Need advice on radical hybrid li-co/li-mn pack design

13 posts / 0 new
Last post
xyster
Offline
Last seen: 12 years 4 months ago
Joined: Thursday, November 16, 2006 - 20:11
Points: 137
Need advice on radical hybrid li-co/li-mn pack design

Here's the deal: due to cold weather, plus maybe aging, my 16s15p 2.2ah 1.5C lithiums are only putting out 30amps total after 25% of pack capacity is drained. Extra voltage droop under 30 amp load is also preventing my formerly 35mph bike from going more than 30mph -- unacceptable!

Here's my radical hybrid idea: Add in parallel one 3ah emoli lithium mangenese cell to each of the 16 parallel subpacks of lithium cobalt cells. So the total pack would be 16s16p (15p li-co + 1p emoli)

Projected benefit: With it's much lower impedance, the single extra emoli per subpack (impedance of these cells supposedly does not increase over the life of the cell as it does for lithium cobalt) I'll be sure to get my full 35 amps back, plus less voltage droop, plus a bit longer run time, plus less wear on the lithium cobalts which apparently are not really 1.5C cells after all.

Possible problems: charging you should not present any problems, but discharging might. As the other lithium cobalts will only give 30 amps max, the emoli cells may be at risk of draining to zero if I'm pulling 35 amps for an extended period. Perhaps there's other problems I'm not anticipating?

What do you guys think?
Fetcher, Leeps, etc, what sayeth the electrical engineering wizards?

Thanks!

Leeps
Offline
Last seen: 12 years 10 months ago
Joined: Friday, November 17, 2006 - 13:51
Points: 23
I would have a look at the

I would have a look at the load balancing on your current battery pack. It looks to me in the photos on your other thread that you had three 4 cell parallel packs. I noticed that the top one is tapped in the middle, the others appear to be tapped on the side. I would have the batteries all go to a common point with the same length of wire on each one. For instance in your case if you tapped the batteries from the right side the current from the left side has the length of the other three batteries to go through to get to the wire whereas the rightmost battery is right there, the rightmost battery will discharge before the leftmost does. Granted this effect is negligeable it might be significant depending on lengths of wire and the connection methods from your other parallel packs. The damage your incurring could be overcurrent on individual strings and/or over discharge on individual strings. The proper way to wire them up would be to have wires leading to individual strings be all the same length up to a common connection point, this way each string sees the same resistance and voltage drop. Load balancing due to different lengths of wire on parallel batteries is a common problem in off grid houses due to the long lengths of wire associated and they go through great lengths to get it right.
I would shy away from mixing the technologys unless you had a battery management board to keep the different technologys from getting into trouble. In this case it would be overdischarge on the e-molis as they would get discharged first.
best of luck
joe

xyster
Offline
Last seen: 12 years 4 months ago
Joined: Thursday, November 16, 2006 - 20:11
Points: 137
Thanks Joe! Perhaps that's

Thanks Joe! Perhaps that's the root of the problem. The pic you cited was from before I resoldered the pack into (16) 15 cells subpacks.

Can you take a good look at this pic, and tell me if you think I should pull it, maybe add some 12 guage crosswires or something else -- I really don't want to have to pull the whole thing apart and resolder again, in part because the heat may damage the batteries too.

Anybody else?
Fechter, what do you think?

--xyster
moderator

My rides

Leeps
Offline
Last seen: 12 years 10 months ago
Joined: Friday, November 17, 2006 - 13:51
Points: 23
It looks like you have 5

It looks like you have 5 packs of 15 cells each. Some of them dont look too good, adding cross wires would definetely help, im not sure if it will solve the problem. I re-read your other thread about the failures that were occuring, and noticed that they said the batterys have wildly varying impedances, this would aggravate the problem, but theres nothing we can do about it, its just manufacturing differences.
Without resoldering the whole pack i would add connections between all the cells in the pack. In some packs there are cells in the far corners that have to make a zig-zag across the whole width of the pack to make it to the tap, this is what you want to get rid of. On top of this adding a small length of wire from the tap point to the far corners could also help. Seriously i know this must be frustrating you have a lot of money invested in that battery, best of luck.
Joe

xyster
Offline
Last seen: 12 years 4 months ago
Joined: Thursday, November 16, 2006 - 20:11
Points: 137
Thank you, Joe, I'll

Thank you, Joe, I'll definitely do as you suggest before pondering more exotic solutions. This pic is of 5 of the 16 15-cell subpacks. Another five like this sits with this five in the rear box for ten, four are in the mid-frame triangle box, and two sit above the midframe box.
http://visforvoltage.org/blog-entry/4-xysters-x5-images-and-stats

--xyster
moderator

My rides

NickF23
Offline
Last seen: 10 years 5 months ago
Joined: Saturday, November 18, 2006 - 10:25
Points: 184
Hi Xyster, I think it

Hi Xyster,

I think it *could* work really well. Reminds me a bit of aerowatt's sla li-ion cobalt lithium system. You could test it on one subpack with just one emoli cell under different loads before investing in a whole set. I think how well it might work would depend on what kind of riding you are doing. Basically the emoli cell voltage should fall faster than the li-ion cobalt. so your li-ion pack would top up the emoli when your not sucking high amps.

But like Joe said, start with the basic stuff first esp the wiring. You should weed out he low capicty (low voltage after discharge) and high resistance cells first to help balance the packs.

xyster
Offline
Last seen: 12 years 4 months ago
Joined: Thursday, November 16, 2006 - 20:11
Points: 137
Well, I did as Joe

Well, I did as Joe suggested, pulled the batteries, checked for broken connections (there were none) and soldered on more crosswires. It didn't make any difference -- still pulling only 32 amps at 3.94v in my 50F garage, dropping to 30 amps as no-load voltage drops to 3.84.
I don't usually ride full out 35 amps for long, so I was thinking as you do, Nick, it should work. The emoli's effectively have >50% more usable capacity than a single 18650.

Nick, there is no way I'm going to desolder this pack again to weed out any stragglers -- I'd like to, but it's possible the heat from two rounds of soldering has already damaged these cells, causing this problem. I'd simply solder-on an emoli to each of the 16 subpacks, also assembling two more in the process for a 18s16p pack with 1p of the 16p an emoli. If I purchased three 7-cell, Milwuakee V28 packs, I'd have three left over with which to also build a little 3s1p 12v boost pack. decisions, decisions...any more input/ideas/thoughts/concerns would be most appreciated.

Thanks!
--xyster
moderator

My rides

Leeps
Offline
Last seen: 12 years 10 months ago
Joined: Friday, November 17, 2006 - 13:51
Points: 23
I'm really sorry that

I'm really sorry that soldering up more crosswires didnt solve your problem. The e-molis will offer more amps, but they will also discharge quicker as a result of it. When you let off the throttle they will not recharge effectively because it takes more voltage to charge a battery than it takes to discharge it. You mentioned that when your pulling 30 amps you have 3.84 volts per cell. 3.84 volts is well above the nominal voltage of a lithium ion battery and "assuming" that all cells are performing alike i would say the pack is doing what it is supposed too. More current would mean for more voltage drop and im sure that the difference between 30 and 35 amps is the same difference between 3.84 volts and 3.6 volts per cell. Perhaps the difference is in the load. It is a possibility because the 3.84 volts under load seems a bit high for being under load. At this point i would like to hear fetchers opinion just as im sure you would.
Joe

xyster
Offline
Last seen: 12 years 4 months ago
Joined: Thursday, November 16, 2006 - 20:11
Points: 137
I'm really sorry that

I'm really sorry that soldering up more crosswires didnt solve your problem.

I wanted to check the solder connections anyway, so no great trouble.

The e-molis will offer more amps, but they will also discharge quicker as a result of it.

The total current is limited to 35 amps, so the difference will be small (30 amps to 35 amps). But if the other 15 lithium-cobalts in each parallel subpack do not act to recharge the single emoli, then I'll have only a few miles of juice from the emoli's 3ah. I understand emolis can be run down to 100% DoD -- but I do not know what happens after that if they're in parallel with other cells still providing current. What do you guys think, will the emoli be toast? Open circuit or closed circuit? Or will the large, ~3.5v voltage difference keep it charged just above 0%?

When you let off the throttle they will not recharge effectively because it takes more voltage to charge a battery than it takes to discharge it.

If true, why do the 15 cells in each parallal subpack balance themselves when not under severe load?

You mentioned that when your pulling 30 amps you have 3.84 volts per cell.

No, I said 'no-load'...as in the cell rebounds to 3.84volts (from ~3.6v), but at this ~50% DoD, they're only delivering 30 amps when warm. When warm with resting volt above 3.9v (around 40% DoD), the cells deliver the full 35 amps requested by the motor and limited by the controller.

3.84 volts is well above the nominal voltage of a lithium ion battery and "assuming" that all cells are performing alike i would say the pack is doing what it is supposed too.

Hot off the chargers, the cells rest at ~4.15v, and under 35amp load at this 0%DoD, droop to 3.8v or so. I understand there is much more to current delivery than just impedance, such as the max speed of chemical reactions which is limited by temperature as well as cell design and type.

More current would mean for more voltage drop and im sure that the difference between 30 and 35 amps is the same difference between 3.84 volts and 3.6 volts per cell.

35 --> 30 amps constitutes a 14% loss of power; very noticeable. I could get most of it back by adding another two subpacks (8 volts), which is what I was planning on anyway as I add batteries up to 92 volts or so. But it'd still be 14% better to run 92v35amps then 92v30amps!

Perhaps the difference is in the load. It is a possibility because the 3.84 volts under load seems a bit high for being under load. At this point i would like to hear fetchers opinion just as im sure you would.

If the load was different, would the ammeter on the handlebars not show it? I would love to hear sir fechter's opinion on this too. Especially if he thinks the single emoli added to each parallel pack would recharge/stay above zero even if depleted, etc. I got this idea from Aerowhatt's hybrid Lithium/SLA pack, but there the different chemistries are kept separate up until the controller -- i.e.:

His:
the two + - Terminals from SLA1+SLA2+SLA3+SLA4= ~48v
join the two + - Terminals from (paralleled Lithium)*14 = ~48v
to feed the controller.

My proposed hybrid pack:
(15p 2.2ah li-co + 1p 3.0ah li-mn) * 16 = ~64v to the controller

--xyster
moderator

My rides

Leeps
Offline
Last seen: 12 years 10 months ago
Joined: Friday, November 17, 2006 - 13:51
Points: 23
Yeah i just misread the

Yeah i just misread the no-load load thing. It makes much more sense to me now. Current delivery is based on battery impedance and terminal voltage. Your right that its complicated, there are many factors influencing the impedance. Impedance is just apparant resistance, if the temperature drops your impedance goes up.
As far as one cell recharging the other, i suppose it depends on what you mean by recharging, and how much balance you really want. Take any cell for example and plot its charging curve from 0% to 100% over time. Now how much could you charged a dead cell if you connected it to a fresh cell. You could pull your dead cell to the 3.6 volts of the fresh cell but no higher if you never pull it up higher it will not charge higher and they will always stay out of balance somewhat. Or another way to think of it when you plug your pack in how long does it take the cells to rise above 3.6-3.7 volts on the charger. Granted we are talking small differences in a parallel pack but a consistent couple % overdischarge on a single cell will wear it out quicker than necessary. In the same breath if you never took below say 95% DOD you wont run into a problem, im just defending that this is something real that happens. Usually this shows itself in off grid power installations and the difference is 10 feet of cable.
Let me give you a scenario you have your 15 cells from 14 of the cells to a common connection point you have the same impedance but one cell gets thicker wire and has 5 milliohms less impedance. You also have an e-moli cell in parallel that has 10 milliohms less impedance than the 14 reference cells which are the 90 milliohms. you pull 35 amps divided amongst them.
Lets attack this first we have to find the lump sum expected voltage drop (ideal 25C,blah blah conditions). The referance 14 cells has 6.42 milliohms of combined impedance. This will be R1. The other cell has 85 and the emili has 80. The parallel resistance is 1/RT=(1/R1)+(1/R2)+(1/R3), so (1/6.42)+(1/85)+(1/80)=1/.180=5.55 milliohms. With 35 amps you have 194 millivolts of drop. Now you can figure out how much current is coming from each battery. Each normal li-co cell would be giving you
.194/.09=2.15 amps for a total of 30.17 amps. The li-co cell that got a thicker wire would give you .194/.085=2.28 amps for a new total of 32.45 amps. The e-moli cell would be giving you (note i didnt find the datasheet and guessed the 80 milliohms) .194/.08=2.425 amps for the new total of 34.875 amps (rounding error close enough). Between the high and the low li-co you have 130 milliamps of difference, not much after an hour of discharge thats .13 amphours out of balance or 6.5%. The e-moli i realized you wouldnt have to worry about because it is 3 amphours and wouldnt get hurt. This really does depend on its impedance if we dropped it to say 10 milliohms it might just discharge before the others. You also have to factor in the impedance rising as it discharges, so its really hard to tell, i would feel bad telling you to go for it and then making the situation worse, in the above example it would work out. but its just that an example.
Joe

xyster
Offline
Last seen: 12 years 4 months ago
Joined: Thursday, November 16, 2006 - 20:11
Points: 137
Thank you Joe! I really

Thank you Joe! I really appreciate seeing how you do the math -- I'll have to go over this again a couple times, plugging in some different numbers. In college I took a year of physics w/calculus including a term of electrophysics where we covered stuff like this, but that was 12 years ago and I've forgotten most of it now :-(... I did realize the lithium-cobalt cells will not have a chance of recharging the emoli up to full. With 33ah on tap, I don't usually go below 3.8 (no load) on these cells, so if they can just hold the emoli's there (at ~50% DoD) between blasts of 35 amps, that would be most acceptable.

It sounds like the smart thing to do is a little empirical testing. I can buy 28 more li-co's, some emoli's and in a separate box try (14p li-co + 1p emoli)*2s. If the lithium manganese cell becomes fully discharged, what do you think will happen to it? I'm concerned about a closed circuit failure which would short out the entire parallel subpack, potentially with explosive results. Is this likely do you think? With the exception of this possibility, am I correct that the lithium cobalt cells are not at any higher risk in such a set-up?

Here's an emoli profile conducted by a RC enthusiast:
T_EMOLI_26700A_DC.gif
In relation to this discussion, what can you divine from these data?

BTW, Joe, I really appreciate your help too with that schottky diode array for my scooter -- I got it all put together and installed, but alas, the Uni-8 controller was basically DOA, I had to send it back, and they haven't sent me a new one yet. I requested a refund...

NickF23
Offline
Last seen: 10 years 5 months ago
Joined: Saturday, November 18, 2006 - 10:25
Points: 184
Xyster, What your doing here

Xyster,

What your doing here is already quite complicated and new. So take the following points with a big pinch of salt coz I don't know that much. Anyway here goes.

1: The resistance of the emoli's is apparently 25 milliohms. If it wasn't this low the idea wouldn't make sense as the emoli's wouldn't recharge fast or provide enoug boost to your cobalt pack either. Its probably worth testing a single sub pack with one or 2 emoli cells before buying a few and finding out it doesn't work.

2: There are 2 main differences between yours and aerowatts pack that I can think of. First is that his workhourse sla's were 50amp hour and the li-co were 40 amp hour. So the ratio of high low to high internal resistance cells is quite differnt. The second piont it that sla voltage tend to drop quite a lot on discharge so the differnce between the two battery sets will be more pronounced and the sla's will be recharged quicker.

3: Are you sure that sparking pct's will not be a problem? might be worth contacting lg on this question. If it is a problem then raising the pack voltage will likely endanger your pack more.

Leeps
Offline
Last seen: 12 years 10 months ago
Joined: Friday, November 17, 2006 - 13:51
Points: 23
Sorry it's taking me so long

Sorry it's taking me so long to get back. Im only home for the weekends now.
I looked around and couldnt find a datasheet on the e-one moli site, at the same time i havent done too much research on this chemistry. I would assume however that overdischarge should be avoided at all costs, simply because its damaging to the battery and its an expensive investment. As far as a short circuit goes make sure all the connections are well insulated. The results of an untamed short circuit can be very bad, use a fuse or circuit breaker, preferably one on each pack if possible. The best thing is prevention, insulate your connections.

The data above works out to roughly a 20 milliohm impedance which falls in line with what nick said, 25 milliohms. So using 25 milliohms lets redo the calculation. 15 cells at 90 milliohms works out to 6 milliohms. the emoli is 25 milliohms so 1/RT=(1/6)+(1/25)=1/.206=4.83 milliohms with 35 amps .00483*35=.169 volts so 169 millivolts drop .169volts/.025 ohms=6.76 amps the other cells will have .169/.09=1.87 amps
and just to check the math 1.87*15=28.05 amps + 6.76=34.81 amps(rounding error) it works
in this case the emoli cell is working harder whether or not it will recharge enough to keep it out of trouble i dont know i never saw the datasheet, sorry.
best of luck
Joe

Log in or register to post comments


Who's online

There are currently 0 users online.

Who's new

  • skler
  • Jim Caudill
  • nikity
  • slartsa
  • olivergeorge

Customize This