 2.4kW AD-DC 5V bus running to 400+ 5W DC/DC on each cell.
5W AC/DC was a bit harder to find at the time.
Neither way added up.
So the short answer is NO but the long answer is YES.
Help prevent road rage - get outta my way!

Nevilleh
Senior Member
Posts: 773
Joined: Thu, 15 Jan 2009, 18:09
Real Name: Neville Harlick
Location: Tauranga NZ

Charge Shuffling for Li cells

Post by Nevilleh »

I dunno, what I am looking at is whether the cost of a "distributed" charger can compete with a single one and you get cell balancing as a bonus. If you have 100 cells, your charger has to be enormous and expensive else you have two of them which is also expensive. Seems to me I can do a distributed charger for a similar price or less.

What I'm thinking of at the moment is a 48V converter powered by either the mains for charging, or the battery for balancing. That means running only a 48V bus around the place which doesn't need permits, certificates, creepage distances and the whole shemozzle involved with higher voltages. Each cell then has a small flyback converter bolted directly to it also incorporating the bms stuff and powered by the 48V supply. Its very much easier to design a flyback converter to produce 3.6V from 48V than it is from 140 or 365 volts! and much cheaper. At the moment, my tentative costing says I can do the celltop bms/converter for around $20 each (cost) and an 1800 watt, 48v converter for about $320. That would give me a 45 cell charger plus high power balancer for a total of $1220 - ball park figures, but it looks quite competitive. Add a second 48V converter at $320 for twice the power. New Zealand dollar figures.
Last edited by Nevilleh on Wed, 14 Dec 2011, 02:01, edited 1 time in total.

Post Reply