Making use of low C rate LiFePo4 cells

I've got a variety of batteries that I want to put to the best use. Low C rate packs aren't much use to me, but I also have a bunch of A123 M1's, which have a higher useful C rate than I require. I'm thinking that I can combine the 2 types in about a 50/50 capacity relationship and they will protect each other.

My thought process is this. Both cell types would be paralleled at the cell level. The low power cells will sag under heavy load, so the A123's would deliver the bulk of the high power demands. When stopped or under light load, the low rate cells would actually recharge the A123's to some extent. It seems to me that since I would never demand continuous high power levels that this would enable me to have high power current for more than 50% of total pack capacity even though the A123's are only 50% of capacity, with the A123's protecting the low rate cells from high discharge rates. It would also enable me to fully utilize the A123's instead of much of their usefulness going to waste if the whole pack was all A123's.

I think the low rate cells would also protect the A123's by preventing their voltage from falling off the cliff. When I'd go past 50% DOD, which is fairly rare, at some point the voltage sag would become quite noticeable as the bulk of the power has to be delivered by the lower power cells. Then just like with my Konion packs I'd know to take it easy, but still have plenty of capacity to get home or a charge spot.

Anyone see any holes in the plan? Should I leave the low power cells connected to each other in series, or should I connect them only to the A123's with beefy parallel connections only at the cell level? ie Each parallel group of A123's cells has its own stand alone low power buddy, recharging the A123s to some extent, enabling them to safely deliver more than 100% of their rated capacity, and supporting the A123's when they're substantially depleted by delivering a lower hobble home current and preventing the A123's from ever falling to a dangerously low voltage.

Let's say I have 20ah of A123's and 20ah of low power LiFePo4's. I could build a 20ah A123 pack that rarely sees more than 10ah of discharge, effectively wasting half of the A123's. I'd also have a 20ah pack of low power cells that I'd rapidly kill with my high power demands. If I split it into 2 mixed packs, I see it as I get 2 high power packs of equal capacity with the only compromise being that on longer rides I really have to watch the power draw below 60% or so DOD. On long rides I typically go easy anyway to extend range making that a non-issue. I also get to use voltage as a fuel gauge (despite voltage being a terrible fuel gauge with Lifepo4). That would include a large reserve tank once voltage sag becomes prominent.

John

didn't someone (was it, dare i say, safe?) do something like this with NiMh and NiCd? seems like it was something about preventing sag iirc. i know, different chemistry, but it seems like the principle might be the same. again, noob here, but can you hook a 48v20a and a 48v10a set of batteries in parallel and get a 48v30a system? likewise, could you hook a 48v20a and a 24v20a in series and get a 72v20a battery? CR, your idea might work really well with like lipo and lead. use the lipo for high discharge, for acceleration, and let the lead sustain you on cruise and trickle charge the lipo. just seems like you could offset a lot of the peukert effect.

I'm not an expert here but i do think this could work, given that the voltage curves are probably quite close.

Instead of having say, a 20C rate pack.. you'd have a 21C rate pack, lol..
The A123 or lifepo4 won't exactly recharge each other, as their voltage cannot differ.
What will happen is that the A123 and lifepo4 will dish out equally what they can put out.

What you create here is a pack with higher amp hours and maybe a fraction less of voltage sag.

ps, i'm not advocating lead, but it seems like the tech development would be in the BMS area, with the possibility of compartmentalizing the batteries, which might lead to a system where any chemistry, any size, could be plug and play. i think this might help with 1) being able to easily increase pack size, and 2) being able to upgrade, piecewise, when new battery tech comes along, so you don't just scrap the whole pack (which represents a large portion of your investment) in favor of the new one, and 3) allow you to easily replace batteries that are subpar or nonfunctional.

John in CR said:

I've got a variety of batteries that I want to put to the best use. Low C rate packs aren't much use to me, but I also have a bunch of A123 M1's, which have a higher useful C rate than I require. I'm thinking that I can combine the 2 types in about a 50/50 capacity relationship and they will protect each other.

My thought process is this. Both cell types would be paralleled at the cell level. The low power cells will sag under heavy load, so the A123's would deliver the bulk of the high power demands. When stopped or under light load, the low rate cells would actually recharge the A123's to some extent. It seems to me that since I would never demand continuous high power levels that this would enable me to have high power current for more than 50% of total pack capacity even though the A123's are only 50% of capacity, with the A123's protecting the low rate cells from high discharge rates. It would also enable me to fully utilize the A123's instead of much of their usefulness going to waste if the whole pack was all A123's.

I think the low rate cells would also protect the A123's by preventing their voltage from falling off the cliff. When I'd go past 50% DOD, which is fairly rare, at some point the voltage sag would become quite noticeable as the bulk of the power has to be delivered by the lower power cells. Then just like with my Konion packs I'd know to take it easy, but still have plenty of capacity to get home or a charge spot.

Anyone see any holes in the plan? Should I leave the low power cells connected to each other in series, or should I connect them only to the A123's with beefy parallel connections only at the cell level? ie Each parallel group of A123's cells has its own stand alone low power buddy, recharging the A123s to some extent, enabling them to safely deliver more than 100% of their rated capacity, and supporting the A123's when they're substantially depleted by delivering a lower hobble home current and preventing the A123's from ever falling to a dangerously low voltage.

Let's say I have 20ah of A123's and 20ah of low power LiFePo4's. I could build a 20ah A123 pack that rarely sees more than 10ah of discharge, effectively wasting half of the A123's. I'd also have a 20ah pack of low power cells that I'd rapidly kill with my high power demands. If I split it into 2 mixed packs, I see it as I get 2 high power packs of equal capacity with the only compromise being that on longer rides I really have to watch the power draw below 60% or so DOD. On long rides I typically go easy anyway to extend range making that a non-issue. I also get to use voltage as a fuel gauge (despite voltage being a terrible fuel gauge with Lifepo4). That would include a large reserve tank once voltage sag becomes prominent.

John

Click to expand...

I am doing the same.

I will have a 36v 14Ah bank of genuine, "known-quantity" A123 M1 cell-man cells on one side of my BMX, and a 36v 14Ah of the ebay A123, which are good cells but not anywhere near as good, on the other side. Both packs would be then paralleled in conjunction with each other to make a 72v 14Ah pack.

My basic thinking was that the M1's would take the sheer brunt of the momentary stress of the 65A-rated system, with the inferior cells contributing "their bit" but rather acting as a "charger" for the M1's

I think LFP used to talk about mixing these cells at cell-level being an option.

I think it would work fine, paralelling the cells together in cell groups. I think you'd need to keep the c rate relativley low though, if there was 50% 2c cells. So you'd have a pack that would last if the c rate was kept pretty close to 2c. Say 4c spikes, but at cruise the discarge rate would not be much over 2c.

As the c rate increased, the weaker cells would still try to put out more, and get damaged if the c rate was real high. The weak cells would not really reach 2c, and then go Ok a123's, you do the rest. They would eventually because of internal resistiance, but by then they'd be not liking it much. So it would work perfect if the controller and motor were sized better for the low c rate cells in the first place. Just guessing at numbers, I would think that 10 ah of a123 paralelled with 10 ah of 2c cells would handle 3c fine, but perhaps not so fine with 5c. If the ratio of A23 cells was higher, then you ought to be able to get away with more, or at least care less about what happens to just a few weak cells.

The other approach would be bms type. One pack of A123's with a bms allowing a c rate it can do, paralelled with another pack that has a bms or some kind of amp limiter preventing it from discharging past it's c rate. Then you might see the bulk of the power provided by the A123's, but still be able to benifit from a few amps from the weak pack. If sized the same, the the weak pack would last longer, since it would provide less of the watts.

Re Mighty volt. Unless I'm misreading or misunderstanding, you are going to murder your weaker pack. 36v paralelled with 36v is 36v. But you say paralelled to 72v? If you are series connecting the two packs, the weak one will see all 65 amps, and go bye bye fairly quick. Perhaps the ebay cells can handle it, perhaps not, but the real a123's won't help any unless paralelled with the weaker cells.

dogman said:

Re Mighty volt. Unless I'm misreading or misunderstanding, you are going to murder your weaker pack. 36v paralelled with 36v is 36v. But you say paralelled to 72v? If you are series connecting the two packs, the weak one will see all 65 amps, and go bye bye fairly quick. Perhaps the ebay cells can handle it, perhaps not, but the real a123's won't help any unless paralelled with the weaker cells.

Click to expand...

Yeah I typed that out wrong. I am tired from staying up till all hours watching the Rugby World Cup.

I will have two 36v 14Ah batteries joined in series.
I think I might also make my "weaker" pack a few Ah bigger.

John, I think you are going to be just fine. In fact I am planning something similar with the Makita cells and Nano packs. I believe what you said is exactly what will happen. Under high demand the powerful cells will keep the voltage up and supply the majority of the power, then when you back off the throttle, the hi power cells will be at a slightly lower voltage than the low power cells. The low power cells will then start to equalize voltage, by charging the hi power cells, and supplying most of the power to the motor.

Mighty volt BAD idea. By putting them in series they become the weak link in the chain. and devalue the hi power cells.

ptd said:

didn't someone (was it, dare i say, safe?) do something like this with NiMh and NiCd? seems like it was something about preventing sag iirc. i know, different chemistry, but it seems like the principle might be the same. again, noob here, but can you hook a 48v20a and a 48v10a set of batteries in parallel and get a 48v30a system? likewise, could you hook a 48v20a and a 24v20a in series and get a 72v20a battery? CR, your idea might work really well with like lipo and lead. use the lipo for high discharge, for acceleration, and let the lead sustain you on cruise and trickle charge the lipo. just seems like you could offset a lot of the peukert effect.

Click to expand...

I believe that is what GGoodrum is doing in his golf cart. He has made the BMS already and is testing it for a combo of Lipo and Lead.
otherDoc

I think it will work for John too. Even if the weaker cells do get a bit of strain on em, it will be shared with the other cells. Just making the pack bigger may be good enough to bring the amps avaliable up to what he needs at cruise.

I don't think the good cells will automaticly pull so much current there is no strain on the weak ones. The weaklings will start squealing and heating up before they get so much resistance they stop putting out current. If it does discharge the A123 side more, then the weaklings will recharge the strong side. There goes any rest for the weak cells.

But I do think that it still will work good enough, even if the weak cells do get warm on the amp spike, and then warmer still trying to recharge the good side. What you will see, is a pack that seems to stay balanced in voltage on both sides. And if it's big enough, you will have some amps to use. Maybe just put a temp sensor on the low c rate cells, and back off if it's really getting hot, like hotter than body temp.

If it fails miserably, what's he lost? Some cells he pretty much has no good use for. So go for it. It shouldn't be able to harm the A123's any.

What I just said about temp might work good for the series connected pack. If it gets all hot, you know it's not working.

ptd,
I am not putting lead on my bike, period. I've got over 10kwh of lithium batteries, and the only things my 2kwh of lead is good for is running my RC chargers/dischargers and storing energy from my solar panels, since I don't have a good solution yet for lithium as my solar batt bank.

mightvolt,
Unless you take those packs apart and rearrange them each to be 72V7ah, then when you parallel them you'll have 36V28ah, not 72V. You can't series them either to be 72V, because the low power pack can't deliver the full current.

Neptronix,

I'm not so sure about cell voltages in parallel always being equal. Once they reach an equilibrium, yes, but not during use. When the low power cells are asked to deliver current, they will sag more, so when I let off the throttle or stop the voltage recovers and be higher so current should flow to the A123s. Whether that really happens doesn't matter that much. During a bit more than 50% of capacity, I should have a nice high power pack, and a low C rate pack in the bottom half of discharge. I say more than 50%, since in the top half the low power cells will deliver a portion of the energy.

Worst case I still have a 20ah pack, just a different C rate in the top half of discharge than the bottom. I also get to more fully utilize the A123's in the pack, so I can use A123's capacity that would be otherwise going to waste in a big pack of only A123's. Plus my A123's should be better protected, because once I start seeing significant sag, then the pack still has significant useful capacity instead of voltage falling off a cliff.

During discharge the transition from a low sag high power pack to a high sag low power pack may be more gradual than I'm thinking, but it will still be pronounce. eg When A123's get down to 3V there's not much capacity left, but even under light loads the low power cells sag down to 3V while they still have lots of capacity.

The sag is so pronounced that I think my 20s Konion packs are actually a better match in parallel with the 23s A123 packs I'm building, though they of course wouldn't be paralleled at the cell level. The Konions would actually provide the power at the very beginning, and then the A123's would take over most of the power delivery with the Konions delivering at the end with sag increasing in the last 1/3rd of total capacity, so I get that voltage fuel gauge.


Dogman,

There's no need to worry about how much current I draw, at least in the 1st 50% of discharge, because the low power cells won't be delivering it. They won't because the A123's will not permit the kind of sag needed for significant current delivery to take place from the low power cells. Once the A123's are mostly discharged, then their voltage will sag enough for the low power cells to deliver a substantial portion of the current. That's when I'll need to be careful about current draw, so I don't kill them prematurely, but I rarely go that deep in discharge and when I do I'm going for range anyway and go easy on the throttle.

I've still got the question out there regarding whether or not to eliminate the series connections of the low power cells to each other. I don't know if eliminating that pathway makes any difference, except maybe variances in their internal resistance might make a difference. I lean toward cutting those connections and parallel to the A123's at the cell level only.

Cells in parallel cannot differ in voltage.
It's like two wheels linked via a solid axle. The two wheels must move together.. if one slips, the other will correct it instantly.

So the lifepo4 will give what it can, and the A123 will give what it can, in equal proportions.

neptronix said:

Cells in parallel cannot differ in voltage.
It's like two wheels linked via a solid axle. The two wheels must move together.. if one slips, the other will correct it instantly.

So the lifepo4 will give what it can, and the A123 will give what it can, in equal proportions.

Click to expand...

So as long as the M1/A123 part of the pack can supply the peak amperage, the LiFepo4 won't be too distressed?

My plan is to have a 14Ah system and only ever use 10Ah.

Theoretically, the repeated high-discharges of the A123 should induce the Peukert effect sooner rather than later, but if I build a large pack and stay inside a limit, I can always stay ahead of the game, no?

Yeah, I see what you mean, the pack should be perky at first for sure. At worst, it would behave like the same size A123 pack, without any other weak cells paralelled to it for awhile. But I also saw some of what neptronix is talking about happen when I paralelled lead and nicad. I couldn't see much difference between how the bike acted when I paralelled lead with lead, or lead with nicad. Both ways got less hot, and had less sag than when by themselves. But try as I might, I never caught one pack at a lower voltage than the other. Any difference was evened up dang quick. But in that case, different chemistries with more similar c rates. I think any difference in voltage will vanish about as fast as you roll the throttle back.

I think the packs will stay very equal in voltage the whole time, but as John says, it won't perform as good when part discharged as when full. But I think the reason why will primarly be from both packs sagging once the pack is partly discharged. I think it will be hard on the weak cells, exactly like having a runt cell in a paralell string of identical cells is hard on that cell. The weaker cells will sag more, and produce less wh while the throttle is on. But they won't save themselves. They'll kill themselves trying to keep up. They'll see the rest of the amps flow when the spike is over as they recharge the better cells, so the spike they see will be longer. The weak cells won't be protected by thier lower c rate from seeing a high amp draw. They will be protected from high c rate primarily from being used in a larger size pack.

Again though, so what? Cells sitting around with no good use. May as well play with em and see. It should definitely work if you keep the amps down to a reasonable level. Something like 3-4c rate on the whole pack should be no problem. Easy enough to just back off the throttle some after about 50% dod.

On the other hand, here I go again, not knowing shit, going bla bla bla.

Do you have some little RC wattmeters? It would be really interesting to see what they said, if you built the pack, or even just a 1s version of it. The put a wattmeter on each type of cells before paralelling them. Then put the load on and see what really happens. Then you'd really know what to expect when you build the whole pack. Data from just a handfull of cells should scale up to the full pack the same way.

The Mighty Volt said:

neptronix said:

Cells in parallel cannot differ in voltage.
It's like two wheels linked via a solid axle. The two wheels must move together.. if one slips, the other will correct it instantly.

So the lifepo4 will give what it can, and the A123 will give what it can, in equal proportions.

Click to expand...

So as long as the M1/A123 part of the pack can supply the peak amperage, the LiFepo4 won't be too distressed?

My plan is to have a 14Ah system and only ever use 10Ah.

Theoretically, the repeated high-discharges of the A123 should induce the Peukert effect sooner rather than later, but if I build a large pack and stay inside a limit, I can always stay ahead of the game, no?

Click to expand...

It depends on how much lifepo4 you have and how much A123 lifepo4 you have, of course.
The more 1C-2C lifepo4 you have, the more you 'water down' the pack's total discharge rating.

Now that i think about it, i am not 100% sure how the current sharing would work. I think a real life test is warranted.

P.S. The peukert effect relates to lead acid batteries. It gets a bit muddy when we start talking about another chemistry.

Yeah, please run a test if you have two wattmeters. My gut instinct is that the weak cells will supply current at a rate higher than thier spec for c rate, and die trying. If the better cells can "protect" the weaker ones from high amps, then this could be something very usefull. Build a large pack but only half of it with 90c nanos's for example.

neptronix said:

P.S. The peukert effect relates to lead acid batteries. It gets a bit muddy when we start talking about another chemistry.

Click to expand...

It does affect all chemistries; it's just most pronounced in LA, out of all the ones we are likely to use.

BTW, the only bad thing I can envision about building a pack of low rate cells paralleled with high rate ones is if one or more cells in a group of low rate ones fails like the ones in my Vpower pack. The high rate cells will be able to supply current into them pretty quick, and the total energy they dissipate might be enough to become problematic if they die badly shorted inside.

It's an unlikely scenario, but a possibility.

If you parallel using a high(er) resistance connection between the low and high C cells, the extra resistance will cause more of the peak load to be taken by the high C cells. Then during low load situations equalization will occur but the current will be low enough that the extra resistance doesn't consume many watts (heat) and even during high load conditions the extra consumed watts are small.

neptronix said:

Cells in parallel cannot differ in voltage.
It's like two wheels linked via a solid axle. The two wheels must move together.. if one slips, the other will correct it instantly.

So the lifepo4 will give what it can, and the A123 will give what it can, in equal proportions.

Click to expand...

I see paralleled cells with voltage differences every time I build a Konion pack. When charging equal size blocks of cells, I put all of the blocks in parallel, and despite a rats nest of parallel cross structure of alligator clip wires. The blocks aren't at identical voltage during charging or immediately after, so I leave them connected in parallel for hours to be sure they are at identical voltage before put them all in series for a relative capacity discharge test. The first time I parallel charged about 30 2p4s packs, I arranged the parallel structure just simple and uniform, from one pack to the next, and connected the charger to the first pack. It took a full day for all of the packs to get to an identical voltage. When there is current flow and resistance exists, then voltage differentials are inevitable. We even see it within the cells themselves in the form of voltage sag.

I agree though that the lower power cells will deliver some of the energy at all times, but the % will vary depending upon the SOC of the A123's.

dogman said:

On the other hand, here I go again, not knowing shit, going bla bla bla.

Do you have some little RC wattmeters? It would be really interesting to see what they said, if you built the pack, or even just a 1s version of it. The put a wattmeter on each type of cells before paralelling them. Then put the load on and see what really happens. Then you'd really know what to expect when you build the whole pack. Data from just a handfull of cells should scale up to the full pack the same way.

Click to expand...

This is interesting enough that it might be worth me finally doing a video. Parallel four A123's and one of the 10ah prismatic lifepo4 cells with my multimeter in the parallel connection to the low power cell, then do some charging and discharging to see if the current flow between the cell types and what happens when the discharging stops and how it changes at different SOCs.

gogo said:

If you parallel using a high(er) resistance connection between the low and high C cells, the extra resistance will cause more of the peak load to be taken by the high C cells. Then during low load situations equalization will occur but the current will be low enough that the extra resistance doesn't consume many watts (heat) and even during high load conditions the extra consumed watts are small.

Click to expand...

The low power cells themselves have a higher internal resistance that causes the whole issue that I'm trying to twist to my advantage. I would never intentionally use a high resistance connection in a battery pack, at least in a place I expected any significant current. If I did cut the series connection of the low power cells to each other, then I would definitely need to beef up the parallel tie to the A123's.

John in CR said:

gogo said:

If you parallel using a high(er) resistance connection between the low and high C cells, the extra resistance will cause more of the peak load to be taken by the high C cells. Then during low load situations equalization will occur but the current will be low enough that the extra resistance doesn't consume many watts (heat) and even during high load conditions the extra consumed watts are small.

Click to expand...

The low power cells themselves have a higher internal resistance that causes the whole issue that I'm trying to twist to my advantage. I would never intentionally use a high resistance connection in a battery pack, at least in a place I expected any significant current. If I did cut the series connection of the low power cells to each other, then I would definitely need to beef up the parallel tie to the A123's.

Click to expand...

The point of using a "higher" resistance connection would be to limit the peak current draw on the low C cells and thereby protect them in an easy and passive way. The energy cost of the "higher" resistance connection would be negligible.

I have a hard time wrapping my head around the idea that the voltage could be different but i'll accept it for the sake of argument since i haven't tested it.

I do know that my lipos in parallel will equalize any voltage difference instantly. Maybe it is due to the super low internal resistance..